HRM

RESEARCH IN PERSONNEL
AND HUMAN RESOURCES

MANAGEMENT
RESEARCH IN PERSONNEL
AND HUMAN RESOURCES
MANAGEMENT
Series Editors: Hui Liao, Joseph J. Martocchio and
Aparna Joshi
Recent Volumes:
Volumes 110:
Edited by Kendrith M. Rowland and

Gerald R. Ferris
Volumes 1120: Edited by Gerald R. Ferris

Supplement 1:
International Human Resources

Management Edited by Albert Nedd
Supplement 2:

Management Edited by James B. Shaw
and John E. Beck
Supplement 3:

Management Edited by James B. Shaw,
Paul S. Kirkbridge and Kendrith M. Rowland
Supplement 4:
International Human Resources Management

in the Twenty-First Century Edited by
Patrick M. Wright, Lee D. Dyer,
John W. Boudreau and George T. Milkovich
Volume 21:
Edited by Joseph J. Martocchio and

Gina Ferris
Volume 22:
Edited by Joseph J. Martocchio and

Gerald R. Ferris
Volumes 2327: Edited by Joseph J. Martocchio

Volume 28:
Edited by Joseph J. Martocchio and Hui Liao
RESEARCH IN PERSONNEL AND HUMAN RESOURCES

MANAGEMENT VOLUME 29
RESEARCH IN
PERSONNEL AND
HUMAN RESOURCES
MANAGEMENT
EDITED BY
HUI LIAO
University of Maryland, USA
JOSEPH J. MARTOCCHIO
University of Illinois, USA
APARNA JOSHI
University of Illinois, USA
United Kingdom North America Japan

India Malaysia China
Emerald Group Publishing Limited

Howard House, Wagon Lane, Bingley BD16 1WA, UK
First edition 2010
Copyright r 2010 Emerald Group Publishing Limited
Reprints and permission service
Contact: booksandseries@emeraldinsight.com
No part of this book may be reproduced, stored in a retrieval system, transmitted in any
form or by any means electronic, mechanical, photocopying, recording or otherwise
without either the prior written permission of the publisher or a licence permitting
restricted copying issued in the UK by The Copyright Licensing Agency and in the USA
by The Copyright Clearance Center. No responsibility is accepted for the accuracy of
information contained in the text, illustrations or advertisements. The opinions expressed
in these chapters are not necessarily those of the Editor or the publisher.
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN: 978-0-85724-125-2
ISSN: 0742-7301 (Series)
Awarded in recognition of
Emeralds production
departments adherence to
quality systems and processes
when preparing scholarly
journals for print
CONTENTS
LIST OF CONTRIBUTORS
vii
WORKPLACE SAFETY: A MULTILEVEL,

INTERDISCIPLINARY PERSPECTIVE
Michael J. Burke and Sloane M. Signal
EXECUTIVE PAY AND FIRM PERFORMANCE:
METHODOLOGICAL CONSIDERATIONS AND
FUTURE DIRECTIONS
Beth Florin, Kevin F. Hallock and Douglas Webber
A TIME-BASED PERSPECTIVE ON EMOTION
REGULATION IN EMOTIONAL-LABOR
PERFORMANCE
Michelle K. Duffy, Jason D. Shaw, Jenny M. Hoobler
and Bennett J. Tepper
INSIGHTS FROM VOCATIONAL AND
CAREER DEVELOPMENTAL THEORIES:
THEIR POTENTIAL CONTRIBUTIONS FOR
ADVANCING THE UNDERSTANDING OF
EMPLOYEE TURNOVER
Peter W. Hom, Frederick T. L. Leong and
Juliya Golubovich
HOW DID YOU FIGURE THAT OUT? EMPLOYEE
LEARNING DURING SOCIALIZATION
Jaron Harvey, Anthony Wheeler,
Jonathon R. B. Halbesleben and M. Ronald Buckley
49
87
115
167
vi
CONTENTS
COMPARING APPLES AND ORANGES:

TOWARD A TYPOLOGY FOR ASSESSING
E-LEARNING EFFECTIVENESS
N. Sharon Hill and Karen Wouters
201
ABOUT THE AUTHORS
243
M. Ronald Buckley
Price College of Business, Division of

Management, University of Oklahoma,
Norman, OK, USA
Michael J. Burke
Freeman School of Business, Tulane

University, New Orleans, LA, USA
Michelle K. Duffy
Carlson School of Management, University

of Minnesota, Minneapolis, MN, USA
Beth Florin
Pearl Meyer & Partners, Southborough,

MA, USA
Juliya Golubovich
Department of Psychology, Michigan State

University, East Lansing, MI, USA
Jonathon R. B.
Halbesleben
Department of Management and Marketing,

Culverhouse College of Commerce and
Business Administration, University of
Alabama, Tuscaloosa, AL, USA
Kevin F. Hallock
Cornell University and NBER, ILR School,

Ithaca, NY, USA
Jaron Harvey
Department of Management and Marketing,

Culverhouse College of Commerce and
Business Administration, University of
Alabama, Tuscaloosa, AL, USA
N. Sharon Hill
School of Business, The George Washington

University, Washington, DC, USA
Peter W. Hom
Department of Management, W.P. Carey

School of Business, Arizona State University,
Tempe, AZ, USA
vii
viii
Jenny M. Hoobler
College of Business Administration,

University of Illinois at Chicago, Chicago,
IL, USA
Frederick T. L.
Leong
Department of Psychology, Michigan State

University, East Lansing, MI, USA
Jason D. Shaw
Carlson School of Management, University

of Minnesota, Minneapolis, MN, USA
Sloane M. Signal
Freeman School of Business, Tulane

University, New Orleans, LA, USA
Bennett J. Tepper
J. Mack Robinson College of Business,

George State University, Atlanta, GA, USA
Douglas Webber
Cornell University, ILR School, Ithaca,

NY, USA
Anthony Wheeler
Schmidt Labor Research Center, College of

Business Administration, University of Rhode
Island, Kingston, RI, USA
Karen Wouters
Robert H. Smith School of Business,

University of Maryland, College Park,
MD, USA
WORKPLACE SAFETY:
A MULTILEVEL,
INTERDISCIPLINARY
PERSPECTIVE
Michael J. Burke and Sloane M. Signal
ABSTRACT
While research on workplace safety spans across disciplines in medicine,
public health, engineering, psychology, and business, research to date has
not adopted a multilevel theoretical perspective that integrates theoretical
issues and ndings from various disciplines. In this chapter, we integrate
research on workplace safety from a variety of disciplines and elds to
develop a multilevel model of the processes that affect individual safety
performance and safety and health outcomes. In doing so, we focus on
cross-level linkages among national, organizational, and individual-level
variables in relation to the exhibition of safe work behavior and
occurrence of individual-level accidents, injuries, illnesses, and diseases.
Our modeling of workplace safety is intended to ll a theoretical gap in
our understanding of how the multitude of individual differences and
situational factors interrelate across time to inuence individual level
safety behaviors and the consequences of these actions, and to encourage
research to expand the limits of our knowledge.
Research in Personnel and Human Resources Management, Volume 29, 147

Copyright r 2010 by Emerald Group Publishing Limited
All rights of reproduction in any form reserved
ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029003
MICHAEL J. BURKE AND SLOANE M. SIGNAL
An impressive body of literature on workplace safety that spans across

disciplines in medicine, public health, engineering, psychology, and business
is developing. Within these disciplines, researchers are often attending to the
study of relations between variables at either the individual or situational
level of analysis. With few exceptions (e.g., Wallace & Chen, 2006; Zohar &
Luria, 2005), research to date has not adopted a multilevel theoretical
perspective to the study of workplace safety nor has it attempted to integrate
ndings from various disciplines. Recognizing that individuals and situations
are interdependent, we argue that adopting a multilevel perspective that
incorporates aspects of workplace safety from multiple disciplines is needed
to fully understand the processes that lead to safety-related work behavior
and its consequences. In doing so, this chapter integrates research advances
within human resource management, organizational behavior, safety
engineering, and various elds of medicine and public health to focus on
cross-level linkages among national, organizational, and individual-level
variables in relation to the exhibition of safe work behavior and occurrence
of individual-level accidents, injuries, and illnesses. Our focus is not intended
to diminish the importance of emergent phenomena, processes, and
outcomes at higher levels of analysis such as at the work group and
organization levels, but rather our intent is to ll a theoretical gap in our
understanding of how the multitude of individual differences and situational
factors interrelate across time to inuence individual-level safety behaviors
and the consequences of these actions.
Hence, we have three goals for this chapter. First, we present a multilevel
model of workplace safety in relation to individual-level outcomes. Here, we
emphasize the study of construct domains where advances have been made in
identifying the appropriate taxonomic components or constructs relevant to
the study of individual-level outcomes. As such, our discussion is not
exhaustive of potentially relevant constructs at each level of analysis. Second,
we discuss how relationships between construct domains within the model
have been studied and summarize what is known about these linkages.
Throughout the discussion, our third goal is to discuss the limits of our
current state of knowledge and practice and the types of research needed to
expand these limits.
A MULTILEVEL MODEL OF WORKPLACE SAFETY

A problem faced in the rapprochement of different disciplines and bodies of
literature as they relate to workplace safety is where to begin in terms of a
Workplace Safety: A Multilevel, Interdisciplinary Perspective
theoretical foundation. For our purposes, we begin with a discussion of safety

performance, the safety-related actions or behaviors that workers exhibit in
almost all types of work to promote their safety and the safety and health of
others. While disciplines vary considerably in terms of the relative emphasis
placed on the study of safe work behavior, all disciplines recognize the need
for safe work behavior in preventing or reducing negative outcomes such as
injuries and illnesses. For instance, researchers in industrial ergonomics may
focus on the pacing of actions to avoid visual and muscular disturbances (e.g.,
Salvendy, 1998; M. J. Smith, 1998), kinesthesiologists and occupational
therapists study the biomechanics of postures and movements in efforts to
reduce musculoskeletal disorders (e.g., Gagnon, 2003), applied psychologists
and human resource management researchers focus on the consistency or
frequency of actions taken to avoid accidents (e.g., Burke, Bradley, & Bowers,
2003), community health researchers are often concerned with the percent of
time actions are engaged in so as to preclude the onset of disease (e.g., Forst
et al., 2004; Mayer et al., 2007), and industrial hygienists frequently stress the
correct sequencing of actions to avoid exposure to toxic substances (e.g.,
Perry & Layde, 2003). Although disciplines within business, psychology,
engineering, medicine, and public health focus on different aspects of action
or behavior, the safety-related actions that workers engage in can be
conceptualized relative to a core set of dimensions that apply across jobs and
industries, regardless of safety performance measurement considerations
within any discipline.
At the broadest level, the safety-related actions that workers engage in
can be placed into two broad content categories akin to notions of task
and contextual performance in the job performance literature (Motowidlo,
Borman, & Schmit, 1997): safety compliance and safety participation,
respectively. Safety compliance refers to generally mandated safety
behaviors, whereas safety participation refers to safety actions that are
more discretionary in nature (see Neal, Grifn, & Hart, 2000). In large
part, research on behavioral aspects of safety across disciplines has relied on
the notion of safety compliance and either explicitly or implicitly (via how
safety performance was measured) treated safety compliance as a unitary
factor. Although research on safety participation is less frequent, this
literature also explicitly or implicitly (e.g., Hofmann, Morgeson, & Gerras,
2003) views safety participation as a unitary factor. These points
are noteworthy, as Marchand, Simard, Carpentier-Roy, and Ouellet
(1998) found that a two-factor model of safety performance (with factors
relating to safety compliance and safety initiative) did not provide a good
t to the data.
Given the literature on job performance, Marchand et al.s (1998) factor

analytic ndings are not surprising. The job performance literature has
produced ample evidence that multiple dimensions underlie the analogous
domains of task (requisite) and contextual (discretionary) work behaviors
(e.g., Campbell, McHenry, & Wise, 1990; Motowidlo et al., 1997).
Supporting the multidimensionality of behavioral factors in regard to
notions of safety compliance and safety participation, Burke, Sarpy, Tesluk,
and Smith-Crowe (2002) conrmed, across 23 jobs, a grounded theoretical
model of general safety performance. Two of their conrmed factors, labeled
using personal protective equipment and engaging in work practices to
reduce risk, would clearly fall within the domain of safety compliance;
whereas the other two conrmed factors, communicating health and safety
information and exercising employee rights and responsibilities, are closer to
the notion of safety participation. Notably, these factors were conrmed for
individuals working in dyads, work groups, and teams. Together, conceptual
and empirical research on the factor structure of behavioral safety
performance would suggest that researchers across disciplinary boundaries
may benet from attending to potentially useful construct distinctions in the
measurement and study of workers safety-related actions.
Another distinction in relation to safety performance, which is recognized
in the medical literature, is that of a work-around (Ash, Berg, & Coiera, 2004;
Kobayashi, Fussell, Xiao, & Seagull, 2005). Work-arounds are actions to
address a block or disruption in a work system that involve bypassing safety
procedures or protocols and can be conceptually distinguished from
deviance, errors, and mistakes in regard to motive (see Halbesleben,
Wakeeld, & Wakeeld, 2008). That is, the motive is to complete the task
by getting around the block. Although the term shortcut has been used
interchangeably with work-around, a shortcut is a specic case of a workaround intended to deal with a perceived time block. To date, the
measurement of work-arounds has been qualitatively approached (via
workers responses to open-ended questions) within human factors research
(e.g., Charlton, 2002). Additional research is needed to capture the behavioral
dimensionality of work-arounds as an area of safety performance and to
produce measures that can assist in studying work-arounds in relation to
their antecedents and consequences in a model such as one shown in Fig. 1.
To develop the multilevel model of the processes through which national/
regional, organizational/group, and individual-level factors affect safety
performance and its consequences as depicted in Fig. 1, we rely broadly on
the disciplinary literatures that consider behavioral aspects of workplace
safety. More specically, at the individual level, we ground our modeling in
Fig. 1.
A Multilevel Model of Workplace Safety. Notes: Solid lines with arrows represent expected direct effects. Dashed
lines with arrows designate expected moderation. Arced lines with arrows represent feedback loops.

5
the applied psychology literature that identies knowledge and motivation

as proximal antecedents to performance and more stable individual
differences as distal antecedents to performance. In the applied psychology
eld, the inuence of distal antecedents on performance is often posited to
be mediated by either knowledge or motivation. At the organizational/
group level of analysis, our theoretical modeling is, in large part, based on
theoretical advances within the domain of work climate. Further, at this
level of analysis, we argue for an occupational safety conceptualization of
workplace hazards that we believe offers considerable promise for
reintroducing the study of objective workplace hazards into the human
resource management and applied psychology literatures. At the national/
group level of analysis, we ground our work in research advances within the
domain of public health as well as theoretical notions of culture and values
within the management and organizational behavior literature.
Finally, in regard to the consequences of safety performance, referred to as
safety outcomes, we focus on tangible events and results that are considered
within a variety of disciplines including illness and diseases discussed within
the literature on occupational medicine. Furthermore, we incorporate
notions from the literature on education and learning to suggest how
workers reect on and learn from perceived safety performanceoutcome
associations. In this sense, our model conceptualizes workplace safety as an
ongoing process, where workers learn and think in and by action.
ANTECEDENTS OF SAFETY COMPLIANCE

AND SAFETY PARTICIPATION
In terms of situational factors, we classied antecedents of safety compliance
and safety participation at the national/regional, organizational/group, and
individual levels of analysis. While social, institutional, and legal considerations exist at a global level in relation to workplace safety (see Nuwayhid,
2004), we begin our modeling of safety performance and its consequences at
the national/regional level of analysis. Our rationale is that at this level of
analysis, one can more readily view and study the role of cultural values and
differences between nations in regard to occupational safety and health
political and research agendas. In adopting this approach, we recognize that
an understanding of safe work behavior and its consequences cannot be
divorced from cultural factors and the broader challenges of occupational
health and safety in both developing locations such as Bangladesh, Central
America, and South Africa (Ahasan, Mohiuddin, Vayrynen, Ironkannas, &

Quddus, 1999; Joubert, 2002; Wesseling et al., 2002) and within regions of
developed countries such as in the northern and southern states of the United
States. (Richardson, Loomis, Bena, & Bailer, 2004). In addition, although
one could develop arguments for other possible levels of analysis (i.e.,
occupational, union, or industry levels), the three situational levels that we
consider comprise a more parsimonious and inclusive model than other
possible situational breakdowns. That is, our theoretical arguments at each
situational level apply across occupational, industrial, and union boundaries.
To unfold the discussion of our model in Fig. 1, we begin with a
presentation of conceptual arguments for why cultural values inuence the
political economy of nations and states and follow with a discussion of
several cross-level linkages between national/regional factors and organizational/group-level factors. Subsequently, we discuss how key factors
interrelate at the organization/group level, and how and why these factors
would be expected to affect processes at the individual level of analysis.
Given the multitude of construct domains that we consider as antecedents of
safety compliance and safety participation, we will refer to factors not only
in terms of level of analysis but also with respect to their distance or
expected causal ordering in relation to these safety performance domains.
In doing so, we discuss antecedents as distal and proximal to safety
performance to assist in organizing the discussion of these factors.
Proximal antecedents will include safety motivation and safety knowledge. Safety motivation refers to ones regulatory behavior primarily in
relation to exerting effort when on the job to enact safety behaviors;
whereas, safety knowledge refers to an understanding of both safety-related
facts and procedures, and can be of an implicit or anticipatory nature (see
Broadbent, Fitzgerald, & Broadbent, 1986; Burke, Scheuer, & Meredith,
2007; Gardner, Chmiel, & Wall, 1996). Other individual differences, which
are expected to affect safety performance (through safety motivation or
safety knowledge) as well as situational variables will be referred to as distal
antecedents and will be dened below.
National/Regional Level Antecedents

Recently, several studies have examined how nations differ in regard to
workplace safety (e.g., Burke, Chan-Seran, Salvador, Smith, & Sarpy,
2008; Gyekye & Salminen, 2005; Havold, 2007; Infortunio, 2006). In large
part, these investigations have examined how cultural differences relate to
employees perceptions of workplace safety and accident involvement (i.e.,

fatal accident rates and perceived responsibility for accident causation)
across nations. For instance, in a study with data collected for 43 countries,
Infortunio (2006) found that fatal accident rates were negatively correlated
with Hofstedes (2001) cultural dimension of individualism/collectivism.
As another example, Havold (2007) found that for workers from 10
countries, Hofstedes cultural dimensions of power distance, individualism,
and uncertainty avoidance were positively related to aggregated workers
perceptions of workplace safety (safety climate). While evidence of bivariate
associations between cultural dimensions (that reect cultural values) and
criteria such as accident rates are notable, these types of studies do not
necessarily provide insight into the processes by which national culture
affects aspects of workplace safety. Nevertheless, these studies have been
helpful in pointing to the value of Hofstedes cultural dimensions for
conceptualizing the role of culture in workplace safety.
Although other researchers have developed taxonomies and frameworks
for describing cultural differences (e.g., Schwartz, 1999), considerable
conceptual and empirical support has been generated for Hofstedes cultural
dimensions including the stability of national scores on these dimensions
(see Hofstede & McCrae, 2004; Sndergaard, 1994). Hofstedes (1980, 1991,
2001) work and research based on his taxonomy has focused on ve
dimensions to explain national cultural differences: power distance, which
describes the degree to which power is distributed equally; individualism/
collectivism, which relates to how much an individual values his/her own
needs (individualistic) vs. those of a group (collectivistic); masculinity/
femininity, which refers to the distribution of masculine values (assertive) vs.
feminine values (modest and caring); uncertainty avoidance, which describes
the level of tolerance a society has for ambiguity; and long-term/short-term
orientation, which refers to how much a society values thrift and
perseverance (long-term) vs. fulllment of social obligations (short-term).
Several of these dimensions have important implications for understanding
how and why leaders and organizations within nations differ in their
orientation toward workplace safety.
At the national and regional levels (e.g., state or territory) within nations,
cultural values would be expected to underlie several aspects of the political/
economy that, in turn, would be expected to inuence safety-related working
conditions within organizations. Here, we refer to the political economy as
the political, economic, and legal systems of a country. More specically,
cultural values would be expected to drive the social justice orientation, the
scal capacity of nations and states, and the capacity for labor to organize:
characteristics of the political economy that may signicantly impact the

safety of working conditions within organizations. Our focus on cultural
values as antecedents of aspects of the political economy is not intended to
imply that the political economy of nations is a sole function of cultural
values. The political, economic, and legal systems of nations have roots in
historical/colonial ties, available natural resources, and international agreements to name just a few other antecedents (e.g., Golub, 1991; Jammeh &
Delgado, 1991; Schad, 2001). Our focus on cultural values is meant to
highlight these basic assumptions of national culture as the primary drivers
of aspects of the political economy that relate to workplace safety.
Social justice rests on the premise that all individuals belonging to a society
(a nation or region within a nation for the purposes of this chapter) should
enjoy equal rights, responsibilities, and benets (Turner, Pope, Ellis, &
Carlson, 2009). In effect, the social justice orientation of a nation reects the
extent to which the equal rights of individuals are valued and individuals are
responsible for their actions. As such, the cultural dimensions of power
distance, individualism/collectivism, and uncertainty avoidance would be
expected to directly affect the social justice orientation of nations.
In terms of a social justice orientation, differences can be observed across
nations in institutions and legal considerations that relate to human
resources in general and workplace safety in particular (LaDou, 2002). At a
more general level, many democratically oriented nations have laws that
relate to equal employment opportunities. Moreover, in regard to safety,
institutions within these types of nations often put into play research
agendas such as the National Occupational Research Agenda in the United
States (National Institute for Occupational Safety and Health (NIOSH),
2006) and engage in assessment practices that lead, through legislative and
enforcement processes, to a focus on the rights of workers and safe work
conditions. For instance, in a number of developed nations that emphasize
individual rights, such as Canada, Germany, Sweden, and France,
regulations have been formulated that guide not only human resource
practices, but also workplace health and safety considerations (see Burke &
Sarpy, 2003). A case in point is the United States where the Occupational
Safety and Health Administration (OSHA) has requirements and guidelines
that relate to the maintenance of safe work conditions and the development
(in terms of training) and protection of individual workers (OSHA, 1998).
Notably, the OSHA guidelines not only reect a concern for individual
rights, but they also emphasize uniformity in adherence to these requirements across organizations. The importance placed on standardization and
uniformity in workplace safety practices and work conditions is undergirded
10
by the cultural value of uncertainty avoidance. These arguments form the

basis for the expected indirect effects, as specied in Fig. 1, of cultural values
on organizational policies and practices through an aspect of the political
economy (i.e., social justice orientation) of nations.
Cultural values such as power distance and masculinityfemininity are
also expected to have meaningful inuences on national wealth and,
particularly, the distribution of wealth within nations (see Husted, 1999). In
comparison to cultures with low-power-distance orientation and lower
masculinity, cultures having higher power distance and masculinity
orientations would be expected to have greater inequities in the distribution
of wealth, where the majority of the wealth in the country would be under
the control of a select few. The role and size of government would also be
expected to be restricted in these cultures (Husted, 1999). As such, nations
characterized by high vs. low power distance and masculinity would likely
have less scal capacity and fewer regulatory bodies to focus on workplace
safety issues. To the extent that nations and states have the scal capacity to
effectively monitor and regulate occupational health and safety, organizational policies and practices and the consequent safety climates of
organizations are likely to improve. Here, organizational safety climate
refers to work environment characteristics in relation to safety matters that
affect members of the group or organization. In line with this expectation,
Pfeffer and Salancik (1978) argued that in states with high scal capacity,
organizations respond to regulatory uncertainty by reducing conditions that
lead to sanctions. On the other hand, high national or state debt may
impede an entitys ability to effectively monitor and regulate occupational
health and safety.
We caution that lower scal capacity alone may not explain declines in
efforts to enforce safety-related laws and regulations. This point is apparent
in gaps in organizational compliance with child labor laws and the
concurrent striking decline in child labor law enforcement activities, which
may reect changing enforcement policy as much as scal considerations
(see Rauscher, Runyan, Schulman, & Bowling, 2008). At the regional level,
another example is the lax and questionable enforcement policies in the case
of Nevadas OSHA, which has allowed unsafe conditions to persist
(Associated Press, 2009). Despite such cases, enforcement activities are
expected to be positively associated with improvements in organizational
safety policies and practices and organizational safety climate (e.g., lower
hazardous orders violations dened as using a piece of equipment
prohibited by law), as organizations attempt to conform to government
standards and at the same time avoid sanctions.
11
Furthermore, cultural values would be expected to inuence the

organizing capacity of labor. Traditionally, labor unions were founded on
the premise of protecting the rights of individual workers (Rosenberg, 2009),
where valuing the equal rights of individuals would be considered an
important driver of unionization. However, in his analysis of hybrid
regimes, Robertson (2007) demonstrates that the premise of protecting the
rights of workers as a basis for unionization is specic to liberal
democracies. In authoritarian regimes, unions are known to be used as a
means to suppress or control workers or where the ability to unionize at all
has been taken away from workers (Robertson, 2007). These considerations
lead us to posit that nations higher in power-distance orientation would be
more likely to suppress the unionization of their labor forces or use labor
unions as a means of control.
To the extent that labor can organize within a nation or state, we would
expect organizations to maintain sound organizational policies and practices
in relation to safety and more positive safety climates and safe working
conditions (e.g., reduce hazardous equipment violation). In effect, organized
labor may force the maintenance of higher safety standards in efforts to
safeguard the health and well-being of its members. In the United States,
this general expectation is consistent with research that shows that
unionized workplaces tend to be more compliant with Occupational Safety
and Health Act regulations (Weil, 2001). However, to the extent that
societal factors including prejudice impede workers efforts to organize for
improvements in work conditions (e.g., in the case of immigrant workers;
Grifth, 1988), the work environment is likely to be less safe and associated
with higher occupational injury rates (Dong & Platner, 2004; Richardson
et al., 2004).
In nations whose political economic climate favors industry over labor,
we would also expect organizations to have potentially more hazardous
work environments (Loomis et al., 2009; Richardson et al., 2004). That is,
when economic development is coupled with less attention to workplace
safety, workers are likely to face greater potential exposures to general
workplace hazards (e.g., biological, chemical, radiological, and noise). Tied
to this point, in nations that have less developed economies, jobs will be
concentrated more in industries with higher levels of potentially severe
hazardous events and exposures (Koh & Chia, 1998). As a result, we would
expect the political economy of nations and states to directly affect the
nature of workplace hazards and, indirectly, the occurrence of accidents,
injuries, and illnesses/diseases. This expectation implies that, depending on
the political-economic climate of nations or regions within nations, negative
12
individual level safety outcomes and rates would differ across workers in the
same jobs and same industries.
Related to the above discussion, Holzberg (1981) has discussed how
cultural values inuence social stratication through the political economies
of nations. This point is relevant to our discussion, as such processes can
operate in a manner where different racial/ethnic or socioeconomic groups
are at times disproportionately exposed to hazardous work environments.
The reader is also referred to Baum, Be`gin, Houweling, and Taylor (2009)
and the Commission on the Social Determinants of Health (CSDH, 2008) for
more general discussions of the social determinants of health inequities
within and across nations. For instance, within the United States, AfricanAmerican men have higher cancer rates than White men and these
differences are, in part, attributed to differential exposures to hazardous
occupational conditions (see Briggs et al., 2003). As another example, in both
declining types of work (e.g., hand harvesting of row crops) and expanding
labor-intensive types of work (e.g., construction and landscaping) in the
southern United States, a pattern has emerged where these types of work are
done by Hispanic workers who are experiencing some of the nations highest
fatal occupational injury rates (Richardson et al., 2004). Also, in comparison
to non-Hispanic White men, African-American and Hispanic men miss more
work days due to injury (Strong & Zimmerman, 2005). These developments
signal a strong need, as discussed in detail below, to reconsider how
educational/development activities can take into account backgrounds of
workers along with information on the severity of workplace hazards to
enhance worker motivation to learn about and avoid such hazards (some of
which may seem quite benign, such as wood dust).
Together, we expect political/economic factors to have meaningful direct
effects on organizational policies and practices in relation to safety and
workplace hazards that are present in these organizations. Furthermore,
while some research has examined direct relationships between political/
economic factors and occupational injury rates (e.g., Loomis et al., 2009),
we view the effect of political/economic factors on safety performance and
safety outcomes such as fatal occupational injuries to be indirect through
the environment (workplace hazards and safety climate) that workers
experience. To date, research has not been directed at addressing expected
cross-level mediated relationships involving factors at the national and
organizational levels of analysis as specied in Fig. 1.
Finally, as indicated in Fig. 1, we expect cultural values to directly affect
leadership in organizations. As noted above, evidence indicates that cultural
values, such as those focused on equality and respect for individual rights,
13
become personalized in the sense that they remain relatively stable over
time. In addition, Hofstedes cultural dimensions, in particular those
focused on individualism and masculinity, have been shown to have a
greater effect on workers perceptions of the ethicality of actions within
organizations than ethicality as specied in company codes of conduct
(Arnold, Bernardi, Neidermeyer, & Schmee, 2007). As such, we expect
cultural values to affect organizational safety climate indirectly through the
actions of organizational leaders embedded within those nations. We will
comment further on this expectation below.
Organizational/Group-Level Antecedents
The above discussion recognized that elements of the national/regional
environment can have meaningful direct and indirect impacts on organizational functioning insofar as the personal values of leaders and safety
climate of the organization are concerned. As indicated in Fig. 1, at the
organization/group level of analysis, we expect societal values (e.g., justice)
that become personalized over time as well as organizationally espoused
values to underlie leadership, where leadership refers to the actions that
managers take to achieve organizational and group objectives. Organizationally espoused values such as Safety is Job One or Create and Sustain
a Safe Work Environment signal what is strategically important to leaders
of organizations. To the extent that these values are enacted by leaders, they
guide the various organizational policies and practices (i.e., safety-related
policies and production technology in the organization as well as human
resource management policies and practices in relation to workplace safety).
Thus, we expect leader behavior to mediate the inuence of cultural values
and organizationally espoused values toward safety on the human resource
and safety-related policies of the organization.
Furthermore, the linkage between leadership and organizational or grouplevel climate has long been recognized in the literature (Lewin, Lippitt, &
White, 1939). That is, leaders are directly or indirectly (as noted above
through the establishment of policies and procedures) instrumental in
shaping and reinforcing an organizational climate for safety. However, if the
leadership of an organization or group is decient in terms of establishing a
positive safety climate (e.g., in terms of supervisory support, work pressure,
and reward practices), then a weak safety climate will result irrespective of
formal policies. Further, if one were to view leadership behaviors and leader
member interactions along a continuum of support for the welfare or
14
concern of employees, then different forms of leadership (e.g., transformational) would be expected to have stronger effects on safety climate in
comparison to other forms of leader behavior (e.g., corrective leadership)
(Zohar, 2003).
Noting that production technology and potential exposures to workplace
hazards differ across departments of an organization, the possibility exists
that safety climate may also differ across these organizational units. This
possibility is likely realized when top management beliefs about safety are
not uniformly conveyed to lower levels of management or when supervisors
of organizational subunits hold different beliefs about the importance of
safety (Cox & Cox, 1991; Williamson, Feyer, Cairns, & Biancotti, 1997). In
particular, when supervisors of work groups hold different beliefs about who
is responsible for safety or attribute accidents to either internal or external
factors, then they are likely to execute organization-level policies and
procedures differently, regardless of risks involved (Zohar, 2000; Zohar &
Luria, 2004). Related to this point, we note that the safety climate dimension
of managerial support, which is included in almost all models of safety
climate, refers to the support that employees immediate supervisor provides
in carrying out their work. In this sense, our model in Fig. 1 leaves open the
possibility that organizational safety subclimates may be operating. Nevertheless, the expected causal paths in the model would be expected to hold for
possible differences in organizational climates as well as for differences in
organizational safety subclimates.
Safety climates are also likely to vary across subunits of an organization
due to the fact that members of different organizational subunits often face
different hazards or risks. For instance, nurses within a hospital emergency
room may face exposure to harmful substances such as bloodborne
pathogens; whereas nurses in a rehabilitation unit may confront situations
that lead to overexertion. Notably, different social groups or units within an
organization can have different interpretations of risk, even when the
potential hazardous event or exposure is the same (Weyman & Clarke,
2003). In fact, Pidgeon (1991) and others (e.g., Fleming, Flin, Mearns, &
Gordon, 1998; Perez-Floriano & Gonzalez, 2007; Rundmo, 1996) have
discussed notions of safety subcultures and these groupings inuence both
the perception of risk and safety-related behavior.
However, noting that perceptions of risk are socially constructed, we
believe that arguments in the literature to essentially abandon the study of
objective risks in favor of the study of perceived risk (e.g., Morrow & Crum,
1998; Pidgeon, 1991) unnecessarily promote a strict person-based approach
to the study of risk. The weakness of this approach is apparent from the
15
considerable evidence showing that differences between individuals often

lack cross-situational consistency (Hattrup & Jackson, 1996; Mischel, 1990).
If, in fact, risk perception is socially constructed, a point that we agree with,
then understanding how risk perceptions and safety-relevant motivations
(e.g., motivation to learn about and avoid hazardous events or exposures)
develop in situ is of paramount concern. Yet, such investigations would
necessitate a cross-level or interactionist approach that explicitly measures or
takes into account the properties of potential hazardous events and
exposures. We will return to this point in our discussion of educational/
development activities.
Returning to our discussion of organizational safety climate, a number of
authors have discussed how organizational climate might moderate relationships between individual difference variables (Brown, 1981; Tracey, Tannenbaum, & Kavanagh, 1995). Notably, a body of literature on climates-forsomething suggests that organizational climate moderates relationships
between individual difference variables to the extent that the organization
promotes a strategically focused climate (i.e., a climate that is aligned with
organizational goals). This research is premised on the assumption that
individuals strive to achieve an adaptive t with their work environments. As
noted by Smith-Crowe, Burke, and Landis (2003), one would expect that if an
organization has a climate supportive of safety, then workers will exert effort
to act safely and transfer the knowledge that they have acquired through
educational/developmental experiences. On the other hand, in a comparable
organization without a climate focused on safety, one would expect a lower
relationship between safety knowledge and safety performance as well as
between safety motivation and safety performance because workers would not
necessarily be willing and able to exhibit acquired knowledge. This argument
applies to both safety compliance and safety participation. The result being
that we would expect safety climate to moderate relationships between safety
knowledge and safety performance and safety motivation and safety
performance, with respect to both safety compliance and safety participation
criteria. Surprisingly, beyond Smith-Crowe et al. (2003), little research has
been directed at examining safety climate as a moderator of such relationships
with no research to our knowledge on the degree to which safety climate
moderates safety motivationsafety performance relationships.
Although we did not emphasize a discussion of the expected direct effects
of human resource policies and safety-related policies on an organizations
safety climate, these policies and associated practices (e.g., selection
practices, training procedures, and safety meetings) are well recognized as
important contributors to organizational safety climates (Neal & Grifn,
16
2004). Notably, within supply chains, Cantor (2008) discusses how human
resource practices as well as physical resources and production technologies
can enable organizations to better track and monitor activities and promote
a strong safety climate. In an age of global sourcing practices, we note that
the issue of workplace safety in organizational supply chains is an
understudied area, where unsafe practices in one leg of the chain or
organization can have catastrophic adverse consequences in another part of
the supply chain (see Cantor, 2008).
Distal Individual Level Antecedents

Psychological (Safety) Climate
Prior to discussing the role of employees perceptions of work environments
characteristics (i.e., psychological climate), we note key distinctions between
organizational climate and psychological climate. Organizational climate
was dened above as work environments characteristics in relation to
matters (here, safety) that affect members of the group or organization.
Shared or aggregated perceptions of work environment characteristics may
serve as indicators of group or organizational climate (see James et al.,
2008). While this point is widely recognized in the literature, what is not
recognized is that organizational climate need not be strictly operationally
(and conceptually) dened in terms of shared perceptions of work
environment characteristics. For instance, Burke et al. (2008) and SmithCrowe et al. (2003) have discussed how subject matter expert judgments and
content analyses of archival data may also serve as useful indicators or
provide meaningful descriptions of an organizations climate. Furthermore,
as discussed above within the domain of workplace safety, an organizations
safety climate is primarily determined by leadership practices, human
resource policies and practices, safety-related policies and practices, and
workplace hazards. On the other hand, as discussed in more detail below,
employees perceptions of work environment characteristics (psychological
climate) are conceptualized as personal value-based appraisals of work
environment characteristics. This distinction between organizational and
psychological climates is nontrivial as meaningful and differential variation
in climate scores (for similar dimensions) may exist at both levels of analysis
and would be expected to have different implications for understanding the
role of climate in individual safety performance and its consequences.
Finally, we note that our reference to psychological safety climate, which
refers to individual perceptions of characteristics of the work environment
17
in relation to safety, is distinct from Edmondsons (1999) notion of

psychological safety that relates to a teams shared belief that members are
free to engage in risk taking.
In recent years, a number of studies have examined relationships at the
individual level of analysis between dimensions of climate, safety compliance
factors, safety participation factors, and accident involvement across a wide
range of occupations and industries (see Clarke, 2006a). A general nding is
that the magnitude of safety climatesafety participation relationships is
greater than the magnitudes of the associations between safety climate
dimensions and safety compliance dimensions (Christian, Bradley, Wallace, &
Burke, 2009; Clarke, 2006a). However, with a few exceptions (Christian et al.,
2009; Grifn & Neal, 2000), researchers have not conceptually or empirically
addressed the underlying mechanisms through which psychological climate
perceptions affect safety performance and safety outcomes.
In an effort to advance our understanding of the linkages between
psychological climate and safety performance, we note that researchers have
offered varied denitions and measures of safety climate at the individual
level of analysis (see Flin, Mearns, OConnor, & Bryden, 2000) with some
denitions and measures being more narrowly focused on supervisory
practices (e.g., Zohar & Luria, 2004). In our view, conceptualizing safety
climate (at either the individual or organizational level) with respect to
climate factors that have been conrmed across a wide range of occupations
and industries (i.e., means emphasis, goal emphasis, management support,
etc.; see Burke, Borucki, & Kaufman, 2002; James et al., 2008) as well as
with respect to several more safety-specic issues that apply across
organizations (e.g., work pressure; see Neal & Grifn, 2004) provides a
comprehensive, yet parsimonious denition of the factor space of work
environment characteristics. Furthermore, we view psychological (safety)
climate as an employees perception of these work environment characteristics that affect not only the employees personal well-being, but also the
well-being of relevant stakeholders (e.g., customers, suppliers, and the
public). For instance, if we were studying safety climate in the nuclear
hazardous waste industry, employees can provide useful information on
aspects of the work environment (e.g., safe handling, storage, and disposal
of nuclear waste) that not only affect them personally, but also their
perceptions of factors that might adversely affect the public (such as the
failure to maintain protective barriers in underground storage tanks, the
training provided to materials handlers, and so on).
In adopting a multiple stakeholder perspective to the study of safety
climate, we also view safety climate perceptions as being hierarchically
18
arranged. However, unlike other researchers who have posited a single,

higher-order factor tied to employees perceptions of how the work
environment affects their personal well-being (see Christian et al., 2009;
Grifn & Neal, 2000; Neal & Grifn, 2004), we conceptualize the higherorder factors as employees assessments of well-being with respect to
relevant stakeholder groups that they interact with in the organizations task
environment. That is, to the extent that organizationally espoused values
reect concern for the safety and well-being of multiple stakeholders and
organizational practices reinforce these values, then we would argue that
employees cognitively appraise their work environment with respect to the
impact of work environment characteristics on personal well-being as well
as with respect to the well-being of the other relevant stakeholder groups.
Empirical support for a multiple stakeholder conceptualization to psychological climate has been found in both business (see Burke, Borucki, &
Hurley, 1992) and educational settings (see Vaslow, 1999), and discussed in
detail relative to workplace safety (Burke et al., 2002).
A multiple stakeholder conceptualization of safety climate holds promise
for broadening the domain and measurement of safety climate and may lead
to an improved understanding of the effects of safety-related work contexts
on individuals and groups other than employees. This point is important as
safety contexts, perhaps more so than any other type of work environment,
have the potential to affect the well-being (i.e., in the broadest psychological
and physical sense) of employees, their families, customers/clients, and the
public. One need not look far for striking cases of how work environment
characteristics that were initially studied in relation to occupational illnesses
and disease quickly became major public health concerns (e.g., in the cases
of silicosis, asbestosis, lead toxicity, and pesticide poisoning; Corn, 1992;
Kipen, 1994; Nuwayhid, 2004; Rosner & Markowitz, 1991). We know very
little about how workers perceive characteristics of work environments in
relation to safety and health of other stakeholders.
Studying safety climate from a multiple stakeholder perspective will also
be important as the nature of work changes. While applied psychology and
management scholars have studied many aspects of the changing nature of
work in regard to individual and group outcomes (see Ilgen & Pulakos,
1999), virtually no research has been directed at examining how the
changing nature of work affects workers climate perceptions and, in
particular, their safety climate perceptions. Yet, changes in the nature of
safety-related work often bring with them not only new production
technologies, which have new associated injury risks, but also potential
for new exposures that could result in illnesses and diseases. These changes
19
may also have important implications for how workers perceive the safety of
their work environments with respect to their well-being and the well-being
of others.
As an example of how the changing nature of work can potentially affect
safety climate, we will briey discuss a few implications of the massive public
works effort to repair the national highway system in the United States. This
change from a focus on building highways to an emphasis on the repair of
highways has brought with it a new method of repair, which uses large crews
to cut, break-up, and remove large blocks of concrete. This process results in
the generation of large amounts of dust and increased risk of silicosis for
workers (Valiante, Schill, Rosenman, & Socie, 2004). Silicosis is a disabling,
nonreversible lung disease (NIOSH, 2002). Given the context of this work,
the threat is also a public health concern. While health-monitoring processes
such as epidemiological exposure assessments can provide valuable
information related to potential work exposures of this nature (Ott, 1998),
broader safety climate assessments (at either the individual or organizational level of analysis) have the potential for identifying where more
immediate actions and interventions could be initiated to safeguard various
constituents. This is just one example where improved conceptualizations
and measures of safety climate from a multiple stakeholder perspective may
provide valuable data for understanding safety-related outcomes.
Personality
A substantial amount of empirical evidence exists concerning relationships
among personality characteristics, safety performance, and accident involvement (Christian et al., 2009; Clarke & Roberston, 2008). While many studies
have employed measures of general personality characteristics that could
be classied within the Big 5 framework (e.g., conscientiousness, Geller,
Roberts, & Gilmore, 1996; extraversion, Iverson & Erwin, 1997), research has
also focused on several more specic aspects of personality (i.e., propensity
for risk taking, Frone, 1998; locus of control, Salminen & Klen, 1994).
In terms of safety performance and accident involvement, a number of
studies have examined locus of control as a predictor (e.g., Brown, Wilis, &
Prussia, 2000; Eklof, 2002; Hsu, Lee, Wu, & Takano, 2008; Rundmo, 2001).
Locus of control is the extent to which an individual believes that events are
under his or her control as opposed to being the result of situational factors.
Meta-analytic ndings are consistent with the expectation that individuals
who are higher in terms of internal control are more motivated to learn
about workplace safety, engage in higher levels of both safety participation
and safety compliance, and have fewer accidents (see Christian et al., 2009).
20
Notably, preliminary evidence indicates that locus of control may be a

relatively strong predictor of safety participation and a good predictor of
accident involvement. These ndings are consistent with the motivational
implications of believing that one has control over events, and have
important implications for worker selection and training. However, more
research is needed before rm conclusions can be made about the relative
utility of locus of control as a predictor of safety participation vs. safety
compliance. Furthermore, research examining the expected indirect effect of
locus of control on safety performance and accident involvement through
safety motivation is needed.
Conscientiousness has also received a fair amount of attention as a
predictor of safety performance and accident involvement (Haaland, 2006).
Those individuals who score high on this personality dimension are more
likely to be trustworthy and dependable, which would lead them to being
motivated to engage in appropriate safety behavior and have fewer negative
safety outcomes. Despite this general expectation, conscientiousness has a
somewhat modest (low) relationship with safety performance (see Christian
et al., 2009). This result may be due to the fact that conscientiousness would
be expected to primarily affect safety performance and safety outcomes
through safety motivation. Future research examining the extent to which
conscientiousness indirectly affects safety performance (with an emphasis on
the study of participatory behaviors) and safety outcomes would be
informative. To date, only a few studies have examined conscientiousness as
an antecedent to safety participation (e.g., Geller, 1996).
In regard to accident involvement, neuroticism has been found to be
consistently related, albeit in a low positive manner, to accident involvement
(e.g., Davids & Mahoney, 1957; Frone, 1998; Hansen, 1989; Salminen,
Klen, & Ojanen, 1999). The general reasoning is that individuals higher in
neuroticism are more likely than those with lower levels of neuroticism to
experience negative affective states, which would lead them to have lower
levels of safety participation, more lapses of attention, and be predisposed to
making more mistakes. Although neuroticism has received a fair amount of
research attention relative to other personality characteristics in regard to
accident involvement, the available evidence would suggest that conscientiousness and locus of control play more central roles in the explication of
negative safety outcomes.
Although several studies have examined the role of extraversion in
accident involvement and found it to have a low negative association (see
Christian et al., 2009), fewer studies have incorporated measures of openness
to experience and agreeableness. Individuals who are high in agreeableness
21
are more friendly, good natured, and likely to conform to social norms.
These characteristics should lead to higher levels of safety motivation and,
consequently, higher levels of safety participation and safety compliance and
fewer accidents. On the other hand, we do not have a strong basis for
suggesting how openness to experience will affect safety outcomes. Yet, to
the extent that elevated levels of openness to experience and extraversion are
related to thrill seeking or the propensity to take risks (an amalgamation of
Big 5 traits; see Nicholson, Soane, Fenton-OCreevy, & Willman, 2005), we
would expect safety motivation to decrease and consequent unsafe work
behavior to increase. In effect, the thrill seeking would be expected to
undermine the desire or need to engage in safe work behavior.
Education/Development Experiences
Educational and development experiences broadly dened are expected to
lead to both the development of safety motivation and acquisition of safety
knowledge. Formal on-the-job (Baird, Holland, & Deacon, 1999; Seibert,
1999) as well as informal or off-site (Curwick, Reeb-Whitaker, & Connon,
2003; Marsick & Watkins, 1997) educational activities within the safety
domain focus heavily on the development of factual (often referred to as
declarative) knowledge and procedural knowledge and skills related to using
personal protective equipment, engaging in work practices to reduce risk,
communicating health and safety information, and exercising employee
rights and responsibilities. From here onward, we will use the term safety
knowledge interchangeably with the concepts of declarative and procedural knowledge. In addition, educational activities can be directed at the
development of worker attitudes and regulatory activities in efforts to
enhance safety motivation (Ford & Tetrick, 2008). Thus, we would expect
these educational experiences to directly relate to safety motivation.
The primary means for developing safety knowledge is formal training
(Colligan & Cohen, 2004), which is occasionally guided by the application of
a particular learning theory. For instance, the literature is replete with
applications of stage learning theories (e.g., Azizi et al., 2000), reinforcement theory (e.g., Cooper, 2009; Lingard & Rowlinson, 1997), and
principles of social and experiential learning theories (e.g., Lueveswanij,
Nittayananta, & Robison, 2000) to safety knowledge development. The later
interventions often employ more hands-on, experiential training methods
such as role-plays, demonstrations with practice, and simulations
involving individuals, dyads, and teams. Along with the use of multiple
training methods in the delivery of training, many of the training programs
based on social and experiential learning theories emphasize individualized
22
feedback and dialogue in small groups (e.g., Luskin, Somers, Wooding, &
Levenstein, 1992).
In 2006, Burke et al. reported on a meta-analysis that examined the
relative effectiveness of safety and health training methods according to the
extent to which trainees participated in the learning process. Their metaanalytic ndings were consistent with the theoretical argument that as the
method of safety and health training becomes more engaging (going from
passive, less engaging methods such as lecture to experiential-based, highly
engaging methods such as hands-on training that incorporate dialogue), the
effect of training is greater for knowledge acquisition, safety performance,
and the reduction of accidents and injuries. Importantly, their ndings point
to needed research on the usefulness of incorporating more active forms of
participation into traditionally structured safety training and development
efforts including the commonly employed computer-based and distance
instructional methods.
Burke et al.s (2006) ndings also suggest that the unbridled promotion of
reinforcement or operant theory as Behavioral Safety be tempered. The
reader is referred to Geller (1996) and McSween (2002) for discussions of
operant theory that underlies moderately engaging feedback interventions.
As discussed elsewhere (Burke et al., 2007; Cooper, 2009), applications of
operant theory are most effective when work is relatively static (i.e., involves
primarily routine actions) and where the intended target behaviors relate to
safety compliance. Also, as discussed by Olson and Winchester (2008), the
workplace literature on behavioral self-monitoring (BSM, another name
associated with applications of operant theory) is theoretically unfocused
and has neglected relevant scholarly work.
A particular deciency in the literature on educational efforts to develop
safety knowledge and safety motivation is that we know little about learning
conditions that promote dialogue and reective thinking. Arguably, dialogue
and reection are critical elements of the learning process and efcacy
formation (Burke, Holman, & Birdi, 2006; Gorsky & Caspi, 2005; Holman,
2000a, 2000b). Dialogue involves discussion with others including virtual
others (interpersonal dialogue) or ones self (intrapersonal dialogue), often
with respect to actions taken or considered. Dialogue is characterized by
thought-provoking activities such as questioning, explaining, and evaluating
issues or problems at hand. Reection is a systematic thought process
concerned with simplifying experience (i.e., thinking about contradictions,
dilemmas, and possibilities). Practitioners and researchers alike could
consider different forms of dialogue (e.g., inquiry, debate, argumentation,
storytelling; see Cullen & Fein, 2005; Gorsky & Caspi, 2005) and different
23
structural considerations for promoting intrapersonal dialogue (e.g., selfinstruction/review materials, tutorial sessions, website materials) and
interpersonal dialogue (e.g., actual and computer or web-based discussions;
see Gorsky, Caspi, & Trumper, 2004). The form, structure, and instructional
activities would likely be somewhat specic to the nature of the safety
training intervention, the level of skill being acquired, and the size of the
training group (Frederiksen, 1999; Gorsky, Caspi, & Trumper, 2006;
McConnell, 1997). Nevertheless, research on the role of dialogue and
reection has considerable potential for advancing our understanding of how
to optimally develop safety knowledge and safety motivation.
More recently, Burke et al. (2009) discussed how hazardous events and
exposures might interact with developmental activities to inuence safety
motivation and safety knowledge acquisition. Burke et al. (2009) posited that
for hazardous events and exposures of an ominous nature (e.g., res and
explosions, exposure to toxic chemicals, radiation, and human immunodeciency virus; see Mullet, Ciutad, & Riviere-Shaghi, 2004), the action,
dialogue and considerable reection that take place in highly engaging
developmental activities such as simulation training would be expected to
engender a dread factor, a realization of the actual dangers and feelings of
dread. Furthermore, they argued that this realization and the experienced
feelings and negative affect should play a primary role in motivating
individuals to learn about how to avoid exposure to such hazards. However,
the dread factor would not necessarily be produced in (a) highly engaging
development activities targeted at hazardous events and exposures that
typically do not have severe injury potential (e.g., contact with objects and
equipment, excessive physical effort, and repetitive bodily motion) or (b) in
lesser engaging development activities (e.g., lectures), irrespective of the level
of hazard. Their theoretical arguments are consistent with that of many
theorists who have given affect a direct and primary role in motivating
behavior, especially in regard to unpleasant feelings, which arguably
motivate action that people anticipate will avoid such feelings or associated
consequences (see Schwartz & Clore, 1988; Slovic & Peters, 2006).
Furthermore, their arguments are in line with social-cognitive perspectives
concerning workers willingness to participate in safety interventions and the
development of safety motivation (see Cree & Kelloway, 1997; Floyd,
Prentice-Dunn, & Rogers, 2000; Ford & Tetrick, 2008; Goldberg, Dar-El, &
Rubin, 1991).
The above arguments are the basis for the expected interaction between
objective workplace hazards and educational/development activities in the
formation of safety motivation and safety knowledge. That is, the above
24
arguments suggest that objective hazards and development activities of either

a formal or informal nature should jointly affect the motivation to engage in
safe work behavior and the knowledge of how to do so. In effect, the
motivational and learning benets of more engaging education/development
activities should be enhanced when individuals face hazards of a particularly
ominous nature, with no necessary enhanced benets of more engaging
developmental activities when hazard severity is relatively low or when the
developmental activity is less engaging. However, we caution that when
choosing educational/development activities, practitioners and researchers
may need to take into account the backgrounds of workers who may be
concentrated in particular types of hazardous work. For instance, indigenous
farm workers from Mexico and Guatemala who are working in the northwest
United States are not of Hispanic or Latino descent and come from regions
with unique cultural and linguistic traditions that could affect what and how
they learn (Farquhar, Shadbeh, Samples, Ventura, & Goff, 2008).
Notably, the workplace hazard measurement system discussed in Burke
et al. (2009), which is based, in part, on the Bureau of Labor Statistics
Occupational Injury and Illness Classication System (OIICS) (also see
Biddle, 1998), hierarchically arranges workplace hazards to reect the
increasing potential for severe illness, injury, or death due the hazardous
event or exposure. This workplace hazard measurement system permits the
scoring of hazard event/exposure in terms of low and high severity hazards,
and roughly corresponds to breaking the Bureau of Labors ranking system at
the midpoint of the OIICS hierarchy. Such a dichotomy highlights the point
at which the consequences of hazards go from being less severe (e.g., slips,
overexertion, repetitive motion within the third most severe hazard category:
bodily reaction and exertion) to being more severe (e.g., the contraction of
hepatitis or HIV resulting from needle sticks within the fourth most severe
hazard category: exposure to harmful substances and environments). Moreover, this workplace hazardous measurement system offers considerable
potential for reintroducing the study of objective hazards into the human
resource management and organizational behavior literature to illuminate
how workplace hazards interact with individual characteristics to affect
workplace safety.
Returning to a discussion of educational activities per se, several studies
have pointed to educational status (i.e., educational level, school attendance,
high school drop out) as a predictor of work injuries (e.g., Breslin, 2008;
Breslin et al., 2007). These studies, several of which are longitudinal,
indicate that young workers (aged 1624) with less than a high school
education are up to three times more likely to have a work disability absence
25
than those with at least a high school diploma. Notably, these ndings are
maintained when controlling for the type of work and number of hours
worked. These results imply that individual differences such as those
modeled in Fig. 1 might be fruitfully investigated to further explicate the
role of individual differences associated with educational attainment that
are impacting workplace safety for those with more vs. those with less
formal education. Finally, although various approaches have been
considered for imparting work-relevant safety knowledge to high school
students and young workers (e.g., through vocational/technical education,
skill standards, career clusters initiatives, and apprenticeships), we know
very little about how these efforts affect safety performance and safety
outcomes (see Schulte, Stephenson, Okun, Palassis, & Biddle, 2005).
Cognitive Abilities
While general mental ability is well recognized as an antecedent to job
knowledge as specied in Fig. 1 (Colquitt, LePine, & Noe, 2000; Schmidt,
Hunter, & Outerbridge, 1986), the study of cognitive abilities in safety
research has been largely delimited to the role of cognitive abilities in
accidents (e.g., Arthur, Barrett, & Alexander, 1991; Lawton & Parker, 1998;
Wallace & Vodanovich, 2003). Here, research indicates that selective
attention and cognitive failures are meaningful predictors of safety
performance and accident involvement (Wallace & Vodanovich, 2003).
Cognitive failure refers to a breakdown in cognitive functioning, which
results in an error or mistake in task execution.
Cognitive failures have origins in the organization of work (e.g., 12-hour
shifts in units with stafng shortages for nurses; see Smith, Folkard, Tucker, &
Macdonald, 1998), sleep opportunities (Dawson & McCulloch, 2005), and
relatively stable individual differences. In regard to stable individual
differences, some personality characteristics may predispose individuals to
being more susceptible to experiencing cognitive failures than others (Wallace,
Kass, & Stanny, 2002). While research has examined expected interactions
between cognitive failures and conscientiousness on accidents, examination of
how the interactive effect of cognitive failure and personality characteristics
such as conscientiousness operate through safety compliance on safety
outcomes including accidents and near misses has not been made.
In addition, learning disabilities, a general term used to describe a variety
of information-processing problems, would be expected to directly relate to
the acquisition and retention of safety knowledge. Depending on the learning
disability including attention-decit/hyperactivity disorder (ADHD), problems may arise with respect to reading, memory, abstract reasoning, and
26
spatial orientation. As a result, individuals with such disabilities may, on

average, acquire fewer job-relevant skills in comparison to those without
such disabilities and be more likely to hold physically demanding and more
hazardous jobs. While there is some evidence to support this assertion (see
Mannuzza, Klein, Bessler, Malloy, & Hynes, 1997), the cognitive consequences of their disability alone would place them at greater risk of
workplace injury and illness even for comparable jobs with people without
such disabilities. In this sense, learning disabilities and ADHD can lead to
difculties reading instructions or remembering previously taught material
(Schaeffer, 2004). Thus, we would expect learning disabilities to primarily
affect safety outcomes such as injury indirectly through safety knowledge
and safety compliance (e.g., completing tasks in a required sequence).
Indirect evidence for this proposed causal sequence comes from a large-scale
study of workers with self-reported dyslexia (a learning disability characterized by problems in reading, writing, and spelling) where there was an
89% increase in work injury risk among workers with self-reported
dyslexia in comparison to workers reporting no learning disability (Breslin &
Pole, 2009).
Although not depicted in Fig. 1, we would expect some learning
disabilities to interact with aspects of safety climate (e.g., time pressure or
workload demands) to affect safety performance and subsequent safety
outcomes. That is, the expected negative effect of learning disabilities on
safety performance would be greater under more restrictive organizational
safety climate conditions (e.g., high time pressure, high workload) than
under less restrictive and more supportive safety climates. Decrements in
safety performance would be expected to carry through in terms of increases
in workplace accidents, injuries, and near misses.
Physical Abilities and Physiological Aspects of Work
The study of physical abilities and physiological aspects of safety-related
work has been approached in several ways with considerable research being
generated on the muscular nature of work (Smolander & Louhevaara, 1998),
postures at work (Daltroy et al., 1997; Kuorinka, 1998), the ways the body
produces force and generates movement (Darby, 1998; Hultman, Nordin, &
Ortengren, 1984), and the role of fatigue (Robb, Sultana, Ameratunga, &
Jackson, 2008). These efforts have contributed to an understanding of the
physical-related predictors of safe work behavior and the design of worker
and workplace interventions to reduce occupational injuries and illnesses
(e.g., interventions to reduce the incidence of musculoskeletal problems
related to repetitive work or inappropriate postures). In large part, we would
27
expect physical abilities and physiological variables to directly relate to the

actions that workers engage in, and indirectly through these actions (i.e.,
safety performance) to occupational injuries and illnesses. This expectation is
supported by ndings that functional measures, such as balance, reaction
time, and isometric muscle strength, are related to performance on simulated
tasks (Davis, Dotson, & Santa Maria, 1982; Guralnik & Ferrucci, 2003) and
that increasing the physical capacity of workers can be benecial in reducing
their mist with a work system (Genaidy, Karwowski, & Shoaf, 2002; Tuncel
et al., 2008). In the case of non-exercise education/development efforts
related to physical and physiological aspects of work, we expect these
interventions to have direct impacts on workers safety knowledge especially
of a procedural nature.
We note that exibility and strength generally decline with increases in
age (Brandon, Boyette, Lloyd, & Gaasch, 2004; Peate, Bates, Lunda,
Francis, & Bellamy, 2007; Sherrington, Lord, & Finch, 2004). This general
nding has important implications for understanding safety performance
and injury rates in occupations such as reghting, where the execution of
job tasks can require maximal physical performance (see Womack, Green, &
Crouse, 2000). Furthermore, these ndings suggest the need for continued
efforts to evaluate the efcacy of interventions to improve exibility and
core strength (hip complex strength) among workers in highly demanding
physical work, especially where the work conditions are changing and not
under the workers control.
Finally, in relation to the physical characteristics of workers and their
health, little is known about the effects of sickness presenteeism (i.e.,
attending work while sick) on safety performance and subsequent safety
outcomes. Sickness presenteeism has been linked to risk of serious coronary
events, theoretically as a result of the cumulative stress burden of working
while sick (Kivimaki et al., 2005). Yet, sickness presenteeism would be
expected to have more immediate impacts on safety performance especially in
relation to working with others and the effects on coworker health and safety.
Proximal Individual Level Antecedents

Safety Knowledge
Consistent with general models of performance (e.g., Campbell, 1990) and
more specic models of workplace safety (Neal & Grifn, 2004; Christian
et al., 2009), safety knowledge is posited as a direct antecedent to safety
performance. In particular, Burke and Sarpy (2003) discuss how and why
28
knowledge in four content areas is critical to engaging in safe work

behavior: fundamental knowledge and skills (e.g., related to using
personal protective equipment), recognition and awareness knowledge and
skills (e.g., collecting information about workplace hazards), problemsolving skills (e.g., using union and management resources), and decisionmaking skills (e.g., negotiating effective health and safety contract
language). A fair amount of research indicates that safety knowledge in
these content areas has a relatively strong relationship with safety
compliance (see Christian et al., 2009). Importantly, empirical evidence
indicates that these relatively strong relationships between safety knowledge
and safety compliance are maintained across self and supervisory ratings of
job performance as well as across industry and occupational boundaries
(Burke, Sarpy et al., 2002; Grifn & Neal, 2000). In addition, safety
knowledge has been found to be a strong predictor of safety participation
(Christian et al., 2009).
While we have a reasonably good understanding of how safety knowledge
develops and relates to safety compliance and safety participation, we know
relatively little about how safety knowledge decays over time. Perhaps some
of the best evidence related to declines in knowledge and skill comes from
simulation research on healthcare education, where notable decays in
patient safety knowledge occurred from two to eight months (Laschinger
et al., 2008). Related to this point, goal setting and feedback interventions
have been employed in the posttraining contexts to encourage retention and
application of knowledge and skills in relation to more routine, housekeeping-type activities (see Hickman & Geller, 2003). These types of
interventions have been effective in maintaining the display of routine, task
behaviors across different types of work (Geller, 2001; Sulzer-Azaroff &
Austin, 2000). However, little, if any, research attention has been given to
the role of dialogue and action-focused reection in the maintenance of
knowledge and skills and particularly of an advanced procedural nature.
This issue is important as severe injuries, illnesses, and fatalities occur more
often in non-routine types of work (Kriebel, 1982; Peterson, 1998), where
more complex skills are often required (Gardner et al., 1996). In addition,
this issue of how best to promote the retention of safety knowledge is
important due to the fact that efforts to maintain safety knowledge (via
safety training) are mandated for some occupations. To date, we rely more
on scientically uninformed legislative/political processes and legislators to
determine when (in regard to timing/frequency) and how to maintain critical
safety knowledge than on sound theoretical and empirical research bases for
such recommendations.
29
Safety Motivation
Along with safety knowledge, safety motivation is posited as a direct
antecedent to safety performance. This expectation is consistent with
recognized models of workplace safety (Neal & Grifn, 2004; Christian
et al., 2009). Given that safety participation is characterized as behavior that
is more volitional in nature, researchers have expected safety motivation to
be more strongly related to safety participation than to safety compliance.
Some evidence, where participation motivation was measured, indicates that
motivation is more strongly related to safety participation than to safety
compliance with safety knowledge being the primary determinant of safety
compliance (Grifn & Neal, 2000). Nevertheless, Christian et al.s (2009)
meta-analytic ndings indicate that safety motivation is an important
antecedent of safety performance, where performance was primarily
measured with respect to safety compliance.
As specied in Fig. 1, safety motivation would be expected to mediate the
relationship among psychological work climate, personality, and education/
development variables on safety performance. In regard to climate, this
general expectation has been examined in several studies where climate and
conscientiousness were expected to predict safety motivation, which in turn
was expected to relate to safety performance (Christian et al., 2009; Wallace &
Chen, 2006). Notably, these expected mediated effects were found when safety
motivation was conceptualized and measured in a more general manner (see
Christian et al., 2009) as opposed to a more specic conceptualization and
measurement (see Wallace & Chen, 2006, with respect to regulatory focus
theory), and when climate was conceptualized and measured at the individual
vs. group levels in these respective studies.
For the most part, research attention in the domain of workplace safety
has been directed at how safety motivation and safety knowledge mediate
the effects of personality characteristics and climate variables on safety
performance. This research has not examined mediation involving cognitive
abilities and education/development experiences as exogenous variables.
Research incorporating constructs and measures from the latter domains
would inform the relative causal inuence of distal antecedents of both a
cognitive and affective nature on safety performance. This point is
important as safety motivation would be expected to primarily mediate
the effects of affectively oriented distal antecedents on safety participation;
whereas safety knowledge would be expected to primarily mediate the
effects of more cognitively oriented distal antecedents on safety compliance.
More recent research points to the need to examine safety motivation
in relation to two sometimes-competing performance goals: safety and
30
production. Although both types of goals are important in organizational

contexts, understanding if and when employees focus on one type of goal at
the expense of the other type is important (given the consequence of errors in
either domain). Wallace and Chen (2006) refer to the focus on accomplishing
more tasks, more quickly as a promotion focus, and performing tasks
accurately and in accordance with ones duties as a prevention focus (also see
Forster, Higgins, & Bianco, 2003). Wallace and his colleagues research has
indicated that prevention and promotion foci relate positively and negatively
to safety and productivity, respectively, with task complexity serving as a
possible moderator (Wallace, Little, & Shull, 2008). That is, when task
complexity is high, a promotion focus has been found to negatively relate to
safety and a prevention focus negatively to production. The extent to which
Wallace et al.s (2008) laboratory ndings, which examined changes within
an experimental task, generalize to changes in work complexity within and
across jobs, are needed future research directions.
CONSEQUENCES OF SAFETY PERFORMANCE

Accidents, Near Misses, Illness/Disease, and Injury
With a few exceptions (e.g., Barling, Loughlin, & Kelloway, 2002), models of
workplace safety conceptualize safety-related behavior as a direct antecedent
of accidents, near misses, injury, and illness (e.g., Neal & Grifn, 2004). Tests
of structural equation models of workplace safety have lent strong support to
the conceptualization of safety behavior as direct antecedents to accidents
and injuries (see Christian et al., 2009; Paul & Maiti, 2007). Our modeling of
workplace injuries, however, considers accidents as partially mediating the
safety performanceinjury relationship. The rationale for the latter expected
partial mediation is that some injuries result from accidents, but many
injuries such as cumulative trauma, musculoskeletal, and knee injuries result
directly from work behavior (e.g., Alnaser, 2007; Brulin et al., 1998; Chen
et al., 2004; Marras, Davis, Kirking, & Bertsche, 1999). That is, different
combinations or repeated exposure to lifting, lowering, pushing, pulling,
and carrying can precede injury in the absence of an accident. This discussion
also suggests that some physical work exposures of either a repetitive
or prolonged nature may interact with worker behavior to produce
occupational injuries and illnesses with delayed onset of symptoms (Cole,
Ibrahim, & Shannon, 2005; Melchior et al., 2005). Furthermore, the indirect
relationship between safety performance and injuries through accidents is
31
important given that experiencing an accident-related injury may result in

qualitatively different thinking about how to avoid the reoccurrence of such
an injury than might be expected from involvement in an accident without
injury or involvement in a near miss.
At the individual level of analysis, safety performance, often measured as
a composite of safety participation and safety compliance items (e.g.,
Barling et al., 2002), has been found to have moderate negative relationships
with accidents, near misses, and injuries (Clarke, 2006b; Hayes, Perander,
Smecko, & Trask, 1998; Hofmann & Morgeson, 1999; Paul & Maiti, 2007;
Probst, 2004; Probst & Brubaker, 2001; Siu, Phillips, & Leung, 2003). While
too few studies have included measures of accidents, near misses, and
injuries to judge possible differential relationships of these outcomes with
safety performance, limited evidence (see Probst, 2004) indicates that safety
performance has a stronger relationship with near misses in comparison to
other outcomes. This nding is not unexpected, as the base rate for near
misses will tend to be greater than the rate of accidents and injuries.
Importantly, the literature on accident involvement and injury generally
supports our theoretical modeling of the role of distal (e.g., personality) and
more proximal antecedents (i.e., safety motivation and safety performance) to
these outcomes. For instance, Siu et al. (2003) found that negative affectivity
(a Big 5 personality factor relating to emotional stability) had a strong
indirect (through safety performance) effect on objectively measured work
injury for workers in two underground coal mines in India. As another
example, Probst and Brubaker (2001) found that safety motivation had
strong indirect effects through safety compliance on separate measures of
workplace accidents and injuries. Finally, Christian et al. (2009) presented
evidence supportive of expected indirect effects of psychological safety
climate on accidents and injuries through safety knowledge, safety motivation, and safety performance. In short, empirical research aimed at testing
causal models of workplace safety that incorporates distal and proximal
antecedents of safety performance and the outcomes of safety-related
behavior is providing process insights in regard to workplace safety, but
this research is at an early stage.
Although we stressed above that causal modeling of accident involvement
and injury would benet from improved conceptualizations and measures of
safety behavior in relation to conrmed safety performance constructs
(Burke et al., 2002), we also believe that the same point holds to some extent
for accidents and injuries. First, accidents and injuries are frequently
measured via self-reports and often in terms of dichotomous items (e.g., the
person has or has not been in an accident or has or has not been injured),
32
which creates ambiguity in terms of the meaning of such measures. Second,

when accidents are more objectively measured via the number of OSHArecordable incidents they involve incidents that required more then basic
rst aid or resulted in lost work days due to injury. Depending on the
purpose of ones investigation, self-reports or OSHA-recordable incidents
alone are likely decient in terms of capturing the nature of accidents and
differentiating more or less severe injuries. The nature of accidents (e.g.,
accidents that do or do not require more than simple rst aid) and injury
severity (viewed in terms of nancial, personal, and social costs) are related,
yet research modeling accident/injury involvement has generally not
attended to such distinctions. Notable exceptions that focus on the
distinction between OSHA- recordable (or Mine Safety and Health
Administration-recordable) accidents and microaccidents that only require
basic rst aid are reported in Wallace and Chen (2006) and Zohar (2000,
2002). Our understanding of the national, organizational, and individuallevel antecedents of accident involvement and injury will likely improve to
the extent that we also improve conceptualizations and measures within
these safety outcomes domains.
For the most part, the literature is silent in regard to estimating
relationships between constructs within most domains in Fig. 1 and
occupational illnesses and diseases. This void is understandable given that
many illnesses and diseases that have their origins in work such as cancer and
lung disease tend to have long latencies before their onset. However, the
literature clearly recognizes the need for appropriate worker action to avoid
exposures that may lead to occupational illnesses and diseases. In this sense,
there is often some basis such as an epidemiological study to support the link
between the health protective actions that workers can take and the possible
reductions in risk associated with such exposures. To the extent that workers
understand the association between such behaviors and health outcomes,
their health protective actions are known to relate to perceptions of control
over potential exposures (see Arcury, Quandt, & Russell, 2002).
Numerous behavioral interventions have been examined, with the aim of
improving workers motivation and knowledge for engaging in actions that
will preclude exposure to substances and conditions that are known to relate
to occupational illnesses and diseases. These interventions have been
effective with respect to improving knowledge and behavior related to
foodborne illness (Finch & Daniel, 2005), HIV infection (Wu et al., 2002),
noise-induced hearing loss (Lusk et al., 2003), and respiratory disease
(Acosta, Chapman, Bigelow, Kennedy, & Buchan, 2005) to name just a few.
In some cases, the interventions have led to not only improvements in
33
knowledge and safe work behavior, but also the reduction of symptoms
themselves as determined via clinical examinations (e.g., Held, Mygind,
Wolff, Gyntelberg, & Anger, 2002, in the case of occupational contact
dermatitis). Consistent with the above discussion, we believe that future
research directed toward examining a hypothesized interaction between the
level of engagement of these types of behavioral interventions and the
potential severity of exposure (which in the case of many occupational
illnesses and diseases is high) on knowledge acquisition and safety
performance would be informative. Along with engineering solutions,
conrming such an expectation would have important practice implications
related to conducting behavioral interventions to reduce risks associated
with occupational illnesses and diseases.
Dialogue and Reection

Unlike other models of workplace safety, we posit that dialogue (of either
intrapersonal or interpersonal nature) and reecting on ones work-related
safety experiences are key elements in learning from these experiences.
As discussed above, reection is a systematic thought process concerned
with simplifying experience as well as a process involving thinking about
contradictions, dilemmas, and possibilities. In particular, intrapersonal and/
or interpersonal dialogue subsequent to the experience of either positive or
negative consequences of safety-related behavior including a near miss is
likely to engender concrete reection in relation to actions taken and the
consequences of those actions. In regard to the severity of injury and illness
that was or could have been experienced as a result of safety-related
behavior, this dialogue would likely focus on the dilemmas faced and
possibilities associated with these critical situations and actions taken.
In considering the situational demands, this reection might also take into
account information on leader behavior and safety climate considerations.
Importantly, dialogue and reection in regard to critical situations,
actions taken, and their consequences would be expected to enhance ones
knowledge and efcacy for handling future events (Burke, 2008). Thus, in
Fig. 1, we propose feedback loops from dialogue and reection to safety
knowledge and safety motivation. Our expectation in regard to self-efcacy
for handling future events of a similar nature is that that self-efcacy
develops, in part, when individuals engage in extensive self-reection on
their experience and the adequacy of their thoughts and actions (Bandura,
1986). In regard to safety knowledge development, our expectation is
34
consistent with arguments within action regulation theory and experiential

theories of learning that individuals learn from experiencing errors and
thinking in terms of actions (Hacker, 2003; Heimbeck, Frese, Sonnentag, &
Keith, 2003; Holman, Pavlica, & Thorpe, 1997). To the extent that
individuals can independently or through other means be encouraged to
engage in dialogue and reection, especially subsequent to an accident or near
miss, we would expect their procedural knowledge and skill for handling such
future events to improve. The reader is also referred to discussions in Tesluk
and Quigley (2003) and Hofmann and Stetzer (1998) for how an open, freeowing discussion over safety issues may enhance learning from accidents,
errors, and near misses (and to Edmondson, Dillon, & Roloff, 2008, for a
related discussion in teams).
In terms of organizational efforts to encourage individual learning from
accidents, injuries, and near misses, several focused efforts could be
considered. As an example, a structured after-action-review akin to the
analysis of critical incidents (Flanagan, 1954) might prove benecial for
generating dialogue and reective thinking on the behalf of workers (Baird
et al., 1999). Relevant questions from an after-action-review might include:
(a) What was the intent or purpose of the action, (b) What exactly occurred
and why, (c) What lessons (in relation to organizational and personal
contributing factors) were learned, and (d) What future short-term and
longer-term actions and plans should be considered by management and
workers? Depending on the nature of the safety and health outcomes, Ellis,
Mendel, and Nirs (2006) work would suggest tailoring the after-actionreview. That is, after successful events, their research would call for an
emphasis on the discussion of incorrect actions, whereas, after unsuccessful
events, their research supports a discussion of both correct and problematic
actions. While Ellis and colleagues research is informative, research is needed
to delineate how dialogue and reection can be considered or encouraged
within after-action-reviews or other types of interventions to optimally
enhance safety knowledge and safety motivation. The reader is referred to
Burke et al. (2007) for a more detailed discussion of needed research on
dialogical aspects of safety knowledge and motivation development.
CONCLUSION
In this chapter, we integrated research on workplace safety from a variety of
disciplines and elds within business, engineering, psychology, public health,
and medicine to develop a multilevel model of the processes that affect
35
individual safety performance and safety and health outcomes. Our

discussion highlighted what is known about key construct relationships
within and among national, organizational, and individual levels of analysis
and identied future research directions for advancing our understanding of
workplace safety. Importantly, our modeling of the situational- and
individual-level determinants of workplace safety reected not only an
interactionist approach, but it also emphasized the social construction of
workplace safety especially in relation to discussing how and why workers
learn and think about safety in and by action. Our emphasis on the social
construction of workplace safety also highlighted key social, political,
economic, and cultural conditions that are expected within and among
nations to serve as important drivers of workplace safety.
Finally, to manage the scope of our review, we delimited our discussions
to key constructs and processes within and between the respective levels of
analysis that over time would be expected to inuence individual safety
behavior and outcomes. Although we delimited our scope in this manner,
our perspective provides a basis for extending multilevel modeling of
workplace safety to include processes and constructs that over time would
affect performance and safety-related outcomes at the organizational/group
and national/regional levels of analysis. In this sense, we hope that our
multilevel, interdisciplinary perspective on workplace safety not only
engenders research aimed at testing expected causal relationships
within the model, but it also encourages efforts to expand multilevel
conceptualizations and efforts to study workplace safety at the organizational/group and national/regional levels of analysis.
REFERENCES
Acosta, M. S. V., Chapman, P., Bigelow, P. L., Kennedy, C., & Buchan, R. M. (2005).
Measuring success in a pesticide risk reduction program among migrant farmworkers in
Colorado. American Journal of Industrial Medicine, 47(3), 237245.
Ahasan, M. R., Mohiuddin, G., Vayrynen, S., Ironkannas, H., & Quddus, R. (1999). Workrelated problems in metal handling tasks in Bangladesh: Obstacles to the development of
safety and health measures. Ergonomics, 42(2), 385396.
Alnaser, M. Z. (2007). Occupational musculoskeletal injuries in the health care environment
and its impact on occupational therapy practitioners: A systematic review. Work, 29,
89100.
Arcury, T. A., Quandt, S. A., & Russell, G. B. (2002). Pesticide safety among farmworkers:
Perceived risk and perceived control as factors reecting environmental justice.
Environmental Health Perspectives Supplements, 110, 233240.
36
Arnold, D. F., Bernardi, R. A., Neidermeyer, P. E., & Schmee, J. (2007). The effect of country
and culture on perceptions of appropriate ethical actions prescribed by codes of conduct:
A western European perspective among accountants. Journal of Business Ethics, 70(4),
327340.
Arthur, W., Barrett, G. V., & Alexander, R. A. (1991). Prediction of vehicular accident
involvement: A meta-analysis. Human Performance, 4(2), 89105.
Ash, J. S., Berg, M., & Coiera, E. (2004). Some unintended consequences of information
technology in health care: The nature of patient care information system-related errors.
Journal of the American Medical Informatics Association, 11(2), 104112.
Associated Press. (2009). Nevada construction deaths prompt review: Safety workplace
program faces scrutiny, USA Today, October 30.
Azizi, E., Flint, P., Sadetzki, S., Solomon, A., Lerman, Y., Harari, G., et al. (2000). A graded
work site intervention program to improve sun protection and skin cancer awareness in
outdoor workers in Israel. Cancer Causes & Control, 11(6), 513521.
Baird, L., Holland, P., & Deacon, S. (1999). Learning from action: Imbedding more learning
into the performance fast enough to make a difference. Organizational Dynamics, 27(4),
1932.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall Inc.
Barling, J., Loughlin, C., & Kelloway, E. K. (2002). Development and test of a model linking
safety-specic transformational leadership and occupational safety. Journal of Applied
Psychology, 87(3), 488496.
Baum, F., Begin, M., Houweling, T. A. J., & Taylor, S. (2009). Changes not for the
fainthearted: Reorienting health care systems toward health equity through action on the
social determinants of health. American Journal of Public Health, 99(11), 19671974.
Biddle, E. (1998). Development and application of an occupational injury and illness
classication system. In: J. Stellman (Ed.), Encyclopaedia of occupational safety and
health (4th ed., Vol. 1, The Body Health Care Management and Policy Tools and
Approaches, pp. 32.1232.18). Geneva, Switzerland: International Labour Ofce.
Brandon, L. J., Boyette, L. W., Lloyd, A., & Gaasch, D. A. (2004). Resistive training and longterm function in older adults. Journal of Aging and Physical Activity, 12(1), 1028.
Breslin, F. C. (2008). Educational status and work injury among young people: Rening the
targeting of prevention resources. Canadian Journal of Public Health, 99(2), 121124.
Breslin, F. C., Pole, J. D., Tompa, E., Amick, B. C., Smith, P., & Johnson, S. H. (2007).
Antecedents of work disability absence among young people: A prospective study.
Annals of Epidemiology, 17(10), 814820.
Breslin, F. C., & Pole, J. P. (2009). Work injury risk among young people with learning
disabilities and attention-decit/hyperactivity disorder in Canada. American Journal of
Public Health, 99(8), 14231430.
Briggs, N. C., Levine, R. S., Hall, H. I., Cosby, O., Brann, E. A., & Hennekens, C. H. (2003).
Occupational risk factors for selected cancers among African American and White men
in the United States. American Journal of Public Health, 93(10), 17481752.
Broadbent, D. E., Fitzgerald, P., & Broadbent, M. H. (1986). Implicit and explicit knowledge in
the control of complex systems. British Journal of Psychology, 77(1), 3350.
Brown, K. A., Willis, P. G., & Prussia, G. E. (2000). Predicting safe employee behavior in the
steel industry: Development and test of a sociotechnical model. Journal of Operations
Management, 18(4), 445465.
37
Brown, S. H. (1981). Validity generalization and situational moderation in the life insurance
industry. Journal of Applied Psychology, 66(6), 664670.
Brulin, C., Gerdle, B., Granlund, B., Hoog, J., Knutson, A., & Sundelin, G. (1998). Physical
and psychosocial work-related risk factors associated with musculoskeletal symptoms
among home care personnel. Scandinavian Journal of Caring Sciences, 12(2), 104110.
Burke, M. J. (2008). On the skilled aspect of employee engagement. Industrial and
Organizational Psychology, 1(1), 7071.
Burke, M. J., Borucki, C. C., & Hurley, A. E. (1992). Reconceptualizing psychological climate
in a retail service environment: A multiple-stakeholder perspective. Journal of Applied
Psychology, 77(5), 717729.
Burke, M. J., Borucki, C. C., & Kaufman, J. D. (2002). Contemporary perspectives on the study
of psychological climate: A commentary. European Journal of Work and Organizational
Psychology, 11(3), 325340.
Burke, M. J., Bradley, J., & Bowers, H. N. (2003). Health and safety training programs. In: J.
E. Edwards, J. C. Scott & N. S. Raju (Eds), The human resources program-evaluation
handbook (pp. 429446). Thousand Oaks, CA: Sage Publications.
Burke, M. J., Chan-Seran, S., Salvador, R., Smith, A., & Sarpy, S. A. (2008). The role of
national culture and organizational climate in safety training effectiveness. European
Journal of Work and Organizational Psychology, 17(1), 133152.
Burke, M. J., Holman, D., & Birdi, K. (2006). A walk on the safe side: The implications of
learning theory for developing effective safety and health training. In: G. P. Hodgkinson
& J. K. Ford (Eds), International review of industrial and organizational psychology 2006
(Vol. 21, pp. 144). Hoboken, NJ: Wiley.
Burke, M. J., Salvador, R., Smith, A., Chan-Seran, S., Smith-Crowe, K., & Sonesh, S. (2009).
A meta-analytic examination of how hazards and safety training inuence training
outcomes. Paper presented at the 8th International Conference on Occupation Stress
and Health: Work, stress, and health, San Juan, Puerto Rico.
Burke, M. J., & Sarpy, S. A. (2003). Improving worker safety and health through interventions.
In: D. A. Hoffman & L. E. Tetrick (Eds), Health and safety in organizations (pp. 5690).
San Francisco, CA: Jossey-Bass.
Burke, M. J., Sarpy, S. A., Smith-Crowe, K., Chan-Seran, S., Salvador, R. O., & Islam, G.
(2006). Relative effectiveness of worker safety and health training methods. American
Journal of Public Health, 96(2), 315324.
Burke, M. J., Sarpy, S. A., Tesluk, P. E., & Smith-Crowe, K. (2002). General
safety performance: A test of a grounded theoretical model. Personnel Psychology,
55(2), 429457.
Burke, M. J., Scheuer, M. L., & Meredith, R. J. (2007). A dialogical approach to
skill development: The case of safety skills. Human Resource Management Review,
17(2), 235250.
Campbell, J. P. (1990). Modeling the performance prediction problem in industrial and
organizational psychology. In: M. D. Dunnette & L. M. Hough (Eds), Handbook of
industrial and organizational psychology (2nd ed., Vol. 1, pp. 687732). Palo Alto, CA:
Consulting Psychologists Press.
Campbell, J. P., McHenry, J. J., & Wise, L. L. (1990). Modeling job performance in a
population of jobs. Personnel Psychology, 43(2), 313333.
Cantor, D. E. (2008). Workplace safety in the supply chain: A review of the literature and call
for research. International Journal of Logistics Management, 19, 6583.
38
Charlton, S. G. (2002). Selecting measures for human factors tests. In: S. G. Charlton &
T. G. OBrien (Eds), Handbook of human factors testing and evaluation (2nd ed., pp. 3753).
Mahwah, NJ: Lawrence Erlbaum Associates Publishers.
Chen, J.-C., Dennerlein, J. T., Shih, T.-S., Chen, C.-J., Cheng, Y., Chang, W., et al. (2004).
Knee pain and driving duration: A secondary analysis of the taxi drivers health study.
American Journal of Public Health, 94(4), 575581.
Christian, M. S., Bradley, J. C., Wallace, J. C., & Burke, M. J. (2009). Workplace safety: A
meta-analysis of the roles of person and situation factors. Journal of Applied Psychology,
94(5), 11031127.
Clarke, S. (2006a). The relationship between safety climate and safety performance: A metaanalytic review. Journal of Occupational Health Psychology, 11(4), 315327.
Clarke, S. (2006b). Safety climate in an automobile manufacturing plant: The effects of work
environment, job communication and safety attitudes on accidents and unsafe
behaviour. Personnel Review, 35(4), 413430.
Clarke, S., & Roberston, I. (2008). An examination of the role of personality in work accidents
using meta-analysis. Applied Psychology: An International Review, 57(1), 94108.
Cole, D. C., Ibrahim, S., & Shannon, H. S. (2005). Predictors of work-related repetitive strain
injuries in a population cohort. American Journal of Public Health, 95(7), 12331237.
Colligan, M. J., & Cohen, A. (2004). The role of training in promoting workplace safety
and health. In: J. Barling & M. R. Frone (Eds), The psychology of workplace safety
(pp. 223248). Washington, DC: American Psychological Association.
Colquitt, J. A., LePine, J. A., & Noe, R. A. (2000). Toward an integrative theory of training
motivation: A meta-analytic path analysis of 20 years of research. Journal of Applied
Psychology, 85(5), 678707.
Commission on Social Determinants of Health. (2008). Closing the gap in a generation: Health
equity through action on the social determinants of health. Final Report of the Commission
on Social Determinants of Health. Geneva, Switzerland: World Health Organization.
Cooper, M. (2009). Behavioral safety interventions: A review of process design factors.
Professional Safety, 54, 3645.
Corn, J. K. (1992). Response to occupational health hazards: A historical perspective. New York:
Van Nostrand Reinhold.
Cox, S., & Cox, T. (1991). The structure of employee attitudes to safety: A European example.
Work and Stress, 5(2), 93106.
Cree, T., & Kelloway, E. K. (1997). Responses to occupational hazards: Exit and participation.
Journal of Occupational Health Psychology, 2(4), 304311.
Cullen, E. T., & Fein, A. H. (2005). Tell me a story: Why stories are essential to effective safety
training (NIOSH Publication No. 2005-152). Cincinnati, OH: National Institute for
Occupational Safety and Health.
Curwick, C. C., Reeb-Whitaker, C. K., & Connon, C. L. (2003). Reaching managers at an
industry association conference: Evaluation of ergonomics training. AAOHN Journal,
51(11), 464469.
Daltroy, L. H., Iversen, M. D., Larson, M. G., Lew, R., Wright, E., Ryan, J., et al. (1997).
A controlled trial of an educational program to prevent low back injuries. New England
Journal of Medicine, 337(5), 322328.
Darby, F. (1998). Biomechanics. In: J. Stellman (Ed.), Encyclopaedia of occupational safety and
health (4th ed., Vol. 1, The Body Health Care Management and Policy Tools and
39
Davids, A., & Mahoney, J. T. (1957). Personality dynamics and accident proneness in an
industrial setting. Journal of Applied Psychology, 41(5), 303306.
Davis, P. O., Dotson, C. O., & Santa Maria, D. L. (1982). Relationship between simulated re
ghting tasks and physical performance measures. Medicine And Science In Sports And
Exercise, 14(1), 6571.
Dawson, D., & McCulloch, K. (2005). Managing fatigue: Its about sleep. Sleep Medicine
Reviews, 9(5), 365380.
Dong, X., & Platner, J. W. (2004). Occupational fatalities of Hispanic construction workers
from 1992 to 2000. American Journal of Industrial Medicine, 45(1), 4554.
Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams.
Administrative Science Quarterly, 44, 350383.
Edmondson, A. C., Dillon, J. R., & Roloff, K. S. (2008). Three perspectives on team learning:
Outcome improvement, task mastery, and group process. The Academy of Management
Annals, 1, 269314.
Eklof, M. (2002). Perception and control of occupational injury risks in shery A pilot study.
Work and Stress, 16(1), 5869.
Ellis, S., Mendel, R., & Nir, M. (2006). Learning from successful and failed experience:
The moderating role of kind of after-event review. Journal of Applied Psychology, 91(3),
669680.
Farquhar, S., Shadbeh, N., Samples, J., Ventura, S., & Goff, N. (2008). Occupational
conditions and well-being of indigenous farmworkers. American Journal of Public
Health, 98(11), 19561959.
Finch, C., & Daniel, E. (2005). Food safety knowledge and behavior of emergency food relief
organization workers: Effects of food safety training intervention. Journal of
Environmental Health, 67(9), 3034.
Flanagan, J. C. (1954). The critical incident technique. Psychological Bulletin, 51(4), 327358.
Fleming, M., Flin, R., Mearns, K., & Gordon, R. (1998). Risk perceptions of offshore workers
on UK oil and gas platforms. Risk Analysis, 18(1), 103110.
Flin, R., Mearns, K., OConnor, P., & Bryden, R. (2000). Measuring safety climate: Identifying
the common features. Safety Science, 34(1), 177192.
Floyd, D. L., Prentice-Dunn, S., & Rogers, R. W. (2000). A meta-analysis of research on
protection motivation theory. Journal of Applied Social Psychology, 30(2), 407429.
Ford, M. T., & Tetrick, L. E. (2008). Safety motivation and human resource management in North
America. International Journal of Human Resource Management, 19(8), 14721485.
Forst, L., Lacey, S., Chen, H. Y., Jimenez, R., Bauer, S., Skinner, S., et al. (2004). Effectiveness
of community health workers for promoting use of safety eyewear by Latino farm
workers. American Journal of Industrial Medicine, 46(6), 607613.
Forster, J., Higgins, E. T., & Bianco, A. T. (2003). Speed/accuracy decisions in task
performance: Built-in trade-off or separate strategic concerns? Organizational Behavior
and Human Decision Processes, 90(1), 148164.
Frederiksen, C. H. (1999). Learning to reason through discourse in a problem-based learning
group. Discourse Processes, 27(2), 135160.
Frone, M. R. (1998). Predictors of work injuries among employed adolescents. Journal of
Applied Psychology, 83(4), 565576.
Gagnon, M. (2003). The efcacy of training for three manual handling strategies based on the
observation of expert and novice workers. Clinical Biomechanics (Bristol, Avon), 18(7),
601611.
40
Gardner, P. H., Chmiel, N., & Wall, T. D. (1996). Implicit knowledge and fault diagnosis in the
control of advanced manufacturing technology. Behaviour & Information Technology,
15(4), 205212.
Geller, E. S. (1996). The psychology of safety: How to improve behaviors and attitudes on the job.
Radnor, PA: Chilton Book Company.
Geller, E. S. (2001). Behavior-based safety in industry: Realizing the large-scale potential of
psychology to promote human welfare. Applied & Preventive Psychology, 10(2), 87105.
Geller, E. S., Roberts, D. S., & Gilmore, M. R. (1996). Predicting propensity to actively care for
occupational safety. Journal of Safety Research, 27(1), 18.
Genaidy, A., Karwowski, W., & Shoaf, C. (2002). The fundamentals of work system
compatibility theory: an integrated approach to optimization of human performance at
work. Theoretical Issues in Ergonomics Science, 3(4), 346368.
Goldberg, A. I., Dar-El, E. M., & Rubin, A.-H. E. (1991). Threat perception and the
readiness to participate in safety programs. Journal of Organizational Behavior, 12(2),
109122.
Golub, S. S. (1991). The political economy of the Latin American debt crisis. Latin American
Research Review, 26(1), 175215.
Gorsky, P., & Caspi, A. (2005). Dialogue: A theoretical framework for distance education
instructional systems. British Journal of Educational Technology, 36(2), 137144.
Gorsky, P., Caspi, A., & Trumper, R. (2004). Dialogue in a distance education physics course.
Open Learning, 19(3), 265277.
Gorsky, P., Caspi, A., & Trumper, R. (2006). Campus-based university students use of
dialogue. Studies in Higher Education, 31(1), 7187.
Grifn, M. A., & Neal, A. (2000). Perceptions of safety at work: A framework for linking safety
climate to safety performance, knowledge, and motivation. Journal of Occupational
Health Psychology, 5(3), 347358.
Grifth, B. S. (1988). The crisis of American labor: Operation Dixie and the defeat of the CIO.
Philadelphia: Temple University Press.
Guralnik, J. M., & Ferrucci, L. (2003). Assessing the building blocks of function: Utilizing
measures of functional limitation. American Journal of Preventive Medicine, 25(3
Suppl 2), 112121.
Gyekye, S. A., & Salminen, S. (2005). Responsibility assignment at the workplace: A Finnish
and Ghanaian perspective. Scandinavian Journal of Psychology, 46(1), 4348.
Haaland, D. E. (2006). Safety rst: Hire conscientious employees to cut down on costly
workplace accidents. Nations Restaurant News, 40(16), 2224.
Hacker, W. (2003). Action regulation theory: A practical tool for the design of modern
work processes? European Journal of Work and Organizational Psychology, 12(2),
105130.
Halbesleben, J., Wakeeld, D., & Wakeeld, B. (2008). Work-arounds in health care settings:
Literature review and research agenda. Health Care Management Review, 33(1), 212.
Hansen, C. P. (1989). A causal model of the relationship among accidents, biodata, personality,
and cognitive factors. Journal of Applied Psychology, 74(1), 8190.
Hattrup, K., & Jackson, S. E. (1996). Learning about individual differences by taking situations
seriously. In: K. R. Murphy (Ed.), Individual differences and behavior in organizations
(pp. 507547). San Francisco: Jossey-Bass Publishers.
Havold, J. I. (2007). National cultures and safety orientation: A study of seafarers working for
Norwegian shipping companies. Work and Stress, 21(2), 173195.
41
Hayes, B. E., Perander, J., Smecko, T., & Trask, J. (1998). Measuring perceptions of workplace
safety: Development and validation of the work safety scale. Journal of Safety Research,
29(3), 145161.
Heimbeck, D., Frese, M., Sonnentag, S., & Keith, N. (2003). Integrating errors into the training
process: The function of error management instructions and the role of goal orientation.
Personnel Psychology, 56(2), 333361.
Held, E., Mygind, K., Wolff, C., Gyntelberg, F., & Anger, T. (2002). Prevention of work related
skin problems: An intervention study in wet work employees. Occupational and
Environmental Medicine, 59(8), 556561.
Hickman, J. S., & Geller, E. S. (2003). A safety self-management intervention for mining
operations. Journal of Safety Research, 34(3), 299308.
Hofmann, D. A., & Morgeson, F. P. (1999). Safety-related behavior as a social exchange: The
role of perceived organizational support and leader-member exchange. Journal of
Applied Psychology, 84(2), 286296.
Hofmann, D. A., Morgeson, F. P., & Gerras, S. J. (2003). Climate as a moderator of the
relationship between leader-member exchange and content specic citizenship: Safety
climate as an exemplar. Journal of Applied Psychology, 88(1), 170178.
Hofmann, D. A., & Stetzer, A. (1998). The role of safety climate and communication in accident
interpretation: Implications for learning from negative events. Academy of Management
Journal, 41(6), 644657.
Hofstede, G., & McCrae, R. R. (2004). Personality and culture revisited: Linking traits and
dimensions of culture. Cross-Cultural Research: The Journal of Comparative Social
Science, 38(1), 5288.
Hofstede, G. H. (1980). Cultures consequences: International differences in work-related values.
Beverly Hills, CA: Sage Publications.
Hofstede, G. H. (1991). Cultures and organizations: Software of the mind. London: McGrawHill.
Hofstede, G. H. (2001). Cultures consequences: Comparing values, behaviors, institutions, and
organizations across nations (2nd ed.). Thousand Oaks, CA: Sage Publications.
Holman, D. (2000a). Contemporary models of management education in the UK. Management
Learning, 31(2), 197218.
Holman, D. (2000b). A dialogical approach to skill and skilled activity. Human Relations,
53(7), 957980.
Holman, D., Pavlica, K., & Thorpe, R. (1997). Rethinking Kolbs theory of experiential
learning in management education: The contribution of social constructionism and
activity theory. Management Learning, 28(2), 135148.
Holzberg, C. S. (1981). The cultural context of the Jamaican national system: Ethnicity and
social stratication reconsidered. Anthropologica, 23(2), 157179.
Hsu, S. H., Lee, C.-C., Wu, M.-C., & Takano, K. (2008). A cross-cultural study of
organizational factors on safety: Japanese vs. Taiwanese oil renery plants. Accident
Analysis and Prevention, 40(1), 2434.
Hultman, G., Nordin, M., & Ortengren, R. (1984). The inuence of a preventive educational
programme on trunk exion in janitors. Applied Ergonomics, 15(2), 127133.
Husted, B. W. (1999). Wealth, culture, and corruption. Journal of International Business Studies,
30(2), 339359.
Ilgen, D. R., & Pulakos, E. D. (1999). The changing nature of performance: Implications for
stafng, motivation, and development (1st ed.). San Francisco: Jossey-Bass Publishers.
42
Infortunio, F. A. (2006). An exploration of the correlations between fatal accident rates across
nations and the cultural dimensions of power distance, uncertainty avoidance,
individuality, and masculinity. Dissertation Abstracts International (67), 1809.
Iverson, R. D., & Erwin, P. J. (1997). Predicting occupational injury: The role of affectivity.
Journal of Occupational and Organizational Psychology, 70(2), 113128.
James, L. R., Choi, C. C., Ko, C.-H. E., McNeil, P. K., Minton, M. K., Wright, M. A., et al.
(2008). Organizational and psychological climate: A review of theory and research.
European Journal of Work and Organizational Psychology, 17(1), 532.
Jammeh, S., & Delgado, C. (1991). The political economy of Senegal under structural adjustment.
SAIS study on Africa, p. 219, Praeger (Praeger in cooperation with the School of
Advanced International Studies).
Joubert, D. M. (2002). Occupational health challenges and success in developing countries: a
South African perspective. International Journal of Occupational and Environmental
Health, 8(2), 119124.
Kipen, H. M. (1994). The next asbestos: The identication and control of environmental
and occupational disease. Advanced and Modern Environmental Toxicology, 22,
425430.
Kivimaki, M., Head, J., Ferrie, J. E., Hemingway, H., Shipley, M. J., Vahtera, J., et al. (2005).
Working while ill as a risk factor for serious coronary events: The Whitehall II Study.
Kobayashi, M., Fussell, S., Xiao, Y., & Seagull, F. J. (2005). Work coordination, workow,
and workarounds in a medical context. In: CHI late breaking results (pp. 15611564).
New York: ACM Press.
Koh, D., & Chia, K. (1998). Surveillance in developing countries. In: J. Stellman (Ed.),
Encyclopaedia of occupational safety and health (4th ed., Vol. 1, The Body Health
Care Management and Policy Tools and Approaches, pp. 32.3932.12). Geneva,
Switzerland: International Labour Ofce.
Kriebel, D. (1982). Occupational injuries: Factors associated with frequency and severity.
International Archives of Occupational and Environmental Health, 50(3), 209218.
Kuorinka, I. (1998). Postures at work. In: J. Stellman (Ed.), Encyclopaedia of occupational safety
and health (4th ed., Vol. 1, The Body Health Care Management and Policy Tools and
LaDou, J. (2002). The rise and fall of occupational medicine in the United States. American
Journal of Preventive Medicine, 22(4), 285295.
Laschinger, S., Medves, J., Pulling, C., McGraw, R., Waytuck, B., Harrison, M. B., et al.
(2008). Effectiveness of simulation on health profession students knowledge, skills,
condence and satisfaction. International Journal of Evidence-Based Healthcare, 6(3),
278302.
Lawton, R., & Parker, D. (1998). Individual differences in accident liability: A review and
integrative approach. Human Factors, 40(4), 655671.
Lewin, K., Lippitt, R., & White, R. K. (1939). Patterns of aggressive behavior in experimentally
created social climates. The Journal of Social Psychology, 10, 271299.
Lingard, H., & Rowlinson, S. (1997). Behavior-based safety management in Hong Kongs
construction industry. Journal of Safety Research, 28(4), 243256.
Loomis, D., Schulman, M., Bailer, A., Stainback, K., Wheeler, M., Richardson, D., et al.
(2009). Political economy of U.S. states and rates of fatal occupational injury. American
Journal of Public Health, 99(8), 14001408.
43
Lueveswanij, S., Nittayananta, W., & Robison, V. A. (2000). Changing knowledge, attitudes,
and practices of Thai oral health personnel with regard to AIDS: an evaluation of an
educational intervention. Community Dental Health, 17(3), 165171.
Lusk, S. L., Ronis, D. L., Kazanis, A. S., Eakin, B. L., Hong, O., & Raymond, D. M. (2003).
Effectiveness of a tailored intervention to increase factory workers use of hearing
protection. Nursing Research, 52(5), 289295.
Luskin, J., Somers, C., Wooding, J., & Levenstein, C. (1992). Teaching health and safety:
Problems and possibilities for learner-centered training. American Journal of Industrial
Medicine, 22(5), 665676.
Mannuzza, S., Klein, R. G., Bessler, A., Malloy, P., & Hynes, M. E. (1997). Educational and
occupational outcome of hyperactive boys grown up. Journal of the American Academy
of Child and Adolescent Psychiatry, 36(9), 12221227.
Marchand, A., Simard, M., Carpentier-Roy, M. C., & Ouellet, F. (1998). From a
unidimensional to a bidimensional concept and measurement of workers safety
behavior. Scandinavian Journal of Work, Environment & Health, 24(4), 293299.
Marras, W. S., Davis, K. G., Kirking, B. C., & Bertsche, P. K. (1999). A comprehensive analysis
of low-back disorder risk and spinal loading during the transferring and. Ergonomics,
42(7), 904926.
Marsick, V. J., & Watkins, K. E. (1997). Lessons from informal and incidental learning. In:
J. Burgoyne & M. Reynolds (Eds), Management learning: Integrating perspectives in
theory and practice (pp. 295311). Thousand Oaks, CA: Sage Publications Inc.
Mayer, J. A., Slymen, D. J., Clapp, E. J., Pichon, L. C., Eckhardt, L., Eicheneld, L. F., et al.
(2007). Promoting sun safety among US Postal Service letter carriers: Impact of a 2-year
intervention. American Journal of Public Health, 97(3), 559565.
McConnell, D. (1997). Computer support for management learning. In: J. Burgoyne &
M. Reynolds (Eds), Management learning: Integrating perspectives in theory and practice
(pp. 283294). Thousand Oaks, CA: Sage Publications Inc.
McSween, T. E. (2002). The values-based safety process: Improving your safety climate with a
behavioral approach (2nd ed.). New York: Wiley.
Melchior, M., Krieger, N., Kawachi, I., Berkman, L. F., Niedhammer, I., & Goldberg, M.
(2005). Work factors and occupational class disparities in sickness absence:
Findings from the GAZEL cohort study. American Journal of Public Health, 95(7),
12061212.
Mischel, W. (1990). Personality dispositions revisited and revised: A view after three decades.
In: L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 111134).
New York: Guilford Press.
Morrow, P. C., & Crum, M. R. (1998). The effects of perceived and objective safety risk on
employee outcomes. Journal of Vocational Behavior, 53(2), 300313.
Motowidlo, S. J., Borman, W. C., & Schmit, M. J. (1997). A theory of individual differences in
task and contextual performance. Human Performance, 10(2), 7183.
Mullet, E., Ciutad, N., & Riviere-Shaghi, S. (2004). Cognitive processes involved in the
assessment of health hazards severity. Health, Risk & Society, 6(3), 277288.
National Institute for Occupational Safety and Health (NIOSH). (2002). Health effects of
occupational exposure to respirable crystalline silica (DHHS publication 2002-129).
Cincinnati, OH: NIOSH.
National Institute for Occupational Safety and Health (NIOSH). (2006). The team document:
Ten years of leadership advancing the National Occupational Research Agenda
44
(DHHS (NIOSH) publication 2006-121). Washington, DC: U.S. Department of Health

and Human Services.
Neal, A., & Grifn, M. A. (2004). Safety climate and safety at work. In: J. Barling &
M. R. Frone (Eds), The psychology of workplace safety (pp. 1534). Washington, DC:
American Psychological Association.
Neal, A., Grifn, M. A., & Hart, P. M. (2000). The impact of organizational climate on safety
climate and individual behavior. Safety Science, 34(1), 99109.
Nicholson, N., Soane, E., Fenton-OCreevy, M., & Willman, P. (2005). Personality and
domain-specic risk taking. Journal of Risk Research, 8(2), 157176.
Nuwayhid, I. A. (2004). Occupational health research in developing countries: A partner for
social justice. American Journal of Public Health, 94(11), 19161921.
Occupational Safety and Health Administration. (1998). Training requirements in OSHA
standards and training guidelines. (OSHA 2254). Washington, DC: U.S. Department of
Labor.
Olson, R., & Winchester, J. (2008). Behavioral self-monitoring of safety and productivity in the
workplace: A methodological primer and quantitative literature review. Journal of
Organizational Behavior Management, 28, 975.
Ott, M. G. (1998). Exposure assessment. In J. Stellman (Ed.), Encyclopaedia of occupational
safety and health (4th ed., Vol. 1, The Body Health Care Management and
Policy Tools and Approaches, pp. 28.2628.29). Geneva, Switzerland: International
Labour Ofce.
Paul, P. S., & Maiti, J. (2007). The role of behavioral factors on safety management in
underground mines. Safety Science, 45(4), 449471.
Peate, W. F., Bates, G., Lunda, K., Francis, S., & Bellamy, K. (2007). Core strength: A new
model for injury prediction and prevention. Journal of Occupational Medicine and
Toxicology, 2, 39.
Perez-Floriano, L. R., & Gonzalez, J. A. (2007). Risk, safety and culture in Brazil and
Argentina: The case of TransInc Corporation. International Journal of Manpower, 28,
403417.
Perry, M. J., & Layde, P. M. (2003). Farm pesticides: Outcomes of a randomized controlled
intervention to reduce risks. American Journal of Preventive Medicine, 24(4), 310315.
Peterson, D. (1998). Safety management. Des Plaines, IL: American Society of Safety Engineers.
Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource
dependence perspective. New York: Harper & Row.
Pidgeon, N. F. (1991). Safety culture and risk management in organizations. Journal of CrossCultural Psychology, 22(1), 129140.
Probst, T. M. (2004). Safety and insecurity: Exploring the moderating effect of organizational
safety climate. Journal of Occupational Health Psychology, 9(1), 310.
Probst, T. M., & Brubaker, T. L. (2001). The effects of job insecurity on employee safety
outcomes: Cross-sectional and longitudinal explorations. Journal of Occupational Health
Psychology, 6(2), 139159.
Rauscher, K., Runyan, C., Schulman, M., & Bowling, J. (2008). U.S. child labor violations in
the retail and service industries: Findings from a national survey of working adolescents.
Richardson, D. B., Loomis, D., Bena, J., & Bailer, A. J. (2004). Fatal occupational injury rates
in southern and non-southern states, by race and Hispanic ethnicity. American Journal of
Public Health, 94(10), 1756.
45
Robb, G., Sultana, S., Ameratunga, S., & Jackson, R. (2008). A systematic review of
epidemiological studies investigating risk factors for work-related road trafc crashes
and injuries. Injury Prevention, 14(1), 5158.
Robertson, G. (2007). Strikes and labor organization in hybrid regimes. The American Political
Science Review, 101(4), 781798.
Rosenberg, J. (2009). Organized labors contribution to the human services: Lessons from
the past and strategies for the future. Journal of Workplace Behavioral Health, 24(1),
113124.
Rosner, D., & Markowitz, G. E. (1991). Deadly dust: Silicosis and the politics of occupational
disease in twentieth-century America. Princeton, NJ: Princeton University Press.
Rundmo, T. (1996). Associations between risk perception and safety. Safety Science, 24(3),
197209.
Rundmo, T. (2001). Employee images of risk. Journal of Risk Research, 4(4), 393404.
Salminen, S., & Klen, T. (1994). Accident locus of control and risk taking among forestry and
construction workers. Perceptual and Motor Skills, 78(31), 852854.
Salminen, S., Klen, T., & Ojanen, K. (1999). Risk taking and accident frequency among Finnish
forestry workers. Safety Science, 33(3), 143153.
Salvendy, G. (1998). Work pacing. In J. Stellman (Ed.), Encyclopaedia of occupational safety
and health (4th ed., Vol. 2, Hazards, pp. 34.2534.26). Geneva, Switzerland:
International Labour Ofce.
Schad, G. D. (2001). Colonialists, industrialists, and politicians: the political economy of
industrialization in Syria, 19201954. Dissertation Abstracts International, 62(5), 1924.
Schaeffer, B. (2004). Learning disabilities and attention decits in the workplace. In: J.
C. Thomas & M. Hersen (Eds), Psychopathology in the workplace: Recognition and
adaptation. New York: Burnner-Routledge.
Schmidt, F. L., Hunter, J. E., & Outerbridge, A. N. (1986). Impact of job experience and ability
on job knowledge, work sample performance, and supervisory ratings of job
performance. Journal of Applied Psychology, 71(3), 432439.
Schulte, P. A., Stephenson, C. M., Okun, A. H., Palassis, J., & Biddle, E. (2005). Integrating
occupational safety and health information into vocational and technical education
and other workforce preparation programs. American Journal of Public Health, 95(3),
404411.
Schwartz, N., & Clore, G. L. (1988). How do I feel about it? Informative functions of affective
states. In: K. Fiedler & J. P. Forgas (Eds), Affect, cognition, and social behavior: New
evidence and integrative attempts (pp. 4462). Toronto: C.J. Hogrefe.
Schwartz, S. H. (1999). A theory of cultural values and some implications for work. Applied
Psychology: An International Review, 48(1), 2347.
Seibert, K. W. (1999). Reection-in-action: Tools for cultivating on-the-job learning conditions.
Organizational Dynamics, 27(3), 5465.
Sherrington, C., Lord, S. R., & Finch, C. F. (2004). Physical activity interventions to prevent
falls among older people: Update of the evidence. Journal of Science and Medicine in
Sport, 7(1), 4351.
Siu, O. L., Phillips, D. R., & Leung, T. W. (2003). Age differences in safety attitudes and safety
performance in Hong Kong construction workers. Journal of Safety Research, 34(2),
199205.
Slovic, P., & Peters, E. (2006). Risk Perception and Affect. Current Directions in Psychological
Science, 15(6), 322325.
46
Smith-Crowe, K., Burke, M. J., & Landis, R. S. (2003). Organizational climate as a moderator
of safety knowledge-safety performance relationships. Journal of Organizational
Behavior, 24(7), 861876.
Smith, L., Folkard, S., Tucker, P., & Macdonald, I. (1998). Work shift duration: A review
comparing eight hour and 12 hour shift systems. Occupational and Environmental
Medicine, 55(4), 217229.
Smith, M. J. (1998). Ergonomic factors. In J. Stellman (Ed.), Encyclopaedia of occupational
safety and health (4th ed., Vol. 2, Hazards, pp. 34.2234.23). Geneva, Switzerland:
International Labour Ofce.
Smolander, J., & Louhevaara, V. (1998). Muscular work. In J. Stellman (Ed.), Encyclopaedia of
occupational safety and health (4th ed., Vol. 1, The Body Health Care Management and
Policy Tools and Approaches, pp. 29.2829.31). Geneva, Switzerland: International
Labour Ofce.
Sndergaard, M. (1994). Research note: Hofstedes consequences: A study of reviews, citations
and replications. Organization Studies (Walter de Gruyter GmbH & Co. KG.), 15, 447456.
Strong, L. L., & Zimmerman, F. J. (2005). Occupational injury and absence from work among
African American, Hispanic, and non-Hispanic White workers in the national
longitudinal survey of youth. American Journal of Public Health, 95(7), 12261232.
Sulzer-Azaroff, B., & Austin, J. (2000). Does BBS Work? Behavior-based safety and injury
reduction: A survey of the evidence. Professional Safety, 45(7), 1924.
Tesluk, P. E., & Quigley, N. R. (2003). Group and normative inuences on health and safety.
In: D. A. Hoffman & L. E. Tetrick (Eds), Health and safety in organizations (pp. 31172).
San Francisco, CA: Jossey-Bass.
Tracey, J. B., Tannenbaum, S. I., & Kavanagh, M. J. (1995). Applying trained skills on
the job: The importance of the work environment. Journal of Applied Psychology, 80(2),
239252.
Tuncel, S., Genaidy, A., Shell, R., Salem, S., Karwowski, W., Darwish, M., et al. (2008).
Research to practice: Effectiveness of controlled workplace interventions to reduce
musculoskeletal disorders in the manufacturing environment Critical appraisal and
meta-analysis. Human Factors and Ergonomics in Manufacturing, 18(2), 93124.
Turner, S. L., Pope, M., Ellis, C. M., & Carlson, J. (2009). Counseling with North Americas
indigenous people. In: C. M. Ellis & J. Carlson (Eds), Cross cultural awareness and social
justice in counseling (pp. 185209). New York: Routledge/Taylor & Francis Group.
Valiante, D. J., Schill, D. P., Rosenman, K. D., & Socie, E. (2004). Highway repair: A new
silicosis threat. American Journal of Public Health, 94(5), 876880.
Vaslow, J. (1999). The multiple stakeholder model of psychological climate: Beyond employees
and customers. Unpublished doctoral dissertation. Tulane University, New Orleans, LA.
Wallace, C., & Chen, G. (2006). A multilevel integration of personality, climate, self-regulation,
and performance. Personnel Psychology, 59(3), 529557.
Wallace, J. C., Kass, S. J., & Stanny, C. J. (2002). The cognitive failures questionnaire revisited:
Dimensions and correlates. Journal of General Psychology, 129(3), 238256.
Wallace, J. C., Little, L. M., & Shull, A. (2008). The moderating effects of task complexity on
the relationship between regulatory foci and safety and production performance. Journal
of Occupational Health Psychology, 13(2), 95104.
Wallace, J. C., & Vodanovich, S. J. (2003). Workplace safety performance: Conscientiousness,
cognitive failure, and their interaction. Journal of Occupational Health Psychology, 8(4),
316327.
47
Weil, D. (2001). Assessing OSHA performance: New evidence from the construction industry.
Journal of Policy Analysis and Management, 20(4), 651674.
Wesseling, C., Aragon, A., Morgado, H., Elgstrand, K., Hogstedt, C., & Partanen, T. (2002).
Occupational health in Central America. International Journal of Occupational and
Environmental Health, 8(2), 125136.
Weyman, A. K., & Clarke, D. D. (2003). Investigating the inuence of organizational role on
perceptions of risk in deep coal mines. Journal of Applied Psychology, 88(3), 404412.
Williamson, A. M., Feyer, A.-M., Cairns, D., & Biancotti, D. (1997). The development of a
measure of safety climate: The role of safety perceptions and attitudes. Safety Science,
25(1), 1527.
Womack, W. W., Green, S. G., & Crouse, S. F. (2000). Cardiovascular risk markers in
reghters: A longitudinal study. Cardiovascular Reviews and Reports, 21(10), 544548.
Wu, Z., Detels, R., Ji, G., Xu, C., Rou, K., Ding, H., et al. (2002). Diffusion of HIV/AIDS
knowledge, positive attitudes, and behaviors through training of health professionals in
China. AIDS Education and Prevention, 14(5), 379.
Zohar, D. (2000). A group-level model of safety climate: Testing the effect of group climate on
microaccidents in manufacturing jobs. Journal of Applied Psychology, 85(4), 587596.
Zohar, D. (2002). The effects of leadership dimensions, safety climate, and assigned priorities on
minor injuries in work groups. Journal of Organizational Behavior, 23(1), 7592.
Zohar, D. (2003). The inuence of leadership and climate on occupational health and safety.
In: D. A. Hofmann & L. E. Tetrick (Eds), Health and safety in organizations: A multilevel
perspective (1st ed., pp. 201230). San Francisco, CA: Jossey-Bass.
Zohar, D., & Luria, G. (2004). Climate as a social-cognitive construction of supervisory safety
practices: Scripts as proxy of behavior patterns. Journal of Applied Psychology, 89(2),
322333.
Zohar, D., & Luria, G. (2005). A multilevel model of safety climate: Cross-level relationships
between organization and group-level climates. Journal of Applied Psychology, 90(4),
616628.
EXECUTIVE PAY AND FIRM

PERFORMANCE:
METHODOLOGICAL
CONSIDERATIONS AND
FUTURE DIRECTIONS
Beth Florin, Kevin F. Hallock and Douglas Webber
ABSTRACT
This paper is an investigation of the pay-for-performance link in executive
compensation. In particular, we document main issues in the pay
performance debate and explain practical issues in setting pay as well as
data issues including how pay is disclosed and how that has changed over
time. We also provide a summary of the state of CEO pay levels and pay
mix in 2009 using a sample of over 2,000 companies and describe main
data sources for researchers. We also investigate what we believe to be at
the root of fundamental confusion in the literature across disciplines
methodological issues. In exploring methodological issues, we focus on
empirical specications, causality, xed-effects, rst-differencing, and
instrumental variable issues. We then discuss two important but not yet
well-explored areas, international issues, and compensation in non-prots.
We conclude by examining a series of research areas where further work
can be done, within and across disciplines.

ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029004
49
50
BETH FLORIN ET AL.
Executive compensation is a topic that has interested researchers for nearly

a century. Bearle and Meanss (1932) pioneering discussion of the modern
corporation, Roberts (1956) rst empirical study, Murphys (1985) rst
discussion of considering panel data, Jensen and Murphys (1990) discussion
of the way executives are paid, and Hall and Liebmans (1998) study of the
link between wealth and rm performance and a steady stream of academic
papers since have focused on the link between pay of CEOs and corporate
performance. This subject is perhaps even more relevant today with the
dramatic changes in the economy and the outcry from various constituent
groups to examine the pay of senior executive more closely. The subject has
interested researchers from a variety of elds including Human Resource
Management, Economics, Accounting, Finance, Law, Sociology, Psychology,
and Industrial Relations. It is remarkable that, although hundreds of papers
have been written on the subject, there is no real consensus on the relationship
between executive pay and rm performance. This is due, in part, to the
diverse set of disciplines involved in the study, the wide variety of methods
used to investigate the main questions and the diversity in knowledge about
the institutions that matter in this area. This paper is an attempt to bridge
these gaps and consolidate some of the understanding of this important issue.
This paper reviews some of the literature, documents the current empirical
facts, explains the data available, discusses pay and performance, discusses
the varied empirical methods and possible reasons for differences in
results across studies, identies international issues, explores the current
regulatory environment, and considers avenues for future work in the area.
The paper is almost exclusively focused on CEO pay, although there is some
discussion of the top executive team. Although it should be clear from the
context, in most instances when we say executive we primarily are
referring to the CEO. The paper is not a comprehensive review of all of the
literature on executive compensation, nor is it a review of the pay-forperformance literature, and we apologize, in advance, to the authors whose
papers we have not discussed.1 However, this paper highlights certain
studies that are examples of the issues we explore.
In the next section, we describe the CEO pay-for-performance debate and,
in general terms, why it has not yet been resolved. Following that is another
section, where we describe a series of important data issues. Part of the rise
of the eld has been due to public disclosure of executive compensation data
in publicly traded rms in the United States. However, there have been
substantive changes in disclosure over time, and we explore the implications
of that for our work. We discuss types of executive compensation data and
changes over time and then go on to describe the specic sources most
Executive Pay and Firm Performance
51
frequently used by academics and practitioners today. We nally discuss pay

levels and pay mix across a set of 2,108 CEOs using data from 2009 proxy
statements.
We then have a section on practical matters of setting executive
compensation and reasons why that may affect the pay-for-performance
debate. This includes a discussion of proxy advisory services and the role
they play in setting executive compensation.
We then follow with a discussion of methodological issues, which we feel
are a central part of this work. In fact, we feel that issues of methodology
are specically important in this area of research since researchers across
disciplines (and within) use similar, yet distinctly different empirical models,
and these can have profound implications for their ndings. We feel that
these considerations can help clarify past ndings and perhaps suggest future
directions. The particular functional form of the empirical specications used
in the executive pay literature differ widely. We describe examples of each
and discuss implications of these models for interpreting pay-for-performance measures. In fact, we believe that a part of the problem in differing
interpretations of pay-for-performance results comes from problems in
appropriately interpreting empirical models. This paper includes discussion
of the right functional form, xed-effect and rst-difference models,
instrumental variables and the implications for executive compensation
research, and a section on thinking about causality. Next, we explore
different measures of rm performance including accounting measures,
nancial measures, subjective measures, and relative performance. We also
include a discussion of the little-studied area of international compensation.
Executive compensation in non-prot organizations is also briey discussed.
As difcult as it may be to consider the pay of managers in for-prot rms,
at least it is clear that there is a bottom line. This section describes
compensation of senior managers in a variety of non-prot organizations
including non-prot hospitals, labor unions, and non-prots in general.
Finally, we offer a concluding section where we summarize our work, discuss
various new government reforms (including say on pay and managing risk),
and offer a plan for future work in the area.
A BRIEF HISTORY OF THE CEO PAYPERFORMANCE

DEBATE, AND WHY IT ISNT RESOLVED
Summarizing the massive literature on the CEO payperformance debate is
not an easy task. However, in this section, we offer a brief summary of some
52
BETH FLORIN ET AL.
of the issues and ndings over the past few decades. This will set the stage
for our later discussion of functional forms, empirical problems, and other
issues.
The study of executive compensation goes back at least to Roberts (1956)
and even Bearle and Means (1932). There were also notable papers decades
ago such as Masson (1971), Lewellen and Huntsman (1970), and Coughlin
and Schmidt (1985) among others. Although the eld really took off with
the availability and use of better data (both in terms of quantity and quality)
and Murphys (1985) landmark study.
Murphy (1985) collected data on the compensation and performance of
461 executives at 71 rms over a number of years. But, rather than
estimating simple cross-sectional relationships (which showed no relationship between CEO pay and performance), Murphy (1985) introduced
xed-effects models (described below) and found a strong relationship
between pay and performance. This empirical method was not novel in
economics at the time, but it had not been applied to the CEO pay literature
and was an interesting and important advance for reasons we discuss below.
Murphy (1985), documenting a relationship between pay and performance
also wrote a paper in the Harvard Business Review at the time stating that
CEOs are worth every nickel they can get.
Five years later Jensen and Murphy (1990) wrote an important paper
using rst-difference methods (also described below). In that paper, they
found that for every $1,000 increase in shareholder value (measured as a
change in the market value of equity), CEO pay went up by $3.25. Their
interpretation of this was that, although there was a relationship between
pay and performance, the relationship was rather weak and could be
strengthened. In part, due to Jensen and Murphys work, and due to calls
from practitioners, this leads to the extraordinary rise in the use of stock
and stock options in executive compensation contracts. Options and stock
became much more important components of executive pay packages
starting in the early 1990s.2
Later in the 1990s, Hall and Liebman (1998) asked whether CEOs were
paid like bureaucrats? They collected unique data on stock and stock
options (that were at the time not more formally disclosed) and found
stronger relationships between pay and performance than found by Jensen
and Murphy (1990), on the order of $5.29 for every $1,000 increase in
shareholder wealth. They conclude that while this may still seem like
quite a weak relationship, their work suggests that even small changes in
performance can have very large effects on the lifetime wealth of an
executive.
53
More recently, Bebchuk and Fried (2004, 2006) wrote a provocative book
called Pay Without Performance. This book carefully articulates the
difference between the often-discussed arms-length bargaining framework and what they call the managerial power perspective, where, in
essence, boards are captured by CEOs. They discuss many reasons why they
think the system for setting CEO pay needs reform.3
Answering the pay-for-performance debate in executive compensation is
obviously a difcult question. There are many complications. For example,
researchers use different data sources, companies have different compensation and business strategies (even in the same industry), and there are many
potential factors that are not easily measured by academic researchers.
However, one of the main reasons we think the debate has not yet been
resolved is methodological issues that we explore in the rest of this paper.
But, rst we turn to some practical, institutional issues.
DATA ISSUES
In this section, we will outline several data issues that have confronted
researchers who study executive compensation. This will include a
discussion of the types of compensation data, Securities and Exchange
Commission (SEC) rules changes on reporting compensation, and specic
ways rms are required to report today. We will also discuss major sources
of data used by practitioners and academics. We will then go on to briey
describe pay levels and pay mix for CEOs across industries and rm sizes
from a set of 2,108 companies.
Types of Executive Compensation Data and Changes Over Time

One of the reasons for the meteoric rise in the number of papers published
on executive compensation is the availability of data. Data on executive pay
have been widely available since 1992, when the SEC through new disclosure
rules required rms to report on top ofcer compensation in a systematic
way. However, the way data were reported was not, until very recently
(a 2006 SEC change that we discuss below), entirely satisfying in terms of
really understanding the way executives were paid at a point in time. For
example, prior to 2006, we may have known an executives salary, bonus,
value of stocks sold, and options exercised in a given year. However, those
options and that stock may have been accumulated over many years so the
54
BETH FLORIN ET AL.
information reported in a given year was a combination of some compensation for the most recent year (e.g., the salary) and stock and options
accumulated for perhaps as long as a decade earlier. This may be one of the
reasons researchers have had trouble linking pay to performance since the
performance may have been for a given year or the year prior while the pay
was from a hybrid of data that may have covered a large number of years.
We discuss this problem with the payperformance relationship below.
Of course, there have been exceptional cases of authors who have gone to
great lengths to collect data on executive compensation prior to the most
recent SEC change. Examples include Murphy (1985) perhaps the rst real
classic in the literature Hall and Liebman (1998), and a more recent
effort by Frydman and Saks (2010). In the rst case, Murphy (1985) collected
data from 76 manufacturing rms (71 were used in the analysis sample) to
investigate the relationship between pay and performance. In the second,
Hall and Liebman (1998) collected detailed data on stock options (that were
not previously collected in one place) to demonstrate a relationship between
ownership in rms and rm performance. Frydman and Saks (2010) is an
ambitious example of careful data collection. In this paper, the authors
investigate executive compensation from a set of rms from 1936 to 2005.
They nd that the median real value of executive compensation was quite at
from the late 1940s through the 1970s, showing a weak overall link between
CEOs pay and rms growth. The collected the data by hand from company
proxy reports from 1936 through 1991 and then used ExecuComp (explained
in more detail below) for data beyond 1991.
In 2006, the SEC began requiring publicly traded rms to disclose
compensation for the CEO, Chief Financial Ofcer, and three other most
highly paid Named Executive Ofcers (NEOs). The new guidelines both
claried and standardized the elements of compensation as well as the time
period for reporting. Among the information rms are now required to
report are salary, bonus, nonequity incentive compensation, stock, stock
options, changes in pension and nonqualied deferred compensation, and
other compensation. It is worth being clearer about these seven main
components of compensation. Salary, of course, is the annual, xed, and
guaranteed compensation for the executive. Bonus and nonequity incentive
compensation are sometimes confused and, intuitively, both can be
considered a type of bonus. Strictly speaking, the bonus as listed in the
table is formula-based pay beyond cash salary. On the other hand, nonequity
incentive compensation can be both short- or long-term pay that is based on
some preset criteria (based on performance) whose outcome is uncertain.
Stock compensation is the value of the stock granted over the prior year, as
55
of the time it is granted. Stock options represent the value of the options
granted over the prior year. Stock options pose a unique problem in valuing
executive compensation contracts. The numbers included in rms proxy
statement summary compensation tables are accounting-based numbers
and do not necessarily reect the value of the options at the time of the grant.
Therefore, we recommend and most researchers use the value of stock
options from the stock option grant summary tables, which are also included
in rm proxy statements.4 Finally, other compensation refers to amounts
of perquisites of $10,000 or more or to tax gross-ups, company contributions
for security, private use of aircraft, nancial planning, etc.
Table 1 is an example of a Summary Compensation Table for the General
Electric for 2009. Several features of the table are noteworthy. The table lists
compensation for the CEO, CFO, and ve other executives. As noted above,
rms are required to list the CEO, CFO, and at least three others. One reason
for listing more than ve executives is the fact that some may have retired or
otherwise left the rm during the year. Another (which is not the case
for GE) is the example of co-CEOs. It is also clear from the table that
information is included for each of the last three years. This is the third year
since the new SEC regulations came into place so is, therefore, the rst time
outsiders can see three years of compensation information all in the same
proxy statement. Table 1 also shows the seven different pay components that
are required to be reported for each executive.5 It is also interesting to see
that at GE, the CEO was not the highest paid executive (at least as reported
in the most recent proxy statement). In fact, as reported in Table 1, three
others earned more than the CEO. Hallock and Torok (2010) report that of
2,108 rms they studied, in only 81% CEO was the highest paid executive.
There are many reasons why the CEO may not be the highest paid, including
one-time signing bonus, larger than normal option grants (commonplace
when hiring new executives) or severance, for example. In the case of GE in
Table 1, the CEO received no bonus, option awards, or nonequity incentive
payout. All three executives who were paid more than the CEO had non-zero
values for each of these elements of compensation. Finally, it is interesting to
see, in Table 1, the diversity of compensation across pay elements and to see
the diversity of pay within the top management team.
Main Data Sources Used by Academics and Practitioners

There are three major commercial data sources on executive pay at the
person- and rm-level that are now relatively widely used. The rst,
1,350,000
916,667
2,750,000
2,500,000
2008
2007
2006
2008
2007
2006
2008
2007
2006
2008
2007
2006
2008
2008
2008
2007
2006
Jeffrey R. Immelt,
Chairman of the Board
and CEO
Keith S. Sherin,
Vice Chairman and
CFO
Michael A. Neal,
Vice Chairman
Stock
Awards2
($)
2,783,000
7,590,000
6,900,000
1,310,000
214,664
574,322
1,597,537
1,714,833
2,225,749
1,475,945
1,457,839
1,759,672
1,597,537
1,714,833
2,225,749
1,239,568
1,943,665 1,303,005
2,516,712 2,473,683
10,148,300
0
0
4,169,500
0
0
0
2,555,300
0
0
2,933,900
0
0
5,615,400
0
0
4,000,200
Option
Nonequity
Awards4 Incentive Plan
($)
Compensation
($)
6,777,594 2,731,013
0 6,860,3183
5,800,000 9,802,3593
5,000,000 7,404,2093
2,550,000 2,987,493
3,000,000 3,076,095
2,550,000 2,808,919
2,900,000 3,512,898
3,880,000 4,212,201
3,300,000 3,906,929
2,700,000 3,659,090
3,000,000 4,406,900
2,550,000 4,122,437
1,850,000 2,284,110
Bonus
($)
1,208,099
1,072,075
2,422,714
5,911,944
3,563,466
78,290
1,036,908
2,503,541
1,281,453
1,564,398
3,484,939
2,979,130
3,032,927
3,328,715
1,852,735
2,183,677
1,432,870
Change in Pension
Value and
Nonqualied
Deferred
Compensation
Earnings5 ($)
2,080,058
1,314,005
1,010,780
190,426
372,819
396,267
548,013
288,718
275,400
308,222
344,044
343,674
294,872
261,073
393,825
335,866
250,857
All Other
Compensation6
($)
17,136,124
15,972,750
17,823,889
22,440,477
14,096,603
19,591,580
17,863,452
13,982,589
10,701,948
10,682,288
16,301,726
14,422,844
13,694,400
18,811,815
12,918,293
12,817,729
12,257,605
Total
($)
Note: Table from 2009 proxy statement of General Electric Company.

1
Messrs. Sherin and Rice deferred a portion of their salaries under the 2006 Executive Deferred Salary Plan. They were not named executives
at the time this plan was initiated. The amounts are also part of the Nonqualied Deferred Compensation table on page 30. In addition, each
of the named executives contributed a portion of his salary to the companys 401 (k) savings plan.
2
This column represents the dollar amounts recognized for the 2007 and 2006 scal years for the fair value of PSUs and RSUs granted in those
years, as well as prior scal years, in accordance with SFAS 123R. Pursuant to SEC rules, the amounts shown exclude the impact of estimated
forfeitures related to service-based vesting conditions. For RSUs, fair value is calculated using the closing price of GE stock on the date of grant.
Brackett B. Denniston,
Senior Vice President,
General Counsel and
Secretary
David R. Nissen,
Former President &
CEO, GE Money
Robert C. Wright,
Former Vice Chairman
John G. Rice,
Vice Chairman
3,300,000
3,300,000
3,300,000
1,500,000
1,354,167
1,225,000
1,650,000
1,550,000
1,400,000
1,650,000
1,550,000
1,400,000
1,200,000
Year
Salary1
($)
Proxy Statement for General Electric Company (2008 Summary Compensation Table).
Name and Principal

Position
Table 1.
56
BETH FLORIN ET AL.
As Mr. Wright is eligible for retirement, the fair value of his awards that have been held for more than a year have already been fully expensed.
For additional information, refer to note 23 of the GE nancial statements in the Form 10-K for the year ended December 31, 2007, as led with
the SEC. For information on the valuation assumptions with respect to grants made prior to 2007, refer to the note on Other Stock-Related
Information for the GE nancial statements in the Form 10-K for the respective year-end. Refer to note 3 below for a discussion of the
calculation of the fair value of PSUs. See the Grants of Plan-Based Awards table for information on grants awarded in 2007. These amounts
reect the companys accounting expense, and do not correspond to the actual value that will be realized by the named executives.
3
This amount represents the companys accounting expense for PSUs pursuant to SFAS 123R and SEC rules. It reects the expense for all
previously granted PSUs, not only those granted in 2006 or 2007. The actual value received depends on performance: 50% of the PSUs
converts into GE stock only if GEs cash ow from operating activities, adjusted to exclude the effect of unusual events, has grown an average
of 10% or more per year over the performance period, and 50% converts into GE stock only if GEs total shareowner return meets or exceeds
that of the S&P 500 over the performance period. Accordingly, Mr. Immelt may receive 0%, 50%, or 100% of each PSU grant. For example,
as described in the Compensation Discussion and Analysis on page 17, Mr. Immelt did not earn 50%, or a total of 215,000 shares, from the
PSUs granted to him in September 2003 and February 2006 because the total shareowner return condition was not met. Although the PSUs
not earned by Mr. Immelt were cancelled, the related accounting expense of $4.3 million has been disclosed as compensation to Mr. Immelt
over the performance period. In measuring fair value, SFAS 123R distinguishes between the PSU vesting condition related to the companys
stock price and the nonstock price-related performance condition. The restrictions on the PSUs lapse at the MDCC meeting in February
following the end of the performance period.
4
This column represents the dollar amounts recognized for the 2007 and 2006 scal years for the fair value of stock options granted in those
years, as well as prior scal years, in accordance with SFAS 123R. Pursuant to SEC rules, the amounts shown exclude the impact of estimated
forfeitures related to service-based vesting conditions. As Mr. Wright is eligible for retirement, the fair value of his awards that have been held
for more than a year have already been fully expensed. For information on the valuation assumptions, refer to note 23 of the GE nancial
statements in the Form 10-K for the year ended December 31, 2007, as led with the SEC. For information on the valuation assumptions with
respect to grants made prior to 2007, refer to the note on Other Stock-Related Information for the GE nancial statements in the Form 10-K
for the respective year-end. See the Grants of Plan-Based Awards table for information on options granted in 2007. These amounts reect the
companys accounting expense and do not correspond to the actual value that will be realized by the named executives.
5
This column represents the sum of the change in pension value and nonqualied deferred compensation earnings for each of the named
executives. The change in pension value in 2007 was $99,861, $1,221,780, $2,913,282, $1,759,575, and $488,342 for Messrs. Immelt, Sherin,
Neal, Rice, and Wright, respectively. The negative value for Mr. Immelt was primarily due to an increase in the discount rate used to calculate
the present value of his benet, partially offset by an additional year of pension accrual. In accordance with SEC rules, the amount included in
this column relating to the change in pension value for Mr. Immelt is $0. See the Pension Benets table on page 28 for additional information,
including the present value assumptions used in this calculation. In 2007, the above-market earnings on the executive deferred salary plans in
which the named executives participated were $78,290, $59,673, $65,848, $93,160, and $583,733 for Messrs. Immelt, Sherin, Neal, Rice, and
Wright, respectively. Above-market earnings represent the difference between market interest rates determined pursuant to SEC rules and the
8.5% to 14% interest contingently credited by the company on salary deferred by the named executives under various executive deferred
salary plans in effect between 1987 and 2007. See Nonqualied Deferred Compensation beginning on page 29 for additional information.
6
See the All Other Compensation table below for additional information.

57
58
BETH FLORIN ET AL.
ExecuComp (Executive Compensation database) is produced by Standard

and Poors Corporation and is surely the most widely used source of data
for research on executive pay by academics. This source has data available
from 1992 to present on the compensation of the top ve highest paid
employees of U.S. publicly traded rms (who have managerial control) in
roughly 1,500 rms per year. These rms include those listed in the Standard
and Poors 500, the Standard and Poors SmallCap 600, and the Standard
and Poors MidCap 400. The data source starts in 1992, which was (until
three years ago) the last time there was a major change in executive pay
disclosure rules.
Two other commercial executive pay sources are equilar.com and
salary.com. Each also provides comprehensive datasets of executive
compensation but have a larger focus on marketing to the for-prot rm
and compensation consulting market. These sources are frequently used by
compensation design practitioners and consultants to help design executive
pay plans (and to set comparison groups). Some academics are using data
from these sources but they are much more widely used by practitioners.6
The three data sources fundamentally report the same basic information.
equilar.com and salary.com started after ExecuComp and perhaps academics
have used ExecuComp in part due to inertia. It may also be due to cost
considerations. Equilar.com and salary.com provide many interface features
for, for example, making comparison groups easy and for presentation
purposes. Most academic dont need these features of their products.
Pay Levels and Pay Mix across Industries and Size Groups in 2009
This section is designed to set the basic context for the kinds of pay levels,
mix (types of pay across different components of compensation), and pay
distributions.7 The data for this section are from salary.com and comprise
2,108 publicly traded rms who reported executive compensation information in their proxy statements as of June 2009.
Fig. 1 displays two measures of compensation. The rst is dened as
cash and is the sum of salary, bonus, and nonequity incentive. The second
measure is total compensation. This is dened as the sum of salary, bonus,
nonequity incentive, stock, stock options, change in pension and nonqualied deferred earnings, and other. Fig. 1 displays the median cash
compensation and total compensation for CEOs by industry for each of 22
different industries. Notice the dramatic heterogeneity in compensation levels
for the median CEO across industries. For example, the median CEO in
Fig. 1.
59
CEO Compensation by Industry. Source: From Hallock and Torok (2010).

Data from salary.com
60
BETH FLORIN ET AL.
Commercial Banks earned about $581,000 in cash pay and $906,000 in total
compensation. At the other extreme is the Food and Tobacco industry where
the median CEO earned $2.28 million in cash compensation and $5.80
million in total compensation. These statistics alone mask another level of
heterogeneity. Consider, for example, the Food and Tobacco industry
(numbers not reported in tables or gures). There, the CEO at the 10th
percentile earned $575,000 in cash pay and $901,632 in total compensation
but the CEO at the 90th percentile of that industry earned $5.6 million in
cash and $14.9 million in total compensation (Table 2).
It may seem strange that Commercial Banks represent the industry with
the lowest paid median CEO. It must be kept in mind that these numbers
do not control for the size of the organization. In fact, there are a large
number of Commercial Banks in the sample and many of them are quite
small. Organization size (e.g., revenue and employees) is highly positively
correlated with the compensation of the senior leaders. Fig. 2 is a case in
point. In this gure, the 2,108 companies are sorted by their level of annual
revenue. The smallest 10 percent are in decile 1, the next 10 percent in
decile 2, and up to the largest 10 percent in decile 10. It is clear that the
median level of compensation rises monotonically with organization size.
In particular, for the smallest 10 percent of companies (those with annual
revenues below $155 million, the median CEO earned $522,000 in cash pay
and $1.04 million in total compensation. This rises monotonically up to the
largest 10 percent of rms (those with annual revenues above $9.6 billion)
where the median CEO earned $3.03 million in cash compensation and
$11.3 million in total compensation. Again, the median masks the larger
distribution. For example, for the largest 10 percent of companies, the CEO
at the 10th percentile earned $4.2 million in total compensation but the CEO
at the 90th percentile earned $25 million in total compensation.
Understanding the levels of pay for CEOs is interesting and important but
misses a more interesting and important part of executive compensation,
how executives are paid. In particular, we now explore how executives are
paid across the seven components of compensation discussed above. Fig. 3
shows a great deal of heterogeneity across compensation components by
industry. In fact, it is quite reasonable to expect diversity in compensation
mix within industry. Fig. 4 reports the pay mix distribution on rm size
deciles (the same deciles reported in Fig. 2). Notice that as the average rm
gets larger a smaller fraction of the total compensation is paid in salary and
a larger fraction is paid in stock and stock options. For example, for the
smallest 10 percent of companies, the fraction of total compensation paid
in salary is 43.91 percent but for the largest 10 percent of companies, the
500,000
463,040
332,083
678,501
654,711
370,000
833,750
395,159
465,500
500,000
575,000
416,678
450,000
612,435
602,077
479,938
500,000
437,396
654,647
429,504
548,035
605,000
88
183
185
85
78
137
26
153
97
58
50
109
162
89
40
132
91
127
20
59
86
53
10th
1,352,478
1,585,984
696,411
1,867,734
1,585,526
1,220,244
2,100,333
1,223,686
1,628,969
1,860,484
Mean
1,071,563
1,063,750
581,250
1,524,583
1,137,002
875,308
1,353,457
878,333
1,199,756
1,009,856
Median
1,637,340
1,966,500
875,000
2,434,725
1,732,500
1,375,350
2,386,318
1,427,375
2,025,597
2,740,276
75th
894,072
788,987
631,222
800,000
700,000
852,332
640,000
850,962
891,033
1,993,162
1,373,753
1,277,607
1,308,878
1,553,112
1,505,563
1,516,370
1,694,643
1,774,548
1,375,000
1,124,018
900,000
1,003,167
1,116,926
1,325,500
1,040,000
1,287,665
1,450,059
2,839,000
1,692,627
1,708,600
1,600,000
1,700,000
2,292,103
1,800,126
2,050,000
2,075,000
10th
25th
Mean
Median
Total Compensation
75th
90th
4,100,000
941,741
2,691,200
936,277
2,501,325
803,130
2,163,333
836,708
3,167,087
657,296
3,477,798
984,558
3,600,000
710,625
3,457,587
874,643
2,841,266 1,185,812
1,830,011
1,393,705
1,169,791
1,192,087
1,330,447
1,450,150
1,405,963
1,800,794
1,629,529
5,683,356
3,588,637
3,831,050
3,530,677
4,186,537
3,268,828
3,941,940
4,551,456
3,575,034
3,269,665
3,313,487
2,346,803
2,459,920
2,579,077
2,053,119
2,398,471
3,217,168
2,888,031
7,325,265 14,900,000
5,167,598 6,929,589
4,958,694 7,682,034
4,387,239 6,765,583
5,259,613 9,182,022
6,072,780 8,106,963
5,293,130 7,904,901
6,571,790 10,700,000
4,674,634 6,929,124
901,632 2,025,222 6,946,929 5,804,652 9,575,570 14,900,000

865,876 1,307,296 3,036,322 2,271,405 4,053,217 6,983,402
769,648 1,399,934 4,999,320 2,959,994 6,875,570 13,000,000
2,493,708
686,254 1,199,744 3,438,851 2,232,751 5,430,194 7,282,921
3,526,250
823,005 1,403,982 4,666,154 2,686,526 6,454,073 11,500,000
1,158,041
437,261
580,432 1,762,620
905,673 1,895,970 3,493,447
3,284,000 1,102,336 1,686,500 5,099,530 3,606,402 6,353,118 10,800,000
3,577,845
819,035 1,518,311 4,293,980 2,849,066 5,615,136 11,900,000
2,671,250
589,629 1,132,758 3,285,691 1,939,824 3,845,512 7,676,077
5,005,000 1,705,216 2,252,217 5,297,176 4,349,201 7,411,561 11,000,000
2,548,000
638,707 1,256,280 3,277,515 2,307,811 3,979,451 7,614,834
4,400,600
838,168 1,292,547 5,011,330 2,761,803 6,986,162 16,200,000
4,950,000
840,944 1,254,669 4,619,841 3,416,057 7,606,430 15,100,000
90th
1,147,342 2,922,108 2,277,237 4,046,612 5,614,329

652,000 1,222,678 1,000,000 1,590,750 2,330,144
717,750 1,841,272 1,204,883 2,250,000 4,000,002
683,542
661,770
402,500
961,299
780,000
555,024
1,000,000
550,000
646,154
800,000
25th
Cash Compensation
CEO Compensation by Industry.
Note: From Hallock and Torok (2010). Data from salary.com
Business services
Chemicals
Commercial banks
Commodities
Communications
Computer services
Construction
Electronics
Energy
Financial services
(nonbanks)
Food and tobacco
Holding companies
Industrial and
transportation
equipment
Insurance
Lumber and paper
Other manufacturing
Other services
Retail trade
Textile and apparel
Transportation
Utilities
Wholesale trade (S)
Table 2.

61
62
Fig. 2.
BETH FLORIN ET AL.
CEO Compensation by Company Size. Source: From Hallock and Torok

(2010). Data from salary.com
fraction of total compensation paid in salary is only 13.5 percent.

Conversely, the average CEO in the smallest 10 percent of companies
earned 35.56 (21.55 14.01) percent of his or her total compensation in
stock and options. But the average CEO in the largest 10 percent of
companies earned 53.05 (24.81 28.24) percent of his or her total
compensation in stock and stock options.
Now that we have set the stage for where the data come from and how
CEOs are paid in the United States today, we turn to some of the practical
issues in setting CEO pay.
PRACTICAL ISSUES IN CEO PAY

In this section, we briey describe some of the institutions that are important
in setting CEO and other senior executive compensation in the United
States. In particular, we briey mention the role of the Board of Directors,
the Compensation Committee of the board, and the role of executive
compensation consultants. We go on to highlight some of what are known as
Fig. 3.
CEO Compensation Mix by Industry. Source: From Hallock and Torok (2010). Data from salary.com

63
Fig. 4.
CEO Compensation Mix by Size Group. Source: From Hallock and Torok (2010). Data from salary.com
64
BETH FLORIN ET AL.
65
proxy advisory services, and the roles they play in executive compensation
in the United States today.
What Happens in Board Rooms, and How Pay is Set

Very little academic work formally discusses the role of the Board of
Directors or the Compensation Committee of the Board, even though
these are the organizations that formally set pay of executives in publicly
traded companies.8 The Board is formally responsible for executive
compensation but most boards have a Compensation Committee (a subset
of the board) who set pay of the CEO and his or her top team. Hallock,
Tonello, and Torok (2010) show that the median Boards across 10 deciles
of rm size have between 8 and 12 members. The median compensation of
Board members from the top decile of rms in 2009 was $191,000 (Hallock
et al., 2010).
Formally the Board is required to set compensation of the executive and
expected to be independent. Many Boards hire compensation consultants
to facilitate the development of compensation strategy and bring relevant
data to help frame pay decisions for CEOs and other top managers. These
compensation professionals consider the rms business strategy, compensation strategy, industry, organization size, composition of pay of like rms,
and make recommendations to the board.
Many in the press, government, and academia (most notably Bebchuk &
Fried, 2006) have been critical of the formal system that exists in the
United States for setting pay for executives. Bebchuk and Fried (2006) argue
that there is no arms-length negotiation between boards and the CEO
over executive contracts. On the other hand, they argue that boards (and
consultants) are captured by CEOs who try to surround themselves with
those who advocate for higher pay or more favorable compensation mix.
To ameliorate this potential conict, some have argued that compensation
consultants to boards should be independent. An example of this is large
multipurpose HR consulting companies that provide executive compensation consulting services, while at the same time providing others HR and
benets administration and consulting services. Since the revenue to the HR
consulting rm from providing executive compensation consulting services
may be only a small fraction of the total revenue received from that
particular rm, some argue that the executive compensation consultants
may want to give the CEO higher pay in implicit exchange for the large
volume of other services provided to their company.
66
BETH FLORIN ET AL.
Proxy Advisory Services, Investors, and Executive Pay

Many proxy advisory services have emerged in the past decade including
Risk Metrics (formerly Institutional Shareholder Services), Glass Lewis, and
GovernanceMetrics International. These organizations market and sell their
services to smaller investors, rms, and institutional investors. Among the
services they provide is a type of scorecard by rm for thousands of
publicly traded companies. They might, for example make recommendations on votes before the company, provide information about directors and
also provide some information about executive compensation.
The kinds of information these organizations provide about executive
compensation is varied but quite limited. For example, in 2008, the typical
Risk Metrics report was on the order of 15 pages, one of which was devoted to
executive compensation. This included a total of three charts showing (1) the
CEOs total compensation relative to a peer group median, (2) salary, bonus,
and nonequity compensation of the CEO relative to the median of a peer
group, and (3) stock and option awards of the CEO relative to a peer group.
A comparable 10-page report from Glass Lewis for the same time period
devoted one page to executive compensation. This included (1) a grade (AF)
and historical compensation score, (2) two charts comparing various
components of compensation with sector groups, and (3) a chart on shareholder
wealth and business performance. GovernanceMetrics International for the
same time period produced a three-page report for each rm that was entirely
devoted to executive compensation. This included some charts on pay levels
plotted against shareholder returns, comparisons to industry, charts on returns
and pay, relative to industry, a chart on pay mix, and a one-page narrative.
While we think these companies provide easily understood summaries to
investors and rms with respect to CEO pay, rm performance, and comparison groups, the analysis is extremely simple and largely reports numbers
straight out of individual rm proxy statements (except for some median
comparisons by industry). This serves as an example where the chasm
between academic work on executive compensation and the practical world is
enormous. Almost none of what has been learned in the past decades about
executive compensation, pay, or performance is included in these sources.
METHODOLOGICAL ISSUES
In this section, we explore two main issues. First, there have been a wide
variety of empirical specications in diverse research on pay and
67
performance for CEOs of publicly traded rms. We explore the diverse set
of specication and how one might make different interpretations of results
depending on specication used. Second, we discuss the issue of causality
in empirical research on the payperformance relationship for CEOs.
Our intent is to help consolidate understanding and hopefully help
researchers synthesize the diverse results.
Empirical Specications
One of the difculties in comparing the results of all papers in the executive
pay literature is that very few of them estimate the exact same model. While
there does not appear to be a consensus on the best specication, there are
many commonalities that appear over and over again.
We begin our discussion with how compensation is actually dened. The
most popular denition is to use total compensation (see our discussion
above) as the dependent variable, but many studies also focused on
option (Almazan, Hartzell, & Starks, 2005), bonus (Fattorusso, Skovoroda,
Buck, & Bruce, 2007), or basic cash compensation (Comprix & Muller,
2006). One desirable trend in many papers is the analysis of several models,
dening executive compensation differently each time, and demonstrating
their results are robust to multiple forms of pay. Clearly, it is natural to
expect certain forms of pay to be more strongly related to certain measures
of rm performance than others. For example, it is reasonable to expect that
stock and stock option compensation are more highly correlated with rm
performance than salary.
One important feature of a papers empirical specication is the
functional form chosen to represent key variables. Specically, the natural
log-transformation, which is used to deal with skewed data, is commonly
applied to the dependent and many independent variables. While the
interpretation of coefcients is slightly less straightforward following a logtransformation, it is especially important for valid statistical inference when
dealing with variables with a very skewed distribution such as compensation
and sales data (Hallock, 1997 provided evidence that the log-transformation
is important in executive compensation settings). Unfortunately, slightly less
than half of the empirical papers we surveyed used the log-transformation
on the dependent variable. This omission has the potential to seriously alter
the magnitude and interpretation of results.
One generally agreed-upon aspect is the need to control for a rms size
when estimating an executives pay. While not unanimous, nearly every study
68
BETH FLORIN ET AL.
we examined controls for size in some way. By far, the two most common
ways to control for rm size were to use a measure of sales (revenue) or assets
held by a rm. Another popular measure is the total number of employees.
These three variables are commonly highly correlated, except in banks and
other nancial institutions, where assets are obviously substantially higher
than sales and employees, relative to many other industries.
When examining the link between pay and performance, a crucial choice
authors must make is how exactly to dene performance. There are a
number of accepted ways this is done, although the most convincing studies
present the results for several different measures (Abowd, 1990). The most
common performance measure is a rms return on assets, followed by the
return on common stock. Other measures include the return on equity,
shareholder wealth, or rm prots. This point will be revisited in more detail
later in this section.
One important factor when looking at the effect of performance on pay,
which only about a third of studies control for, is the variability of
performance. The intuition being that some industries naturally have very
volatile performance indicators, likely weakening the payperformance
(observed) relationship because a high (or low) value may not necessarily
be a signal of the executives ability but rather a random shock. Studies
that control for volatility generally use either the standard deviation
(or variance) of performance (e.g., Garen, 1994) or the cumulative
distribution function (CDF) of performance (e.g., Garvey & Milbourn,
2003; Dee, Lulseged, & Nowlin, 2005). The CDF appears to be the better
option since it is a measure of the entire distribution of performance rather
than just the spread.
One class of variables conspicuously absent from the majority of models
is demographic information about the executive (Kostiuk, 1990; Hallock,
1997; Bertrand & Mullainathan, 2001; Bertrand & Hallock, 2001 are
notable exceptions). Very few studies include standard control variables
such as gender, race, or age. Age and tenure (and their squared terms) seem
particularly important to include in any sort of wage equation (as they are
standard practice dating back to the original Mincerian wage equations).
However, whether due to data constraints or omission, these variables
appear in less than a fth of the empirical studies we examined.
Another important feature of many papers (particularly those without
rm xed-effects) is the inclusion of some measure of corporate governance
(such as Core, Holthausen, & Larcker, 1999; Cornett, Marcus, & Tehranian,
2008). Evidence suggests that the strength of a rms corporate governance
is positively related to the association between pay and performance.
69
While not all studies use panel data, and some that do dont take full
advantage of their panels, the papers that fully utilize panel data tend to be
the most convincing works. To begin our brief panel discussion, it seems
that including industry xed-effects are a bare minimum (this is true in
a cross-sectional framework as well). Nearly every paper with rms in
different industries either includes industry indicators or runs separate
regressions for different industries. The most common panel data approach
taken by studies in this literature is a rst-difference approach (Boschen &
Smith, 1995; Anderson & Bizjak, 2003; Becker, 2006 just to name a few),
with relatively few papers using a rm/executive xed-effect strategy
(such as Aggarwal & Samwick, 1999a, 1999b, 2003; Cichello, 2005; Murphy,
1985). In many cases, this is likely due to a lack of data or insufcient
variation within executive pay/performance measures. That said, the papers
that are able to employ a xed-effects/differencing strategy are most likely
to come close to obtaining the best estimates of the true causal effect
of performance on executive pay. We discuss this in more detail below in a
section on causality.
In total, about half of the empirical papers we examined attempt to
fully exploit the panel nature of their data to varying degrees of success.
The rst-difference model was used by about a third of our sample of
empirical studies, with CEO/rm xed-effects used by slightly less than
a fth of papers. A discussion of which of these approaches is more
appropriate (they are equivalent only in the case of two time periods) and is
beyond the scope of our project, and rests on distributional assumptions and
serial correlation of the error term (Table 3).
To give the reader an idea of the relative frequency of the approaches to
estimating executive compensation, we have compiled Table 4, which breaks
down the studies by type of model (xed-effects, rst-difference, etc.) and
whether the log-transformation was applied to the dependent variable.
It should be noted that the 49 studies that appear in the table represent only
a sample of the executive pay literature (we believe a representative sample
of the top research), and there are undoubtedly a number of other papers
that could appear in the table but are not included.
Thinking More Seriously about Causality

The following section is meant as a brief introduction to the problem of
establishing causality in empirical work, with specic emphasis on the CEO
payperformance problem. One of the reasons that there is such diversity in
396,000
412,961
500,000
642,310
690,000
838,235
1,000,000
1,031,279
1,233,333
1,694,000
25th
616,804
694,353
827,883
1,041,286
1,163,253
1,452,193
1,616,560
2,037,876
2,264,928
3,334,802
Mean
521,703
560,800
700,625
897,188
948,147
1,223,614
1,421,101
1,703,739
1,875,000
3,025,857
Median
765,290
765,164
967,500
1,090,939
1,334,500
1,723,000
1,950,913
2,440,838
3,008,750
4,468,336
75th
975,849
1,090,000
1,418,027
1,690,000
1,741,250
2,399,341
2,977,725
3,750,000
4,325,333
7,016,500
90th
449,349
418,168
521,101
726,737
854,755
1,078,276
1,535,356
1,567,596
2,114,767
4,223,182
10th
Mean
Median
Total Compensation
75th
90th
676,779 1,382,037 1,034,623 1,631,245 2,500,199

621,750 1,465,745 1,043,271 1,501,970 2,906,560
807,011 1,787,558 1,309,888 2,138,803 3,562,378
1,067,464 2,218,387 1,694,451 2,905,494 4,158,242
1,484,095 2,729,881 2,118,466 3,234,420 4,677,025
1,681,375 3,246,883 2,648,624 4,031,500 6,461,061
2,249,450 4,118,238 3,632,436 5,238,242 7,186,328
2,664,599 5,434,129 4,662,380 6,702,930 9,479,940
3,315,934 6,673,713 6,108,194 8,477,317 12,900,000
6,478,689 11,600,000 11,300,000 16,100,000 25,000,000
25th
CEO Compensation by Revenue.
Note: From Hallock and Torok (2010). Data from salary.com
0o116
326,774
116o219
330,000
219o360
385,467
360o600
473,525
600o912
530,173
912o1371
638,230
1371o2209 727,282
2209o3974 800,000
3974o9637 900,000
9637
1,000,000
10th
Cash Compensation
Table 3.
70
BETH FLORIN ET AL.
Industry FE
Firm FE
Garen (1994)
Panel B: Studies that use a log-transformation on the dependent variable

Ang, Lauterbach, and Chhaochharia and
Bertrand and
Vu (2003)
Grinstein (2009)
Mullainathan (2001)
Cosh and Hughes
Core and Guay (1999)
Bebchuk and Grinstein
(1997)
(2005)
Fattorusso, Skovoroda,
Cornett et al. (2008)
Buck, and Bruce (2007)
Hall and Murphy (2002) Cunat and Guadalupe
(2009)
Hallock (1997)
Gabaix and Landier
(2008)
Hallock (2002)
Kostiuk (1990)
Carpenter and Sanders

(2002)
Chen, Steiner, and
Whyte (2006)
Dee et al. (2005)
Anderson and Bizjak (2003)
Almazan et al. (2005)
Aggarwal and Samwick (1999a)
Diff in Diff or Lagged Term
Murphy (1985)
Hall and Liebman (1998)

Kato and Kubo (2006)
Leonard (1990)
Hall and Knox (2004)
Gibbons and Murphy (1990)
Coughlin and Schmidt (1985)
Becker (2006)
Abowd (1990)
David et al. (1998)

Frye, Nelling, and Webb (2006)
Garvey and Milbourn (2003)
Girma, Thompson, and Wright
(2007)
Hambrick and Finkelstein (1995)
Jensen and Murphy (1990)
Lippert and Moore (1994)
Rajgopal, Shevlin, and Zamora
(2006)
Hartzell and Starks (2003) Boschen and Smith (1995)
Aggarwal and Samwick

(1999b)
Aggarwal and Samwick
(2003)
Cichello (2005)
CEO FE
Breakdown of Empirical Research on Executive Compensation (Sample).
Panel A: Studies that do not use a log-transformation on the dependent variable

Antle and Smith (1986) Aboody, Barth, and
Comprix and Muller
Kaszmik (2006)
(2006)
Baker and Hall (2004) Core et al. (1999)
Harford and Li (2007)
No FE
Table 4.
71
72
BETH FLORIN ET AL.
the answer to the payperformance problem is the diversity of

empirical methods and the dearth of papers that consider the issue
of causality at all. For example, trying to consider causal relationships in
a simple cross section using ordinary least squares (OLS) specications is
clearly impossible. Given the wide range of statistical approaches and results
in the executive pay literature, we feel that having a basic notion of causality
is crucial for the reader to evaluate the relative merits of each paper.
We have attempted to make this discussion as accessible as possible to
readers across elds.
The ultimate goal of most empirical research is to nd the true causal
effect of some independent variable X on an outcome Y. In the rare case that
X is exogenously determined (randomly assigned), a simple regression of
Y on X will give us the causal impact of X on Y. In the case of the pay-forperformance literature, this would be the equivalent of estimating Eq. (1):
where a is a constant, b is the change in pay associated with a one-unit
change in performance, and e is a random error term.
CEO Pay a bPerformance
(1)
The reason this equation is never estimated is because in this model,

we view performance as endogenous. By this, we mean that performance
measures were not randomly assigned to executives, and that the same
process that determines performance may also be related to the process that
determines executive compensation (in a statistical sense, this means that
performance is correlated with the error term).
A commonly discussed analogy is to think of the effect of schooling on
future earnings (Griliches, 1976; Card, 1995). It is an accepted fact that
people with higher levels of schooling have higher incomes, but it is also
generally true that, on average, individuals with higher innate ability have
higher levels of schooling. So we must then ask how much of the increase in
income is due to increased schooling and how much is due to a higher innate
ability or some other unobserved variable related to schooling.
The most common way to address this issue is to estimate Eq. (2), where
we have added a vector of control variables Z along with their associated
coefcients d.
CEO Pay a bPerformance Zd
(2)
These control variables may include measures such as the size of the rm,
the age of the executive, or any other factors that we believe has an impact
on an executives pay. In this case, for us to assign a causal interpretation to
73
the coefcient on performance, we are assuming that performance is

randomly assigned after conditioning on the covariates in Z.
If performance cannot be considered exogenous, then the results we
obtain will be considered correlational rather than causal. Hence, we could
say that a one-unit increase in performance is on average associated with
a b increase in compensation. We would, however, be unable to make the
causal claim that if we increased a performance measure by one unit, then
we would observe an increase in pay of b. This is an important distinction
because it is possible to have a strong correlation between performance and
pay, yet there is no causal link. Unfortunately, many authors in the
executive pay literature fail to make this distinction, discussing their results
in a causal context when they have not fully addressed potential endogeneity
concerns.
Establishing a causal link is not an easy task, and many will argue it is
impossible without a randomized experiment. There are several approaches,
however, which can greatly improve the accuracy of the basic regression
estimates mentioned above.
First-Difference and Fixed-Effects

The essence of the rst-difference and xed-effect methods is to exploit the
panel nature of certain datasets, namely the fact that we may observe
the same rm/executive different times. For instance, assume that there is
some xed but unobserved factor that is correlated with both executive
compensation and rm performance (this could be the executives ability,
rm culture, etc.) denoted by g. If we observe many executives in periods
t and t 1, then assume that the true wage equations are as follows:
CEO Payt1 a bPerformancet1 Z t1 d g t1
(3)
CEO Payt a bPerformancet Z t d g t
(4)
Subtracting Eq. (3) from Eq. (4) will then give us Eq. (5), which we can
estimate since all variables in Eq. (5) are observed, and an unbiased estimate
of b. This is known as the rst-difference method for obvious reasons.
The most famous example of this in the empirical executive compensation
literature is Jensen and Murphy (1990). The xed-effect method is illustrated
in Eq. (6), where instead of taking the difference between two periods we
include a dummy variable for each CEO/rm in the sample. Interestingly,
in the two-period case, the rst-difference and xed-effect methods are
74
BETH FLORIN ET AL.
algebraically identical. In panels with more than two years the two methods
differ, but the statistical assumptions underlying the difference between
these methods is beyond the scope of this chapter (although should be taken
seriously by the researcher). Murphys (1985) classic is the rst example of
this method in the empirical CEO pay literature.
CEO Payt CEO Payt1 bPerformancet Performancet1
Z t Z t1 d t t1
(5)
CEO Pay a bPerformance Zd Firm
(6)
It is always important to note in any model where the identication

is coming from (exactly which observations are contributing to which
estimates). In a standard OLS regression context, all observations contribute
equally to the estimated parameters. However, in a rst-difference/xedeffect context, the parameters are estimated based on changes within a rm.
In other words, if a rms performance does not change, then it will not
contribute to the estimate of b. Practically, this is important because if there
is not much variation in a dataset, the estimates could be driven entirely by
a small number of rms, or could even be the result of random noise.
While it is not always possible to control for rm xed-effects, including a
set of industry dummy variables (assuming the dataset being used contains
executives from more than one industry) is an absolute must. To not control
for industry would implicitly assume that, conditional on the covariates,
rms in different industries have the exact same pay structure.
Instrumental Variables
Sometimes researchers do not have panel data available to them, or do
not believe the assumptions implicit in a xed-effects framework (these
are discussed briey in the section Empirical Problems). In these cases,
an instrumental variables (IV) framework can theoretically identify causal
effects. We will not go into the mechanics of IV, but the intention is to
nd an exogenous variable that does not belong in the compensation
equation correlated with the potentially endogenous variable of interest
(performance), and identify b from this exogenous variation.
The best example in the executive compensation literature is from
Bertrand and Mullainathan (2001). In their study of oil companies, they
use exogenous shocks to the price of oil to instrument for performance.
75
In this example, these shocks clearly do not belong in the compensation

equation, but are highly correlated with performance indicators, making
them an ideal instrument.
Empirical Problems
The two methods for establishing causality outlined above are by no means
a panacea and have many problems that must be addressed in practice.
In fact, many researchers use these methods blindly and are lulled into a false
sense of security that all problems have been solved via use of these methods.
Clearly that is not the case. For a rst-difference/xed-effects approach,
having a large dataset is crucial, since multicollinearity is exacerbated in this
setting. In an IV framework, strong instruments are often hard to nd, and
must be rigorously justied in order to be accepted as valid.
When surveying the pay-for-performance, or any empirical literature for
that matter, it is important to keep this causal thought process in mind.
Many studies will come to wildly different conclusions, for a variety of
different reasons, and this line of thinking is key to discerning between the
studies you should believe and those you shouldnt.
Performance Metrics
As mentioned before, a wide variety of measures, and categories of
measures, are used to proxy for performance throughout the executive pay
literature. The main categories of performance metrics are accounting,
economic/market, relative performance, and subjective. The most commonly used measure, an accounting measure, is the after-tax return on
assets (used by Bebchuk & Grinstein, 2005; Carpenter & Sanders, 2002;
Chhaochharia & Grinstein, 2009; David, Kochhar, & Levitas, 1998 to name
a few). This variable is preferable both because of its availability and
straightforward interpretation/construction.9
The next most common measure of performance falls under the market
category, shareholder return on common stock. Dened below,10 many
researchers (such as Cosh & Hughes, 1997; Hall & Knox, 2004; Harford &
Li, 2007) prefer this metric because it most directly measures the progress
of the chief mission of the rm, to nancially benet its shareholders.
Another accounting measure, the after-tax return on equity, is also widely
used (Hambrick & Finkelstein, 1995; Leonard, 1990).
76
BETH FLORIN ET AL.
As one might expect, the link between executive pay and performance
appears to be inuenced somewhat by the performance metric used. For
instance, Abowd (1990) found that the link was much stronger for market
measures than for accounting measures.
Another class of performance metrics deals with the possibility that rms
reward (or punish) their executives based on how the rm performs relative
to other comparable rms in the same industry. The most common way to
test for the presence of relative performance evaluation is simply to control
for the difference between a rms performance measure and the market
average for that same measure. Two of the papers that focus explicitly on
this type of performance metric are Antle and Smith (1986) and Gibbons
and Murphy (1990).
As mentioned earlier, some studies explore the link between pay and
performance using less quantitative and more subjective measures of
performance. Denis, Hanouna, and Sarin (2006) nd a positive link between
allegations of securities fraud and executive stock options. In a similarly
themed paper, Efendi, Srivastava, and Swanson (2007) nd a positive
correlation between misstated nancial statements and executive stock
options which were in the money. Finally, McGuire, Dow, and Argheyd
(2003) nd that executive compensation is unrelated to various measures of
social performance.
INTERNATIONAL ISSUES
Generally, very little is known about international issues in compensation
and in executive compensation in particular. Part of the problem with the
study of executive compensation and the pay-for-performance issue across
nations is related to differences in disclosure and, therefore, availability of
data across countries. There are two main issues we discuss here. First is
why there may be differences in compensation across countries. Second, is
the state of the payperformance literature using various within-country
data sources.
CEO Pay Differences across Countries

Making CEO pay comparisons across countries is extremely difcult.
Fortunately there is one source that we know of that sheds some light on
this issue. In a recent paper, Gabaix and Landier (2008) have some data
77
collected from the consulting company Towers Perrin on average levels of

pay for CEOs across various countries as well as average levels of company
size in those countries.
We have reproduced one of their gures as Fig. 5. The horizontal axis in
the gure is the average log company size in terms of company annual
revenue. The vertical axis is the average log total compensation for CEOs.
The points represent the different countries. For example, it is clear from the
gure that CEOs in the United States are paid more, on average, than CEOs
in any other country. At the same time, rms in the United States are larger
(in terms of annual revenue) than rms in any of the other countries. The
gure also has an OLS regression line plotted through it. Intuitively this
suggests that if there is a world market for pay and if rm size is the only
relevant characteristic in predicting pay, then countries with points above
the line have CEOs who are, on average, over paid, relative to their
average rm size and countries with points below the line have CEOs who
are, on average, under paid, relative to their average rms size.
This gure is interesting since it tells us several things. First, perhaps one
of the reasons CEOs in the United States are paid so much, relative to their
counterparts in other countries, is due to the fact that companies in the
United States are so large.11 Second, even though rm size is one important
Fig. 5. CEO Compensation versus Firm Size across Countries. Source: From
Gabaix and Landier (2008). With permission from MIT Press Journals.
78
BETH FLORIN ET AL.
predictor of CEO pay (e.g., Rosen, 1992), rm size doesnt completely

explain why CEOs in the United States are paid so much more than
elsewhere. We should note that the data for this gure were provided by one
Human Resource consulting rm and their clients may not be representative
of the universe of rms.
International Pay and Performance within Countries

Several authors have tried to investigate the CEO payperformance
relationship, or CEO pay more generally, in countries other than the
United States. We highlight some of these studies here. In this section, we
will note studies from the UK, Germany, Japan, Sweden, and China.
The one country (other than the United States) where most work has been
done is the UK. In one paper, Conyon and Murphy (2000) nd that CEOs
are paid more in the US than the UK but much of that is due to stock-based
pay that has a stronger link to performance. Conyon and Sadler (2001)
nd that in the UK, the sensitivity of pay-for-performance increases with
organization levels. They go on to show a link between stock ownership and
subsequent performance. In a more recent paper, Fattorusso et al. (2007)
use UK data and nd little link between bonus and rm performance. They
argue that bonuses are, therefore, in essence guaranteed. Finally, Girma
et al. (2007) nd little link between CEO pay and performance, on average,
in the UK. However, they nd that for rms with many employees, there is
a small payperformance link.
In Germany, Edwards, Eggert, and Weichenrieder (2009) nd that rms
with low concentrations of investor ownership have only a small link
between CEO pay and rm prots. However, those with highly concentrated
ownership have no link at all. Using some data from the UK and Germany,
but in a mostly theoretical paper, Bruce, Buck, and Main (2005) discuss
region-specic social norms, which others have argued are important to
international compensation but very difcult to operationalize. In Japan,
Kato and Kubo (2006) nd a strong link between CEO pay and rm
performance using a ten-year panel , especially if using accounting measures
of performance. Becker (2006) nds that, in Sweden, incentives decrease
with CEO wealth. Finally, Firth, Fung, and Rui (2007) study China.
Clearly we need more research on the link between pay for CEOs and rm
performance internationally. Given the difculties in matching data sources
across rms, the differences in disclosure requirements and tax rules across
79
countries, and vastly different social norms in some countries, this may
prove to be difcult yet interesting work.
OTHER ORGANIZATIONAL FORMS

Studying executive compensation in the for-prot world is clearly difcult,
as this paper has stressed. However, in for-prot rms the bottom line is
quite clear. On the other hand, in non-prot organizations, it is not always
clear what the bottom line mission of the organization really is. If we
thought the true measure of performance for a particular non-prot
was alleviating poverty or helping those in need or caring or
trustworthiness, how would we measure this?
A series of recent papers have begun to investigate the compensation of
managers in non-prot organizations. Oster (1998) and Hallock (2002) both
show that the strong link between rm size and executive pay that exists in
for-prot rms is also present in non-prots. Additionally, Hallock (2002)
points out that the revenue from government grants has no effect on
executive compensation among non-prots once rms xed-effects are
controlled for. The same study also notes that a higher proportion of
compensation at non-prots is in the form of benets rather than salary.
As noted in both Oster (1998) and Hallock (2002), determinants of pay at
non-prots vary greatly by the type industry or charity. Examples include
the nursing home industry as studied by Weisbrod and Schlesinger (1986),
or universities that are examined by Ehrenberg, Cheslock, and Epifantseva
(2001). Looking specically at non-prot hospitals, Bertrand, Hallock, and
Arnould (2005) examine the link between executive pay and prot-based
performance measures following the introduction of health maintenance
organizations (HMOs) into the market. They nd that HMO penetration
strengthened the link between pay and prot-based performance measures,
and also increased the likelihood of turnover in less protable non-prot
hospitals.
Some types of non-prots have more easily measurable performance
measures. In their recent examination of the compensation structure for the
heads of labor unions, Hallock and Klein (2009) nd that the union
membership and the wage of union members are strongly related to the pay
of the union executive.
Finally, Frye et al. (2006) use a matching strategy to compare rms that
are socially responsible to the more traditional for-prot rm. While the
analysis does not explicitly concern non-prots, the basic idea that some
80
BETH FLORIN ET AL.
rms may put more emphasis on performance measures not easily observable
still holds. Consistent with the non-prot literature, this study nds that
socially responsible rms (as measured by the Domini Social Index) have
a much weaker link between pay and nancial performance, and that option
grants do not appear to induce the same risk-taking behavior in socially
responsible rms that is observed in nonsocially responsible rms.
CONCLUDING COMMENTS AND RESEARCH

QUESTIONS FOR THE FUTURE
This paper is an investigation of the pay-for-performance link in executive
compensation. In particular, we have explored data issues including how
pay is disclosed and how that has changed over time, a summary of the state
of CEO pay levels and pay mix in 2009 using a sample of over 2,000
companies, described main data sources, documented main issues in the
payperformance debate and explained practical issues in setting pay.
We also investigated what we believe to be at the root of fundamental
confusion in the literature across disciplines methodological issues. In
exploring methodological issues, we focused on empirical specications,
causality, xed-effects, rst-differencing, and instrumental variables issues.
We ended with a discussion of two important but not yet well-explored
areas, international issues, and compensation in non-prots.
We think there are several promising avenues for future work in executive
compensation. The dramatic advances in data availability, consistency, and
quality will likely lead to less diversity in results in future studies. We hope
that researchers across elds take advantage of this. We also believe that
a serious exchange of ideas on methods across (and within) disciplines
needs to happen. We nd it odd that researchers across disciplines have such
trouble talking to, understand, or working with those outside their niche
areas. Additionally, we think that the walls between academic research
and practice should be brought down generally, but in the area of executive
compensation in particular.
There are also a host of regulatory issues on the horizon, and we suspect
(hope!) that some will lead to the potential for interesting exogenous and
unexpected changes that could help us to better focus on issues of causality
we mention above. In addition, certain legislation pending in Congress
right now including legislation related to Say on Pay could be very
interesting. Say on Pay refers to the opportunity for a rms shareholders
81
to have an annual, nonbinding, vote on the compensation of the rms

senior executives. We suspect that study of this sort of proposal could lead
to clever new research on executive compensation across disciplines. We also
believe more work on issues of corporate governance would be investigated,
including work on the separation of the CEO and Chair of the Board.
The difcult issue of severance and so-called change-in-control agreements are also gaining increased attention by practitioners and we suspect
will garner more attention from academics soon.
There is too little work on international issues in compensation and in
executive compensation in particular. This is due, in part, to data availability
problems and in part due to the inherent difculties in doing work across
countries and cultures. We hope that in the near future researchers will
embrace the challenge of working on international issues.
Another area we feel is ripe for exploration and analysis by academic
researchers is that of risk. While there has been some work in this area,
the recent nancial crisis has focused this discussion much more sharply.
We expect analysis focusing on risk, risk-adjusted pay, and the like will
occupy many of colleagues in the future.
This paper is focused on many areas of executive compensation including
the pay-for-performance debate, the current issues, and the state of pay
today. But compensation internationally had a large focus on methodological issues and cross-disciplinary problems. Our hope is that this work will
not only help researchers focused on CEO pay and performance but also
motivate researchers in other areas of Human Resources to explore outside
their own discipline and work more closely with those from other elds.
NOTES
1. See for example Devers, Cannella, Reilly, and Yoder (2007) and Murphy (1999)
for recent reviews.
2. One other reason for the dramatic increase in the use of options could have
been due to the accounting treatment of the options. Until recently, most standard
employee options did not have to count as an expense in the company balance sheet.
3. See Kay and Van Putten (2007) for a rejoinder written by experienced
practitioners in the eld.
4. The subject of how to value stock options for executives and other employees is
an interested issue for debate. We do not focus on it in this paper. The interested
reader can nd discussion of this in Lambert, Larcker, and Verrecchia (1991),
Hall and Murphy (2002), and Hallock and Olson (2010).
5. Recall that for most studies, researchers dont use the stock options data from
this table but use the stock option grants table that appears later in proxy statements.
82
BETH FLORIN ET AL.
6. Hallock and Olson (2010) provide a much more comprehensive description of

data sources for research on executive compensation and employee stock options.
7. Substantially more detail than provided in this section can be found in Hallock
and Torok (2010). This section is based on that work.
8. There are some exceptions including Hallock (1997, 1999) who investigates the
relationships between reciprocally interlocking boards of directors (CEO of rm
A is a member of rm Bs board at the same time CEO of rm B is a member of rm
As board) and executive compensation.
9. Dened as the net income plus interest, divided by the average total assets over
the previous year (adjusted for the corporate tax rate).
10. Calendar-year return (dividends plus capital gains) per share of common stock.
11. Note that the axes are in logarithms so a step from 3 to 4 is, for example,
substantially smaller than a step from 7 to 8.
ACKNOWLEDGMENT
We are grateful to Sherrilyn Billger and Catherine McLean for helpful
suggestions. We thank the Compensation Research Initiative at Cornell
University for support.
REFERENCES
Aboody, D., Barth, M., & Kasznik, R. (2006). Do rms understate stock option-based
compensation expense disclosed under SFAS 123. Review of Accounting Studies, 11(4),
429461.
Abowd, J. (1990). Does performance-based managerial compensation affect corporate
performance? Industrial and Labor Relations Review, 43, 5273.
Aggarwal, R., & Samwick, A. (1999a). Executive compensation, strategic competition, and
relative performance evaluation: Theory and evidence. Journal of Finance, 54(6), 1999
2043.
Aggarwal, R., & Samwick, A. (1999b). The other side of the trade-off: The impact of risk on
executive compensation. Journal of Political Economy, 107(1), 65105.
Aggarwal, R., & Samwick, A. (2003). Performance incentives within rms: The effect of
managerial responsibility. Journal of Finance, 58(4), 16131649.
Almazan, A., Hartzell, J., & Starks, L. (2005). Active institutional shareholders and costs of
monitoring: Evidence from executive compensation. Financial Management, 34(4), 534.
Anderson, R., & Bizjak, J. (2003). An empirical examination of the role of the CEO and the
compensation committee in structuring executive pay. Journal of Banking and Finance,
27(7), 13231348.
Ang, J., Lauterbach, B., & Vu, J. (2003). Efcient labor and capital markets: Evidence from
CEO appointments. Financial Management, 32(2), 2752.
Antle, R., & Smith, A. (1986). An empirical investigation of the relative performance evaluation
of corporate executives. Journal of Accounting Research, 24(1), 139.
83
Baker, G., & Hall, B. (2004). CEO incentives and rm size. Journal of Labor Economics, 22(4),
767798.
Bebchuk, L., & Fried, J. (2004). Pay without performance: The unfullled promise of executive
compensation. Cambridge, MA: Harvard University Press.
Bebchuk, L., & Fried, J. (2006). Pay without performance: Overview of the issues. Academy of
Management Perspectives, 20(1), 524.
Bebchuk, L., & Grinstein, Y. (2005). The growth of executive pay. Oxford Review of Economic
Policy, 21(2), 283303.
Becker, B. (2006). Wealth and executive compensation. Journal of Finance, 61(1), 379397.
Berle, A., & Means, G. (1932). The modern corporation and private property. New York, NY:
Harcourt, Brace, and Word Inc.
Bertrand, M., & Hallock, K. (2001). The gender gap in top corporate jobs. Industrial and Labor
Relations Review, 55(1), 321.
Bertrand, M., Hallock, K., & Arnould, R. (2005). Does managed care change the mission of
nonprot hospitals? Evidence from the managerial labor market. Industrial and Labor
Relations Review, 58(3), 494514.
Bertrand, M., & Mullainathan, S. (2001). Are CEOs rewarded for luck? The ones without
principals are. The Quarterly Journal of Economics, 116(3), 901932.
Boschen, J., & Smith, K. (1995). You can pay me now and you can pay me later the dynamicresponse of executive-compensation to rm performance. Journal of Business, 68(4),
577608.
Bruce, A., Buck, T., & Main, B. (2005). Top executive remuneration: A view from Europe.
Journal of Management Studies, 42(7), 14931506.
Card, D. (1995). Using geographic variation in college proximity to estimate the return
to schooling. In: L. Christodes, E. Grant & R. Swidinsky (Eds), Aspects of labor
market behaviour: Essays in honour of John Vanderkamp. Toronto: University of
Toronto Press.
Carpenter, M., & Sanders, W. (2002). Top management team compensation: The missing link
between CEO pay and rm performance? Strategic Management Journal, 23(4), 367375.
Chen, C., Steiner, T., & Whyte, A. (2006). Does stock option-based executive compensation
induce risk-taking? An analysis of the banking industry. Journal of Banking and Finance,
30(3), 915945.
Chhaochharia, V., & Grinstein, Y. (2009). CEO compensation and board structure. Journal of
Finance, 64, 231261.
Cichello, M. (2005). The impact of rm size on pay-performance sensitivities. Journal of
Corporate Finance, 11(4), 609627.
Comprix, J., & Muller, K. (2006). Asymmetric treatment of reported pension expense and
income amounts in CEO cash compensation calculations. Journal of Accounting and
Economics, 42, 385416.
Conyon, M., & Murphy, K. (2000). The prince and the pauper? CEO pay in the United States
and United Kingdom. Economic Journal, 110(467, Special Issue F), F640F671.
Conyon, M., & Sadler, G. (2001). Executive pay, tournaments and corporate performance in
UK rms. International Journal of Management Reviews, 3(2), 141168.
Core, J., & Guay, W. (1999). The use of equity grants to manage optimal equity incentive levels.
Journal of Accounting and Economics, 28(2), 151184.
Core, J., Holthausen, R., & Larcker, D. (1999). Corporate governance, chief executive ofcer
compensation, and rm performance. Journal of Financial Economics, 51(3), 371406.
84
BETH FLORIN ET AL.
Cornett, M., Marcus, A., & Tehranian, H. (2008). Corporate governance and pay-forperformance: The impact of earnings management. Journal of Financial Economics,
87(2), 357373.
Cosh, A., & Hughes, A. (1997). Executive remuneration, executive dismissal and institutional
shareholdings. International Journal of Industrial Organization, 15(4), 469492.
Coughlin, A., & Schmidt, R. (1985). Executive compensation, management turnover and rm
performance. Journal of Accounting and Economics, 7, 4366.
Cunat, V., & Guadalupe, M. (2009). Globalization and the provision of incentives inside the
rm. Journal of Labor Economics, 27(2), 179212.
David, Kochhar, R., & Levitas, K. (1998). The effect of institutional investors on the level and
mix of CEO compensation. Academy of Management Journal, 41(2), 200208.
Dee, C., Lulseged, A., & Nowlin, T. (2005). Executive compensation and risk: The case of
internet rms. Journal of Corporate Finance, 12(1), 8096.
Denis, D., Hanouna, P., & Sarin, A. (2006). Is there a dark side to incentive compensation?
Journal of Corporate Finance, 12(3), 467488.
Devers, C., Cannella, A., Reilly, G., & Yoder, M. (2007). Executive compensation: A multidisciplinary review of recent developments. Journal of Management, 33(6), 10161072.
Edwards, J., Eggert, W., & Weichenrieder, A. (2009). Corporate governance and pay for
performance: Evidence from Germany. Economics of Governance, 10, 126.
Efendi, J., Srivastava, A., & Swanson, E. (2007). Why do corporate managers misstate nancial
statements? The role of option compensation and other factors. Journal of Financial
Economics, 85(3), 667708.
Ehrenberg, R., Cheslock, J., & Epifantseva, J. (2001). Paying our presidents: What do trustees
value? The Review of Higher Education, 25(1), 1537.
Fattorusso, J., Skovoroda, R., Buck, T., & Bruce, A. (2007). UK executive bonuses and
transparency a research note. British Journal of Industrial Relations, 45(3), 518536.
Firth, M., Fung, P., & Rui, O. (2007). How ownership and corporate governance inuence chief
executive pay in Chinas listed rms. Journal of Business Research, 60(7), 776785.
Frydman, R., & Saks, R. (2010). Executive compensation: A New View from a long term
perspective. Review of Financial Studies, Still under publication.
Frye, M., Nelling, E., & Webb, E. (2006). Executive compensation in socially responsible rms.
Corporate Governance: An International Review, 14(5), 446455.
Gabaix, X., & Landier, A. (2008). Why has CEO pay increased so much? Quarterly Journal of
Economics, 123, 49100.
Garen, J. (1994). Executive-compensation and principal-agent theory. Journal of Political
Economy, 102(6), 11751199.
Garvey, G., & Milbourn, T. (2003). Incentive compensation when executives can hedge the
market: Evidence of relative performance evaluation in the cross section. Journal of
Finance, 58(4), 15571581.
Gibbons, R., & Murphy, K. (1990). Relative performance evaluation for chief executive
ofcers. Industrial and Labor Relations Review, 43(3), s30s51.
Girma, S., Thompson, S., & Wright, P. (2007). Corporate governance reforms and
executive compensation determination: Evidence from the UK. Manchester School,
75(1), 6581.
Griliches, Z. (1976). Wages of very young men. Journal of Political Economy, 84, 569586.
Hall, B., & Knox, T. (2004). Underwater options and the dynamics of executive
pay-performance sensitivities. Journal of Accounting Research, 42(2), 365412.
85
Hall, B., & Liebman, J. (1998). Are CEOs really paid like bureaucrats? Quarterly Journal of
Economics, 123(3), 653692.
Hall, B., & Murphy, K. (2002). Stock options for undiversied executives. Journal of Accounting
and Economics, 33(1), 342.
Hallock, K. (1997). Reciprocally interlocking boards of directors and executive compensation.
Journal of Financial and Quantitative Analysis, 32(3), 331344.
Hallock, K. (1999). Dual agency: Corporate boards with reciprocally interlocking relationships.
In: J. Carpenter & D. Yermack (Eds), Executive compensation and shareholder value:
Theory and evidence (pp. 5575). New York, NY: Kluwer.
Hallock, K. (2002). Managerial pay and governance in American nonprots. Industrial
Relations, 41(3), 377406.
Hallock, K., & Klein, F. (2009). Executive compensation in American unions. Compensation
Research Initiative Working Paper no. 2009-007. Ithaca, NY: Cornell University.
Hallock, K., & Olson, C. (2010). New data for answering old questions regarding employee
stock options. In: K. Abraham, J. Spletzer & M. Harper (Eds), Labor in the new
economy. Cambridge, MA: University of Chicago Press for the National Bureau of
Economic Research.
Hallock, K., Tonello, M., & Torok, J. (2010). Directors compensation and board practices in
2009. New York: The Conference Board.
Hallock, K., & Torok, J. (2010). Top executive compensation in 2009. Research Report
R-1454-10-RR. The Conference Board, New York.
Hambrick, D., & Finkelstein, S. (1995). The effects of ownership structure on conditions at the
top-the case of CEO pay raises. Strategic Management Journal, 16(3), 175193.
Harford, J., & Li, K. (2007). Decoupling CEO wealth and rm performance: The case of
acquiring CEOs. Journal of Finance, 62(2), 917949.
Hartzell, J., & Starks, L. (2003). Institutional investors and executive compensation. Journal of
Finance, 58(6), 23512374.
Jensen, M., & Murphy, K. (1990). Performance pay and top management incentives. Journal of
Political Economy, 98(2), 225264.
Kato, T., & Kubo, K. (2006). CEO compensation and rm performance in Japan: Evidence
from new panel data on individual CEO pay. Journal of the Japanese and International
Economies, 20(1), 119.
Kay, I., & Van Putten, S. (2007). Myths and realities of executive pay. New York: Cambridge
University Press.
Kostiuk, P. (1990). Firm size and executive compensation. Journal of Human Resources, 25(1),
90105.
Lambert, R., Larcker, D., & Verrecchia, R. (1991). Portfolio considerations in valuing executive
compensation. Journal of Accounting Research, 29, 129149.
Leonard, J. (1990). Executive pay and rm performance. Industrial and Labor Relations Review,
43(3, Special Issue), S13S29.
Lewellen, W., & Huntsman, B. (1970). Managerial pay and corporate performance. American
Economic Review, 60, 710720.
Lippert, R., & Moore, W. (1994). Compensation contracts of chief executive ofcers
determinants of pay-performance sensitivity. Journal of Financial Research, 17(3),
321332.
Masson, R. (1971). Executive motivations, earnings, and consequent equity performance.
Journal of Political Economy, 79(6), 12781292.
86
BETH FLORIN ET AL.
McGuire, J., Dow, S., & Argheyd, K. (2003). CEO incentives and corporate social performance.
Journal of Business Ethics, 45(4), 341359.
Murphy, K. (1985). Corporate performance and managerial remuneration: An empirical
analysis. Journal of Accounting and Economics, 7(13), 1142.
Murphy, K. (1999). Executive compensation. In: O. Ashenfelter & D. Card (Eds), Handbook of
labor economics (Vol. 3B, pp. 24852563). Oxford, UK: Elsevier Science Publishers.
Oster, S. (1998). Executive compensation in the nonprot sector. Nonprot Management &
Leadership, 8(3), 207221.
Rajgopal, S., Shevlin, T., & Zamora, V. (2006). CEOs outside employment opportunities and
the lack of relative performance evaluation in compensation contracts. Journal of
Finance, 61(4), 18131844.
Roberts, D. (1956). A general theory of executive compensation based on statistically tested
propositions. Quarterly Journal of Economics, 70, 270294.
Rosen, S. (1992). Contracts and the market for executives. In: L. Werin & H. Wijkander (Eds),
Contract economics (pp. 181211). Oxford: Blackwell.
Weisbrod, B., & Schlesinger, M. (1986). Public, private, nonprot ownership and the response
to asymmetric information: The case of nursing homes. In: S. Rose-Ackerman (Ed.),
The economics of nonprot institutions: Studies in structure and policy (pp. 133151).
New York: Oxford University Press.
A TIME-BASED PERSPECTIVE
ON EMOTION REGULATION
IN EMOTIONAL-LABOR
PERFORMANCE
Michelle K. Duffy, Jason D. Shaw,
Jenny M. Hoobler and Bennett J. Tepper
ABSTRACT
We extend emotional-labor research by developing a time-based theory
of the effects of emotion regulation in emotional-labor performance.
Drawing on Grosss (1998a) process model, we argue that antecedentand response-focused regulatory styles can be used to make differential
predictions about outcomes such as performance, health, and antisocial
behavior and that these effects differ in shorter- and longer-time windows.
We discuss the theoretical implications and address the strengths and
limitations of our approach.
Displays of task-appropriate emotions are particularly crucial facets of

performance in occupations requiring employeecustomer interaction and
are the dening features of many jobs in the growing service economy
(Bernhardt, Morris, Handcock, & Scott, 2001). The task conditions of such
jobs, the process of regulating emotion, and the emotional displays are often
ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029005
87
88
MICHELLE K. DUFFY ET AL.
studied under the label of emotional labor (Beal, Trougakos, Weiss, &
Green, 2006). The increasing importance of emotional labor is reected not
only in its presence in todays workplace, but also in the surge of research
interest in understanding its consequences (e.g., Cote & Morgan, 2002;
Grandey, Fisk, & Steiner, 2005). On one hand, researchers have found that
emotional labor not only negatively affects individuals physical and mental
health (e.g., Pugliesi, 1999), but also imposes staggering total societal
costs (e.g., Landy, 1992; Morris & Feldman, 1996). Some research has also
shown occupation-level wage penalties for performing emotional labor,
especially in jobs lower in other forms of cognitive complexity (e.g., Glomb,
Kammeyer-Mueller, & Rotundo, 2004). On the other hand, effective
emotional displays have been often argued to relate positively to individual
and organizational performance (e.g., Ashforth & Humphrey, 1993).
Although the past decade witnessed much progress, the emotionallabor literature has been beset by conceptual inconsistencies and an
absence of inquiry into the processes underlying the appropriate display
of emotions in business settings. Indeed, the conceptual literature has split
between focusing on the task characteristics that elicit emotional labor
(e.g., Morris & Feldman, 1996) and the appropriate display of emotions
(e.g., Ashforth & Humphrey, 1993), while the empirical literature has
centered around what Beal et al. (2006) referred to as affective delivery or
the maintenance of required emotional displays at work. As those authors
pointed out, a detailed understanding of how employees successfully
regulate their emotional expressions at work seems necessary (p. 1053).
We take steps in this direction in this paper.
In her ambitious essay, Grandey (2000) linked the emotional-labor
literature to more basic psychological theory on emotion regulation or the
ways individuals inuence which emotions they have, when they have
them, and how they experience and express these emotions (Gross, 1999,
p. 557). Regulation strategies can take various forms, but can be broadly
categorized as antecedent focused (regulating the emotion before it is fully
formed) or response focused (regulating the emotion after it is fully formed).
She proposed that emotion-regulation research could serve to guide our
understanding of the regulatory processes underlying emotional-labor
performance. In this paper, we extend her work on emotion regulation in
emotional-labor performance by developing a theory of the differential
effects of emotion-regulation strategies on individual outcomes in work
settings. Our themes are individuals must regulate their emotions when
they are confronted with emotion-labor tasks and required-display rules,
regulatory styles play a role in determining important employee-related
Time and Emotion Regulation
89
outcomes, and these regulation processes have different consequences for

employee outcomes in different time windows.
Developing an emotion-regulation-based perspective on important
work-related outcomes is important for several reasons. First, research has
shown that emotion regulation occurs commonly and daily. Thus, the
question is not whether individuals regulate their emotions, but how they
go about doing so. We focus on how individuals regulate their emotions and
the implications these processes have for individuals and organizations.
Following Grandeys (2000) guiding work, we use basic psychological
theory on emotion regulation as a framework for understanding a wide
cross-section of individual outcomes (job performance, mental health,
physical health, and the consequences of being targeted for and perpetrating
aggression). Third, tremendous interest has grown recently in studying
the consequences of emotional labor or the displays of task-appropriate
emotions, especially in jobs in the growing service sector where these
displays are dening features (Beal et al., 2006; Brotheridge & Grandey,
2002; Glomb, et al., 2004; Grandey, 2003; Morris & Feldman, 1996).
Our approach is unique in three different ways. First, in addition to
examining the consequences of emotion regulation across several important
work-related outcomes, we incorporate the role of time and make
differential predictions for different combinations of emotion-regulation
strategy and time frame. Second, we depart from extant literature, the
preponderance of which suggests that antecedent-focused emotion regulation produces better outcomes than response-focused emotion regulation,
by arguing that the relative efcacy of antecedent- and response-focused
emotion regulation is time dependent. We distinguish between antecedentand response-focused approaches across outcomes. Third, the study of
emotions in the organizational literature is nascent but rapidly growing,
yet it has been hampered historically by what some see as the intellectual
debasement of emotions in the workplace (Muchinsky, 2000, p. 802). Our
approach also differs from recent attempts to incorporate emotions into the
organizational literature in that we do not focus on the outcomes of specic
emotions (e.g., Geddes & Callisters, 2007 ambitious model of anger and
antisocial behavior at work). Instead we focus on the processes individuals
use to regulate their emotions.
In the following sections, we (1) provide a foundation for our theory
building by briey reviewing conceptual and denitional issues with respect to
emotional labor; (2) review Grosss (1998a, 1998b) process model of emotion
regulation and Grandeys (2000) application of it to the study of emotional
labor; (3) derive propositions concerning emotion regulation and short- and
90
long-term individual consequences in work settings; and (4) discuss the

implications of the framework for future theory, research, and practice.
EMOTIONAL LABOR AND THE

EMOTION-REGULATION PROCESS
Perspectives on Emotional Labor
The study of emotional labor is rooted in Hochschilds (1979, 1983)
sociological perspective on the management of feelings in public situations.
Drawing on Goffmans (1959) inuential work on self-presentation, she
offered a dramaturgical denition of emotional labor where she saw
individuals as managing emotions through surface- or deep-level acting.
Surface-level actors manage their emotional expressions or displays; deep-level
actors consciously modify feelings in order to express the desired emotion
(Grandey, 2000, p. 96). She argued further that both surface-level actors and
deep-level actors nd the acting to be effortful and distasteful; as a result,
emotional labor should relate positively to emotional exhaustion and strain.
In contrast, Ashforth and Humphrey (1993) proposed the emotional-labor
model, which downplayed the role of emotion management and focused
almost singularly on observable behaviors or emotional displays. These
authors suggested that many emotional displays are automatic or effortless,
so emotional-labor performance may not manifest itself in stress-related
outcomes such as burnout and exhaustion. Rather, these authors argued,
the primary outcome of effective emotional labor is task performance, and
this relationship should be positive to the extent that customers perceive
emotional displays as genuine. In a rather marked contrast, Morris and
Feldman (1996) offered a third perspective by focusing not on emotion
management or the effectiveness of displays, but on task characteristics.
They argued that emotional labor can be dened by the frequency, intensity,
duration, and variety of required emotional displays, as well as the
dissonance these tasks create.
In her synthesis and extension of these founding works, Grandey (2000)
argued that the Ashforth and Humphrey (1993) and Morris and Feldman
(1996) frameworks described parts of the emotional-labor process effectively,
but not necessarily the construct of emotional labor itself. In particular,
task characteristics are factors that create emotional-labor situations: they
are precursors to emotional labor, while effective displays of emotion are the
91
proximal outcomes of emotional labor. She proposed a return to emotional

labors roots to a variation of Hochschilds (1983) framework and the key
issue of emotion regulation. Dening emotional labor as the process of
regulating both feelings and expressions for organizational goals, Grandey
(2000) used Grosss (1998a, 1998b) emotion-regulation theory as a guiding
framework for the study of emotional labor. She argued that understanding
the effects of emotional labor is more straightforward if viewed through the
lens of the prolonged physiological and cognitive arousal that denes
emotion as well as the processes and strategies used to regulate this arousal.
We adopt Grandeys (2000) denition of emotional labor in this paper.
The Emotion-Regulation Framework

Emotion regulation is typically characterized as a conscious process, but
it resides on a continuum of consciousness; some regulatory processes are
nearly automatic and require little conscious processing (e.g., hiding
disappointment in a social context), while others require signicant effort
and involve higher-level conscious processing. As a number of authors have
noted, the process of emotion regulation extends well beyond the boundaries
of emotional-labor performance; indeed it is a common, everyday process
(Morris & Reilly, 1987).
Gross (1999) argued that the process of emotion regulation commences
with exposure to and evaluation of environmental cues that then trigger or
stimulate a set of response tendencies designed to shape emotional responses.
When considered in a workplace context, these signals have much in
common with Morris and Feldmans (1996) conceptualization of emotional
labor as a set of task characteristics. These tendencies can be broadly
categorized in terms of whether an individual attempts to adjust emotions by
modifying perceptions of the situation (or perhaps even the situation itself )
or manipulates the emotional response after the emotion is fully formed and
experienced. At least ve different emotion-regulatory strategies have been
identied. Four of these strategies (situation selection, situation modication,
attention deployment, and cognitive change) are referred to as antecedentfocused approaches because they occur in advance of full emotion generation
(Gross, 1998a). Situation selection and modication involve niche picking
(Scarr & McCartney, 1983) or avoiding certain people or situations that
are likely to have a strong emotional impact. Although these types of
regulation may occur frequently, they are less relevant for our purposes
because, as Grandey (2000) pointed out, many workers, especially in jobs rife
92
with emotional-labor demands, have few alternatives for situation selection

or modication.
More appropriate for our purposes are the two antecedent-focused
approaches that Gross (1999) labeled attentional deployment and cognitive
change. With attentional deployment, in a given emotion-eliciting situation
an individual can choose to regulate emotions by redirecting attention from
the cue and toward unrelated memories or events or by concentrating on
something other than the current situational cue. Grandey (2000) offered the
example of a restaurant server who whistles arias to avoid being overwhelmed by negative or difcult customers. Cognitive change, in contrast,
involves lessening the impact of a situational cue either by changing how
one thinks about the situation or about ones capacity to manage the
demands it poses (Gross, 1999, p. 560). Although these approaches have
differences, the key issue with antecedent-focused emotion regulation is that
individuals take action to lessen the impact of environmental cues before
they fully affect the emotional experience and that the actions involve
deep-level modication of personal thoughts (attentional deployment) and
external appraisals (cognitive change). These antecedent-focused approaches
seem to share some construct space with Hochschilds (1983) concept of
deep-level acting.
The fth type of emotion regulation is often referred to as responsefocused emotion regulation; it involves inuencing or altering responses after
emotions arise. The basic process underlying response-focused emotion
regulation is suppression, such as hiding anger at a rude customer or calming
frustration by breathing deeply and slowly (Gross, 1999). Suppression,
therefore, requires active inhibition of the emotion-expressive behavior
that is generated as the emotion unfolds (Gross, 2001, p. 216). Grandey
(2000) noted that suppression can involve not only adjusting the intensity
(e.g., acting very happy when you are only moderately happy) but also
faking the emotional display (e.g., pretending to be happy when you are
actually angry). Suppression seems to share some construct space with
Hochschilds (1983) concept of surface-level acting.
Empirical research on these forms of emotion regulation has increased in
recent years. The implicit assumption in much of the empirical research
on emotion regulation is that individual- and work-related outcomes of
antecedent-focused regulation are more positive than outcomes of suppression. Some evidence has supported this notion (Gross & John, 2003).
Gross (2002), when discussing reappraisal as a form of antecedent-focused
regulation, argued that efforts to down-regulate emotion through
reappraisal should alter the trajectory of the entire emotional response,
93
leading to lesser experiential, behavioral, and physiological responses

(p. 283). In samples of US and French workers, Grandey et al. (2005) found
that frequent suppression was strongly and positively related to emotional
exhaustion. Among a sample of employed college students, Cote and
Morgan (2002) found that suppression of negative emotions was negatively
related to job satisfaction and positively related to intentions to quit. In a
novel study of cheerleaders, Beal et al. (2006) found that participants who
performed high levels of surface acting while experiencing negative emotions
reported higher levels of difculty at maintaining required display rules.
Two rare studies of deep acting (similar to the attentional deployment form
of reappraisal) and surface acting (similar to suppression) also provided
support for the positive effects of reappraisal and the negative effects
of suppression. Grandey (2003) found that surface acting was positively
related to self-reports of emotional exhaustion and negatively related to
coworker ratings of affective delivery (e.g., warmth and friendliness), while
deep acting was positively related to coworker ratings of affective delivery.
Moreover, deep acting was not signicantly related to self-reports of
emotional exhaustion. Among a convenience sample of workers in Canada,
Brotheridge and Grandey (2002) found that surface acting was positively
related to exhaustion and depersonalization and negatively related to
perceptions of personal accomplishment. Consistent with Grandeys (2003)
results, deep acting was not related to exhaustion, but was positively related
to perceptions of personal accomplishment.
When viewed in toto, initial ndings have suggested fairly uniform effects
of antecedent- (positive) and response-focused (negative) emotion regulation on individual and workplace outcomes and, indeed, this is somewhat
consistent with experiment-based empirical literature in the basic psychology elds (e.g., Gross, 2001; Gross & John, 2003). But are the consequences
of emotion regulation always uniformly positive (antecedent focused) or
negative (response focused) in terms of individual outcomes in the
workplace? Although the advances in the past few years have been
consequential, the literature has tended to view task conditions as uniform
in terms of the emotional labor required emotional labor is either required
or not and has tended to take short-term views of the dynamics of emotion
regulation. We argue that much of the speculation concerning emotion
regulation and outcomes implicitly confounds the effects of emotion regulation on short-term consequences with long-term consequences. Cote and
Morgans (2002) study represented an initial attempt to study regulation
processes over time, but the one-month time lag they examined may not
have been sufcient for differential effects to emerge. Gross (2001, 2002)
94
concluded his reviews of emotion-regulation research by suggesting that

researchers explore the potential differential consequences of different
regulatory styles and pursue a research agenda focused on the long-term
consequences of differing emotion regulation strategies (p. 218). We argue
below that emotion-regulation-based predictions may hold only in certain
time windows and only for certain outcomes. We focus on three aspects
of emotion regulation attentional deployment, cognitive change, and
suppression and in the following sections, elucidate our predictions.
WORK-RELATED OUTCOMES OF EMOTION

REGULATION IN EMOTIONAL-LABOR
PERFORMANCE
In this section, we derive propositions for the consequences of emotion
regulation in emotional-labor performance. Drawing on the denitions
described above, we categorize work-related outcomes in terms of time and
draw distinctions or moderators with regard to task type where theory
dictates. The range of outcome variables included in this conceptualization
should be considered a subset of all possible outcomes inuenced by
emotional-labor contexts and the process of emotion regulation. In this
sense, the proposed framework should be viewed as a point of departure
rather than as an all-inclusive theoretical framework regarding the impact
of emotion-regulation issues. We make two general assumptions in terms
of the boundary conditions of our theory building. First, we assume that
environmental cues trigger the regulation of emotion in emotional-labor
contexts (Grandey, 2000). This triggering process can be a function of the
work itself (e.g., the task characteristics that create emotional-labor
situations; Morris & Feldman, 1996), or of specic environmental events
(e.g., interactions with angry customers; Rafaeli & Sutton, 1987, 1990), or
both (Sutton, 1991). Second, we use the notion of time in a general sense,
making differential derivations and proposals for short and long time
frames, respectively. We do not intend to establish a specic window.
Rather, we suggest that short-term propositions concern emotion-regulation
strategies in single emotional-labor events (e.g., interacting with a customer
at a service desk). Long-term propositions concern the effects of emotionregulation strategies in repeated emotional-labor events over time (e.g.,
working in a customer-service role for a year). In the following sections we
derive short-term predictions regarding attentional deployment, cognitive
95
change, and suppression in terms of job performance and mental health, and
long-term predictions in terms of job performance, physical health, and
exposure to accidents, abusive situations, and aggression.
Short-Term Consequences of Emotion Regulation

Job Performance
In an initial attempt to link emotion-regulation strategies to work-related
outcomes, Grandey (2000) proposed that individuals using antecedentfocused regulation would perform better in customer-service roles than
those using suppression regulation. She argued that emotional suppression
may relate to observers perceptions that the individual is faking or is
disingenuous. Alternatively, those engaging in attentional deployment or
cognitive change have either reevaluated their emotional state and/or
recall other emotions to get through the moment (Gross, 1998a, 1998b).
Accordingly, observers are less likely to detect their true feelings. On the
surface, these generalizations appear accurate, but a closer examination of
the theoretical foundation of emotion regulation reveals that we can derive
more specic propositions.
First, in the emotional-labor context, the type of emotion regulation may
affect job performance depending on how deeply one interacts with others.
To illustrate, suppression, which requires that an individual inhibits a
generated emotion, may powerfully impact individuals (Gross, 1998a).
In contrast, antecedent-focused regulation is a front-end approach in which
an individual interprets an emotional cue in unemotional terms before it
is fully generated. The process of emotion suppression results in slightly
higher levels of sympathetic activation (Gross & Levenson, 1997) than the
process of emotion regulation through attentional deployment and cognitive
change; the activation difference may mean that the suppressor is more
engaged (e.g., energetic or on edge) in surface-level or shallow interactions
than is the antecedent-focused regulator. Antecedent-focused regulation, by
denition, means that the individual is more detached from the environmental or emotional cue (recall the example of the aria-whistling server),
and observers may perceive them as being less engaged, and hence
performing more poorly. In many occupations rife with emotional labor,
the level of interpersonal interaction with others is often quite shallow
(e.g., a receptionist directing trafc in a busy ofce, or a food-service
employee managing a drive-through window). Because of heightened
physiological activation, these situations provide opportunities for higher
96
job performance among those engaging in suppression regulation compared

with those using with antecedent-focused regulation.
In more interaction-rich interpersonal contexts, substantial benets may
accrue for job performance for those engaging in antecedent-focused
regulation. Interactions that last longer provide greater opportunities for
emotional leakage (e.g., Ekman & Friesen, 1969) to betray the performer
and also more opportunities for observers to detect that the employee is
faking emotions. By manipulating the environmental and emotional input,
those engaging in attentional-deployment or cognitive-change forms of
regulation may be slightly less physiologically activated, but the payoffs
include that they appear to be interacting more genuinely. The antecedentfocused process partly concerns perpetuating an illusion that the actual
situation is really something different (e.g., misperceiving that a situation is
not so ominous when it is actually threatening) such that the appraiser is
convinced that the expressed emotion is real. Taylor and Brown (1999)
argued that illusions are a critical component of the ability to care for others
and are especially useful in threatening situations. In emotional-labor
situations where single interactions are deeper and last longer, this process
may be markedly advantageous in terms of job performance. To illustrate,
some jobs requiring emotional labor are not particularly complex but require
prolonged interactions with others. For example, an auto salesperson and a
customer may haggle over the price of the car and trade-in for several hours.
The time-exposure pressures may cause the salesperson to leak suppressed
negative emotions, and ultimately reveal to the customer that the salesperson, as a suppression regulator, has been disingenuous. By way of
contrast, the reappraisal regulator may be slightly less activated initially, but
should be capable of successfully interacting with the client for a longer time.
These arguments are summarized in the following propositions:
Proposition 1. Response-focused emotion regulation (suppression) will
result in better short-term job performance than antecedent-focused
emotion regulation (attentional deployment and cognitive change) in
situations where interpersonal interactions are shallow and/or of short
duration.
Proposition 2. Antecedent-focused emotion regulation (attentional
deployment and cognitive change) will result in better short-term job
performance than response-focused emotion regulation (suppression) in
situations where interpersonal interactions are deep and/or of longer
duration.
97
Another important issue with regard to emotion regulation and shortterm task performance concerns the level of cognitive activity needed in
tasks requiring emotional labor. A great deal of research has concerned
the loss of cognitive functioning during acute suppression episodes. For
example, Baumeister, Bratslavsky, Muraven, and Tice (1998) found that
emotion regulation through suppression degrades performance on cognitive
assignments. The logic behind the nding is that individuals have a limited
supply of resources for performing cognitive tasks; the challenge of
suppressing a negative emotion exhausts some of the resources, leaving
few available for performing the task successfully. In addition, some
researchers have shown that suppression degrades memory, a key facet of
performance on cognitively complex tasks (Richards & Gross, 1999). These
authors hypothesized and found that the conscious suppression of emotions
increases self-focus and diminishes the ability to encode new information
(e.g., Ellis & Ashbrook, 1988; Pyszczynski & Greenberg, 1987). Reappraisal
regulation, by contrast, stops emotional cues before they consume valuable
cognitive resources.
As mentioned before, many jobs that require emotional labor are not
particularly complex, but the implications of these arguments extend beyond
jobs that demand constant emotional labor to those that require emotion
regulation in only limited circumstances, for example, managing emotive
processes in periodic dealings with supervisors or working in teams and on
task forces charged with cognitive tasks. More mundanely, restaurant
servers engaging in emotional suppression may perform less well on aspects
of their job that require more cognitive functioning (e.g., remembering the
food and drink orders from a large table). Thus, we suggest the following
proposition:
deployment and cognitive change) will result in better short-term job
performance than response-focused emotion regulation (suppression) in
emotional-labor situations with high cognitive demands.
Mental Health
Having an accurate perception of reality is often seen as a hallmark of
mental health and well-being, but considerable evidence has suggested that
individuals may enjoy increased mental health by maintaining positive
illusions (Alloy & Ahrens, 1987; Taylor, 1983). As Taylor and Brown
noted, the happy person appears to have the enviable capacity to distort
reality (1988, p. 204) to enhance their views of control and optimism.
98
The connections are clear between this literature base and reappraisal as an
emotion-regulation style. In a manner similar to reappraisal, individuals
may protect their well-being by using lters that distort or modify negativity
in the environment. Returning to the whistling-arias anecdote, individuals
who use antecedent-focused (attentional deployment and cognitive change)
emotion-regulation approaches create, in essence, positive illusions
about situations so that they experience the emotion in a direction that
enhances self-esteem, maintains beliefs in personal efcacy, and promotes an
optimistic view of the future (Taylor & Brown, 1988, p. 204). Empirical
evidence has suggested that at least in the short-term, antecedent-focused
(attentional deployment and cognitive change) approaches are effective;
individuals who are asked to reappraise emotions rather than suppress them
tend to report lower subjective levels of that emotion (e.g., Gross, 1998a).
Thus, we suggest the following proposition:
deployment and cognitive change) will result in better short-term mental
health than response-focused emotion regulation (suppression).
Long-Term Consequences of Emotion Regulation

Job Performance
Several theoretical perspectives posit that the long-term consequences
for job performance will be more severe among those using responsefocused rather than antecedent-focused emotion-regulation strategies.
First, self-discrepancy theory (e.g., Higgins, Klein, & Strauman, 1985) holds
that individuals are motivated to achieve a match between self-concept
and a number of personally relevant self-guides. Mismatches between the
actual-self (the attributes you believe you possess) and the ought-self
(the attributes you believe you should possess) can cause discomfort, guilt,
and the tendency toward self-punishment (Higgins, 1987). Conformance to
scripts can cause employees to deviate from their ought-self to another
socially prescribed actual-self. Following the self-discrepancy model,
individuals often suffer guilt and shame when they transgress or fail to
meet their self-imposed standards. When a salesperson suppresses negative
emotions and pretends to be chipper as if the customer is always right,
negative outcomes may surface. Beyond these outcomes, Higgins (1987)
reviewed a substantial body of evidence that related these forms of selfdiscrepancies to irritation, lethargy, and disinterest over time, all factors
99
that should reduce an employees ability to perform well on the job. Possibly
antecedent-focused regulation may also cause emotional laborers to display
diminished performance over time, but following a self-discrepancy
approach, these negative effects on job performance in the long term
should be more pronounced under a suppression regulatory style.
A second theoretical issue concerns enhanced accessibility of emotional
cues over time. Polivy (1998) and others (e.g., Rachman, 1980) argued
that emotions are a cue for appropriate behaviors and that repeated
inhibition or suppression may eventually produce a surge in target thoughts
during the attempted suppression as well as behavioral excess over
time. This effect is particularly plausible for long-term job performance,
because research has demonstrated that the surge in thoughts about the
emotion may increase dramatically after lapses occur in mental control
either voluntarily (a work break, for example) or involuntarily (an acute
increase in cognitive demands). Just as restrained dieters may suddenly lose
control and overindulge, individuals regulating their response-focused
emotion may, over time, lose their strength to suppress emotions and
ultimately become incapable of doing so. Wegner and Schneider (1989)
called this phenomenon the rebound effect. Repeated suppression
attempts and frequent relinquishment of mental control over time should
weaken performance.
have more severe negative effects on job performance over time than
antecedent-focused emotion regulation (attentional deployment and
cognitive change).
Physical Health
Although emotion regulation in the form of suppression reduces the
intensity of emotional displays, it does not exempt the person from
experiencing the emotion physically and psychologically. Studies have
suggested that suppression increases physiological activation and elevates
indicators such as nger pulse, nger temperature, and skin conductance
(Gross, 1998a; Gross & Levenson, 1997). Although these indicators may not
be hazardous in isolation, the cumulative effects of prolonged sympathetic
activation are related to a number of damaging health problems over
time (Krantz & Manuck, 1984). The physical consequences of emotion
regulation suppression in particular can relate to cardiovascular and
immunological system malfunctioning (Schaubroeck & Jones, 2000). Each
time the stress response is engaged, the immune system may be selectively
100
inhibited (OLeary, 1990). In terms of cardiovascular reactions, emotion

suppression has been linked to increased sympathetic nervous system
activation (including cardiovascular symptoms).
result in more physical health problems over time than antecedent-focused
emotion regulation (attentional deployment and cognitive change).
Targeting of Antisocial Behavior
As noted above, some evidence has indicated that antecedent-focused
regulation strategies (attention deployment and cognitive change) may be
preferable to suppression in terms of short-term mental health benets and
perhaps in terms of job performance as well. However, we propose that the
employee who must habitually use attentional deployment and cognitive
change may pay a signicant cost in terms of being exposed to antisocial
behavior at work, such as social undermining (e.g., Duffy, Ganster, &
Pagon, 2002; Duffy, Ganster, Shaw, Johnson, & Pagon, 2006) and abusive
supervision (e.g., Tepper, 2000; Tepper, Duffy, & Shaw, 2001). We suggest
two primary reasons that antecedent-focused strategies may increase
the risk that an individual will be targeted for antisocial behavior at work.
First, individuals who use attentional deployment (e.g., distraction) may
inadvertently signal would-be aggressors that the target individual may be
easily taken advantage of. As Marx, Heidt, and Gold (2005) pointed out,
different forms of attentional deployment may help an individual stave off
acute negative emotions associated with problematic situational cues, but
others who note the observable signs of dissociation or altered consciousness may be quick to take advantage (p. 80). Moreover, individuals who
are distracted and concentrating on things beyond the emotional-labor
event as ways of constricting emotional expression have been shown to
signal vulnerability for targeting (e.g., Luterek, Orsillo, & Marx, 2002).
Beyond attentional deployments attraction to aggressors, this avenue for
emotional regulation may also interfere with an individuals ability to process
threatening warnings in the environment. Emotions consist of response
tendencies that are meant to coordinate behavior in times of challenge or
danger. Gross (1998a, 1998b) suggested that one long-term consequence of
antecedent-focused regulation may be that one begins to deny important
features of ones environment through the use of unrealistic or inexible
views (e.g., insisting that everything is okay when it is not). Individuals can
often achieve attentional deployment through distraction or by focusing
attention on nonemotional or cross-emotional aspects of the situation.
101
In addition, they can redirect their attention from the cue demanding higher
levels of concentration to another task, which consumes cognitive resources
that they could use to process the emotion fully. Both avenues for attentional
deployment may impede the process for detecting and dealing with threatrelevant cues in the work context (Marx et al., 2005).
Our arguments concerning antecedent-focused regulation and workplace
targeting involve not only the threat-identication arguments above, but also
the idea that attentional deployment and cognitive change can compromise
defense activation or reex systems. The key interference point for defensive
reexes is during the postencounter stage what emotion-regulation
researchers have called the emotion-soliciting situational cue. During this
stage, normal responses would include freezing, focusing attention on the
cue, and considering the potential threat. This stage, which precedes the
more commonly discussed ght-or-ight stage, is critical not only in terms
of evaluating the situational cue but also in handling the situation. Ignoring
situational warnings or reevaluating ones capacity to manage the situation
interferes with, and perhaps disables, this process. In essence, these
antecedent-focused strategies allow an individual to get through the moment
but may inadvertently result in a failure to utilize valuable information
that the task or emotional cue contains. An individual using distraction to
fend off a distressing emotion may be able to function, but may fail to see
potential hazards in the social environment even under the circumstances
in which such arousal might be directly linked to such threat cues (Marx
et al., 2005, p. 81). Although unrelated to interpersonal deviance, research
examining safety at work has linked distraction and proneness to
distractibility as signicant predictors of accidents (e.g., Hansen, 1989).
Likewise, an individual who engages in cognitive change while dealing
with irritating or hostile coworkers or customers may start to believe
that the situation is not so bad or under control, when in reality the
situation should be activating defensive reexes. These individuals,
confronted with signs that others are bullying or aggressive, may ignore
the signals or reevaluate them as harmless. By ignoring their adaptive
emotion-based defenses they may become more vulnerable to future acts of
aggression by coworkers or customers. This risk may be heightened by the
fact that these employees may also present themselves more passively in
social interactions (i.e., they do not feel a need to appear assertive because
they do not perceive danger). Recent research has suggested that individuals
who appear to be more passive and are low in self-determination are more
likely to report being victimized at work (e.g., Aquino, Grover, Bradeld &
Allen, 1999).
102
In terms of response-focused emotion regulation, we expect that suppression will not increase susceptibility in becoming a workplace target for two
primary reasons. First, in response-focused regulation such as suppression,
emotions are fully formed before they are modied. The formation of the
emotion ensures that the key processing stage of the defense reexes system
is not altered. Although the emotion is later modulated, emotion formation
allows the appropriate defense mechanisms to be activated and the
situational cue processed correctly, which should aid in accurate threat
detection. In addition, Wegners (1994) ironic processing model suggests
that suppression of thoughts and emotions involves two distinct mental
processes an operating process that attempts to create the desired state
and a monitoring process that continuously supervises the suppression and
searches for lapses of mental control. Important in this view is that the
processes underlying emotion suppression are attentional processes that
initially orient the individual toward the stimulus or situational cue. In the
case of workplace stimuli, the suppression-processing model would suggest
that after attention is focused on the situational cue and the emotion is
developed, the choice to regulate through suppression starts an operating
process to bring desired emotions to the surface. After the operating phase is
activated, an unconscious monitoring process continues to scan for signs
that the system is working properly including sensations and thoughts that
are inconsistent with the achievement of successful control (Wegner, 1994,
p. 38). Wegner (1994) concluded, anything that is not the target of the
operating process, after all, indicates failure of the operating process and
should be monitored (p. 40). It is reasonable, then, to expect that focused
regulation will lower the likelihood of victimization because not only are
emotions fully formed before suppression, thereby activating defense reexes,
but the unconscious monitoring system will continue to scan for lapses of
mental control or other factors that may cause emotional control to fail.
deployment and cognitive change) will relate to a higher risk of being
the target of antisocial behavior at work relative to response-focused
emotion regulation (suppression).
Engaging in Antisocial Behavior at Work
Although the arguments above suggest that attentional deployment and
cognitive change increase vulnerability, evidence has also suggested that
individuals who habitually use response-focused regulation may be more
aggressive than those who use antecedent-focused approaches such as
103
attentional deployment and cognitive change. Constrained or suppressed

negative emotion is often redirected or displaced toward less-powerful or
more-available targets (e.g., Marcus-Newhall, Pedersen, Carlson, & Miller,
2000). The frustration hypothesis suggests that those individuals who
suppress their emotions will be more likely to express their frustration
through antisocial behaviors directed at others in the workplace. Theory
and empirical ndings on displaced aggression are also consistent with these
arguments and further highlight that antisocial behavior can be triggered by
a simple, daily event such as a happy dog jumping on the owner returning
from work or a coworker in a break room commenting on an unrelated
topic (Miller, Pedersen, Earleywine, & Pollock, 2003).
In terms of direct antisocial behavior, when emotions are suppressed
during emotional-labor work, emotional leakage may occur, especially when
the social interactions are longer or deeper (e.g., Ekman & Friesen, 1969).
Companies may have strict rules limiting emotional displays and prohibiting
aggression toward customers. Fear of retaliation or punishment serves
as a major constraining factor for the expression of negative emotions. Most
jobs that exert strong emotional-labor demands on employees (e.g., airport
ticketing agents, bar servers, social workers) require that they constrain their
emotions when they encounter hostile or abusive customers or clients. But it
is also likely that suppressed workers will displace their aggression,
expressing it toward coworkers or even further downstream toward others
outside the workplace.
The ironic processing model noted above also shows that response-focused
regulation will increase the likelihood an individual will engage in antisocial
behavior at work. Recall that in the ironic processing model, an individual
achieves mental control through an operating process that creates the desired
state and a monitoring process that continuously supervises the suppression
and searches for lapses of mental control. As the irony label suggests, a
consequence of these mental processes is that the emotion and the precipitating
event surrounding the emotions may actually become more accessible over
time (Polivy, 1998; Wegner & Schneider, 1989). Elevated accessibility may
lower the threshold for future aggressive behavior in response to minor
annoyances (Marcus-Newhall et al., 2000). Thus, individuals who suppress
emotions at work may explode at seemingly trivial or minor triggers in a
manner that seems incommensurate with the social interaction that preceded
the reaction. Although evidence of these effects is sparse, Christoforou (2008)
found among a sample of sales representatives that individuals using responsefocused regulation strategies to deal with interactions from abusive customers
engaged in higher levels of emotional deviance in the workplace.
104
In terms of antisocial work behavior, a value of antecedent-focused

regulation is that repeated regulation from attentional deployment and
cognitive change is less likely to increase the risk of antisocial behavior
directed at individuals responsible for the emotion-triggering cue or
indirectly at innocent coworkers, family members, and friends. Because
negative emotions are not formed fully with ex ante regulation strategies,
defensive reexes such as ghting or eeing and the mental control processes
charged with operating- and monitoring-suppressed emotions are not
activated. In addition, the ironic processing system is circumvented, which
lessens the likelihood that the individual will easily recall the precipitating
event, will extensively ruminate on the emotion and the cue, or will
pervasively feel frustration associated with attempts to suppress the emotion.
Attentional deployment and cognitive change should be less likely, then,
to cause direct and displaced antisocial behavior. Indirect support for these
ideas can be found in a recent series of studies by Bushman, Bonacci,
Pedersen, Vasquez, and Miller (2005). These authors found that compared
with experimental participants who were distracted, participants who
ruminated about a provocation were more likely to engage in displaced
aggression after being exposed to a minor annoyance or trigger. Thus, we
suggest the following proposition:
relate to a higher likelihood of engaging in direct antisocial behavior at
work and displaced aggression than the antecedent-focused emotion
regulation (attentional deployment and cognitive change).
CONCLUSIONS AND FUTURE

RESEARCH DIRECTIONS
The concept of emotional labor has received increasing attention in the
organizational literature in recent years, but the process of emotion
regulation and its consequences in the workplace have rarely been explored.
Research in psychology disciplines has conrmed the importance of emotion
regulation in assessing the behavior, attitudes, and well-being of individuals,
but inconsistencies have marred the ndings. In this paper, we propose
the factor of time (short term versus long term) as critical, and typically
missing, in conceptualization of emotion regulation. We then develop an
initial framework for predicting time-based work-related consequences of
regulation processes.
105
Our exposition is but a rst attempt to delineate the relevant workplace

aftereffects of employed regulation strategies. It is impossible to point to one
emotion-management tactic as preferable in the absence of rich contextual
and personal factors. Indeed, our framework highlights the positive and
negative aspects of both approaches to emotion regulation; neither is
universally effective, neither is completely unsound. Given the expectations
and demands businesses are placing on the ever-increasing service sector
of the labor force, it has become crucial to know how, when, and why
individuals use various emotional strategies (Tarvis, 1984). Moreover,
further information regarding the relative benets and detriments of various
emotion-regulation strategies would benet managers and employees alike.
As Stone-Romero (1994) suggested, it is necessary to highlight the
boundary conditions, both of the overall framework presented and of the
specic propositions offered. First, we make no differential predictions
regarding the type of emotion being regulated. Clearly, the type of emotion
(e.g., disgust, sadness, anger, fear, enthusiasm, pride) may impact not only
the choice of emotion-regulation strategies but also their consequences.
Research on the asymmetry of positive and negative social interactions
(e.g., Duffy et al., 2002) may help future researchers to develop specic
propositions for the outcomes of the regulation of individual emotions. We
also use the concept of time in a general sense (short term versus long term).
We distinguish between single emotional-labor events (short term) versus
the same regulation strategy repeated over time (long-term). Although this
allows us to decipher inconsistent theory and research ndings regarding
emotion-regulation processes, future researchers should explore the timing
issues our hypotheses capture. Perhaps research on behavioral spirals and
additional exploration of contagion effects would clarify the organizational
impact of short- and long-term regulations.
We also rely on the results of experimental research and, to some extent,
other conceptual work, to develop our propositions. This is necessary
because experimental psychology has conducted most emotion-regulation
research; longitudinal eld studies of emotion regulation in organizational
settings are largely unavailable. Finally, we examine only consequences of
emotion regulation in the performance of emotional labor the antecedents
remain largely unexplored. For instance, little is known about why people
choose to suppress rather than use attentional deployment, whether some
individuals oscillate between emotion-regulation strategies, and how
frequently they choose consciously between the two options.
At this point, we highlight the need for additional theory development
and empirical research to test and extend the basic tenets of our theorizing.
106
The propositions delineated here are clearly testable, although not without
challenges. As noted above, much of the empirical testing of emotionregulation theory tenets has been conducted in short-term laboratory
settings where emotion-regulation strategies have been manipulated
experimentally. But the nature of emotion regulation and outcomes such
as performance, physical health, and antisocial behavior lend themselves to
longitudinal or multiwave studies. These work dynamics may play out in
immediate contexts or within certain interactions for example, a customer
directly targeting an employee but in many cases will unfold over time.
One-to-one correspondence will not appear between an employee engaging
in attentional deployment or cognitive change and the outcomes of interest;
rather our thesis suggests that these regulation approaches increase the
likelihood or probability that an individual will experience better or worse
performance, health, and antisocial-behavior-related outcomes over time.
Similarly, even in longer time windows, we would expect that not every
incidence of suppression would result in direct or displaced antisocial
behavior, for example, but that repeated attempts to suppress negative
emotions would increase the likelihood of these outcomes. Gross (2001,
2002) concluded his reviews of emotion-regulation research by suggesting
that researchers explore the potential differential consequences of different
regulatory styles and pursue a research agenda focused on the long-term
consequences of differing emotion regulation strategies (p. 218). The Cote
and Morgan (2002) study represented an initial attempt to study regulation
processes over time, but they used a one-month time lag, which may be
insufcient for differential effects to emerge. On the positive side, several
researchers, including Cote and Morgan (2002) and Grandey et al. (2005),
have developed and usefully employed measures of emotion regulation in
eld settings. Thus, the key challenges to overcome in terms of adequate
tests of our propositions would seem to be designing and executing a study
that would allow these dynamic processes to unfold.
Woven throughout our analysis is the apparent critical need for examining
outcomes of emotion regulation in eld research where dynamics related
to time are in play. As a number of authors (e.g., Goodman, Lawrence,
Ancona, & Tushman, 2001) have pointed out, time has multiple meanings
and can be conceptualized in different ways. In terms of understanding the
temporal dynamics of emotion regulation, the problems are no less complex.
Marks, Mathieu, and Zaccaro (2001), in their theory of time and team
processes, conceptualized team performance episodes as distinguishable
periods of time over which performance accrues and feedback is available
(p. 359). Emotion-regulation episodes could be similarly viewed as the time
107
between the situational cue that initiated the regulation process and the end
of the regulated interaction or, perhaps, at the point when the emotion was
no longer being regulated. In longer-duration and interaction-rich interpersonal contexts, emotional leakage has more opportunity to emerge
(e.g., Ekman & Friesen, 1969), and observers have more opportunity to
detect whether and how individuals are regulating their emotions. Time in an
emotion-regulation context can also be conceptualized as the mapping
of multiple regulation events in terms of cycles, frequency, and rhythm
(Ancona, Okhuysen, & Perlow, 2001) a conceptualization that Marks et al.
(2001) labeled a recurring phase model of time. It may be possible to capture
this view of time by simply operationalizing it as job tenure. Because our
propositions concerning long-term outcomes implicitly assume greater or
lower likelihood of outcomes over time, it is reasonable to assume that
job tenure moderates these relationships such that the relationships are
stronger when job tenure is high. But, although convenient, job tenure as an
operationalization of recurring phases falls somewhat short of capturing the
full nature of the construct. Marks et al. (2001) pointed out that when time is
conceptualized as a series of episodes, it can be deconstructed into action
antecedent- or response-focused regulating in our case and transition phases.
Across jobs, the frequency and duration of the emotion-regulation episodes
and the length of transition phases may differ markedly. Auto salespersons
may face only one or two daily incidences that require them to regulate
emotions, although, as noted, the incidences may last a relatively long time but
be alleviated by long transition or recovery phases in between. Other
salespersons such as call-center employees may have numerous, short, daily
interactions that require emotion regulation and multiple, short-transition
periods while they wait for their next call. The rhythms of these cycles are
markedly different and likely have different implications for workplace
outcomes, implications that in all likelihood will be masked by simple time
measures such as tenure. Beyond these frequency and duration issues, some
research has also suggested that the content of transition periods also affects
future behaviors. Harinck and De Dreu (2008) found that when they
distracted their experimental participants with another task during a break, the
participants later reached higher-quality negotiated agreements, as compared
with participants who continued to reect on the ongoing negotiations during
the break. Thus multiple periods of distracting transitions may diffuse or
ameliorate the effects of response-focused regulation and aggression.
A third way to view time concerns regulators temporal perceptions,
which can range from perceptions of time passage (e.g., time ying, time
passing, or time dragging; McGrath & Kelly, 1986), but also perceptions
108
about the novelty or originality of a given moment. It is possible to view

attentional deployment as a means of managing the passing of time.
Grandeys (2000) example of a server whistling arias to deploy attention
from negative or difcult customers might also be seen as way of passing
time. Aside from our arguments about threat detection and defensive
reexes above, attentional deployment may also change individuals
understanding and knowledge about time acquired through the senses
(Ancona et al., 2001, p. 519). In general, as noted by Gross (1999) and
others, the long-term consequences of emotion-regulation strategies are
essentially unexplored. We encourage future researchers to explore and
extend our propositions by considering how and when these relationships
strengthen, weaken, or change in other ways across different views of time.
A nal issue concerns the role of display rules in the emotion-regulation
process. On one hand, display rules could be viewed as antecedents to
emotion regulation and as useful in predicting what emotion-regulation
strategies individuals choose. In some organizations, display rules are
implicit and passed along through high-performance expectations (Zapf,
2002). In others, display rules are formal, strict, and regulated by direct
supervisors (e.g., Wilk & Moynihan, 2005). On the other hand, the strength
of organizational or supervisor-level display rules may change the nature of
the relationships between emotion regulation and antisocial behavior. For
example, Grandey et al. (2005) argued that response-focused regulation
relate positively to employee burnout, but they also argued and found that
this relationship is stronger when individuals lack personal control over the
situation. Similarly, Wilk and Moynihan (2005) demonstrated that display
rules varied at the supervisor level, drained resources as they increased, and
related positively to emotional exhaustion (see also Sutton, 1991). In terms
of antisocial behavior at work, strict and regulated display rules are likely to
squelch certain forms of direct aggression that may result from emotionregulation processes, but may increase the likelihood that response-focused
regulation will displace aggression toward coworkers or even outside
the workplace. Understanding how display rules relate to different forms
of emotion regulation but also shape antisocial responses to regulation
strategies are important areas for future research.
Implications for HRM Practice

Our theory building is helpful as an academic exercise, but it also has
practical implications. First, the effects of emotion-regulation strategies on
109
employees abilities to stave off accidents may be critical to occupational

health. In the United States, 21 million individuals work in occupational
sectors such as wholesale and retail trade where emotional labor
predominates (National Occupational Research Agenda, 2008), making the
potential for accidents and violence an important public health concern.
If antecedent-focused strategies leave employees in high-risk workplaces
such as convenience stores and gas stations susceptible to social situations
where others may harm them, perhaps training and awareness of more
appropriate regulation strategies would protect this at-risk population.
Second, uncovering emotional-labor processes helps us understand the
factors that determine individual performance in jobs high in emotional
labor. Hochschild (1983) told of an airline that screens for warm
personalities during interviews for ight attendants, explaining that, in
general, selling products and services involves selling your personality.
It is often assumed that being good at emotional labor stems from certain
stable personality characteristics; organizations might screen for such
characteristics and select employees based on their emotional-labor skills.
Yet our framework suggests that it is not personality that determines
individual performance but the deployment of regulation strategies in
response to emotion-eliciting situations. As such, perhaps high performance
in these jobs comes from training or practice in situation assessment and
regulation-strategy deployment.
In conclusion, while the current literature provides many interesting
insights about the role of emotion regulation in emotional-labor
performance, there is much more to be understood about the advantages and disadvantages of emotion regulation for individuals and
organizations. We provide this set of time-based propositions as a point
of departure for future research in this area and also as a roadmap for
future theory-building and empirical testing. The study of emotion
regulation in emotional-labor performance is fraught with challenges
that researchers will have to overcome, especially in terms of research
design and measurement. We hope, however, that this review will
encourage investigators to take up these challenges and move the literature
forward.
ACKNOWLEDGMENT
The authors thank Jacquelyn Thompson for editorial assistance.
110
REFERENCES
Alloy, L., & Ahrens, A. (1987). Depression and pessimism for the future: Biased use of
statistically relevant information in predictions of self vs. others. Journal of Personality
and Social Psychology, 52, 11291140.
Ancona, D. G., Okhuysen, G. A., & Perlow, L. A. (2001). Taking time to integrate temporal
research. Academy of Management Review, 26, 512529.
Aquino, K., Grover, S., Bradeld, M., & Allen, D. (1999). The effects of negative affectivity,
hierarchical status, and self-determination on workplace victimization. Academy of
Management Journal, 42, 260273.
Ashforth, B. E., & Humphrey, R. H. (1993). Emotional labor and identity. Academy of
Management Review, 18, 88115.
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998). Ego depletion:
Is the active self a limited resource? Journal of Personality and Social Psychology, 74,
12521265.
Beal, D. J., Trougakos, J. P., Weiss, H. M., & Green, S. G. (2006). Episodic processes in
emotional labor: Perceptions of affective delivery and regulation strategies. Journal of
Applied Psychology, 91, 10531065.
Bernhardt, A., Morris, M., Handcock, M., & Scott, M. (2001). Divergent paths: Economic
mobility in the new American labor market. New York: Russell Sage Foundation.
Brotheridge, C. M., & Grandey, A. A. (2002). Emotional labor and burnout: Comparing two
perspectives on people work. Journal of Vocational Behavior, 60, 1739.
Bushman, B. J., Bonacci, A. M., Pedersen, W. C., Vasquez, E. A., & Miller, N. (2005). Chewing
on it can chew you up: Effects of rumination on triggered displaced aggression. Journal
of Personality and Social Psychology, 88, 969983.
Christoforou, P. T. (2008). Antecedents and consequences of emotion regulation and emotional
deviance. Paper presented at the annual meetings for the Academy of Management,
Anaheim, CA.
Cote, S., & Morgan, L. M. (2002). A longitudinal analysis of the association between emotion
regulation, job satisfaction, and intentions to quit. Journal of Organizational Behavior,
23, 947962.
Duffy, M. K., Ganster, D. C., & Pagon, M. (2002). Social undermining in the workplace.
Academy of Management Journal, 45, 331352.
Duffy, M. K., Ganster, D. C., Shaw, J. D., Johnson, J. L., & Pagon, M. (2006). The social
context of undermining behavior at work. Organizational Behavior and Human Decision
Processes, 101, 105121.
Ekman, P., & Friesen, W. V. (1969). Nonverbal leakage and clues to deception. Journal for the
Study of Interpersonal Processes, 32, 88106.
Ellis, H. C., & Ashbrook, P. W. (1988). Resource allocation model of the effects of depressed
mood states on memory. In: K. Fiedler & J. F. Forgas (Eds), Affect, cognition, and social
behavior (pp. 2543). Gottingen, Germany: Hogrefe.
Geddes, D., & Callister, R. R. (2007). Crossing the line(s): A dual threshold model of anger in
organizations. Academy of Management Review, 32, 721746.
Glomb, T. M., Kammeyer-Mueller, J. D., & Rotundo, M. (2004). Emotional labor
demands and compensating wage differentials. Journal of Applied Psychology, 89,
700712.
Goffman, E. (1959). Presentation of self in everyday life. New York: Overlook.
111
Goodman, P. S., Lawrence, B. S., Ancona, D. G., & Tushman, M. L. (2001). Introduction.
Academy of Management Review, 26, 507511.
Grandey, A. A. (2000). Emotion regulation in the workplace: A new way to conceptualize
emotional labor. Journal of Occupational Health Psychology, 5, 95110.
Grandey, A. A. (2003). When the show must go on: Surface acting and deep acting as
determinants of emotional exhaustion and peer-rated service delivery. Academy of
Grandey, A. A., Fisk, G. M., & Steiner, D. D. (2005). Must service with a smile be stressful?
The moderating role of personal control for American and French employees. Journal of
Gross, J. J. (1998a). Antecedent and response-focused emotion regulation: Divergent
consequences for experience, expression, and physiology. Journal of Personality and
Social Psychology, 74, 224237.
Gross, J. J. (1998b). The emerging eld of emotion regulation: An integrative review. Review of
General Psychology, 2, 271299.
Gross, J. J. (1999). Emotion regulation: Past, present, and future. Cognition and Emotion,
13, 551573.
Gross, J. J. (2001). Emotion regulation in adulthood: Timing is everything. Current Directions in
Psychological Science, 10, 214219.
Gross, J. J. (2002). Emotion regulation: Affective, cognitive, and social consequences.
Psychophysiology, 39, 281291.
Gross, J. J., & John, J. (2003). Individual differences in two emotional regulations processes:
Implications for affect, relationships, and well-being. Journal of Personality and Social
Psychology, 85, 348362.
Gross, J. J., & Levenson, R. W. (1997). Hidden feelings: The acute effects of inhibiting negative
and positive emotion. Journal of Abnormal Psychology, 106, 95103.
Hansen, C. (1989). A causal model of the relationship among accidents, biodata, personality,
and cognitive factors. Journal of Applied Psychology, 74, 81105.
Harinck, F., & De Dreu, C. K. W. (2008). Take a break! Or not? The impact of mindsets
during breaks on negotiation processes and outcomes. Journal of Experimental Social
Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review,
94, 319340.
Higgins, E. T., Klein, R., & Strauman, T. (1985). Self-concept discrepancy theory: A
psychological model for distinguishing among difference aspects of depression and
anxiety. Social Cognition, 3, 5176.
Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of
Sociology, 85, 551575.
Hochschild, A. R. (1983). The managed heart. Berkeley: University of California Press.
Krantz, D. S., & Manuck, S. B. (1984). Acute psychophysiologic reactivity and risk of
cardiovascular disease: A review and methodologic critique. Psychological Bulletin,
96, 435464.
Landy, F. J. (1992). Work design and stress. In: G. P. Keita & S. L. Sauter (Eds), Work and
well-being: An agenda for the 1990s (pp. 119158). Washington, DC: American
Psychological Association.
Luterek, J., Orsillo, S., & Marx, B. P. (2002). Emotional responding in adult childhood sexual
abuse survivors: Emotional experience, written and facial expression to positive and
112
negative lm stimuli. Paper presented at the 35th annual meeting of the Association for
the Advancement of Behavior Therapy, Reno, NV.
Marcus-Newhall, A., Pedersen, W., Carlson, M., & Miller, N. (2000). Displaced aggression is
alive and well: A meta-analytic review. Journal of Personality and Social Psychology,
78, 670689.
Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally based framework and
taxonomy of team processes. Academy of Management Review, 26, 356376.
Marx, B. P., Heidt, J. M., & Gold, S. D. (2005). Perceived uncontrollability and unpredictability,
self-regulation, and sexual revictimization. Review of General Psychology, 9, 6790.
McGrath, J. E., & Kelly, J. R. (1986). Time and human interaction: Toward a social psychology
of time. New York: Guilford Publications Inc.
Miller, N., Pedersen, W. C., Earleywine, M., & Pollock, V. E. (2003). A theoretical model of
triggered displaced aggression. Personality and Social Psychology Review, 7, 7597.
Morris, J. A., & Feldman, D. C. (1996). The dimensions, antecedents, and consequences of
emotional labor. Academy of Management Journal, 21, 9861010.
Morris, W. N., & Reilly, N. P. (1987). Toward the self-regulation of mood: Theory and
research. Motivation and Emotion, 11, 215249.
Muchinsky, P. M. (2000). Emotions in the workplace: The neglect of organizational behavior.
Journal of Organizational Behavior, 21, 801805.
National Occupational Research Agenda. (2008). Wholesale and retail trade agenda,
06/13/2008. National Wholesale and Retail Sector Council. Available at: http://www.
cdc.gov/niosh/nora/comment/public/WholRetTradeDraftJune2008/pdfs/WholRet
TradeDraftJune2008.pdf
OLeary, A. (1990). Stress, emotion, and human immune function. Psychological Bulletin,
108, 363382.
Polivy, J. (1998). The effects of behavioral inhibition: Integrating internal cues, cognition,
behavior, and affect. Psychological Inquiry, 9, 181204.
Pugliesi, K. (1999). The consequences of emotional labor: Effects on work stress, job
satisfaction, and well-being. Motivation and Emotion, 23, 125154.
Pyszczynski, T., & Greenberg, J. (1987). Self-regulatory preservation and the depressive selffocusing style: A self-awareness theory of reactive depression. Psychological Bulletin,
102, 122138.
Rachman, S. (1980). Emotional processing. Behaviour Research and Therapy, 18, 5160.
Rafaeli, A., & Sutton, R. I. (1987). Expression of emotion as part of the work role. Academy of
Rafaeli, A., & Sutton, R. I. (1990). Busy stores and demanding customers: How do they affect
the display of positive emotion? Academy of Management Journal, 33, 623637.
Richards, J. M., & Gross, J. J. (1999). Composure at any cost? The cognitive consequences of
emotion suppression. Personality and Social Psychology Bulletin, 25, 10331044.
Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of
genotype-environment effects. Child Development, 54, 424435.
Schaubroeck, J., & Jones, J. R. (2000). Antecedents of workplace emotional labor dimensions
and moderators of their effects on physical symptoms. Journal of Organizational
Behavior, 21, 163183.
Stone-Romero, E. (1994). Construct validity issues in organizational behavior research. In:
J. Greenberg (Ed.), Organizational behavior: The state of the science (pp. 155177).
Hillsdale, NJ: Lawrence Erlbaum.
113
Sutton, R. (1991). Maintaining norms about expressed emotions: The case of bill collectors.
Administrative Science Quarterly, 36, 245268.
Tarvis, C. (1984). One the wisdom of counting to ten: Personal and social dangers of anger
expression. Review of Personality and Social Psychology, 5, 170191.
Taylor, S. (1983). A theory of cognitive adaptation. American Psychologist, 21, 1161180.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective
on mental health. Journal of Personality and Social Psychology, 103, 193210.
Taylor, S. E., & Brown, J. D. (1999). Illusion and well-being: A social psychological perspective
on mental health. In: R. F. Baumeister, et al. (Eds), The self in social psychology. Key
readings in social psychology (pp. 4368). Philadelphia, PA: Psychology Press/Taylor &
Francis.
Tepper, B. J. (2000). Consequences of abusive supervision. Academy of Management Journal,
43, 178190.
Tepper, B. J., Duffy, M. K., & Shaw, J. D. (2001). Personality moderators of the relationship
between abusive supervision and subordinates resistance. Journal of Applied Psychology,
86, 974983.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 3552.
Wegner, D. M., & Schneider, D. J. (1989). Mental control: The war of the ghosts in
the machine. In: J. S. Uleman & J. A. Bargh (Eds), Unintended thought (pp. 287305).
New York: Guilford.
Wilk, S. L., & Moynihan, L. M. (2005). Display rule regulators: The relationship between
supervisors and work emotional exhaustion. Journal of Applied Psychology, 90, 917927.
Zapf, D. (2002). Emotion work and psychological well-being: A review of the literature and
some conceptual considerations. Human Resource Management Review, 12, 237268.
INSIGHTS FROM VOCATIONAL

AND CAREER DEVELOPMENTAL
THEORIES: THEIR POTENTIAL
CONTRIBUTIONS FOR ADVANCING
THE UNDERSTANDING OF
EMPLOYEE TURNOVER
Peter W. Hom, Frederick T. L. Leong and
Juliya Golubovich
ABSTRACT
This chapter applies three of the most prominent theories in vocational
and career psychology to further illuminate the turnover process.
Prevailing theories about attrition have rarely integrated explanatory
constructs from vocational research, though career (and job) choices
clearly have implications for employee affect and loyalty to a chosen job
in a career eld. Despite remarkable inroads by new perspectives for
explaining turnover, career, and vocational formulations can nonetheless
enrich these and conventional formulations about why incumbents stay
or leave their jobs. To illustrate, vocational theories can help clarify why
certain shocks (critical events precipitating thoughts of leaving) drive
attrition and what embeds incumbents. In particular, this chapter reviews
ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029006
115
116
PETER W. HOM ET AL.
Supers life-span career theory, Hollands career model, and social

cognitive career theory and describes how they can ll in theoretical gaps
in the understanding of organizational withdrawal.
Theories and research on why and how employees sever employment ties
have undergone a renaissance in modern times (Holtom, Mitchell, Lee, &
Eberly, 2008; Hom, 2010). After fallow years during the late 20th century
(OReilly, 1991), turnover scholars began introducing a host of innovative
theories (e.g., Lee & Mitchells, 1994 unfolding model; macro-level
models of aggregate quit rates; Kacmar, Andrews, Rooy, Steilberg, &
Cerrone, 2006) and constructs ( job embeddedness, Mitchell, Holtom,
Lee, Sablynski, & Erez, 2001b; movement capital, Trevor, 2001) as well
as new methodologies (latent growth modeling; Bentein, Vandenberg,
Vandenberghe, & Stinglhamber, 2005) to promote understanding and
prediction of organizational withdrawal. These recent developments are
stimulating considerable rethinking and empirical inquiry (Holtom et al.,
2008; Hom, 2010). Despite such theoretical and methodological advances,
however, there remain major gaps in turnover perspectives and unresolved
questions about certain turnover phenomena.
To illustrate, though it is the most comprehensive formulation about
proximal antecedents and processes leading to turnover thus yet conceived
(Hom, 2010), the unfolding model fails to fully elucidate why various critical
events (shocks) precipitate thoughts of leaving (Lee & Mitchell, 1994).
For example, T. W. Lee, Mitchell, Wise, and Fireman (1996) observed that
some nurses quit when they become pregnant, a shock that sets in motion
preexisting plans to exit for childbearing. They also noted that other nurses
exit because their hospital switched from individualized to team-based
patient care, a negative event violating personal career plans or goals. While
specifying that preexisting plans (to quit) and violations of career plans
mediate shocks impact, this theory nevertheless overlooks why employees
adopt certain plans. Quite likely, employees react to the same critical events
differently (not necessarily leaving) based on their assessment of how such
events affect their personal plans (Sweeny, 2008). Other nurses facing
pregnancies or team-based nursing would not necessarily resign if they
plan to work during motherhood or if this hospital policy ts their
professional image. Consequently, greater clarication of how different
individuals appraise critical events is essential because critical events are
prime movers of nonaffective turnover paths in the unfolding model that
Career Theories and Turnover
117
many if not most leavers follow (Holtom, Mitchell, Lee, & Inderrieden, 2005;
T. H. Lee, Gerhart, Weller, & Trevor, 2008; Mitchell, Holtom, & Lee, 2001a).
Similarly, theoretical accounts of many well-documented withdrawal
phenomena remain unspecied or insufcient. To illustrate, empirical
research has long established that age and rm tenure are inversely
related to quits, such associations becoming styled facts in the turnover
literature (Griffeth, Hom, & Gaertner, 2000; Hom, Roberson, & Ellis, 2008).
While often treated as control variables or proxies for proximal antecedents
(Mitchell et al., 2001b; Mobley, Horner, & Hollingsworth, 1978), these
demographic characteristics deserve greater scholarly attention. After all,
they often forecast attrition more reliably than common explanatory
constructs (e.g., job satisfaction; Griffeth et al., 2000) and may reect more
fundamental but neglected distal causes (e.g., life stages). Further, many
scholars (Judge & Watanabe, 1995) and practitioners (Khatri, Fern, &
Budhwar, 2001) believe that the Hobo syndrome (Ghiselli, 1973; Hulin,
Rosnowski, & Hachiya, 1985; Maertz & Campion, 2004), an individuals
job-hopping history, is a reliable signal of future quits. Though the premise
that the best predictor of future behavior is past behavior is deemed a basis
for the hobo syndrome (Griffeth & Hom, 2001), prediction is not, however,
explanation. In short, what explains hobos wanderlust?
To enrich turnover perspectives and clarify certain turnover phenomena,
we draw on theory and research from career and vocational psychology.
Three of the most prominent theories in this discipline are Supers life-span
career development theory, Hollands career model, and social cognitive
career theory (SCCT; Lent, Brown, & Hackett, 1994). These views have long
dominated research on vocational choice and behavior (Leong & Barak,
2001). Turnover investigators nonetheless have rarely capitalized on
insights and ndings from this longstanding research stream. To address
this oversight, our chapter demonstrates how these vocational models
can further illuminate the reasons and manner by which incumbents vacate
their jobs. Because certain SCCT constructs (e.g., newcomer self-efcacy)
have been investigated by socialization scholars as turnover antecedents, we
further elaborate and rene SCCT theory to extend existing socialization
research on newcomer attrition.
DONALD SUPERS DEVELOPMENTAL THEORY

Donald Supers life-span career development theory (1953, 1990) is one of
the pivotal theories in vocational psychology, which is embedded within a
118
PETER W. HOM ET AL.
developmental framework. The main principle of Supers theory, in contrast

to other career theories, is the centrality of the development axis. In his
model comprising ten propositions, Super posits that career development
involves the ongoing implementation of the self-concept across life stages.
He formulated career developmental stages that include a growth
stage followed by exploration, establishment, maintenance, and decline
stages. Within each of these developmental stages are substages that include
developmental tasks and challenges that individuals must meet and
surmount to realize their self-concept.
Life and Career Development Stages

According to Supers (1953, 1990) theory, there are ve life and career
development stages. These ve stages are: (1) Growth stage (birth to 15),
which is concerned with the development of the self-concept and ones
capacity, attitudes, interests, needs as they pertain to performance at school
and at home; (2) Exploration stage (ages 1524), which produces tentative
career/vocational choices and involves career exploration activities including
class selection, work experiences, and hobbies; (3) Establishment stage (ages
2544), which entails entry into the workforce and the start of ones ofcial
career during which initial skill-building and stabilization in ones work
experiences occur; (4) Maintenance stage (ages 4564), which consists of an
adjustment process to advance ones position at work and to nd continuing
satisfaction in work activities that one has mastered at an earlier stage; and
(5) Decline stage (around age 65 onwards), which includes reduced output
and productivity in preparation for retirement. Super later reformulated this
last stage into a more positive Readiness for retirement stage.
Embedded within the major development stages, Super also articulated
certain phases with specic developmental tasks that interact with an
individuals experiences. Because our chapter focuses on turnover, we focus
on phases within career development stages. Within the Exploration stage,
the rst phase involves Crystallization (ages 1418) during which one is
developing and planning a tentative vocational goal. This is followed by
a Specication phase (ages 1821) where one rms up ones vocational goal
and moves on to the Implementation phase (ages 2124), when one seeks
and obtains the requisite education and training for the chosen career or
vocation. The Establishment stage is comprised of two phases. During a
Stabilization phase (ages 2435), individuals begin working and conrming
119
their career choices, while being primarily focused on career advancement

during a Consolidation phase (ages 35).
Career Constructs
Within this life-stage career development framework, Super conceptualized
several constructs that can advance turnover theory and research. The rst is
the notion of the self-concept that is assessed by the Values Scale (VS). The
VS consists of 21 work values that help guide a persons implementation of
his or her self-concept. Attainment and implementation of these values in the
workplace give rise to job satisfaction and career growth and development.
From a turnover point of view then, work environments that fail to provide
opportunities for the satisfaction of employees values will motivate them
to quit. For example, if strongly held values for autonomy, creativity, or
advancement are not satised within a particular work environment, this
value incongruence would likely stimulate turnover cognitions.
Career maturity is the next major construct developed within Supers
theory. His developmental model emphasizes ones readiness to move on to
the next stage and hence the concept of career maturity is one of readiness to
take on appropriate developmental tasks. Given differing maturity, different
people transition from school to work or from early career to mid-career
at differential speeds. According to Super, career maturity comprises the
following components: Awareness of the need to plan ahead; decisionmaking skills; knowledge and use of information resources; general career
information; general world of work information; and detailed information
about occupations of preference. A student of Donald Super, John Crites
(1973) went on to design a career maturity inventory (CMI) to measure
career planfulness, career exploration, and readiness. While much vocational research has examined career maturity among late adolescents and
college students, many implications of career maturity have yet to be
explored by turnover researchers among working adults.
Super (1953, 1990) conceptualized career maturity as a multidimensional
construct that represents a persons readiness to cope with vocationally
related developmental tasks at a particular life-span period. Giving special
attention to adolescent development, Super noted that adolescents level
of career maturity evolves with growing awareness that they must make
an occupational choice and their attitudes toward this task. During this
period, adolescents also develop competencies (i.e., knowledge, abilities, and
skills) that facilitate career decisions and implementations, especially career
120
PETER W. HOM ET AL.
exploration and planning. In general, career maturity is often conceived

of as having both attitudinal and cognitive dimensions. In a stimulus
organismresponse (SOR) paradigm for studying career maturity, these
attitudes and competencies serve as intervening variables (Savickas, 1985).
Successful evolution of career maturity during adolescence is thus expected
to culminate in career choice crystallization and commitment.
In studying career maturity development, vocational researchers have
usually conceptualized change in quantitative terms, such as increases in the
attitudinal and cognitive dimensions, which are operationalized differently
in various measures of career maturity (Savickas, 1985). Whereas much of
the research generated by Super and those following his paradigm have
mostly focused on adolescents and college students (due to availability bias
of counseling psychologists who were primarily academicians), we believe
that the career maturity construct can be readily generalized into adulthood
to help clarify why employees stay or leave jobs.
Indeed, Savickas (1997, 2005) recently conceptualized the concept of
career adaptability a construct akin to career maturity to address
individuals exibility and maturity for coping with workplace challenges.
Building upon Super and Knasels (1981) observation that adaptation is the
central developmental challenge for adults, Savickas (2005) went on to
develop a career construction theory based on a constructivist perspective.
Savickas (2005) viewed career construction as a series of attempts to
implement a self-concept in social roles, focusing attention on adaptation
to a series of transitions from school to work, from job to job, and from
occupation to occupation. In this theory of career construction, career
adaptability plays a central role in shaping the actual problem-solving
strategies and coping behaviors of individuals in their work lives. This model
of career adaptability for adults thus parallels the career maturity model
for adolescents. Within this model, Savickas proposed four dimensions of
adaptability: concern, control, curiosity, and condence. The adaptive
individual is (a) concerned about the vocational future, (b) exerts personal
control over his or her vocational future, (c) displays curiosity by exploring
possible selves and future scenarios, and (d) is condent about pursuing his or
her aspiration. Therefore, increasing a clients career adaptability is an
essential goal of career counseling and career interventions.
Individuals lacking career adaptability might have difculty managing
interpersonal relationships or meeting performance challenges in the
workplace. Therefore, career adaptability might be a useful predictor of
turnover cognitions and actual turnover. It is important to note, however,
that the relationship between career adaptability and turnover may be more
121
complex. For example, Ito and Brotheridge (2005) examined the impact of
organizational support for career adaptability of employees in the form of
career information, advice, and encouragement. Unexpectedly, they found
that career adaptability was positively associated with both organizational
commitment and intentions to leave, suggesting some unintended consequences for management approaches supporting career adaptability
(p. 5). Apparently, promoting greater career adaptability among employees
can increase their ease of movement (March & Simon, 1958).
We return to describing Supers theory and consider the life career
rainbow. In his later formulations, Super began examining the notion of
career salience or work centrality. He recognized that a career might not be
central to everyone and that its centrality may change over time as other life
roles compete for ones attention and energies. Within this later program
of research, he conceptualized the career rainbow in which an individual
has six life roles: worker, student, citizen, husband/wife, parent, and leisure.
The rainbow concept presumes that work becomes central for people at
the middle stages of life but loses centrality due to competing demands of
family, leisure activities, and so forth. Outside the middle life stages, other
life roles may become more central than work.
Career Developmental Assessment and Counseling Model

In 1992, Super and his colleagues articulated the Career Developmental
Assessment and Counseling (C-DAC) Model. C-DAC model emerged from
the life span, life-space theory of careers and integrated key elements
(i.e., the Life Career Rainbow, the Model of Importance, and the Model of
Determinants) from Supers theory. Therefore, the C-DAC model blends
components of differential, developmental, and personal construct theories
into one comprehensive career assessment and counseling system. Super
recommended that counselors implement the C-DAC model in a four-step
process (Super, Savickas, & Super, 1996). Step one consists of an initial
interview to identify a clients presenting concerns. In the interview, the
counselor also reviews any available data from the clients record and
develops a preliminary counseling plan. At the same time, counselors must
understand the importance of work to the client relative to life roles in other
realms (e.g., school, home and family, community, and leisure). Assessing
the clients level of work role salience reveals whether further career
assessment and counseling will be meaningful (high career salience) or not
(low career salience). Clients high in career salience show readiness to
122
PETER W. HOM ET AL.
maximally benet from further career assessment. Clients low in career

salience may, depending on their unique life status, need help either
(a) orienting to the world-of-work prior to further assessment or
(b) exploring and preparing for other life roles.
In the second step, the counselor administers instruments to measure the
clients career stage and concerns, and level of career maturity or career
adaptability. The counselor thus determines the clients readiness for career
decision-making activities, such as identifying and exploring occupational
interests. In step three, the counselor helps clients to objectify their
interests, abilities, and values. Finally, clients progress step four, which
involves subjective self-assessments that identify life themes and patterns
(Super et al., 1996).
In formulating the C-DAC model, Super identied a comprehensive test
battery to provide counselors with exibility in carrying out comprehensive
career assessments. However, he emphasized four core measures for the
C-DAC battery. The rst is The Salience Inventory, a 170-item questionnaire
that measures the extent to which individuals participate in, commit to, and
expect to realize values in ve life roles: student, worker, citizen, homemaker
(including spouse and parent), and leisure. The second core measure is
the Adult Career Concerns Inventory (ACCI; Super, Thompson, Lindeman,
Jordaan, & Myers, 1988). Its 61 items assess planning attitudes, an
important dimension of career adaptability (Savickas, 1997). The ACCI
consists of 4 scales and 12 subscales that measure concerns related to career
stages and developmental tasks: Exploration (Crystallizing, Specifying,
Implementing); Establishment (Stabilizing, Consolidating, Advancing);
Maintenance (Holding, Updating, Innovating); and Disengagement
(Decelerating, Retirement Planning, Retirement Living). The third core
measure for the C-DAC battery is the Career Development Inventory (CDI;
Super, Thompson, Lindeman, Jordaan, & Myers, 1979/1981). Its 120
questions capture readiness for making educational and vocational choices.
The CDI has two parts: (I) Career Orientation and (II) Knowledge
of Preferred Occupation. Part I includes four scales that assess Career
Planning (CP), Career Exploration (CE), Career Decision Making (DM),
and World-of-Work Information (WW). Part II contains one scale
measuring Knowledge of Preferred Occupational Group (PO). Three
composite scores result from summing individual scale scores are as follows:
Career Development Attitudes combines CP and CE; Career Development
Knowledge and Skills combines DM and WW; and Career Orientation
Total combines CDA and CDK. The nal core measure Super specied was
the VS, which includes items that assesses 21 intrinsic and extrinsic values
123
people seek in life. The VS assesses values such as Ability Utilization

(e.g., use all my skills and knowledge) and Economic Security (e.g., be
where employment is regular and secure).
Implications of Supers Theory and Work for

Turnover Theory and Research
In this section, we discuss how constructs and developmental processes
(including vocational inventories designed to assess them) proposed by
Super can improve turnover understanding and prediction. Fundamentally,
Supers theoretical approach suggests a developmental trend in ones career
and life and distinct developmentally linked tasks and challenges at different
stages. Each life and career developmental stage as articulated by Super may
create a distinct set of forces that push or pull incumbents away from their
current jobs (Maertz & Griffeth, 2004). Following this developmental logic,
the reasons why individuals quit likely vary by life stages. Along the career
developmental axis in Supers thinking, different developmental challenges
and concerns at each stage can engender different motives for leaving at the
start of a career versus mid-career or toward the end of ones career. While
turnover researchers readily acknowledge that new and established employees may quit for different reasons (Weller, Holtom, Matiaske, & Mellewigt,
2009), they have not fully or explicitly capitalized on Supers insights
for identifying stage-dependent motives. Supers perspective suggests that
rookies in the Exploration career stage (exploring various job options) may
exit because they do not nd that they are a good t for a particular job
(Hom et al., 2008), whereas veterans at the Establishment career stage
exit because they lack sufcient or timely promotional opportunities that
provide a sense of career progress (Taylor, Audia, & Gupta, 1996).
To test this developmental viewpoint, turnover researchers can systematically explore the impact of these developmental tasks and challenges with
Supers Adult Career Concerns Inventory (ACCI). Using this inventory, they
can test the CDAC implication that a person at the Establishment stage
would more readily leave to advance his or her career prospects elsewhere,
whereas someone at the Disengagement stage are more prone to leave
due to boredom or desire to participate in other life roles (by assuming
another less demanding job). Though stages are somewhat age-dependent,
Supers thinking also recognizes that people progress through career stages
at different rates. For example, some people explore various career options
far longer (such as women bearing and raising children before starting their
124
PETER W. HOM ET AL.
ofcial careers) than the typical Exploration stage, whereas others may settle
on a career at a relatively young age. If so, the ACCI would more precisely
diagnose turnover motives of people in the Exploration stage than would
proxies based on job tenure or work history. After all, Supers theory allows
for the possibility of two individuals with the same length of employment
belonging to different career stages.
Among his other conceptions, Supers notion of self-concept implementation can extend Lee and Mitchells (1994) unfolding model (more fully
described below) by dening the content of matching scripts (preexisting
plans to quit) and internal images (i.e., career plans or goals, which path 2
shocks might violate), increasing its precision for predicting when shocks
evoke various withdrawal paths (shown in Fig. 1). According to Super,
people prefer and choose vocations where they can fulll values embodying
Path 1: Personal, Expected Shocks

Activates
Matching Script
Personal
Event
Quit
Matching
Script
Self-Concept Implementation
Path 2: Negative Workplace Shocks

Negative
Job Event
Image Violation
Of Values Or Goals
Self -Concept Implementation

RIASEC Vocational Fit
Judgement of
Misfit
Quit
Career Images:
Career Plans or Goals
Path 3: Unsolicited Job Inquiry Shocks

Unsolicited Job
Inquiry/Offer
Compare with Current Job

May Pursue Other Jobs
Career
Adaptability
Quit for Better Job
Self-Concept Implementation
RIASEC Vocational Fit
Path 4: Dissatisfaction-Induced Turnover

Gradual
Job Misfit
Job
Dissatisfaction
Seek Other
Jobs
Compare
Job Offers
to Present Job
Poor Vocational Fit
Fig. 1.
How Career Constructs Inuence the Unfolding Model.
Quit
125
their self-concepts. Vocational self-concepts may underpin matching scripts

and images and whether or not they can be implemented in a workplace
may activate various turnover paths. In T. W. Lee et al.s (1996) study, for
example, only nurses who viewed themselves as primarily family caregivers
rather than worker bees would formulate plans to quit when becoming
pregnant (i.e., the personal shock allows them to implement their selfconcept of motherhood in Fig. 1), while those nurses whose vocational
self-concepts were based on their capacity to deliver individualized patient
care would leave when the hospital introduced team-based nursing (i.e., the
negative workplace violates their self-concept implementation in Fig. 1).
Further, Supers construct explains why employees are responsive to
another shock in the unfolding model: unsolicited job inquiries. According
to Fig. 1, job incumbents may leave because they receive an unsolicited job
offer that permits them to better implement their self-concept.
Moreover, career maturity may help clarify why young workers (Griffeth
et al., 2000), new hires (Hom et al., 2008; Weller et al., 2009), and hobos
(Hulin, Roznowski, & Hachiya, 1985) are exit-prone as they may poorly
prepare for jobs or make bad job choices. In contrast, higher career maturity
may account for why older people are more steadfast employees (Griffeth
et al., 2000) and for Booth, Francesconi, and Garcia-Serranos (1999)
observation that British workers quit their fth job less than they do their
rst job. Conceivably, older employees experiences tackling developmental
tasks in advanced life stages or career stages helped them make wiser more
job choices, promoting their greater job stability (Wanous, 1992). Further,
career maturity may explain quit decisions among older adults entering or
reentering the workforce, such as women who raised children, former
members of institutions (e.g., prison, military, priesthood; Ebaugh, 1988), or
those switching careers (who might return to school to learn new skills).
They would undergo (or repeat) Exploration and Establishment stages for
a new career later in life than younger people. Yet they may show greater
career maturity about planning for and choosing jobs (improving job
longevity) as they have successfully surmounted development hurdles of
earlier life stages or career stages in a previous occupational eld.
More research is warranted on the complex relationship between
career adaptability and turnover uncovered by Ito and Brotheridge
(2005). While reducing desirability of movement (March & Simon, 1958)
because adaptable employees may nd greater success and satisfaction
in jobs than their less adaptable counterparts, they also likely have greater
ease of movement (Ito & Brotheridge, 2005). We thus suggest that career
adaptability can broaden the conceptual scope of movement capital or
126
PETER W. HOM ET AL.
the human capital job incumbents possess that enhances job mobility
(Trevor, 2001) to encompass perceived control over ones vocational plans
and self-condence to pursue ones career goals. Although initially dened
in terms of education and occupational skills (Trevor, 2001), movement
capital might also include career adaptability, which enable incumbents
to exit for better career opportunities elsewhere (Ito & Brotheridge,
2005). Additionally, people high in career adaptability may attract more
unsolicited job offers as employers may believe that they can better manage
dynamic and wide-ranging challenges in the hypercompetitive global
marketplace (see Fig. 1).
Further, Supers thoughts about the life career rainbow and how career
or work salience uctuates across life stages can further explicate the
organizational withdrawal process. Multiple competing life roles have rarely
been examined in turnover research (Hom & Griffeth, 1995), though some
authors have alluded to work centrality (Mobley, 1982; Mobley, Griffeth,
Hand, & Meglino, 1979). In particular, the life career rainbow may illuminate
womens higher attrition in female-dominated jobs (e.g., nursing; Hom &
Griffeth, 1991) and male-dominated elds or workplaces (Hom et al., 2008).
Turnover scholars often recognize how womens greater domestic responsibilities (e.g., kinship responsibilities; Price & Mueller, 1981) can impel their
departures but fail to realize that such obligations change across life stages.
When starting new careers in their 20s (Establishment stage), many women
begin bearing children before their biological clock runs out (Booth et al.,
1999). During this period of simultaneous childbearing and career establishment, young women face intense work family conict especially in
professional service rms (e.g., law and accounting rms; Dalton, Hill, &
Ramsay, 1997) and research universities (Fox, 2005) that impose rigorous and
xed up-or-out promotion systems. Turnover experts have noted that such
interrole conict can drive quits (Hom & Kinicki, 2001) but have not formally
recognized how such conict varies across womens life stages, waxing during
Career Establishment (for young mothers beginning careers) and waning
during Career Maintenance (when children have grown up). Indeed, such
implicit static views overstate motherhood as an intrinsic (or deterministic)
explanation of womens higher quit propensity (Fox, 2005). By the same
token, turnover theorists cannot account for why some men take time off
from work or switch to less challenging careers to spend more time with
family or why some leave organizations to have a more balanced investment
of energy across multiple life roles (Hom & Kinicki, 2001; Hulin et al., 1985).
In short, turnover scholars increasingly acknowledge that some leavers exit
the workforce but offer few theoretical explanations for nonemployment
127
destinations other than search unemployment ( job pursuits after leaving;

T. H. Lee et al., 2008).
Further, Supers developmental perspective suggests that turnover scholars
should differentiate between life-cycle and career-stage effects. Though both
are age-dependent, they may exert differential inuence on leaving. Using
chronological age and work experience as rough proxies for life and career
stages, respectively, Booth and associates (1999) observed that age and work
history (e.g., date of entry into the labor market, number of previous parttime jobs) have distinct unique effects on turnover. As mentioned above,
some people may begin a career or a second one later in life (perhaps they
raised children or retired from the military). As a result, 40-year-old new
career entrants may have similar developmental challenges and concerns
about becoming established incumbents as 25-year-old entrants. Nonetheless, their different life stages can affect their progress through career
stages. Because their career spans are truncated, 40-year-old newcomers
may feel greater urgency to move through Career Establishment and
Maintenance stages more quickly than 25-year olds, while more expertly
negotiating developmental tasks if they had transitioned through these career
stages before in a prior career. Forty-year-old women starting new careers
after motherhood may also move through career stages more rapidly and
easily than 25-year-old working mothers because they are liberated from
the responsibilities of household formation. Such developmental speed may
strengthen their loyalty to employing institutions that offer timely career
opportunities (e.g., more rapid promotions; Taylor et al., 1996). At the same
time, mature rookies may face greater age discrimination (slower or denial of
promotional or career opportunities), which would impede their developmental progress and thus inspire them to quit (Johnson & Neumark, 1997).
JOHN HOLLANDS PERSONENVIRONMENT

THEORY
Career choice constitutes a major stream of inquiry for vocational
psychologists. Parsons lay the theoretical groundwork for this research
stream in 1909 by proposing a model of how individuals choose careers
(Tracey & Rounds, 1993). In choosing an occupation, an individual assesses
his or her abilities, interests, goals, and means, analyzes what a particular
occupation requires and offers in return, and considers the analysis
of the self in conjunction with the analysis of a potential occupation
128
PETER W. HOM ET AL.
(Parsons, 1909). Expecting vocational interests to play a role in choice of,

satisfaction with, and performance in occupations, vocational researchers
(e.g., Kuder, 1939; Strong, 1943) constructed interest inventories that would
serve as tools for those attempting to choose occupations as well as those in
positions to advise others making occupational choices (Tracey & Rounds,
1993). Concomitantly, analysis of occupations has been made possible via
publically available vocational information. Sources include the Occupational
Outlook Handbook (U.S. Department of Labor, 2008) and the Dictionary of
Occupational Titles (U.S. Department of Labor, 1991). John Hollands theory
of vocational choice, rst introduced in 1959, provided a means of integrating
self-assessment with occupational assessment.
Hollands Central Theoretical Constructs

John Hollands theory of careers, a seminal inuence in vocational
psychology, relates personality to career choice. One of its main attractions
is arguably its ability to convey a great deal of information via a fairly
parsimonious framework (Holland, 1996). Holland theorized that occupational choice is inuenced by an individuals characteristics, including
personality, ability, and knowledge (Spokane, Luchetta, & Richwine, 2002).
Holland proposed six personality types: realistic, investigative, artistic,
social, enterprising, and conventional. These six personality types and their
hypothesized relationships to one another are summarized in what is
referred to as the RIASEC model. The model is presented visually as a
hexagon. Any given individual is expected to exhibit aspects of the majority
or all of the six personality types, but in varying degree (Spokane et al.,
2002). A letter code summarizes an individuals combination of personality
types. The highest three letters correspond to an individuals most dominant
personality types. In the context of career advisement, these letters are used
to suggest careers that are appropriate given the individuals personality
prole. The organization of career environments along the same RIASEC
model as individuals personalities makes this matching possible.
Hollands theory also features four dimensions: congruence, consistency,
differentiation, and identity. Congruence constitutes how well aligned
an individuals personality is with his or her work environment (Spokane
et al., 2002). Given how obviously broad this conceptualization is, other
researchers have reasoned that several types of congruence may exist,
including avocational congruence (Meir, Melamed, & Abu-Freha, 1990),
skill-utilization congruence (Meir et al., 1990), and within occupational
129
(i.e., occupational specialty) congruence (Meir & Erez, 1981). Differentiation

is the degree to which an individuals interests are spread out or unied
(Weinrach & Srebalus, 1990). Said differently, it reects the extent to which
an individual resembles a single personality type and shows little similarity
with the other personality types (Meir, Esformes, & Friedland, 1994).
Differentiation is calculated by taking the difference between the highest
score and the lowest score of the six (or highest three) types. Consistency
refers to the strength of the relationship between an individuals
dominant two personality types (Spokane et al., 2002). Types that are
positioned next to each other along the hexagon are more consistent than
types positioned farther apart. For example, investigative and artistic types,
neighbors along a side of the hexagon, share the desire for intellectual
stimulation (Furnham, 1994), while realistic and social types, which sit
opposite one another on the hexagon, differ in terms of the former having an
orientation toward things and the latter having an orientation toward people
(Tracey & Rounds, 1997). Though we do not focus on the consistency
dimension in what follows, we mention it here for the sake of completeness.
The nal dimension of Hollands theory, vocational identity, represents
how well formed and stable an individuals goals, interests, and talents are
(Spokane et al., 2002). Before continuing on, we must distinguish Hollands
identity dimension from the identity constructs based on social identity
theory (SIT; Tajfel & Turner, 1979) that are typically studied by career as
well as turnover theorists. As Grote and Raeder (2009) points out, identity
in career research is used interchangeably with concepts like sense of self or
self-concept. Turnover theorists typically split identity into personal identity
and social identity with the former pertaining to personal attributes like
dispositions and abilities and the latter having to do with group categories
such as ethnicity, gender, and company that individuals may identify with
(Mael & Ashforth, 1995). As Randsley de Moura and colleagues (2008)
point out in their literature review, SIT (e.g., the construct of organizational
identity, or perceived oneness with an organization [Mael & Ashforth,
1992, p. 103]) has been used to make and test predictions about a slew
of organizational variables, with turnover being one of them. By contrast,
Hollands vocational identity construct referring to the stability and
crystallization of vocational interests and goals has not been thus applied
by turnover scholars (cf. Hom, 2010).
In addition to laying out the four dimensions described, Hollands theory
offers several basic premises. One premise is that individuals have a
preference for environments that are conducive to the use of their abilities
and skill sets, offer engaging problems and work roles, and permit
130
PETER W. HOM ET AL.
expression of their true attitudes and values (Weinrach & Srebalus, 1990).
A second premise is that an individuals behavior (e.g., career choice, job
changes) is a product of the interaction of his or her personality and
environment (Weinrach & Srebalus, 1990). The next premise says that
individuals experience reinforcement and satisfaction when environments they are in match their personalities. Reinforcing situations make
behavior (e.g., attendance) more stable (Spokane et al., 2002). The fourth
premise addresses situations of an environment-personality mismatch.
Mismatches stimulate behavioral change in order to remove incongruence
(i.e., mismatch). In such circumstances, individuals may search for a
different, congruent work environment or adjust their outlook and behavior
to t their current environment (Spokane et al., 2002).
Vocational-Interest Measurement and Occupational Classication

Vocational interest inventories, like the Strong Interest Inventory (SII)
and Kuder General Interest Survey, that existed and were widely used
before Holland promulgated his theory have been modied to incorporate
Hollands model (Tracey & Rounds, 1993). The SII, which requires
individuals to report their degree of like or dislike for a large variety of
items, proffers proles for respondents at the end. Four sets of scales make
up this prole: general occupational themes, basic interest scales, occupational scales, and personal styles scales. The general occupational themes are
based on Hollands RIASEC model and inform the respondent about his or
her work personality. The basic interest scales separate the aforementioned
general occupational themes into work, school, and leisure activities.
The occupational scales direct the individuals attention to occupations he
or she is expected to t well with based on having received the highest score
for these occupations. The personal styles scales have to do with information about the individuals style in various domains (e.g., work, learning,
leadership, and team participation).
Hollands own inventory, the Self-Directed Search (SDS), presents
respondents with approximately 230 items that pertain to activities,
competencies, occupational preferences, and abilities. Individuals indicate
preferred and disliked activities, activities in which they feel efcacious
and not, interesting and uninteresting occupations, and rate their skills
and abilities. The prole presented to the respondent shows a three-letter
Holland code, which the individual can use to identify careers within a
provided list that match his or her interests and competencies.
131
Vocational Congruence: Methodological Issues and Moderators

Hollands thesis that congruence between ones interests and occupation
should lead to more satisfaction, success, and stability in the chosen
occupation has received much attention in the career-related literature
(e.g., Chartrand & Walsh, 1999). Hollands congruence dimension already
has analogous constructs in the turnover literature. Rather than vocational
t, Chatman (1991) examined how value congruency or how closely an
employees values t with those of the employer induces loyalty. Mitchell
et al. (2001b) also theorized a t dimension in their model of why people
stay in their jobs and dened it as an employees perceived compatibility or
comfort with an organization and with his or her environment (p. 1104).
Their conceptualization is more encompassing than Hollands; two types of
t are considered: organizational t and community t. Yet Mitchell and
Lees (2001) description of on-the-job t fails to explicitly specify vocational
interest t as an embedding force.
The extensive research on congruence has not, however, produced
conclusive ndings; researchers do not always nd the anticipated links
between congruence and outcomes (e.g., Perdue, Reardon, & Peterson,
2007). In their meta-analysis, Assouline and Meir (1987) found correlations
between congruence and satisfaction that range from 0.09 to 0.51, though
reporting a mean corrected correlation of 0.21. Two later meta-analyses
by Tranberg and colleagues (1993) and Tsabari and colleagues (2005)
estimated correlations of 0.174 and 0.166, respectively. In short, congruence
is reliably, though modestly, correlated with satisfaction. We next review
methodological problems that may be responsible for such weak or modest
relationships. To capitalize on Hollands theory as a framework for better
understanding job satisfaction, turnover researchers should design studies
that minimize these problems inuence to give a clearer picture of how
congruence affects satisfaction (and retention).
The lack of consistency and clarity in ndings about congruence is partly
attributed to several methodological shortcomings (e.g., Chartrand &
Walsh, 1999). Tracey (2007) proposes that self-selection may partly account
for weak relationships between personenvironment t and outcomes. That
is, individuals choosing to enter environments that initially t their interests
fairly well and incongruent workers leaving their jobs would limit variance
in congruence, attenuating its effects (Spokane, Meir, & Catalano, 2000;
Tracey, 2007; Tsabari, Tziner, & Meir, 2005). Inadequate measurement
of personenvironment congruence namely, poor assessments of peoples
personalities (or interests) and the environment also weakens empirical
132
PETER W. HOM ET AL.
support (e.g., Chartrand & Walsh, 1999). Environmental measurement has

proven especially difcult. Ambiguity persists over which environmental
levels (e.g., social, cultural) (Furnham, 2001) should be measured and
what the unit of analysis should be (e.g., job title, job tasks, job clusters)
(Chartrand & Walsh, 1999). Thus, adequate methods of measuring the
environment remain lacking (Chartrand & Walsh, 1999).
How job satisfaction is assessed can also impact the congruence
satisfaction relationship. The common use of an overall satisfaction score
can obscure this relationship because an overall satisfaction score
incorporates additional factors beyond how an individual feels about his
or her work activities (e.g., pay, coworkers; Chartrand & Walsh, 1999).
Whether researchers measure actual t (indirect t indices based on
comparisons of separately assessed personal and environmental attributes;
Kristof-Brown, Zimmerman, & Johnson, 2005) or perceived t (direct assessments of compatibility; Mitchell & Lee, 2001) makes a difference as well.
Because it refers to individuals perceptions of how well they t their jobs or
organizations, perceived t is likely more proximal to attitudes and behavior
than is actual t (Cable & DeRue, 2002; Mitchell & Lee, 2001). Sustaining
this view, Kristof-Brown and colleagues (2005) found in their meta-analysis
that perceived t relates more strongly to outcomes (e.g., quit intentions, job
satisfaction, organizational commitment) than does actual t.
Apart from methodological inuences, some researchers have identied
moderators that can affect relationships between interests-job t and job
outcomes (Spokane et al., 2000; Tracey, 2007). In the career literature, much
work has already been done to identify moderators. Group importance
has been shown to moderate the congruencesatisfaction relationship,
such that the correlation between congruence and satisfaction is stronger
when a group is important to an individual (Meir, Keinan, & Segal, 1986;
Meir, Hadas, & Noyfeld, 1997; Meir & Green-Eppel, 1999; Meir, Tziner, &
Glazner, 1997). Age or life stage can moderate congruence effects as well.
A longitudinal study of the work lives of British workers found that average
job tenure increased as jobs accumulated (Booth et al., 1999). Mature
workers stayed longer on their fth job than on their rst job. A metaanalysis by Tsabari et al. (2005) nonetheless concluded that the congruence
satisfaction relationship was actually stronger for those in the 2030 age
group than for those over 30. Their nding implies that congruence matters
less for older workers that older individuals tend to be more satised than
younger ones with their circumstances even if they t the job less (which
accords with the role theory of aging that self-integration, insight, and
positive psychosocial traits grow with age; Yang, 2008). In line with this,
133
Whitbourne (1986) documented age to be inversely related to identity

exibility, dened as deliberate and informed comparison of ones present
identity commitments with other possibilities (p. 164). In other words,
older workers engage in less thinking about alternative commitments (and
prospects of better job t elsewhere). Further support for this idea comes
from research on career stages, which suggests that earlier career stages are
characterized by exploration of different options while later stages are
characterized by stability (Brousseau, 1983).
Further, the time lag between assessments of vocational t and outcomes
may moderate their relationships (Tracey, 2007). Vocational scholars
implicitly dene vocational congruence as a static construct, assuming
stable or xed personal and environmental attributes that can be captured
on one occasion. They typically assess individuals interests at one point in
time and later compare these interests to the occupations (or academic
majors) these individuals choose. Such static research designs fail to
consider how congruence might change over time (Low & Rounds, 2007).
Indeed, Hollands theory posits that t between the individual and his or
her environment is dynamic and likely to change with experience and
progression into new stages of career (e.g., Brousseau, 1983). Furnham
(2001) further elaborates on the dynamic nature of t. Individuals change to
adapt to their environments (e.g., altering personal work style) and can
to some extent alter work environments (e.g., changing the way the job is
done) to better t their needs. As a result of these adaptations, congruence
improves over time. In support, some studies reveal that congruence
increases over time (Meir & Navon, 1992; Tracey, Robbins, & Hofsess, 2005
as cited in Tracey, 2007). That said, measuring t at one point in time does
not provide an indication of how that t is likely to evolve, which some
turnover scholars are acknowledging (Lee & Mitchell, 1994).
Further, Hollands differentiation dimension, which relative to congruence
has received much less scholarly scrutiny (Meir et al., 1994), may impact
associations between interest-occupation congruence and outcomes. Low
differentiation (or more interest exibility) may result in congruence being
less able to predict positive outcomes like job satisfaction. Conceivably,
employees whose interests are less well formed (less differentiated) are better
able to deal with and more willingly grow into occupations that are imperfect
ts with their primary interests, while those whose interests are well-dened
may be less tolerant and willing to adjust to occupations that are not good
ts (Meir et al., 1994). Darcy and Tracey (2003) sustain this moderating
effect, reasoning that individuals with more exible interests have a wider
range of likes than those with less exible interests and thus are able to
134
PETER W. HOM ET AL.
substitute interests that an occupation cannot satisfy with other interests

that it can satisfy. Those with less exibility, however, are more likely to
experience negative outcomes if the chosen occupation does not offer
interesting activities because of their inability to substitute.
Attesting to this idea, Wessel, Ryan, and Oswald (2008) considered
adaptability, a construct which is akin to exibility, as a moderator of the
effects of t in the context of students selection of majors. Adaptability is
willingness and predisposition to adjust well to a changing environment.
Among students who perceived low t with their major, those who were
highly adaptable were more satised with their educational institution than
those who were low on adaptability. For students with high perceived t,
adaptability level did not make much difference in institutional satisfaction.
This nding is consonant with Furnhams (2001) discussion of t as being
subject to change when individuals proactively change themselves or their
environment. Wessel et al.s ndings imply that adaptive individuals who
mist their environment did something to adapt and to thereby remain
satised with their environment.
Finally, as suggested by Spokane (1985), vocational identity (which is
typically measured via Holland, Daiger, and Powers (1980) vocational
identity scale) could prove to be another moderator of congruenceoutcome
relationships; among those with a better sense of identity interest
occupational congruence should more strongly predict occupational outcomes. Healy and Mourton (1985) observed that for women college students
whose interests matched their occupational choice, those high on vocational
identity were more decided and had more general career information.
(No relationships were found for male students). Carson and Mowsesian
(1993) tested Spokanes (1985) idea with a sample of employed adults, using
job satisfaction as the outcome. They did not nd identity to moderate
the congruencesatisfaction relationship. We suggest, however, that further
examination of identity as a moderator is still warranted.
Alternatively, akin to congruence, vocational identity may be a predictor of
job satisfaction in its own right. To illustrate, Carson and Mowsesian (1993)
found that both congruence and vocational identity were signicantly
correlated with job satisfaction. The identitysatisfaction relationship
(r 0.45) was actually stronger than the relationship between congruence
and satisfaction (r 0.18). Commenting on the robust relationship
between vocational identity and job satisfaction (e.g., r 0.70; Holland &
Gottfredson, 1994), Holland (1997) placed particular emphasis on vocational
identity when considering the interaction of individuals with their environments. He described the construct as being applicable not just to individuals
135
but to work environments as well (environmental identity) as an organization can vary in how clear and temporally stable goals, tasks, and rewards are.
He predicted that the interaction of an individual with a well-formed identity
and an environment with a clear identity will make for more predictable
interaction of the two than will be the case for the interactions of individuals
and environments with identities that are less well formed or clear.
Holland (1997) thus suggests that Vocational Identity in conjunction
with the rest of the theory provides a simple and plausible explanation
of career stability or instability (p. 173). Individuals with well-dened
identities have good understanding of what they want (e.g., their goals and
interests) and can offer (e.g., their talents and skills), which increases their
likelihood of selecting jobs that t their prole and of persisting with a job
search until nding a good match. Those with ill-dened identities, on the
other hand, are more liable to choose poor-tting work environments, and
consequently, to engage in job-hopping (Hom et al., 2008). Holland (1997)
argues that some support for these predictions can be drawn from the strong
relationship between vocational identity and job satisfaction, since job
(dis)satisfaction is a well-established antecedent of voluntary turnover.
We must note, however, that issues of measurement creep up here as well.
Carson and Mowsesian (1993) suggest that the relationship between
vocational identity and job satisfaction could be inated due to some
overlap between job satisfaction and what Holland et al.s (1980) vocational
identity scale measures. In summary, though concepts of personenvironment t have been incorporated into some theories about attrition
(Chatman, 1991; Mitchell & Lee, 2001), further theoretical gains are
possible based on elements of Hollands theory. Going forward, turnover
researchers need to address the methodological issues in how (vocational)
t is studied and consider moderators when examining how t drives job
satisfaction and the withdrawal process (Hom & Kinicki, 2001; Lee &
Mitchell, 1994; Mobley et al., 1979; Price & Mueller, 1986; Rusbult &
Farrell, 1983) to best capitalize on Hollands theory. In what follows, we
further describe how his model and measures of its constructs can enrich
theoretical perspectives and research on attrition.
Implications of Hollands Theory for Turnover Theory and Research

Traditional Turnover Models
In this section, we summarize the insights Hollands (1997) theory
offers into causes of turnover and how the turnover process unfolds.
136
PETER W. HOM ET AL.
Traditional models of turnover assume job (dis)satisfaction to be a prime

determinant of turnover (e.g., Mobley et al., 1979; Price & Mueller, 1981,
1986). In contrast to this prevailing preoccupation with proximal causes of
leaving, Hollands theory points to poor vocational choices as a potential
distal cause of job dissatisfaction (and turnover). Conversely, good (in terms
of t) vocational choices should result in positive work outcomes. Though
empirical support for this prediction is mixed, methodological deciencies
(e.g., constrained congruence variance due to self-selection, poor environment measures, global satisfaction indices) in tests and inadequate attention
to moderators likely understated evidence for congruenceoutcome links
(Assouline & Meir, 1987; Tranberg, Slane, & Ekberg, 1993; Tsabari et al.,
2005). By addressing these methodological shortcomings (including
applying Edwards (2002) approach for analyzing difference scores), we
believe that turnover investigators can more rmly demonstrate that
prior vocational choices can shape employees affect and loyalty to their
job (attesting to Hollands theory). By so doing, they can extend
conventional withdrawal formulations that primarily scrutinize the effects
of job satisfaction: how it shapes quit decisions in combination with
determinants, or how mediators convey its inuence onto quitting (Hom &
Kinicki, 2001; Lee & Mitchell, 1994; Rusbult & Farrell, 1983; Steers &
Mowday, 1981). Though Price and Mueller (1981, 1986) specied a broad
array of satisfaction causes, their attention to workplace causes leaves out
more distal causes of satisfaction (and leaving), such as vocational decisions
that occur before job entry. All told, we contend that job satisfaction serves
as the link between Hollands theory (notably, vocational congruence), for
which job satisfaction is a key outcome of interest, and many turnover
theories, for which job satisfaction is a key attrition driver.
The Unfolding Model
What is more, Hollands formulation offers ways to rene newer
nontraditional turnover models (e.g., Lee & Mitchell, 1994; Mitchell &
Lee, 2001), which downplay the centrality of job satisfaction proffered in
older schools of thought (Mobley et al., 1979; Price & Mueller, 1981, 1986).
To illustrate how, we briey summarize Lee and Mitchells (1994) unfolding
model. It proposes four pathways to turnover, three of which do not
involve dissatisfaction (see Fig. 1). In the rst three paths, jarring events
(shocks) initiate the turnover process; in the fourth path, job dissatisfaction is the catalyst, as in conventional models of turnover. In one path
(Fig. 1, path 1), the shock is typically a personal, nonwork event, such as a
graduate-school admission or pregnancy, which activates a preexisting plan
137
(matching script) to quit (what Maertz & Campion, 2004 refer to as

preplanned quits). For example, T. W. Lee et al. (1996) observed that a
nurse learning she was pregnant soon resigned; this shock triggered her plan
to opt out of the labor market. Negative workplace shocks activate another
turnover path (Fig. 1, path 2) when they violate employees values or career
goals (known as image violation). Employees determine whether or not
the shock can be integrated into their values or goals. If not, they exit
without jobs in hand. T. W. Lee and associates (1996) gave the example of
a nurse leaving immediately when the hospital shifted from individualized
patient care her preferred nursing philosophy to team-based nursing.
Unsolicited job offers or inquiries represent a third type of shock
prompting a third withdrawal path (path 3 in Fig. 1). To illustrate, T. W. Lee
et al. (1996) noticed that a nurse quit a hospital when a physician offered her
another position. Leavers taking this path are not unhappy with their current
job; they simply prefer a better alternative. Finally, Lee and Mitchell (1994)
designated a fourth path (path 4 in Fig. 1) to represent the conventional
withdrawal path envisioned by traditional theorists (Hom & Kinicki, 2001;
Price & Mueller, 1986; Steers & Mowday, 1981), in which dissatised
employees pursue other jobs and exit when securing superior ones.
We argue that Hollands theory can rene the unfolding model in several
ways. Earlier we discussed how Hollands views suggest additional
underpinnings of job satisfaction, which can further explain the etiology
behind path 4 in the unfolding model as dissatisfaction is its prime mover.
Though Mitchell et al. (2001a) estimated that only 37% of leavers take
path 4, T. H. Lee et al. (2008) recently determined that 60% of all leavers
from a nationally representative sample left due to dissatisfaction. Fig. 1
thus shows poor vocational t as worsening job t over time, which
in turn increases dissatisfaction and turnover path 4 in the unfolding model
(Lee & Mitchell, 1994). Moreover, incorporating Hollands idea that
individuals have occupational orientations (i.e., realistic, investigative,
artistic, social, enterprising, conventional) into the unfolding model can
clarify why negative workplace shocks in path 3 motivate some employees to
leave (i.e., shocks clash with their occupational orientation). To illustrate,
business professors initially joining research-oriented business schools
because they prefer investigative activities (the I in the RIASEC model)
may face image violation when their colleges later stress vocational
training to attain higher MBA rankings (demanding and rewarding MBA
teaching; S in the RIASEC model) (Morgeson & Nahrgang, 2008). Fig. 1
thus species that violation of RIASEC vocational t can engender turnover
path 2. Likewise, Hollands perspective can elucidate how unsolicited job
138
PETER W. HOM ET AL.
offers (path 3 shocks) impel departures. Using the same running example,
I-oriented business faculty may abandon their academic posts because other
universities recruit them away with the lure of superior opportunities to
fulll I preferences (e.g., supervising PhD candidates rather than teaching
MBA students). In short, alternative jobs represent better matches for their
occupational orientation. Fig. 1 thus shows an inuence of RIASEC
vocational t on employees consideration of alternative job options that
offer better t with occupational preferences.
Job Embeddedness Theory

Departing from earlier viewpoints (including the unfolding model) focusing
on why people leave, Mitchell and colleagues (2001b) promulgated a novel
perspective about why people stay. They came up with a new construct,
called job embeddedness, to explain why some employees are less
inclined than others to leave. They proposed that certain forces embed
employees in their jobs: links (i.e., connections) to ones organization and
larger community, perceived sacrices (i.e., costs) of leaving, and t (or
compatibility) with ones organization and larger community (see Fig. 2).
Links, sacrices, and t combine to cause embeddedness and can combine
differently for different people but still result in equal levels of embeddedness (Mitchell & Lee, 2001). Empirical tests nd that embeddedness
accounts for unique turnover variance beyond attitudes and alternatives
(Mitchell et al., 2001b), encourages higher performance and citizenship
(Lee, Mitchell, Sablynski, Burton, & Holtom, 2004), retains employees by
VOCATIONAL
IDENTITY
VOCATIONAL FIT
RIASEC Match
ON-THE-JOB
LINKS
ON-THE-JOB
FIT
OFF-THE-JOB
LINKS
JOB
EMBEDDEDNESS
OFF-THE-JOB
FIT
VOCATIONAL
DIFFERENTIATION
ON-THE-JOB
SACRIFICE
Fig. 2.
OFF-THE-JOB
SACRIFICE
How Career Constructs Inuence Job Embeddedness.
139
embedding their coworkers (Felps et al., 2009), and mediates how highcommitment human resources management systems affect quit decisions
and rm commitment (Hom et al., 2009).
An obvious point of intersection between Hollands theory and job
embeddedness is the organizational t construct in Mitchell and Lees (2001)
model. Embeddedness theorists describe this construct as a compilation of
various other t constructs considered in previous literature. The broader
nature of their construct is reected in the items used to assess on-the-job t.
Items ask about liking of work group members, similarity to coworkers,
utilization of skills and talents on the job, quality of match with company, t
with company culture, and satisfaction with the level of authority and
responsibility granted (Mitchell et al., 2001b). Though this conceptualization
of t parallels Hollands congruence construct, some elements of congruence
as described by Holland are arguably omitted. Measures about the match
between ones career interests and job, ability to express ones attitudes
and values, and interest level in ones job roles, may be usefully added.
Indeed, other authors have long advocated for more comprehensive
conceptualization and measurement of personenvironment t, arguing that
consideration of individual variability in the types of t people nd most
important can best elucidate how t affects outcomes of interest (Piasentin &
Chapman, 2006). Along with composition of more thorough measures of
organizational t, expansion of the conceptual domain of the organizational
t dimension to encompass vocational congruence is thus a viable direction
for further development of the embeddedness construct. That said, Fig. 2
represents this proposition with a pathway from RIASEC match to
on-the-job t: Greater vocational congruence yields higher on-the-job t.
Also missing from job embeddedness theory and research is consideration
of individual differences that, given equal levels of links and sacrices,
may lead to the same level of t being unequally embedding for different
individuals. We contend that in this regard, Hollands differentiation
construct can make a theoretical contribution to job embeddedness theory.
As we discussed above, persons with low differentiation of interests are
more willing to adapt to situations of low organizational t. Adaptable
individuals whose vocational interests are exible may ultimately experience
a high overall level of embeddedness despite an initial experience of low
organizational t. Over time, these individuals may adjust things about
themselves or their environment so as to attain higher on-the-job t. For
example, Vandenberg and Nelson (1999) suggest that certain organizational
cultures may be amenable to employees taking action to eliminate a source
of discomfort, such as asking to be transferred to a different supervisor.
140
PETER W. HOM ET AL.
Thus, embeddedness researchers might acknowledge the dynamic nature of

(on-the-job) t and assess the level of individuals job embeddedness at more
than one point in time. Turnover research will need to move beyond the
standard practice of surveying employees at one point in time and obtaining
quit data at a second point in time because this static design misses the
dynamic character of certain antecedents (e.g., t) (Steel, 2002). Indeed,
proponents of the unfolding model recognize that incorporating time into
the withdrawal process can resolve ambiguities about the relative duration
of different withdrawal paths (e.g., time lapse between initial deliberations of quitting and its enactment) and can address potential switching
between decision paths over time (T. W. Lee et al., 1996; Lee, Mitchell,
Holtom, McDaniel, & Hill, 1999). Fig. 2 thus species that vocational
differentiation can moderate the inuence of vocational t on on-the-job t.
Vocational congruence might be initially low (as well as on-the-job t)
but those low on vocational differentiation may proactively develop higher
on-the-job (and overall) t.
Finally, another individual difference variable that can enrich job
embeddedness thinking is vocational identity (Holland, 1997). Based on
the aforementioned rationale that persons with well-dened vocational
identities can more effectively choose or secure jobs that match their career
prole (Holland, 1997), we argue that such employees will have greater
on-the-job t. After all, empirical research reveals that such incumbents tend
to feel higher job satisfaction (Carson & Mowsesian, 1993; Holland &
Gottfredson, 1994). Consequently, Fig. 2 adds vocational identity as
another antecedent of on-the-job t.
Future directions for nontraditional turnover models are already evident.
Going beyond qualitative tests based on leavers retrospective accounts
about how shocks induced them to quit (T. W. Lee et al., 1996, 1999),
Kammeyer-Mueller, Wanberg, Glomb, and Ahlburg (2005) furnished
additional corroboration for the unfolding model by demonstrating that
shocks measured before turnover occurrence can predict future turnover.
Embeddedness theorists encouraged research integrating the unfolding
model with the construct of job embeddedness, such as testing whether job
embeddedness moderates the effects of shocks (Holtom & Inderrieden, 2006;
Mitchell & Lee, 2001). In that light, Burton, Holtom, Sablynski, Mitchell,
and Lee (2010) recently showed that low levels of job embeddedness do
make individuals more vulnerable to shocks they encounter. More than this,
Holtom and associates (2008) prescribe applying social network methodology to capture more fully the impact of embedding links. Adding to this
laundry list of suggestions, we urge further renements of the unfolding
141
model and embeddedness theory by incorporating the insights and

methodology of Hollands approach to vocational congruence.
SOCIAL COGNITIVE CAREER THEORY

SCCT is a predominant formulation for explaining career interests,
choice, and preparation based on Banduras (1977, 1982, 1989, 1991) social
cognitive theory (SCT). These theoretical views promulgate self-efcacy or
beliefs in ones capabilities to organize and execute the courses of
action required to produce given attainments (Bandura, 1997, p. 3) as
a fundamental determinant of self-regulating motivation and behavior.
Self-efcacy beliefs can affect choices people make, their perseverance in
goal pursuits during setbacks, whether their thoughts facilitate performance,
their experienced stress when coping with taxing stressors, and their
vulnerability to depression.
Hackett and Betz (1981) and Betz and Hackett (1981) initially applied
Banduras early SCT theory to account for womens career development.
Compared with men, they claim that women fail to fully utilize their
capabilities, talents, and interests in vocational pursuits because their careerrelated self-efcacy expectations are lower (magnitude), weaker (strength or
durability), and less generalized (generality). In their view, such poor selfefcacy perceptions contribute to persistent and ubiquitous occupational
segregation and pay inequity by discouraging women from entering maledominated elds offering higher pay, status, and inuence. Hackett and Betz
(1981) also noted that the traditional socialization of women has denied them
the foundation on which to self-condence to pursue occupations where
men prevail (Bandura, 1977, 1997). Specically, girls and young women
have fewer opportunities to engage in traditionally masculine activities
( performance accomplishments), see fewer female role models in a wide range
of male careers (vicarious learning), and receive little encouragement to
pursue male-dominated careers (verbal persuasion). Lacking such self-efcacy
bases, women can also feel more anxiety when performing male sex-typed
activities (emotional arousal), which reinforces their sense of inefcacy. Initial
tests of the HackettBetz formulation nd that female undergraduates feel
less efcacious about satisfying educational and performance requirements
for stereotypically male jobs (e.g., engineers and lawyers) than do male
students (Betz & Hackett, 1981) and that experimentally induced failure on
a math task can weaken womens feelings of task self-efcacy and interest
(Hackett, Betz, OHalloran, & Romac, 1990).
142
PETER W. HOM ET AL.
Later, Lent et al. (1994) expanded Hackett and Betzs (1981) theory to
fully incorporate Banduras (1991) other constructs and introduced three
interrelated SCCT models corresponding to three career development
stages: (1) formation of career interests; (2) academic and vocational
choices; and (3) performance and persistence in educational and occupational pursuits. Though domain content varies across stages, all models
share Banduras (1991) core constructs: self-efcacy, outcome expectations
(anticipated behavioral outcomes), outcome values, and goals. Specically,
each model posits that anticipated outcomes of behavior, such as tangible
rewards and self-evaluative outcomes (e.g., anticipated self-satisfaction)
shape interests and goal-directed behaviors. Each model further presumes
that self-efcacy is a more potent behavioral determinant than are outcome
expectations (Bandura, 1997). People refrain from performing an act if they
doubt that they can enact it even when they expect rewards from its
enactment (Lent et al., 1994). Conversely, a strong sense of personal efcacy
can motivate individuals to perform behaviors even if they are uncertain
about the prospects of earning rewards. Indeed, a resilient sense of personal
efcacy is most crucial for tackling difcult behaviors because the road
to success is usually strewn with countless impediments (Bandura, 1997,
p. 126). Moreover, each SCCT model positions self-efcacy as the prime
behavioral cause when the quality of performance guarantees particular
outcomes (Lent et al., 1994, p. 84). For such actions, self-efcacy directly
shapes expected outcomes because people perceiving that they can perform
these behaviors know they will likely attain behavioral outcomes. Finally,
self-efcacy beliefs can affect outcome values. For example, students who
feel inefcacious about their academic abilities devalue academic accomplishments. We next describe how the SCCT perspective explains the three
key phases of career development.
Three-Stage SCCT Models

Vocational Interests
To model how career interests arise, Lent and colleagues (1994) postulate
that children and adolescents begin developing a sense of efcacy about
certain career-relevant activities from doing them and observing others do
them. Through experience or vicarious learning, they also learn to expect
certain outcomes from such activities. More formally, Lent et al.s interest
model submits that self-efcacy perceptions and outcome expectations
(especially anticipated self-satisfaction from meeting internal performance
143
standards; Bandura, 1982) together prompt the formation of enduring

interests in activities (Lent et al., 1994). This particular model embraces
Banduras (1997) view that perceived efcacy creates interests through
engrossment in activities and the self-satisfactions derived from fullling
personal challenges that lead to progressive mastery of occupational
activities (pp. 423424). By comparison, vocational interests are stunted
when young people feel inefcacious about career-relevant activities or
foresee few positive outcomes from them. Lent and associates (1994)
further deduced that emergent career interests would in turn initiate the
setting of goals for greater exposure to career-related activities. Accordingly,
more activity participation and practice can feed back and inuence
SSCT determinants by increasing self-efcacy feelings (as one develops
competency) and expectations for positive outcomes (as one derives selfgratication from meeting internal standards). During their formative years,
adolescents may continually repeat this spiraling cycle until their activity
and career interests stabilize.
Vocational Choice
The second SCCT model by Lent and colleagues (1994) addresses
development of career choices and entry once career interests are formed.
Analogous to their interest model, they theorize that self-efcacy (about
ones success in a career rather than a particular activity), outcome
expectations (about rewards available from a career), and goals (career
decision) shape choice actions, such as actions to implement the career
choice. Thus, individuals who feel condent about succeeding in a particular
career, anticipate desirable career rewards, and have strong (preexisting)
interests in activities associated with this career will form a goal (career
choice) to pursue a particular career direction (choosing a vocation or
declaring a college major). Strong intentions translate into actual career
actions, such as enrolling in a training program, selecting an academic
major, or applying for a job in the occupational eld.
Vocational Implementation
For the career implementation stage, Lent et al.s (1994) third SCCT model
seeks to explain career accomplishments (e.g., course grades) and persistence
(e.g., stability of academic major). According to this model, individuals who
believe that they can effectively perform a career-relevant task and expect
rewards from successful performance would set more challenging performance goals. To illustrate, engineering students who feel condent that they
can achieve high grades in engineering courses and who expect rewarding
144
PETER W. HOM ET AL.
outcomes (e.g., more job opportunities) from stellar grades will establish
high GPA goals. Higher goals then promote greater task performance as
hard goals induce more effort and persistence. Finally, this model postulates
that ability and past performance affect task performance directly as well
as indirectly via self-efcacy and outcome expectations. Sustaining the
three SCCT models, an early meta-analysis by Lent and colleagues (1994)
documented that self-efcacy and outcome expectations positively and
moderately correlate with vocational interests, choice, and performance.
Later research also corroborated these models (Betz & Hackett, 2006),
establishing their validity for predicting academic interests and choice goals
(Lent, Lopez, Lopez, & Sheu, 2008) as well as academic performance and
persistence (Brown et al., 2008).
Though SCCT theorists primarily strive to explain students vocational
and academic choices, performance, and persistence, their viewpoints can
also help account for their workplace behaviors once they completed
occupational training. After all, Lent et al. (1994) proposed that an SCCT
perspective can apply across the career life span and elucidate various
kinds of work adjustment (subsequent to schooling), such as occupational
satisfaction and career change. Given its validity for predicting student
attrition from academic elds of study (Harvey & McMurray, 1994), we
extend SCCT theory to deepen insight into why new employees leave
workplaces. Our theoretical extension thus echoes Banduras (1997) assertion
that the sense of efcacy that newcomers bring and further develop during
the course of their occupational training at the beginning stage of their
careers contributes to the success of [the] socialization process (p. 446).
In the section below, we adapt Lent et al.s (1994) SCCT perspective
especially their performance model to advance understanding of the
termination process among the most exit-prone segment of the workforce:
newcomers (Hom et al., 2008).
SCCT Model of Newcomer Coping and Attrition

Job attrition is primarily concentrated among new hires during their
initial period of employment (Griffeth et al., 2000; Hom et al., 2008; Weller
et al., 2009). They face special challenges when adjusting to new work
settings, such as coping with unexpected stressful events, mastering new job
requirements, and earning coworker acceptance (Ashforth, 2001; Feldman,
1976; Hom, Griffeth, Palich, & Bracker, 1998; Kammeyer-Meuller &
Wanberg, 2003). Such challenges are more daunting if they are also new
145
entrants to the labor market (e.g., graduating students) because workplace

realities (e.g., close supervision, rare feedback, and organizational politics)
differ so much from former school roles (Greenhaus, 1987). Given such
demands, most scholars thus implicate assimilation maladjustment as a key
driver of newcomers premature departures (Allen, 2004; Griffeth & Hom,
2001; Kammeyer-Meuller & Wanberg, 2003).
Despite an extensive body of ndings about maladjustment causes and
remedies (Allen, 2004; Ashforth, 2001; Chan & Schmitt, 2000; Wanous,
1992), only a few studies examined how self-efcacy impacts newcomer
adaptation. For example, socialization researchers report that beginning
employees with a strong sense of efcacy often redene work roles to suit
their preferences (Ashforth & Saks, 2000), while those feeling less selfefcacious can most benet from training or certain socialization tactics
(Jones, 1986; Saks, 1995). All the same, these studies neglect pivotal SCCT
constructs, such as personal goals and outcome expectations (Lent et al.,
1994), which likely codetermine socialization success along with selfefcacy (Bandura, 1997). Moreover, socialization experts (Jones, 1986;
Saks, 1995) mainly attend to new hires beliefs about whether they can
fulll performance requirements, neglecting other kinds of demands crucial
for organizational assimilation (e.g., workgroup integration, managing
stressors; Bandura, 1997). For instance, Saks (1995) only scrutinized new
accountants perceptions about efcacy for documenting audit procedures
and maintaining client relationships. Further, some authors investigated
generalized dispositions, such as desire for control (Ashford & Black,
1996) or proactive personality (Kammeyer-Meuller & Wanberg, 2003). By
comparison, SCCT theorists construe perceived self-efcacy as a state-like
belief about ones capacity to perform in a particular domain of functioning
and contend that such self-beliefs better explain for how one approaches and
performs tasks than personality-like traits (Bandura, 1997). In summary,
socialization tests omitted complimentary constructs put forth by Lent et al.
(1994) that may reinforce or condition how newcomer self-efcacys effects
and operationalized newcomer self-efcacy too narrowly (overlooking other
assimilation challenges) or too broadly (focusing on dispositional traits
rather than motivational states).
To further clarify the process by which newcomer self-efcacy sustains
job survival, we extend Lent et al.s (1994) framework to furnish a more
thorough account of how self-perceptions about efcaciousness acting in
concert with other SCCT constructs facilitate newcomer socialization.
Specically, we adapt their SCCT model to focus on how newcomers cope
with the stressful transition from outsider to insider status (Ashforth, 2001;
146
PETER W. HOM ET AL.
Hom et al., 1998; Wanous, 1992). After all, coping represents a persons
constantly changing cognitive and behavioral efforts to manage specic
external and/or internal demands that are appraised as taxing or exceeding
the persons resources (Folkman, Lazarus, Dunkel-Schetter, DeLongis, &
Gruen, 1986, p. 993). We draw from Lazarus and Folkmans (1984)
framework about how people cope with stress to adapt Lent et al.s (1994)
constructs to the domain of coping. Our SCCT model of coping thus
asserts that newcomers condence in surmounting assimilation obstacles
(coping self-efcacy), their expectations about outcomes derived from
meeting these demands (coping outcome expectations), and the goals they
set to meet such demands (coping goals) promote their adaptation to
unfamiliar work settings.
We also strengthen the explanatory power of Lent et al.s (1994) model by
conceptualizing SCCT constructs in terms of well-established socialization
concepts. In particular, our SCCT formulation species coping goals for the
key socialization tasks that many scholars have identied (Feldman, 1976).
Thus, we maintain that newcomers would form separate coping goals to
become effective performers and socially integrated within work groups
(Chan & Schmitt, 2000; Feldman, 1976; Kammeyer-Meuller & Wanberg,
2003). Similarly, our model species how different socialization tactics
(Jones, 1986) can furnish the bases for developing (coping) self-efcacy,
noting how these tactics embody the diverse self-efcacy underpinnings
(e.g., vicarious learning) noted by Bandura (1997) and SCCT theories.
From theory and research on coping with workplace stress (Latack,
Kinicki, & Prussia, 1995), our reformulated SCCT model subsumes coping
strategies to represent how newcomers can manage assimilation difculties.
For instance, new hires might become more procient in satisfying work
role requirements by changing work methods or soliciting advice from
others (Feldman & Brett, 1983; Latack, 1986; Latack et al., 1995). Our
SCCT coping perspective also recognizes that newcomers must regulate
emotional distress because coping thinkers hold that coping success
entails both effective stress management and problem-solving (Bandura,
1997; Lazarus & Folkman, 1984; Sweeney, 2008). By modeling stress
management, we also acknowledge longstanding observations that newcomers feel reality shock upon learning that the job does not live up to
expectations and anxiety when they enter novel workplaces (Hom et al.,
1998; Kramer, 1974; Meglino, DeNisi, Youngblood, & Williams, 1988;
Wanous, 1992). As a result, our theoretical approach asserts that entry
shock can invoke dysfunctional thoughts and emotions among newcomers,
which can derail their adjustment.
147
In short, we adapt Lent et al.s (1994) SCCT theory by integrating

concepts from the coping and socialization literatures to deepen insight
into why newcomers readily abandon jobs. By so doing, we derive a more
thorough and precise account of how newcomers adjust to jobs and survive
the assimilation period. Fig. 3 depicts this integrated model. At the outset,
we note that SCCT theorists prefer more ne-grained models for each
domain of functioning (e.g., different self-efcacy beliefs for task mastery
and social integration). But various domains are, however, combined to
simplify graphic depiction of this model. In what follows, we elaborate the
logic for the various pathways in this SCCT coping model of assimilation.
Coping Self-Efcacy - Coping Goals
Our SCCT framework posits that entering employees harbor self-beliefs
about their capacity to perform new jobs (coping self-efcacy, Saks,
1995), given the vital role that self-efcacy plays in newcomer adjustment
(Ashforth, 2001). Consistent with coping and SCCT thinking, we submit
that these beliefs shape the coping goals that are set and that guide actions
(Latack et al., 1995; Lent et al., 1994). Specically, beginning incumbents
who feel efcacious about their capabilities to overcome socialization
hurdles would form stronger and more resilient goals, while those feeling
less self-efcacious would set modest coping goals. To fully assimilate
(Chan & Schmitt, 2000; Feldman, 1976; Hom et al., 1998), newcomers must
achieve multiple goals to cope with the challenges of performing job tasks
prociently (Kammeyer-Meuller & Wanberg, 2003), understanding role
requirements (e.g., learning role expectations from their role set; Graen,
1976), gaining collegial acceptance (Feldman, 1976), and managing
disruptive emotions aroused by entry shock (Bandura, 1997).
Following SCCT points of view (Bandura, 1997; Lent et al., 1994), our
model species that outcome expectations anticipated outcomes for
attaining coping goals shape goal formation. That is, entry employees who
anticipate greater rewards (e.g., job security, teammate approval, healthcare coverage) from goal attainment would then set higher coping goals
(including becoming insiders sooner) than those looking forward to fewer
rewards.
Coping Goals - Coping Strategies
The current conceptualization submits that coping goals initiate varied
coping strategies for each domain of socialization, consistent with SCCT
(Lent et al., 1994) and coping (Bandura, 1997; Latack et al., 1995) models.
We nonetheless follow Lazarus and Folkmans (1984) scheme for classifying
Fig. 3.
Encounter
Unexpected Conditions
Reality Shock
Face Uncertainty
Transition to New
Workplace
Cognitive
Control Efficacy
Fit
Links
Sacrifice
Shock-Driven
Turnover Path
Problem Management
Work Long Hours
Redefine Job
Seek Task Information
Seek Social Support
Change Work Methods
Stress Management
Cognitive Reappraisal
Constructive Self-Talk
Functional Thinking
Symptom Management
Recreation
Exercise
Family Activities
Coping Strategies
Meeting
Socialization
Challenges
Expanded Social Cognitive Career Theory of Newcomer Adaptation.
Anxiety Arousal
Stress Reactions
Emotional States
Task Mastery
Role Clarification
Social Integration
Manage Disruptive
Thoughts & Emotions
Coping Goals
Internal Attributions for Success

Effort Attributions for Setbacks
Believe Ability is Acquired
Believe Effort can be Repeated
Causal Attributions
Of Performance
Perceived Threat or
Opportunity
Secondary Appraisal
Can I Cope?
Primary Appraisal
Self-Regulatory Skills
Ability/Skills
Performance Accomplishments
Internships
Prior Job Experience
On-the-Job Training
Formal Socialization Tactics
Vicarious Learning
Outcome
Serial Socialization Tactics Expectations
Collective Socialization
Tactics
Verbal Persuasion
Investiture Socialization
Tactics
Emotional Arousal
Coping Efficacy
Sequential & Fixed
Perceived Coping
Socialization Tactics
Efficacy
Orientation
Sources of Efficacy
Job
Embeddedness
148
PETER W. HOM ET AL.
149
coping strategies according to the two functions that they fulll: problem
and stress management. In support, Folkman et al. (1986) noted that a
multitude of distinct coping strategies (e.g., seeking social support, positive
reappraisal) people use to handle stressful encounters in their day-to-day
lives either serve to regulate stressful emotions or modify stressors (or both).
Likewise, Latack (1986) and Latack et al. (1995) interpreted employees
various ways for coping with job stress as embodying a proactive strategy to
resolve problems or a means to manage emotional reactions. Further,
Feldman and Brett (1983) observed that beginning employees cope with new
transitions by changing the environment (e.g., work longer hours, redene
the job, get others to provide help, seek social support) or recalibrating
affective responses to the environment (e.g., overindulge in alcohol, repress
awareness of stress).
Drawing from Lazarus and Folkman (1984) and Bandura (1997), our
theory thus classies coping strategies according to problem management
(actions to handle socialization demands, such as working long hours;
Feldman & Brett, 1983; Morrison, 2002) or stress management (blocking
out dysfunctional thoughts and emotions; Bandura, 1997). These coping
strategies are deployed to achieve specic coping goals and thereby
constitute the means by which newcomers address socialization challenges.
To meet the task-mastery coping goal, new hires might redene tasks and
seek task advice from established incumbents (problem management) as
well as participate in cognitive reappraisal or constructive self-talk (stress
management; Manz & Neck, 1999) when facing frustrations or setbacks
(Bandura, 1997).
Coping Efcacy - Stress Appraisal - Emotions
Because how people interpret stressful events dictates their coping responses
(Latack et al., 1995; Lazarus & Folkman, 1984; Sweeny, 2008), the current
model includes a sense-making mechanism in which newcomers rst
interpret the meaning of stressful events (e.g., hazing by coworkers) along
two dimensions. In primary appraisal, they must rst decide if stressors
potentially threaten career plans or goals or represent opportunities to
advance careers (or competencies; Folkman et al., 1986). They next decide
whether or not they have the resources and ability to cope with threats or
seize opportunities (secondary appraisal).
In our view, newcomers sense of personal efcacy also conditions their
stress appraisal. After all, how they construe stressful events depends on the
match between perceived coping capabilities and potentially hurtful aspects
of the environment says Bandura (1997, p. 140). Entering employees
150
PETER W. HOM ET AL.
perceiving that they can ourish in unfamiliar workplaces see opportunities

for growth and development, whereas those who believe otherwise may
regard socialization challenges as threatening. The latter in turn are prone to
anxiety and dysfunctional thoughts (that fuel such anxiety), according to
perspectives on depression and negative moods (Bandura, 1997; Beck, 1976;
Burns, 1980). If extremely upset, some distressed newcomers may form
coping goals and execute coping strategies to manage dysfunctional
cognitions and affect to lessen their debilitating effects (Bandura, 1997).
They might change their thoughts with cognitive reappraisal (e.g., reconstruing stressors as benign; Latack, 1986) or feelings with symptom management
(e.g., alcohol consumption; Feldman & Brett, 1983).
Cognitive Control Efcacy as Moderator
We further differentiate coping self-efcacy perceived capacity to control
actual threats from cognitive control efcacy perceived capacity to
control intrusive disturbing cognitions (Bandura, 1997). Accordingly, our
model embraces Banduras (1997) contention that newcomers effectively
cope with stress by developing greater condence that they can manage
actual threats as well as manage how they think about them. When people
have a strong sense of efcacy to control their own thinking, they are less
burdened by negative thoughts and experience a low level of anxiety,
claims Bandura (1997, p. 149). We thus posit that cognitive control efcacy
dampens how stress appraisals arouse emotional distress. Persons avoiding
excessive worry and ruminations are less agitated by threats (Burns, 1980;
Manz & Neck, 1999).
Socialization Sources of Self-Efcacy
In keeping with SCCT tradition (Hackett & Betz, 1981; Lent et al., 1994), we
next identify how employers assimilation methods, such as socialization
tactics (Allen, 2004; Ashforth, 2001), cultivate self-efcacy development.
In particular, formal tactics, which segregate newcomers from other
employees during socialization, may foster self-efcacy. Segregated new
hires can develop higher performance accomplishments because they can
practice new skills in safe training environments and acquire graduated
mastery experiences (Ashforth, 2001). Moreover, serial (which employs
established incumbents as socialization agents) and collective (which groups
newcomers and exposes them to common training) tactics may enhance
newcomer self-efcacy by way of vicarious learning. Newcomers observe
how experienced incumbents perform tasks correctly and thereby may feel
more efcacious about their own ability to do the same. Investiture tactics
151
afrm newcomers incoming characteristics, while divestiture attempts to

strip away newcomers former identities (e.g., boot camp; Ashforth, 2001).
When recruiting newcomers for their particular skills and abilities, rms
implementing investiture communicate to recruits that they value them and
believe them capable of meeting job demands. High expectations instill selfcondence in recruits. Finally, sequential (requiring recruits to follow a xed
progression of steps before assuming roles) and xed (setting a timetable for
role preparation) tactics alleviate anxiety. Beginning employees feel more
self-doubt when physiologically agitated by uncertainty, interpreting such
bodily states as indicative of weakened capabilities (e.g., fatigue during
athletic contests) to persist on tasks. Arousal, along with fear that arousal
can hinder performance, also erode self-efcacy (Bandura, 1997).
Feedback Loop via Causal Attributions
This conceptual scheme introduces feedback loops as new employees may
revise initial self-efcacy beliefs and outcome expectations based on early
attributions about how well they initially performed socialization tasks and
experienced rewards. Bandura (1997) assert that peoples interpretations
about feedback about task performance can modify self-beliefs about
efcacy. Simply put, individuals who take credit for success (attributing high
performance to ability or effort) maintain if not raise beliefs about selfcapabilities. Moreover, setbacks do not diminish self-efcacy when people
make effort attributions or assume that ability is cultivated through practice
rather than innate. (Note that those making effort attributions may not
feel self-efcacious if they believe that they cannot repeat their high effort in
future performances). Further, experienced success (or failure) fullling
socialization requirements may modify newcomers expectations about
outcomes derived from task performance. Finally, individuals interpretation of past performance can hinge on their current coping self-efcacy. This
is because people with a high sense of efcacy accept successes as indicants
of their capabilities but dismiss the diagnostic import of failures and
attribute them to external impediments (Bandura, 1997, pp. 154155).
SCCT Constructs - Proximal Turnover Antecedents
We next specify how our SCCT coping perspective dovetails with extant
turnover models. Socialization theory and work regard job survival as a
pivotal assimilation outcome (Ashforth, 2001; Feldman, 1976; Saks, 1995)
but rarely outline the process by which ineffective socialization engenders
turnover (Kammeyer-Meuller & Wanberg, 2003). To address this oversight,
we discuss how job embeddedness can mediate how SCCT constructs affect
152
PETER W. HOM ET AL.
newcomer retention (Hom et al., 1998; Weller et al., 2009). In our

perspective, newcomers who mastered job requirements, understood role
requirements, secured collegial approval, and managed stress reactions
during the adjustment period in turn become embedded through higher
stronger t, links, and sacrice (Mitchell & Lee, 2001). Socialized new
hires achieve on-the-job t because they satisfy performance demands
of the new job (improving ability-job requirement t) and internalize
corporate norms by way of social integration and role clarication (aligning
values to those of the company; Chatman, 1991). By socially integrating
within the workgroup, these newcomers also forge more and stronger
links in the workplace (as well as social capital; Hom et al., 2009). Though
relatively new, assimilated newcomers also experience a growing sense
of sacrice when considering leaving (Weller et al., 2009, p. 1147), such as
giving up new mentors and friends and the hard-won status of surviving
probation (earning them associated perks, such as health and pension
coverage) if they exit.
By comparison, on-the-job t is low for poorly socialized newcomers as
they fail to meet performance standards or internalize corporate values
(Mitchell & Lee, 2001; Weller et al., 2009). Newcomers who have trouble
assimilating also tend to be excluded from organizational social networks,
which deprive them of embedding links. Foundering during socialization,
they may also not accumulate as many corporate perks nor develop much
social capital. As a result, they may incur little or no sacrices upon leaving.
Because unassimilated newcomers are less embedded in the job, they are
thus more liable to quit according to Mitchell and Lee (2001).
All the same, job embeddedness or its absence cannot fully account for
all newcomer attrition. Some beginning employees quit soon after job entry
well before they can (gradually) learn if they are good ts with the job
(Weller et al., 2009). Rather, these leavers may likely encounter shocks
during early employment (Lee & Mitchell, 1994). Weller and colleagues
(2009) note that new employees who are going through a role transition
(outsider to insider) are more likely to experience organizational shocks
(p. 1148). If they experience negative workplace shocks that conict with
career values, goals, or plans (Lee & Mitchell, 1994), they may readily leave
even without job offers in hand (taking turnover path 2; T. W. Lee et al.,
1996, 1999). Upholding this thesis, socialization investigators often observed
that job entrants whose pre-entry expectations are unrealistic (because
recruiters misled them; Wanous, 1992) are more prone to endure reality
shock and quickly exit (Hom et al., 1998). Consequently, our framework
includes shock-driven turnover as another path by which newcomers exit.
153
Implications of SCCT Theory for Explaining

Higher Minority and Female Quits
Though we propose a general framework to explain newcomer attrition, this
conceptualization can be adapted to reect unique socialization problems
confronting different populations. For example, Hom and associates (2008)
documented higher corporate ight among female and minority incumbents
in managerial and professional elds where their identity groups have been
historically underrepresented (Hom & Griffeth, 1995; Roberson, 2004).
Because most women and minorities exit corporate America during their
rst few years of work, inadequate or incomplete organizational socialization may underlie their higher turnover relative to white men (Hom et al.,
2008). We thus extend our general formulation about newcomer coping to
address their special adjustment challenges, offering a more comprehensive
explanation about corporate ight among minorities and women.
Higher Coping Demands
For women and minorities entering nontraditional workplaces or occupational elds, social integration or its absence represents a prominent
socialization challenge because they, demographically dissimilar newcomers,
may endure social isolation (due to tokenism), exclusion from informal
dominant white-male networks (due to demographic dissimilarity; Eagly &
Carli, 2007; Elvira & Cohen, 2001; Riordan, Schaffer, & Stewart, 2005), and
hostility from peers or superiors (e.g., sexual or ethnic harassment; Berdahl
& Moore, 2006; Laband & Lentz, 1998). Moreover, they may encounter
more difculty understanding and mastering new task requirements because
they lack good mentors (due to supervisory bias or scarcity of mentors
of their own ethnicity or sex; Cottrell & Neuberg, 2005; Eagly & Karau,
2002; Fiske, Cuddy, Glick, & Xu, 2002; Thomas, 2001) and receive less
information from colleagues (Forret & Dougherty, 2004; Ibarra, 1992).
Weaker Self-Efcacy Sources
Besides greater challenges, our SCCT formulation suggests that minorities
and women may feel less efcacious about their ability to survive and
ourish in white- or male-dominated workplaces. During their formative
years, they may have experienced fewer sources on which to build strong
self-beliefs that they can succeed in nontraditional career elds (Hackett &
Betz, 1981; Lent et al., 2008). As youth, they knew few women or minority
professionals or managers, for example. Moreover, employing organizations may fail to cultivate a robust sense of self-efcacy among them.
154
PETER W. HOM ET AL.
To illustrate, certain socialization tactics can disadvantage them. For

example, rms applying collective socialization (grouping newcomers and
orienting them together) may address primarily concerns of the majority of
new hires (white men), omitting unique concerns of minorities or women
(e.g., forming productive relationships with white-male mentors; Thomas,
2001). Along these lines, employers deploying serial tactics may assign
veterans from another sex or race to coach women and minorities (due to
scarcity of female and minority mentors; Elvira & Cohen, 2001; Ibarra,
1992), who provide marginal mentoring (Ragins, Cotton, & Miller, 2000).
Further, rms using divestiture tactics try to strip away newcomers
incoming identities to reconstruct a new corporate identity (Ashforth, 2001).
This approach forces women and minorities to assimilate to the dominant
white male culture (Ely & Thomas, 2001), which devalues their cultural
identities and weakens their self-efcacy.
Along with discouraging socialization tactics, women and minorities face
other conditions, well-documented by diversity research (e.g., supervisory
bias, impoverished job duties, glass ceilings; Eagly & Carli, 2007; Hom &
Griffeth, 1995; Stroh, Brett, & Reilly, 1996) that diminish perceived selfefcacy to master socialization requirements. To illustrate, supervisors may
not see a good t between minority and female newcomers attributes and
job requirements (Eagly & Carli, 2007; Heilman, 1983). Rather, they may
believe that individual minorities and women possess traits stereotypical of
their group, which do not resemble the requisite traits for occupations where
whites or men predominate. As a result, superiors may act (or not act) in
ways that decrease womens and minorities self-efcacy (Hom & Griffeth,
1995). Expecting less of them, they may thus offer fewer chances for
performance accomplishments (relegating women and minorities mundane
tasks or jobs), fewer occasions for vicarious learning (by personally
demonstrating how they would handle problems), and engage in less verbal
persuasion (to encourage women and minorities to tackle stretch assignments). Given stereotypes about occupational mist, superiors may thus fail
to create the preconditions that promote self-efcacy among minority and
female recruits (Bandura, 1997).
Stereotype Threat
Our SCCT viewpoint highlights how minorities and women entering
nontraditional vocational elds or workplaces may appraise threats posed
by role transitions differently or more severely than do white men.
Specically, they are prone to stereotype threat (Roberson & Kulik,
2007), feeling anxiety about conrming negative stereotypes about their
155
identity group. When situated in workplaces that employ few members of

their identity group (including leadership positions; Eagly & Carli, 2007;
Hom et al., 2008), they may worry about corroborating implicit stereotypes
that those like them cannot succeed in white-male elds or settings. As a
result, minority and female newcomers feel performance anxiety, which
impedes performance and induces them to vacate jobs where they cannot
succeed (Steele, 1997).
All told, an SCCT perspective can illuminate why minority and female
newcomers more readily exit nontraditional jobs (e.g., managers, army
ofcers; Payne & Huffman, 2005). With few exceptions (Stroh et al., 1996),
turnover researchers have rarely scrutinized proximal or intermediary
psychological states behind minority and female exits (Roberson, 2004).
Rather, they have focused on documenting reliable racial or gender
disparities in attrition or relating disparate workplace conditions (e.g., pay,
promotions) to differential attrition (Daniels, 2004; Greenhaus, Collins,
Singh, & Parasuraman, 1997; Hewlett & Luce, 2005; Lyness & Judiesch,
2001; Lyness & Thompson, 1997; Rosin & Korabik, 1995). Yet investigating
how SCCT constructs mediate between disadvantageous working conditions and turnover can clarify how such conditions occasion greater female
and minority attrition from male- or white-dominated workplaces. Indeed,
mediation tests would add greater credibility to prevailing theories that
workplace discrimination undergirds elevated minority and female corporate ight (Hom et al., 2008) as they are often grounded in circumstantial
evidence (Johnson & Neumark, 1997).
Moreover, SCCT theory might clarify why racial and gender differences
in turnover from nontraditional jobs vary across studies (Hom et al., 2008).
For example, Lyness and Judiesch (2001) found no quit differences between
men and women managers from a nancial services organization, whereas
Hom and associates (2008) observed that women professionals and
managers from large rms representing various industries left at higher
rates than their male counterparts. Women in the former study however
represented 42% of the entire sample, whereas they constituted 28% of the
workforces sampled in the latter test. Quite likely, the sizeable representation of women across the spectrum of jobs in Lyness and Judieschs (2001)
study created (and reected) conditions more hospitable to fostering
womens self-efcacy and job survival. In their nancial services rm,
women could nd more female mentors who can show them the ropes
about how to succeed and more welcoming mixed-gender networks that can
give task advice and social support. Given greater sources of self-efcacy,
more women would feel efcacious and successfully assimilate, thereby
156
PETER W. HOM ET AL.
staying as long as their male counterparts. By contrast, women in Hom and

associates (2008) study felt less efcacious as they lacked more favorable
workplace conditions for building self-efcacy. Their sample of corporate
women (especially short-tenure ones) were poorly socialized into workplaces
and thus more likely to quit than men.
CONCLUDING REMARKS
Our brief review of three leading theories of career and vocational
development suggests that they hold great promise for elucidating why and
how people stay or leave. Currently, the dominant theoretical and empirical
approaches focus on proximal antecedents to turnover, such as quit
intentions, job attitudes (and their intermediate effects; Hom & Kinicki,
2001), shocks (Lee & Mitchell, 1994; T. W. Lee et al., 1999), and job
embeddedness (certain dimensions such as t and links; Mitchell et al.,
2001b). Some turnover authors allude to distal causes, such as image
violations, matching scripts (Lee & Mitchell, 1994), and centrality of nonwork
values or roles (Mobley, 1982; Mobley et al., 1979). Complementing presentday theoretical advances, we suggest greater scholarly attention to more
fundamental causes of attrition, such as the vocational and career constructs
reviewed in this chapter. As we noted, these latter models can more precisely
answer why image violations, matching scripts, or centrality of nonwork roles
motivate turnover. In addition, as more constructs are transported from
vocational and career theories and empirically linked to turnover research,
we can build more comprehensive models within the turnover eld.
In closing, we contend that the twin scientic goals of prediction and
understanding do not always converge, though turnover scholars strive
for both goals (Maertz & Campion, 1998). Though current emphasis on
predictive accuracy can yield greater insight into turnover (cf., Lee &
Mitchell, 1994), such orientation draws attention to immediate turnover
precursors. More basic causes are thus overlooked, such as how otherwise
satisfying jobs can no longer embed incumbents when their career stage
shifts over time or why many people exit the work force to pursue other life
roles, such as full-time parenting or education. Over the years, turnover
theorists increasingly recognize such forms of leaving (Hom & Kinicki,
2001; Hulin et al., 1985; Lee & Mitchell, 1994) but their attempts at
explanation seem supercial. To address this theoretical void, we thus
advocate greater attention to distal constructs such as those formulated by
career scholars. We hope that our discussion of the conceptual richness of
157
these career development models will encourage scholarly explorations

into how career constructs can promote understanding and perhaps
prediction of turnover.
ACKNOWLEDGMENT
We thank Neal Schmitt for his assistance in providing a review of our earlier
drafts.
REFERENCES
Allen, D. (2004). Do organizational socialization tactics inuence newcomer embeddedness and
turnover? Journal of Management, 32, 237256.
Ashforth, B. (2001). Role transitions in organizational life. Mahwah, NJ: Lawrence Erlbaum.
Ashforth, B., & Saks, A. (2000). Personal control in organizations: A longitudinal investigation
with newcomers. Human Relations, 53, 311339.
Ashford, S., & Black, J. (1996). Proactivity during organizational entry: The role of desire for
control. Journal of Applied Psychology, 81, 199214.
Assouline, M., & Meir, E. I. (1987). Meta-analysis of the relationship between congruence and
wellbeing measures. Journal of Vocational Behavior, 31, 319323.
Bandura, A. (1977). Self-efcacy: Toward a unifying theory of behavioral change. Psychological
Review, 84, 191215.
Bandura, A. (1982). Self-efcacy mechanism in human agency. American Psychologist, 37,
122147.
Bandura, A. (1989). Human agency in social cognitive theory. American Psychologist, 44,
11751184.
Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior and
Human Decision Processes, 50, 248287.
Bandura, A. (1997). Self-efcacy: The exercise of control. New York: W.H. Freeman and Company.
Beck, A. (1976). Cognitive theory and the emotional disorders. New York: International
Universities Press.
Bentein, K., Vandenberg, R., Vandenberghe, C., & Stinglhamber, F. (2005). The role of change
in the relationship between commitment and turnover: A latent growth modeling
approach. Journal of Applied Psychology, 90, 468482.
Berdahl, J. L., & Moore, C. (2006). Workplace harassment: Double jeopardy for minority
women. Journal of Applied Psychology, 91, 426436.
Betz, N., & Hackett, G. (1981). The relationship of career-related self-efcacy expectations to
perceived career options in college women and men. Journal of Counseling Psychology,
28, 339410.
Betz, N., & Hackett, G. (2006). Career self-efcacy theory: Back to the future. Journal of Career
Assessment, 14, 311.
Booth, A. L., Francesconi, M., & Garcia-Serrano, C. (1999). Job tenure and job mobility in
Britain. Industrial and Labor Relations Review, 53, 4370.
158
PETER W. HOM ET AL.
Brousseau, K. R. (1983). Toward a dynamic model of person-job relationships: Findings,

research questions, and implications for work system design. Academy of Management
Review, 8, 3345.
Brown, S., Tramayne, S., Hoxha, D., Telander, K., Fan, X., & Lent, R. (2008). Social cognitive
predictors of college students academic performance and persistence: A meta-analytic
path analysis. Journal of Vocational Behavior, 72, 298308.
Burns, D. (1980). Feeling good: The new mood therapy. New York: William Morrow.
Burton, J., Holtom, B., Sablynski, C., Mitchell, T., & Lee, T. (2010). The buffering effects of job
embeddedness on negative shocks. Journal of Vocational Behavior, 76, 4251.
Cable, D. M., & DeRue, D. S. (2002). The convergent and discriminant validity of subjective t
perceptions. Journal of Applied Psychology, 87, 875884.
Carson, A. D., & Mowsesian, R. (1993). Moderators of the prediction of job satisfaction from
congruence: A test of Hollands theory. Journal of Career Assessment, 1, 130144.
Chan, D., & Schmitt, N. (2000). Interindividual differences in intraindividual changes in
proactivity during organizational entry: A latent growth modeling approach to
understanding newcomer adaptation. Journal of Applied Psychology, 85, 190210.
Chartrand, J., & Walsh, W. B. (1999). What should we expect from congruence? Journal of
Vocational Behavior, 55, 136146.
Chatman, J. (1991). Matching people and organizations: Selection and socialization in public
accounting rms. Administrative Science Quarterly, 36, 459484.
Cottrell, C. A., & Neuberg, S. L. (2005). Different emotional reactions to different groups:
A sociofunctional threat-based approach to prejudice. Journal of Personality and
Social Psychology, 88, 770789.
Crites, J. O. (1973). Career maturity inventory. Monterey, CA: CTB/McGraw Hill.
Dalton, D. R., Hill, J. W., & Ramsay, R. J. (1997). Women as managers and partners:
Context specic predictors of turnover in international public accounting rms.
Auditing: A Journal of Practice & Theory, 16, 2950.
Daniels, C. (2004). Young, gifted, blackand out of here. Fortune, 149(May 4), 48.
Darcy, M., & Tracey, T. J. G. (2003). Integrating abilities and interests into career choice:
Maximal versus typical assessment. Journal of Career Assessment, 11, 219237.
Eagly, A., & Carli, L. (2007). Women and the labyrinth of leadership. Harvard Business Review
(September), 210.
Eagly, A., & Karau, S. (2002). Role congruity theory of prejudice toward female leaders.
Psychological Review, 109, 573598.
Ebaugh, H. (1988). Becoming an ex: The process of role exit. Chicago: University of Chicago
Press.
Edwards, J. (2002). Alternatives to difference scores: Polynomial regression analysis and
response surface methodology. In: F. Drasgow & N. Schmitt (Eds), Measuring and
analyzing behavior in organizations (pp. 350400). San Francisco, CA: Jossey-Bass.
Elvira, M. M., & Cohen, L. E. (2001). Location matters: A cross-level analysis of the effects of
organizational sex composition on turnover. Academy of Management Journal, 44, 591605.
Ely, R., & Thomas, D. (2001). Cultural diversity at work: The effects of diversity perspectives
on work group processes and outcomes. Administrative Science Quarterly, 46, 229273.
Feldman, D. (1976). A contingency theory of socialization. Administrative Science Quarterly,
21, 433452.
Feldman, D., & Brett, J. (1983). Coping with new jobs: A comparative study of new hires and
job changers. Academy of Management Journal, 26, 258272.
159
Felps, W., Mitchell, T. R., Heckman, D. R., Lee, T. W., Holtom, B. C., & Harman, W. S.
(2009). Turnover contagion: How coworkers job embeddedness and job search
behaviors inuence quitting. Academy of Management Journal, 52, 545561.
Fiske, S., Cuddy, A., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content:
Competence and warmth respectively follow from perceived status and competition.
Journal of Personality and Social Psychology, 82, 878902.
Folkman, S., Lazarus, R., Dunkel-Schetter, C., DeLongis, A., & Gruen, R. (1986). Dynamics of
a stressful encounter: Cognitive appraisal, coping, and encounter outcomes. Journal of
Forret, M., & Dougherty, T. (2004). Networking behaviors and career outcomes: Differences
for men and women? Journal of Organizational Behavior, 25, 419437.
Fox, M. (2005). Gender, family characteristics, and publication productivity among scientists.
Social Studies of Science, 35, 131150.
Furnham, A. (1994). Personality at work: The role of individual differences in the workplace.
London: Routledge.
Furnham, A. (2001). Vocational preference and P-O t: Reections on Hollands theory of
vocational choice. Applied Psychology: An International Review, 50, 529.
Ghiselli, E. (1973). Some perspectives for industrial psychology. American Psychologist
(February), 8087.
Graen, G. (1976). Role-making processes within complex organizations. In: M. Dunnette (Ed.),
Handbook of industrial and organizational psychology (pp. 12011246). Chicago, IL:
Rand McNally College Publishing Company.
Greenhaus, J. (1987). Career management. New York: Dryden Press.
Greenhaus, J. H., Collins, K. M., Singh, R., & Parasuraman, S. (1997). Work and family inuences
on departure from public accounting. Journal of Vocational Behavior, 50, 249270.
Griffeth, R., & Hom, P. (2001). Retaining valued employees. Thousand Oaks, CA: Sage.
Griffeth, R., Hom, P., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates of
employee turnover: Update, moderator tests, and research implications for the next
millennium. Journal of Management, 26, 463488.
Grote, G., & Raeder, S. (2009). Careers and identity in exible working: Do exible identities
fare better? Human Relations, 62, 219244.
Hackett, G., & Betz, N. (1981). A self-efcacy approach to the career development of women.
Journal of Vocational Behavior, 18, 326339.
Hackett, G., Betz, N., OHalloran, M., & Romac, D. (1990). Effects of verbal and mathematics
task performance on task and career self-efcacy and interest. Journal of Counseling
Harvey, V., & McMurray, N. (1994). Self-efcacy: A means of identifying problems in nursing
education and career progress. International Journal of Nursing Studies, 31, 471485.
Healy, C. C., & Mourton, D. L. (1985). Congruence and vocational identity: Outcomes of
career counseling with persuasive power. Journal of Counseling Psychology, 32, 441444.
Heilman, M. (1983). Sex bias in work settings: The lack of t model. In: L. Cummings &
B. Staws (Eds), Research in organizational behavior (pp. 269298). Greenwich, CT:
JAI Press.
Hewlett, S. A., & Luce, C. B. (2005). Off-ramps and on-ramps. Harvard Business Review
(March), 1726.
Holland, J. L. (1996). Exploring careers with a typology: What we have learned and some new
directions. American Psychologist, 51, 397406.
160
PETER W. HOM ET AL.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work
environments. Odessa, FL: Psychological Assessment Resources.
Holland, J. L., Daiger, D. C., & Power, P. G. (1980). My vocational situation. Palo Alto, CA:
Consulting Psychologist Press.
Holland, J. L., & Gottfredson, G. D. (1994). Career attitudes and strategies inventory. Odessa,
FL: Psychological Assessment Resources.
Holtom, B., Mitchell, T., Lee, T., & Eberly, M. (2008). Turnover & retention research: A glance
at the past, a closer review of the present, and a venture into the future. The Academy of
Management Annals, 2, 231274.
Holtom, B., Mitchell, T., Lee, T., & Inderrieden, E. (2005). Shocks as causes of turnover:
What they are and how organizations can manage them. Human Resource Management
Journal, 44, 337352.
Holtom, B. C., & Inderrieden, E. J. (2006). Integrating the unfolding model and job
embeddedness model to better understand voluntary turnover. Journal of Managerial
Issues, 18, 435452.
Hom, P. (2010). Organizational exit. In: S. Zedeck, H. Aguinis, W. Cascio, M. Gelfand,
K. Leung, S. Parker & J. Zhou (Eds), Handbook of industrial/organizational psychology
(Vol. 2). Washington, DC: American Psychological Association.
Hom, P., & Griffeth, R. (1991). Structural equations modeling test of a turnover theory: Crosssectional and longitudinal analyses. Journal of Applied Psychology, 76, 350366.
Hom, P., & Griffeth, R. (1995). Employee turnover. Cincinnati, OH: South-Western College
Publishing Company.
Hom, P., Griffeth, R., Palich, L., & Bracker, J. (1998). An exploratory investigation
into theoretical mechanisms underlying realistic job previews. Personnel Psychology,
51, 421451.
Hom, P., & Kinicki, A. (2001). Toward a greater understanding of how dissatisfaction drives
employee turnover. Academy of Management Journal, 44, 975987.
Hom, P., Roberson, L., & Ellis, A. (2008). Challenging conventional wisdom about who quits:
Revelations about employee turnover from corporate America. Journal of Applied
Hom, P. W., Tsui, A. S., Wu, J. B., Lee, T. W., Zhang, A. Y., Fu, P. P., & Li, L. (2009).
Explaining employment relationships with social exchange and job embeddedness.
Journal of Applied Psychology, 94, 277297.
Hulin, C., Roznowski, M., & Hachiya, D. (1985). Alternative opportunities and withdrawal
decisions: Empirical and theoretical discrepancies and an integration. Psychological
Bulletin, 97, 233250.
Ibarra, H. (1992). Homophily and differential returns: Sex differences in network structure and
access in an advertising rm. Administrative Science Quarterly, 37, 422447.
Ito, J. K., & Brotheridge, C. M. (2005). Does supporting employees career adaptability lead
to commitment, turnover or both? Human Resource Management, 44, 519.
Johnson, R., & Neumark, D. (1997). Age discrimination, job separations, and employment
status of older workers: Evidence from self-reports. Journal of Human Resources, 32,
779811.
Jones, G. (1986). Socialization tactics, self-efcacy, and newcomers adjustments to organizations. Academy of Management Journal, 29, 262279.
Judge, T., & Watanabe, S. (1995). Is the past prologue?: A test of Ghisellis hobo syndrome.
Journal of Management, 21, 211229.
161
Kacmar, K., Andrews, M., Rooy, D., Steilberg, R., & Cerrone, S. (2006). Sure everyone can be
replaced y but at what costs? Turnover as a predictor of unit-level performance.
Kammeyer-Meuller, J., & Wanberg, C. (2003). Unwrapping the organizational entry process:
Disentangling multiple antecedents and their pathways to adjustment. Journal of Applied
Kammeyer-Mueller, J., Wanberg, C., Glomb, T., & Ahlburg, D. (2005). Turnover processes in a
temporal context: Its about time. Journal of Applied Psychology, 90, 644658.
Khatri, N., Fern, C., & Budhwar, P. (2001). Explaining employee turnover in an Asian context.
Human Resource Management Journal, 11, 5474.
Kramer, M. (1974). Reality shock: Why nurses leave nursing. St. Louis, MO: CV Mosby.
Kristof-Brown, A. L., Zimmerman, R. D., & Johnson, E. C. (2005). Consequences of
individuals t at work: A meta-analysis of person-job, person-organization, persongroup, and person-supervisor t. Personnel Psychology, 58, 281342.
Kuder, G. F. (1939). Kuder preference record. Chicago: Science Research Associates.
Laband, D. N., & Lentz, B. F. (1998). The effects of sexual harassment on job satisfaction,
earnings, and turnover among female lawyers. Industrial and Labor Relations Review, 51,
594607.
Latack, J. (1986). Coping with job stress: Measures and future directions for scale development.
Latack, J., Kinicki, A., & Prussia, G. (1995). An integrative process model of coping with job
loss. Academy of Management Review, 20, 311342.
Lazarus, R., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Guilford.
Lee, T. H., Gerhart, B., Weller, I., & Trevor, C. (2008). Understanding voluntary turnover:
Path-specic job satisfaction effects and the importance of unsolicited job offers.
Lee, T. W., & Mitchell, T. (1994). An alternative approach: The unfolding model of voluntary
model of voluntary employee turnover. Academy of Management Review, 19, 5189.
Lee, T. W., Mitchell, T., Holtom, B., McDaniel, L., & Hill, J. (1999). The unfolding model of
voluntary turnover: A replication and extension. Academy of Management Journal, 42,
450462.
Lee, T. W., Mitchell, T. R., Sablynski, C., Burton, J., & Holtom, B. (2004). The effects of job
embeddedness on organizational citizenship, job performance, volitional absences, and
voluntary turnover. Academy of Management Journal, 47, 711722.
Lee, T. W., Mitchell, T. R., Wise, L., & Fireman, S. (1996). An unfolding model of voluntary
employee turnover. Academy of Management Journal, 39, 536.
Lent, R., Brown, S., & Hackett, G. (1994). Toward a unifying social cognitive theory of
career and academic interest, choice, and performance. Journal of Vocational Behavior,
45, 79122.
Lent, R., Lopez, A., Lopez, F., & Sheu, H.-B. (2008). Social cognitive theory and the prediction
of interests and choice goals in the computing disciplines. Journal of Vocational Behavior,
73, 5262.
Leong, F., & Barak, A. (2001). Contemporary models in vocational psychology: A volume in
honor of Samuel H. Osipow. Mahwah, NJ: Lawrence Erlbaum.
Low, K. S. D., & Rounds, J. (2007). Interest change and continuity from early adolescence
to middle adulthood. International Journal for Educational and Vocational Guidance, 7,
2336.
162
PETER W. HOM ET AL.
Lyness, K. S., & Judiesch, M. K. (2001). Are female managers quitters? The relationships of
gender, promotions, and family leaves of absence to voluntary turnover. Journal of
Lyness, K. S., & Thompson, D. (1997). Above the glass ceiling? A comparison of matched
samples of female and male executives. Journal of Applied Psychology, 82, 359375.
Mael, F., & Ashforth, B. (1995). Loyal from day one: Biodata, organizational identication,
and turnover among newcomers. Personnel Psychology, 48, 309333.
Mael, F., & Ashforth, B. E. (1992). Alumni and their alma mater: A partial test of the
reformulated model of organizational identication. Journal of Organizational Behavior,
13, 103123.
Maertz, C., & Griffeth, R. (2004). Eight motivational forces and voluntary turnover: A
theoretical synthesis with implications for research. Journal of Management, 30, 667683.
Maertz, C. P., & Campion, M. A. (1998). 25 years of voluntary turnover research: A review and
critique. International Review of Industrial and Organizational Psychology, 13, 4983.
Maertz, C. P., & Campion, M. A. (2004). Proles in quitting: Integrating content and process
turnover theory. Academy of Management Journal, 47, 566582.
Manz, C., & Neck, C. (1999). Mastering self-leadership. Upper Saddle River, NJ: Prentice-Hall.
March, J., & Simon, H. (1958). Organizations. New York: Wiley.
Meglino, B., DeNisi, A., Youngblood, S., & Williams, K. (1988). Effects of realistic job
previews: A comparison using an enhancement and a reduction preview. Journal of
Meir, E. I., & Erez, M. (1981). Fostering a career in engineering. Journal of Vocational Behavior,
18, 115120.
Meir, E. I., Esformes, Y., & Friedland, N. (1994). Congruence and differentiation as predictors
of workers occupational stability and job performance. Journal of Career Assessment, 2,
4054.
Meir, E. I., & Green-Eppel, T. (1999). Congruence, skill utilization, and group importance as
predictors of well-being in army reserve units. Journal of Career Assessment, 7, 429438.
Meir, E. I., Hadas, C., & Noyfeld, M. (1997). Personenvironment t in small army units.
Journal of Career Assessment, 5, 2129.
Meir, E. I., Keinan, G., & Segal, Z. (1986). Group importance as a mediator between personality
environment congruence and satisfaction. Journal of Vocational Behavior, 28, 6069.
Meir, E. I., Melamed, S., & Abu-Freha, A. (1990). Vocational, avocational, and skill utilization
congruences and their relationship with well-being in two cultures. Journal of Vocational
Behavior, 36, 153165.
Meir, E. I., & Navon, M. (1992). A longitudinal examination of the congruence hypotheses.
Meir, E. I., Tziner, A., & Glazner, Y. (1997). Environmental congruence, group importance,
and job satisfaction. Journal of Career Assessment, 5, 343353.
Mitchell, T., Holtom, B., & Lee, T. (2001a). How to keep your best employees: Developing an
effective retention policy. Academy of Management Executive, 15, 96108.
Mitchell, T. R., Holtom, B. C., Lee, T. W., Sablynski, C. J., & Erez, M. (2001b). Why people
stay: Using job embeddedness to predict voluntary turnover. Academy of Management
Journal, 44, 11021121.
Mitchell, T. R., & Lee, T. W. (2001). The unfolding model of voluntary turnover and job
embeddedness: Foundations for a comprehensive theory of attachment. In: B. Staw (Ed.),
Research in organizational behavior (Vol. 23, pp. 189246). Oxford, UK: Elsevier Science.
163
Mobley, W. (1982). Employee turnover: Causes, consequences, and control. Reading, MA:
Addison-Wesley.
Mobley, W., Griffeth, R., Hand, H., & Meglino, B. (1979). Review and conceptual analysis of
the employee turnover process. Psychological Bulletin, 86, 493522.
Mobley, W., Horner, S., & Hollingsworth, A. (1978). An evaluation of precursors of hospital
employee turnover. Journal of Applied Psychology, 63, 408414.
Morgeson, F., & Nahrgang, J. (2008). Same as it ever was: Recognized stability in the
BusinessWeek rankings. Academy of Management Learning & Education, 7, 2641.
Morrison, E. (2002). Newcomers relationships: The role of social network ties during
socialization. Academy of Management Journal, 6, 11491160.
OReilly, C. (1991). Organizational behavior: Where weve been, where were going. Annual
Review of Psychology, 42, 427458.
Parsons, F. (1909). Choosing a vocation. Boston: Houghton Mifin.
Payne, S. C., & Huffman, A. H. (2005). A longitudinal examination of the inuence of
mentoring on organizational commitment and turnover. Academy of Management
Journal, 48, 158168.
Perdue, S. V., Reardon, R. C., & Peterson, G. W. (2007). Person-environment congruence, selfefcacy, and environmental identity in relation to job satisfaction: A career decision
theory. Journal of Employment Counseling, 44, 2939.
Piasentin, K. A., & Chapman, D. S. (2006). Subjective person-organization t: Bridging the gap
between conceptualization and measurement. Journal of Vocational Behavior, 69, 202221.
Price, J. L., & Mueller, C. W. (1981). A causal model of turnover for nurses. Academy of
Price, J. L., & Mueller, C. W. (1986). Absenteeism and turnover of hospital employees.
Greenwich, CT: JAI Press.
Ragins, B., Cotton, J., & Miller, J. (2000). Marginal mentoring: The effects of type of mentor,
quality of relationship, and program design on work and career attitudes. Academy of
Randsley de Moura, G., Abrams, D., Retter, C., Gunnarsdottir, S., & Ando, K. (2008).
Identication as an organizational anchor: How identication and job satisfaction
combine to predict turnover intention. European Journal of Social Psychology, early view
published online, DOI: 10.1002/ejsp.553.
Riordan, C. M., Schaffer, B., & Stewart, M. (2005). Relational demography within groups:
Through the lens of discrimination. In: R. Diboye & A. Colella (Eds), Discrimination at
work: The psychological and organizational bases (pp. 3762). Mahwah, NJ: Lawrence
Erlbaum Associates.
Roberson, L. (2004). On the relationship between race and turnover. In: R. W. Griffeth &
P. W. Hom (Eds), Innovative theory and empirical research on employee turnover
(pp. 211229). Greenwich, CT: Information Age Publishing.
Roberson, L., & Kulik, C. (2007). Stereotype threat at work. Academy of Management
Perspectives (May), 2440.
Rosin, H., & Korabik, K. (1995). Organizational experiences and propensity to leave:
A multivariate investigation of men and women managers. Journal of Vocational
Behavior, 46, 116.
Rusbult, C., & Farrell, D. (1983). A longitudinal test of the investment model: The impact on
job satisfaction, job commitment, and turnover of variations in rewards, costs,
alternatives, and investment. Journal of Applied Psychology, 68, 429438.
164
PETER W. HOM ET AL.
Saks, A. (1995). Longitudinal eld experiment of the moderating and mediating effects of selfefcacy on the relationship between training and newcomer adjustment. Journal of
Savickas, M. L. (1985). Career maturity: The construct and its measurement. Vocational
Guidance Quarterly, 32, 222231.
Savickas, M. L. (1997). Career adaptability: An integrative construct for life-span, life-space
theory. Career Development Quarterly, 45, 247259.
Savickas, M. L. (2005). The theory and practice of career construction. In: S. D. Brown &
R. W. Lent (Eds), Career development and counseling: Putting theory and research to
work (pp. 4270). Hoboken, NJ: Wiley.
Spokane, A. R. (1985). A review of research on personenvironment congruence in Hollands
theory of careers (monograph). Journal of Vocational Behavior, 26, 306343.
Spokane, A. R., Luchetta, E. J., & Richwine, M. H. (2002). Hollands theory of personalities in
work environments. In: D. Brown & Associates (Eds), Career choice and development
(pp. 373426). San Francisco, CA: Jossey-Bass.
Spokane, A. R., Meir, E. I., & Catalano, M. (2000). Person-environment congruence
and Hollands theory: A review and reconsideration. Journal of Vocational Behavior,
57, 137187.
Steel, R. P. (2002). Turnover theory at the empirical interface: Problems of t and function.
The Academy of Management Review, 27, 346360.
Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity and
performance. American Psychologist, 52, 613629.
Steers, R., & Mowday, R. (1981). Employee turnover and post decision accommodation
processes. In: L. Cummings & B. Staws (Eds), Research in organizational behavior
(Vol. 3, pp. 235281). Greenwich, CT: JAI Press.
Stroh, L., Brett, J., & Reilly, A. (1996). Family structure, glass ceiling, and traditional
explanations for the differential rate of turnover of female and male managers. Journal
of Vocational Behavior, 49, 99118.
Strong, E. K., Jr. (1943). Vocational interests of men and women. Stanford, CA: Stanford
University Press.
Super, D. E. (1990). A life-span, life-space approach to career development. In: D. Brown &
L. Brooks (Eds), Career choice and development (pp. 197261). San Francisco, CA:
Jossey-Bass Publishers.
Super, D. E., & Knasel, E. G. (1981). Career development in adulthood: Some theoretical
problems and a possible solution. British Journal of Guidance and Counseling, 9, 194201.
Super, D. E., Savickas, M. L., & Super, C. M. (1996). The life-span, life-space approach
to careers. In: D. Brown & L. Brooks (Eds), Career choice and development:
Applying contemporary theories to practice (3rd ed., pp. 121178). San Francisco, CA:
Jossey-Bass.
Super, D. E., Thompson, A. S., Lindeman, R. H., Jordaan, J. P., & Myers, R. A. (1981).
Career development inventory: College form. Palo Alto, CA: Consulting Psychologists
Press.
Super, D. E., Thompson, A. S., Lindeman, R. H., Myers, R. A., & Jordaan, J. P. (1988). Adult
career concerns inventory. Palo Alto, CA: Consulting Psychologists Press.
Super, D. W. (1953). A theory of vocational development. American Psychologist, 8, 185190.
Sweeny, K. (2008). Crisis decision theory: Decisions in the face of negative events. Psychological
Bulletin, 134, 6176.
165
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conict. In: W. G. Austin
& S. Worchel (Eds), The social psychology of intergroup relations (pp. 3347). Monterey,
CA: Brooks-Cole.
Taylor, S., Audia, G., & Gupta, A. (1996). The effect of lengthening job tenure on managers
organizational commitment and turnover. Organizational Science, 7, 632648.
Thomas, D. A. (2001). The truth about mentoring minorities: Race matters. Harvard Business
Review (April), 99107.
Tracey, T. J., & Rounds, J. (1993). Evaluating Hollands and Gatis vocational-interest models:
A structural meta-analysis. Psychological Bulletin, 113, 229246.
Tracey, T. J. G. (2007). Moderators of the interest congruence-occupational outcome relation.
International Journal for Educational and Vocational Guidance, 7, 3745.
Tracey, T. J. G., Robbins, S. B., & Hofsess, C. D. (2005). Stability and change in adolescence:
A longitudinal analysis of interests from grades 8 through 12. Journal of Vocational
Behavior, 66, 125.
Tracey, T. J. G., & Rounds, J. B. (1997). Circular structure of vocational interests.
In: R. Plutchik & H. R. Conte (Eds), Circumplex models of personality and emotions
(pp. 183201). Washington, DC: American Psychological Association.
Tranberg, M., Slane, S., & Ekberg, S. E. (1993). The relation between interest congruence and
satisfaction: A meta-analysis. Journal of Vocational Behavior, 42, 253264.
Trevor, C. (2001). Interactions among actual ease-of-movement determinants and job
satisfaction in the prediction of voluntary turnover. Academy of Management Journal,
44, 621638.
Tsabari, O., Tziner, A., & Meir, E. (2005). Updated meta-analysis on the relationship between
congruence and satisfaction. Journal of Career Assessment, 13, 216232.
U.S. Department of Labor, Bureau of Labor Statistics. (2008). Occupational outlook handbook.
Washington, DC: U.S. Government Printing Ofce.
U.S. Department of Labor, Employment and Training Administration. (1991). Dictionary of
occupational titles (Revised 4th ed.). Indianapolis, IN: JIST Works.
Vandenberg, R. J., & Nelson, J. B. (1999). Disaggregating the motives underlying
turnover intentions: When do intentions predict turnover behavior? Human Relations,
52, 13131336.
Wanous, J. P. (1992). Organizational entry (2nd ed.). Reading, MA: Addison-Wesley.
Weinrach, S. G. & Srebalus, D. J. (1990). Hollands theory of careers. In: D. Brown, L. Brooks,
& Associates (Eds), Career choice and development (pp. 3767). San Francisco, CA:
Jossey-Bass.
Weller, I., Holtom, B., Matiaske, W., & Mellewigt, T. (2009). Level and time effects of
recruitment sources on early voluntary turnover. Journal of Applied Psychology, 94,
11461162.
Wessel, J. L., Ryan, A. M., & Oswald, F. L. (2008). The relationship between objective and
perceived t with academic major, adaptability, and major-related outcomes. Journal of
Vocational Behavior, 72, 363376.
Whitbourne, S. K. (1986). Openness to experience, identity exibility, and life change in adults.
Journal of Personality and Social Psychology, 50, 163168.
Yang, Y. (2008). Social inequalities in happiness. American Sociological Review, 73, 204226.
HOW DID YOU FIGURE THAT OUT?

EMPLOYEE LEARNING DURING
SOCIALIZATION
Jaron Harvey, Anthony Wheeler,
Jonathon R. B. Halbesleben and M. Ronald Buckley
ABSTRACT
In this paper, we suggest a contemporary view of learning during the
process of organizational socialization. The relationship between learning
and socialization is implicit in much of the existing socialization
literature. In an attempt to make this research more explicit, we suggest
a theoretical approach to the actual learning processes that underlie
workers socialization experiences. In order to accomplish this, we review
previous work on socialization, information seeking and feedback seeking
during socialization, and learning. In doing so we describe the learning
process that underlies socialization, highlighting the beginning of the
process, the role of information during the process, and integrating three
different types of learning (planned, deutero, and meta) into the process
of organizational socialization. In addition, we also discuss the
implications of these three types of learning during the process of
socialization and directions in future research on the socialization process.

ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029007
167
168
JARON HARVEY ET AL.
Everyone who has ever been employed in an organization, whether on the

rst day or shortly thereafter, experiences that moment when he or she
questions if he or she will succeed in the job or within that organization.
Be it on the rst day, or shortly thereafter, employees want to know how
they are going to learn everything they need to know to survive in
an organization. A number of questions may race through an employees
mind; how will I learn all of the things I am supposed to do, how will
I be evaluated, how do I get promoted, or how will I t in with my
coworkers? Success in a new job, however success is dened, hinges on
employees learning the ropes, and then using their knowledge to successfully
navigate new challenges that arise in organizational life. Upon initial entry
into an organization there is a burst of learning that helps employees to
develop needed work skills and abilities, along with basic group norms and
values (Feldman, 1981). Following that momentary urry of information
there is a much longer period of adjustment and readjustment during which
employees develop a greater understanding of what is really required to
succeed in the organization.
The process of answering these how questions, or the act of learning
how to be successful in a job is part of organizational socialization (referred
to throughout as socialization). Socialization is the process through
which employees transition from being outsiders in the workplace to
being insiders (Van Maanen & Schein, 1979). During this process
employees develop key attitudes, behaviors, and knowledge concerning
how to successfully function as a member of the organization (Bauer,
Morrison, & Callister, 1998). This process, fundamentally a learning process,
is inuenced both by the organization, through different tactics or programs
it may adopt (Allen, 2006; Van Maanen & Schein, 1979), and by employees
personalities and individual efforts to learn about their role in the
organization (Kammeyer-Mueller & Wanberg, 2003).
Early socialization research focused on developing models that explained
the different stages of socialization (e.g., Feldman, 1981; Graen, 1976;
Simpson, 1967; Van Maanen, 1975, 1976), which in turn explained the
development of attitudinal outcomes such as job satisfaction, organizational
commitment, and turnover intentions (Feldman, 1981; Van Maanen, 1975).
More recent socialization research has explored the tactics implemented
by organizations to facilitate the socialization process (Bauer et al., 1998;
Cable & Parsons, 2001). These two streams of research seek to identify
stages of socialization (e.g. Feldman, 1976; Van Maanen & Schein, 1979),
link these stages to important organizational and employee outcomes
(Fisher, 1986), and determine the best way for organizations to inuence
Socialization and Learning
169
this process for the benet of both individual and organization (Cable &
Parsons, 2001).
During the development of socialization research, scholars have scrutinized the experiences of new employees as a means to gain new insights into
this process. Some researchers have looked at the inuence of newcomer
proactivity (Chan & Schmitt, 2000; Kammeyer-Mueller & Wanberg, 2003),
control (Ashforth & Saks, 2000), and involvement (Bauer & Green, 1994)
during the socialization experience, while others have examined the role
of information and feedback seeking during these experiences (Bauer &
Green, 1998; Morrison, 1993a, 1993b; Ostroff & Kozlowski, 1992, 1993).
Recent reviews (e.g. Ashforth, Sluss, & Harrison, 2007; Bauer et al., 1998)
and meta-analyses (Bauer, Bodner, Erdogan, Truxillo, & Tucker, 2007; Saks,
Uggerslev, & Fassina, 2007) attest to the developed nature of this research
area. However, while the study of socialization is vigorously active, one key
area of the socialization process has yet to receive much scholarly scrutiny:
the learning process that occurs during socialization.
The role of learning in socialization has been evident from the beginning,
because it is implied that employees must be learning as they see the
organizational world and begin to inculcate and understand its traditions
(Van Maanen & Schein, 1979). Stage models addressed learning tangentially
by identifying what an employee could expect to learn at any given time in
the socialization process; with all of it culminating in an employee becoming
not only fully capable of carrying out the duties and functions of the job, but
also being enmeshed in the culture of the organization (cf., Schein, 1978).
The research that has examined portions of the learning process during the
socialization experience has focused on information acquisition (Ostroff &
Kozlowski, 1993), sources of information (Ostroff & Kozlowski, 1992),
content of information (Burke & Bolf, 1986), and some of the factors that
can enhance or inhibit employee learning (Morrison & Brantner, 1992).
However, none of this has addressed the entire process through which
employees learn. Individual learning occurs when people make changes in
their mental associations and their behavior based on information gathered
from personal experiences (Ormand, 1999). The process through which
employees make sense of their surroundings likely has a signicant impact
upon how fully enmeshed in an organization an employee becomes through
socialization. From social learning theory (Bandura, 1977) we know that
employees make sense of their roles in an organization by learning from
the situation, the individuals around them, and the formal organizational
practices. As employees try to understand and adjust to a work situation,
they will engage in different learning processes (e.g., Visser, 2007).
170
JARON HARVEY ET AL.
In this paper, we focus on three specic modes of learning: planned, meta-,

and deutero-learning. Each of these modes of learning is a part of the
underpinnings of workers socialization experiences.
Much of the extant socialization literature focuses on traditional notions
of planned learning. Planned learning refers to the development and
maintenance of learning systems that promote employee learning (Visser,
2007). A focus on planned learning has specic limitations in understanding
the nature of socialization because it emphasizes established socialization
programs (e.g., Klein & Weaver, 2000) and established group behaviors (e.g.,
Chen, 2005). To address these limitations, in addition to planned learning,
we consider the other two modes of learning described by Visser (2007):
meta- and deutero-learning. Meta-learning is the processing of inconsistencies that occur between individuals expectations and the actual consequences
of their actions (Argyris & Schon, 1974, 1996). Deutero-learning is a mode of
learning that is adaptive and unconscious (Bateson, 1972). As employees
interact with coworkers and organizational systems they experience
different consequences. It is from these consequences, combined with the
social context that surrounds them, that employees learn how to adapt their
personal behavior (Bateson, 1972; Visser, 2003, 2007). In addition to planned
learning, these two modes of learning provide researchers with a new lens to
view the learning element of socialization.
By concentrating on learning processes, we have chosen not to focus upon
where socialization research has been, which others have established (e.g.
Ashforth et al., 2007; Bauer et al., 2007; Saks et al., 2007), but on a direction
that may move socialization research forward by providing a clearer picture
of how the socialization process works. Our work extends the research on
socialization by focusing on the process that employees use to learn what it
takes to move from organizational outsider to insider. In doing this we seek
to make three important contributions to the socialization literature.
First, we seek to expand the time frame that is normally considered as
the socialization experience. Typically, researchers have focused on the
socialization experience as occurring during the rst 1218 months of an
employees time with the organization (cf., Payne, Culbertson, Boswell, &
Barger, 2008). We suggest that more important than a specic time
period are the boundary-crossing experiences (Schein, 1971) that individuals
go through during the course of a career. Drawing from a boundarycrossing typology, we argue that individuals may have a socialization
experience at any point in their career when boundaries are crossed.
Second, building on the initial trigger of boundary crossings, we use
uncertainty reduction theory (Berger & Calabrese, 1975) and social
171
information processing (SIP) theory (Salancik & Pfeffer, 1978) to determine

what important elements are part of the learning process during socialization. Specically, we describe how boundary crossings will increase workers
motivation to learn and inuence both the sources and types of information available to workers. We also explore the role of tenure within an
organization on motivation to learn and the number of information
sources an employee will have. Based on motivation to learn, sources of
information, and types of information we then discuss about how each of
these elements will inuence the modes of planned, meta-, and deuterolearning during employees socialization experiences. The process described
here proposes to explain what prompts employees socialization experiences,
boundary crossings, and how these crossings inuence the entire learning
process that workers engage in throughout their socialization experience.
Ultimately, we seek to further understand what inuences learning during
the socialization process, and how this learning, driven by motivation and
the sources and types of information available to employees, occurs; thereby
explaining the learning process that underlies the socialization experience of
workers.
Third and nally, we offer recommendations for integrating learning into
future socialization research. Specically, we suggest exploring the relationships between different tactics of socialization and the different modes of
learning that we discussed in this paper. Additionally, because we suggest
that socialization experiences occur at multiple times during a career,
not just when employees take new jobs, we suggest some avenues through
which the links between socialization experiences and the different modes
of learning that individuals engage in can be explored. Finally, we also
discuss some of the possible measurement issues that researchers may
encounter as they move forward with this work.
HOW WE KNOW WHAT WE KNOW:

SOCIALIZATION AND LEARNING
To better understand the relationship between socialization and employee
learning we draw from existing research and theory in each of these areas to
develop an understanding of how learning occurs during a socialization
experience. We begin by briey reviewing the general socialization literature.
We then concentrate on important elements of learning in the socialization
literature; namely information and feedback seeking. Following this,
172
JARON HARVEY ET AL.
we discuss boundary crossings. In this section, we describe what boundary

crossings are and why they serve as a trigger for the learning process during
socialization experiences. Next, using uncertainty reduction theory (Berger,
1979) and SIP theory (Salancik & Pfeffer, 1978), we consider how boundary
crossings increase employees motivation to learn, as well as the number of
information sources available to employees, and we also discuss the role
tenure may play during this learning process. Finally, we draw from the
learning literature to discuss the role of learning in organizations, paying
particular attention to social learning theory (Bandura, 1977). We then
focus on three modes of learning, planned, meta-, and deutero-learning and
their role of processing information that employees gather during the
socialization experience.
Learning the Ropes: A Socialization Overview

Any anxiety employees experience as they enter organizations is a function
of the uncertainty they are feeling. The more uncertainty individuals
experience, the greater their level of anxiety will be about the situation
(Berger, 1979; Berger & Calabrese, 1975). Employees entering an organization take up identities that reect their new job roles and organizational
surroundings. This process of learning a new role, a new organizational
culture, and the accompanying requirements, and transitioning to the
status of organizational insider is referred to as organizational socialization
(Feldman, 1976). Through this process of learning about their new roles
and understanding organization expectations, employees come to a better
understanding of how to do their jobs and what they need to do to t into
the organizational culture (Van Maanen & Schein, 1979). It is during
socialization that new employees nd out what the organization is like and
determine whether they want to be a part of it. As employees learn about
their job and the organization, any anxiety felt as an outsider gives way to
the knowledge and certainty of becoming an insider.
As the socialization process commences, organizations employ different
types of socialization tactics to inuence employees experiences (Saks &
Ashforth, 1997; Van Maanen & Schein, 1979). Socialization tactics are the
methods used by organizations to socialize their employees (Van Maanen &
Schein, 1979). Van Maanen and Schein (1979) specied six different
socialization tactics: collective-individual, formal-informal, sequentialvariable, xed-random, serial-disjunctive, and investiture-divestiture.
Research suggests that these six different tactics exist on a continuum,
173
which ranges from institutionalized to individualized (Bauer et al., 1998;

Jones, 1986). Institutionalized tactics (e.g., collective, formal, sequential,
xed, serial, and investiture) reect an organized program of socialization,
designed to familiarize employees with the organization and encourage
them to accept existing organizational norms. Organizations use individualized tactics (e.g., individual, informal, variable, random, disjunctive,
and divestiture) in the absence of a structured socialization program.
Individualized tactics place workers in situations where they are responsible
for themselves, pushes them to challenge the status quo, and tackle
situations themselves (Ashforth & Saks, 1996). Broadly speaking, then,
organizations may apply an individual or institutional approach in the
socialization of employees (Jones, 1986; Van Maanen & Schein, 1979).
Depending on the desired outcomes, some organizations may prefer tactics
that allow for customization, while others may desire uniformity in
socialization practices.
Each of the different socialization tactics focuses on a distinct aspect of
the socialization experience (Van Maanen & Schein, 1979). Collective tactics
place employees in groups during the socialization experience, while
individualized tactics create unique learning opportunities for each employee.
Formal tactics focus on separating newcomers from current employees, while
informal tactics place the new person in an on-the-job training situation.
Organizations using sequential tactics employ a predetermined sequence of
events to provide employees with explicit information. On the other hand,
there is no specic order or explicit information communicated to employees
when organizations use random tactics. Organizations that use xed tactics
provide precise information about each timetable for the assumption of a
role, while variable tactics provide no indication about when employees will
be ready to assume the roles of their job. Serial tactics focus on using an
experienced organizational member to provide a role model for employees
to look to as they are socialized. Disjunctive tactics do not use a role model
during the socialization of employees. Finally, to afrm the incoming identity
and personal characteristics of employees, organizations will use investiture
tactics. Conversely, divestiture tactics concentrate on striping away this
identity and these characteristics (Ashforth & Saks, 1996; Cable & Parsons,
2001; Jones, 1986; Van Maanen & Schein, 1979).
Overall, then, socialization is the process by which employees become
familiar with new roles and organizational cultures (Van Maanen & Schein,
1979). Organizations employ different tactics to socialize workers, and
through these tactics, they can create either an institutionalized socialization
experience, individualized socialization experience, or a combination of the
174
JARON HARVEY ET AL.
two. In the next section, we focus on research about the role of information
and feedback seeking during the socialization experience.
Bits and Pieces: The Roles of Information and

Feedback during Socialization Experiences
While little research has been conducted about the role of the actual learning
process in socialization, a good deal of research has been done that is related
to the learning process. The topics of information seeking and feedback
seeking (e.g., Ashford & Black, 1996; Kim, Cable, & Kim, 2005) have
been at the center of several investigations of the socialization process (cf.,
De Vos, Buyens, & Shalk, 2003; Kim et al., 2005; Morrison & Brantner,
1992). These studies are important in understanding the learning process,
which underlies socialization experiences, because they focus on one of
the fundamental element of learning: information. Information is what
individuals gather from their experiences, which then shape thoughts and
behaviors (Ormand, 1999). While none of these studies has focused on the
cognitive processes of learning, they have concentrated on various aspects of
information, such as sources of information, the acquisition of information,
and the context around individuals providing information (e.g., Burke &
Bolf, 1986; Morrison & Brantner, 1992; Godshalk & Sosik, 2003). The
ndings of these studies suggest how important information is to the
socialization experience.
Prior socialization research on information and feedback seeking has
focused on studying these variables in the context of new hires (e.g., De Vos
et al., 2003). Some of this work has concentrated on specic populations of
new hires, like trainees (Fedor, Rensvold, & Adams, 1992) and other work
has examined the mentorprotege relationship (Godshalk & Sosik, 2003).
Most of these studies have looked at the role of information, either the
seeking of it or the type of information, and its relationship with various
socialization outcomes (e.g., De Vos et al., 2003; Fedor et al., 1992;
Godshalk & Sosik, 2003). Because information plays a critical role in the
learning process, without it individuals would not have a reason to change
their behavior (Ormand, 1999), it is only natural that it plays an important
role in socialization research. In a general sense, during a socialization
experience, information aids workers in understanding what they should do
to fulll their role and be successful in their circumstances. Information also
reduces an individuals level of uncertainty about what types of behaviors
are appropriate in his or her new setting (Ashford & Black, 1996).
175
Feedback, which is a specic type of information, helps employees to know

what others think of them, so they can adjust their behavior in a way that
helps them obtain desired outcomes (Ashford & Black, 1996; Greenberger,
Strasser, & Lee, 1988). Employee learning occurs as employees have
experiences that provide both the information and feedback necessary for
them to more fully understand what it takes to survive in their new position
or new organization.
Early studies were unable to nd a strong relationship between information
seeking and attitudinal outcomes such as job satisfaction (e.g., Ashford &
Black, 1996; Bauer & Green, 1998). Later work, which integrated
information seeking as an intermediate step between socialization tactics
and outcomes (e.g., Cooper-Thomas & Anderson, 2002) were more successful
in establishing that a signicant relationship may exist between information
seeking and various socialization outcomes. In addition to searching for links
between information seeking and typical employment outcomes, researchers
have also examined where employees look for information. Studies have
found that employees turn to different sources of information depending on
what type of information they need (Burke & Bolf, 1986; De Vos et al., 2003;
Morrison, 1993b; Ostroff & Kozlowski, 1992, 1993). For example, an
engineer is more likely to turn to a fellow engineer rather than a manager,
when he or she needs help with understanding a technical part of his or her
job. Thus, sources and types of information are inextricably linked to one
another. Other research has shown that workers have different methods for
acquiring information, such as direct questioning or observation (Burke &
Bolf, 1986; Major, Kozlowski, Chao, & Gardner, 1995; Teboul, 1995).
Studies have also shown that there is a social cost associated with acquiring
information (Holder, 1996; Teboul, 1995). Additionally, there are many
positive outcomes, such as increased role clarity, job mastery, job satisfaction,
and organizational commitment, associated with information seeking (e.g.,
Cooper-Thomas & Anderson, 2002; Finkelstein, Kulas, & Dages, 2003;
Holder, 1996; Saks & Ashforth, 1997). From these ndings, it is clear that
information seeking plays an important role in the process of socialization.
Information will always be useful for workers as they seek to reduce the
general uncertainty they have about an organization and their position
within it (Ashford & Black, 1996). However, employees who are concerned
with their individual performance and controlling personal outcomes will
focus on information that is specic to their performance; this type of
information is feedback (Greenberger et al., 1988). The research investigating feedback seeking as part of the socialization process has found that it is
positively related to important attitudinal variables such as job satisfaction
176
JARON HARVEY ET AL.
and intention to turnover (Saks & Ashforth, 1997; Kammeyer-Mueller &

Wanberg, 2003). One important antecedent of feedback seeking is
performance; specically Fedor et al. (1992) found that a low performance
would increase the likelihood that trainees would seek feedback. Thus,
feedback seeking is an important element of the socialization experience as
workers strive to learn what it takes to become accepted as a member of the
organization. As employees have socialization experiences, and nd they
would like better outcomes (such as a higher level of performance or a
greater understanding of the organizational culture) they are more likely to
seek feedback.
Learning occurs as employees acquire knowledge, which changes the way
they think about things or how they behave (Ormand, 1999), and as the
research on information and feedback seeking in socialization demonstrates
many employees actively seek out information that helps them to learn how
to become more fully enmeshed in the organization. In a study of college
graduates who had been employed less than a year, Morrison (1995) found
that employees are more likely to seek out the information they need, rather
than passively waiting for it to be given to them. This study, among several
others, offers a different image of employees socialization experiences.
Much of the early work on socialization assumed that employees were simply
sponges that were passive during the socialization process and just absorbed
whatever information the organization passed to them (Miller & Jablin,
1991). However, several studies have found that employees who proactively
assert themselves during the socialization process have many positive
outcomes (e.g., Gruman, Saks, & Zweig, 2006; Kim et al., 2005; Saks &
Ashforth, 1996, 1997; Kammeyer-Mueller & Wanberg, 2003). Indeed, the
base of much of the research on information seeking and feedback seeking
is employee proactivity. For example, in a study of new employees during
their rst six months in the organization, Morrison (1993a) found a positive
association between the frequency of information seeking and the outcomes
of job mastery, role clarity, and social integration. This work suggests that as
employees are proactive in their efforts to obtain information about their
roles in the organization and feedback about their performance, they are
more likely to obtain the status of organizational insider.
From prior work on information and feedback seeking in the socialization
literature, we can see that employees who have socialization experiences will
seek out information to help them become a part of their organization (e.g.,
Morrison, 1993a). Information plays a key role in the learning process
because it helps workers adjust their thinking and behaviors, which can lead
them to be successful in their new position. As a result of previous work we
177
know what socialization is, and the key role that information plays during
employee socialization experiences (i.e., the more information an employee
has the better his or her socialization experience). In the next section, we
discuss boundary crossings and the role they play in socialization experiences.
The Trigger of Socialization Experiences: Boundary Crossings

While most socialization research has focused on the initial point of
organizational entry (cf., Feldman, 1976; Kammeyer-Mueller & Wanberg,
2003; Van Maanen, 1975), socialization experiences can occur at other
points during an employees career (Ashforth et al., 2007; Van Maanen &
Schein, 1979). In our discussion of boundary crossings, we seek to
emphasize that boundaries can be crossed at any time during an employees
career. Consequently, socialization experiences can occur at any time and
are not limited to just the time during organizational entry. Past research
suggests that socialization occurs at boundary crossings or times of
transition, because it is during these occurrences that employees are most
receptive to prompts about what he or she should be learning to become
more fully entrenched in the organization (Ashforth et al., 2007; Van
Maanen & Schein, 1979). Schein (1971) proposed that individuals cross
vertical, horizontal, and inclusionary boundaries in the work environment.
A vertical boundary crossing is most likely to occur in the context of a
promotion, although demotions could have a similar impact. Horizontal
crossings may occur with a job transfer or any reorganization that shifts the
employee to a new department. The crossing of inclusionary boundaries is
marked by transitioning closer to the core of the organization where power
and decision-making capabilities are located; these crossings transform the
employee into more of an insider (Louis, 1980; OHara, Beehr, & Colarelli,
1994; Schein, 1971). During each of these boundary crossings, there is a
greater likelihood that individuals will be more open to socialization
experiences (Van Maanen & Schein, 1979). This is because they will be
searching for anything that reveals what it takes to succeed in the
organization. The initial introduction to an organization is the most well
documented and intensely scrutinized boundary crossing because it is a
socialization experience involving the crossing of all three boundaries
(Louis, 1980); however, boundaries can be crossed at any time in employees
careers. For this reason, we contend that the socialization process never
stops, and employees are continually learning what it takes to be a successful
contributor to an organization.
178
JARON HARVEY ET AL.
Vertical Boundary Crossings

During the course of individuals careers, they are most likely to experience
vertical boundary crossings, which are the increasing or decreasing of
ones level in the organizational hierarchy (Schein, 1971, p. 403) through
various promotions within the rm. As individuals ascend the organizational ranks, they will acquire new job roles, interact with different
individuals, and nd their existing relationships altered. Conversely, if for
some reason employees are demoted, they will also experience a change in
job roles and a change in their interactions with coworkers and supervisors.
Organizational level events that may precipitate a vertical boundary
crossing are mergers, reorganizations, or even a layoff event. In the event
of a merger or reorganization, employees may be shifted around, given
new job roles, or placed in new supervisory positions. During a layoff an
employee could remain with the organization, but be reassigned to a
different position. Each of these events may cause employees positions in
the organizational hierarchy to be changed. These same changes can occur
for employees who retain their jobs after a layoff. Additionally, research
suggests that survivors of a layoff may have other types of adjustments that
are specic to the organization and coworkers because of the layoff
(e.g., Brockner et al., 1994). The crossing of a vertical boundary creates a
need for individuals to learn or relearn what is expected of them and how
they t in as insiders again.
Horizontal Boundary Crossings
In the ever-changing work environment, employees may nd themselves
shifting around to other positions, working in new teams, or otherwise
experiencing different types of horizontal movements. Horizontal boundaries are crossed when employees functions change or they are moved from
one division, or group, to another (Schein, 1971). There are several different
career events that may cause employees to cross horizontal boundaries:
reorganization, international assignments, repatriation, special assignments,
rotations to other work units, and so forth. Each of these events may result
in employees learning new skills or competencies that are relevant to their
new roles. Crossing a horizontal boundary may result in more than a simple
change in job roles; it may include special training or development that the
employee requires for broader experience within the organization (Schein,
1971). In the case of an international assignment, individuals may be
performing the same job but need to learn a language or other skills to help
them adjust (Black, Mendenhall, & Oddou, 1991). During the repatriation
process, employees cross horizontal boundaries as they move geographic
179
areas, and often they nd themselves facing a different job as a repatriate or

other challenges as an outsider in their new position (Bolino, 2007). Some
employees may be part of management development programs that rotate
them through different functional areas of the organization. These types
of lateral movements can provide upending experiences that create a need
for employees to adjust their behavior to both new coworkers and a new
work environment. Indeed, employees who experience transfers within the
organization tend to seek more feedback (Kramer, 1993; Kramer & Noland,
1999). Workers who are placed in new situations due to horizontal
boundary crossings will need to learn about job duties, those around them,
and whatever else it takes to become an insider.
Inclusionary Boundary Crossings
The nal type of boundary that individuals may cross during their time in the
workplace is the inclusionary boundary. This boundary represents the degree
to which other members of the organization trust and accept an employee
(OHara et al., 1994; Schein, 1971). Crossing inclusionary boundaries may be
the true mark of an individual who is considered an organizational insider.
Many of the career-changing events previously discussed are likely to cause
employees to cross this boundary because it may bring about changes in the
supervisors or coworkers surrounding an employee. While employees can
work toward crossing inclusionary boundaries, it is ultimately not up to them
whether they will cross this boundary or not; it is up to those around them
(OHara et al., 1994). Inclusionary boundary crossings can be more difcult
to recognize than vertical or horizontal crossings, but OHara et al. (1994)
note that shifts in power, access to sensitive information, and decisionmaking abilities are all likely to change during an inclusionary boundary
crossing. It is plausible that employees needing to cross inclusionary
boundaries must be perceived as trustworthy (Mayer, Davis, & Schoorman,
1995) to develop the type of relationships that allow them to nally cross
this boundary. Mergers, reorganizations, job rotations, international assignments, working in ad hoc groups, and other similar movements can push
employees across inclusionary boundaries because they are required to learn
how to interact with new personnel.
A good example of an inclusionary boundary crossing, which does not
occur in conjunction with any vertical or horizontal boundary crossings,
is a change in leadership. A change in leadership may cause a shift in power,
access to sensitive information, and decision making, which will create
a state of uncertainty for many employees. Although employees may have
been insiders prior to a leadership change, this change could shift them to
180
JARON HARVEY ET AL.
outsider status again. Because of the increased frequency in organizational

turnover (Hom, Roberson, & Ellis, 2008), employees are likely to nd
themselves crossing inclusionary boundaries several times during their
careers. We offer a guideline for boundary crossings in future research and
next discuss the consequences of crossing these boundaries in relation to
workers motivation to learn, information, and tenure.
Guideline 1. Researchers should seek to differentiate between types of
boundary crossings in future socialization research.
On the Other Side: Motivation, Information, and Tenure

We suggest that crossing boundaries will have two outcomes for workers:
rst, it will inuence their motivation to learn and second, it will place them
in contact with new sources of information. To gain a better understanding
of boundary crossings, and the effects of these crossings we turn to
uncertainty reduction theory (Berger, 1979) and SIP theory (Salanik &
Pfeffer, 1978). Prior work in the socialization literature has used uncertainty
reduction theory to explain the behaviors of employees during socialization (e.g., Kramer, 1994; Lester, 1987; Mignery, Rubin, & Gorden, 1995),
because it focuses on the reaction of individuals when they experience
uncertainty, and it suggests what individuals will do to reduce their
uncertainty. We also turn to the SIP theory (Salancik & Pfeffer, 1978) to
determine how the work environment around employees is important during
the socialization process. SIP theory holds that coworkers, other individuals, and the workplace surrounding employees provides critical information for employees as they work to make sense of their situation (Jex & Britt,
2008; Salancik & Pfeffer, 1978). Drawing from these two theories, we can
further understand the role of learning in the socialization process.
Uncertainty Reduction Theory
Uncertainty reduction theory proposes that individuals have an aversion
to uncertainty and seek to gather information in an effort to reduce their
uncertainty. People seek to reduce uncertainty because it is associated with
anxiety. Thus, the reduction of uncertainty is critical to lowering the levels
of anxiety that employees are feeling because of the situation they are in
(Berger & Calabrese, 1975). People seek to reduce uncertainty for two main
reasons. First, they want to be able to predict the behavior of the individuals
around them. Knowing what others are going to do reduces the anxiety
181
associated with uncertainty. Second, employees want to be able to explain

the behavior of those around them. Knowing why people have behaved in
a certain way reduces the anxiety individuals experience when interacting
with these people (Berger, 1979; Berger & Calabrese, 1975). Any time an
employee meets a new person or encounters an unknown situation there
is some uncertainty, because he or she is unable to predict or explain the
behavior of this new person or unknown situation. The greater amount of
uncertainty workers feel in their situation the more likely they are to take
actions that will reduce this uncertainty (Berger & Calabrese, 1975). Thus,
individuals who experience new situations, such as when they enter an
organization, are motivated to reduce the amount of uncertainty that
surrounds them (Saks & Ashforth, 1997). For this reason, we believe that
uncertainty reduction theory helps to explain the learning process that is the
foundation of the socialization experience.
Social Information Processing Theory
SIP theory, developed in the late 1970s as a response to job enrichment
models of job satisfaction (i.e., Hackman & Oldham, 1976), proposes that
the satisfaction of workers is inuenced by the context of the situation and
the consequences of past choices (Salancik & Pfeffer, 1978). Fundamentally,
this theory suggests that employees attitudes, behaviors, and beliefs are
inuenced by the work environment around them (Salancik & Pfeffer,
1978). As such, coworkers, friends, and even the situation itself can inuence
an individuals attitudes or behaviors at work. Workers who seek to make
sense of their world will look for salient, relevant, and credible
information from the social environment around them (Zalesny & Ford,
1990). Studies show that this information, which is gathered from the social
environment, inuences the perceptions, attitudes, and beliefs about ones
self, job, and organization (Ibarra & Andrews, 1993). As such, we expect
that SIP theory will serve to guide us in understanding what is important
for learning during socialization experiences.
Motivation to Learn
Returning to boundary crossings, we know that each boundary-crossing
experience creates uncertainty for the employee. We suggest that this desire
to reduce uncertainty will increase workers motivation to learn. Motivation
to learn is a desire to engage in training, learn job-related content, and take
part in development activities (e.g., Carlson, Bozeman, Kacmar, Wright, &
McMahan, 2000). Prior work links motivation to learn with certain
individual characteristics (Major, Turner, & Fletcher, 2006), but we also
182
JARON HARVEY ET AL.
posit that specic situations, such as boundary crossings, will increase

individuals levels of motivation to learn. These boundary crossings prompt
the beginning of the learning process, and because they increase levels of
uncertainty we believe that employees will be more motivated to learn; thus
we offer this guideline for future socialization research.
Guideline 2. Researchers should strive to measure employees motivation
to learn in future socialization research.
Information
The reduction of uncertainty is one of the primary reasons that individuals
seek information during the socialization process (Miller & Jablin, 1991). As
employees strive to become insiders in the organization by reducing the level
of uncertainty they feel about their position and the organization, SIP theory
suggests they feel they will seek to gain information from the environment
around them (Zalesny & Ford, 1990). To nd this information workers will
be able to turn to the new supervisors, coworkers, mentors, and written
documents (Miller & Jablin, 1991; Morrison, 1993b; Ostroff & Kozlowski,
1993) that surround this new position. For example, employees who are
engaged in an institutionalized socialization experience can turn to other
coworkers and ask questions that are relevant and important to gaining
information about their job, work role, or other related content. Mentors can
also be a critical component of the socialization process because they are
an important source of information for employees during socialization
experiences (cf. Ostroff & Kozlowski, 1992).
However, some employees may not always ask others when acquiring
information, because they worry about how this might look to their
supervisor or colleagues (Morrison, 1993b). Instead, they may simply
observe the individuals, and strive to glean relevant information from the
actions of these persons (Morrison, 1993a; Ostroff & Kozlowski, 1993), or
turn to the written documents that are relevant to their position (Miller &
Jablin, 1991). The sources of information available to a worker may be
determined by the tactics the organization uses while socializing the worker.
If an organization uses individual tactics and separates the employee from
his or her coworkers (Cable & Parsons, 2001) he or she will not be able to
use those coworkers as sources of information. Therefore, while boundary
crossings will generally provide new sources of information to employees,
the socialization tactics used by an organization may also inuence the
number of sources available. For this reason, we suggest that future work
differentiate between sources of information.
183
Guideline 3. Researchers should seek to differentiate between sources of

information in future socialization research.
Workers who have more sources of information will also have more types
of information available to them. Miller & Jablin (1991) have identied
three types of information that workers look for during a socialization
experience: referent, appraisal, and relational. Referent information helps
employees know what is essential for them to perform well in their assigned
roles (Hanser & Muchinsky, 1978; Miller & Jablin, 1991; Morrison, 1993b).
The next information type, appraisal information, assists individuals in
understanding if they are successful in their job performance (Miller &
Jablin, 1991). Relational information is the nal type of information, and it
deals with the relationships between employees and other organizational
members (Miller & Jablin, 1991). This type of information helps individuals
to understand their standing within organizational groups, be it with their
boss, fellow coworkers, or subordinates. In these relationships, learning
from context and nonverbal cues is critical to help employees read between
the lines and understand the true nature of the relationship. The different
types of information that an employee collects during the socialization
process has an important inuence on employee learning. Because SIP
theory proposes that worker attitudes, behaviors, and beliefs are shaped
by the social environment surrounding them (Salancik & Pfeffer, 1978),
we believe that these new sources of information and different types of
information are vital to employees as they navigate the socialization process,
and we suggest the following.
Guideline 4. Researchers should seek to differentiate between types of
information in future socialization research.
Tenure
One important factor that will inuence the socialization process over the
career of an employee is tenure, both tenure in the organization and tenure
in a position. Uncertainty reduction theory suggests that time is important
to reduce the anxiety that people feel because of uncertainty (Berger, 1979).
Over the course of time, individuals are able to gain knowledge about
behaviors, which helps to reduce uncertainty (Berger & Calabrese, 1975).
If an individual remains in a single organization for an extended period, his
or her motivation to learn is likely to decrease because he or she will have a
base of knowledge about how things work, which will reduce uncertainty
when crossing boundaries in that organization. Indeed, the more familiar
184
JARON HARVEY ET AL.
individuals are with situations and the people around them the less
uncertainty they will feel (Berger, 1979). As employees spend more time in
an organization, or more importantly a single position, their familiarity with
both the duties of the job and other employees increases. This results in a
reduced level of uncertainty, and we suggest that consequently the employee
will have less motivation to learn.
While tenure may diminish employees motivation to learn, it could
increase the number of sources from which employees can get information.
The more interactions people have and the more knowledge they
gain, the more they will be able to reduce their uncertainty (Berger &
Calabrese, 1975). The longer employees stay in a single organization, the
greater the number of colleagues they will have and the more interactions
they will have had with these individuals. Employees will have more time to
observe, try new behaviors, and learn about the organizational policies
and procedures. The greater a workers familiarity with these sources, the
less likely he or she is to experience uncertainty when faced with boundary
crossings. We expect that tenure in an organization will positively inuence
the number of information sources to which workers have access, which will
be helpful to employees when crossing boundaries during a socialization
experience; therefore we offer the following guideline for future research.
Guideline 5. Future socialization research should seek to account for
tenure interactions when employees are changing to new positions within
the same organization.
The Learning Process: A Brief Review
As workers seek to gain insider status, they must learn how to behave in
their new workplace. Learning is the permanent change that occurs in an
individuals cognitive associations or personal behaviors due to the
experiences he or she has (Ormand, 1999). The behavior of workers in
organizations is a function of employees cognitive processes and the
environment around them (Davis & Luthans, 1980). The interaction between
workers cognitions and the situation around them, learning, is what shapes
how they behave in the workplace. To better understand the learning process
that employees experience during socialization we turn to social learning
theory (Bandura, 1977) to help us grasp how the learning process functions
in organizational settings. Social learning theory is a behavioral theory that
explains how individuals learn in organizations. It incorporates some of
the principles of operant conditioning (Skinner, 1969), along with the social
185
environment around employees. Thus, it addresses both the personal

experiences, which provide direct learning opportunities, along with the
social context around the employees.
Social learning theory (Bandura, 1977) suggests that the learning process,
which is a continuous reciprocal interaction of individuals thoughts, their
behaviors, and their environmental situation, is the one of the best ways
to explain the actions of individuals. Bandura labeled this process a triadic
reciprocality; through this process individuals process information, learn,
and then adjust their behavior accordingly (Bandura, 1977, 1986). It is
important to point out that individuals do not simply react to the
environment around them, nor are their actions determined solely by past
experiences. Individuals consider the situation, and then using the information available from both the environment and past experience they use
forethought in deciding how to behave (Bandura, 1986). According to social
learning theory, workers can learn how to behave both through their own
direct experience and by observing the behavior of others (Bandura, 1986).
Both experiential and observational learning have important implications
for the socialization process, although each of these types of learning is very
different. Because social learning theory incorporates the principles of
traditional operant theories of behavior, which describes learning through
personal experience (e.g., Skinner, 1969), and it proposes that learning also
takes place through the observation of others behavior (Bandura, 1977) it is
especially applicable to socialization experiences. Organizations can adopt
an institutionalized or individualized approach, or some combination of the
two, to the socialization of their employees. These different approaches will
result in both experiential and observational opportunities. If an organization uses informal socialization tactics that force employees to learn on
the job (Cable & Parsons, 2001) it is more likely that employees will be
engaged in experiential learning. However, organizations that use serial and
investiture tactics, which provide employees with role models to watch
(Cable & Parsons, 2001), create situations where observational learning is a
key part of the socialization experience. From these examples, we conclude
that both observational and experiential learning are critical to learning
during socialization.
One of the strongest pieces of research to address the link between
socialization and the actual process of learning was a study of British Army
recruits that demonstrated the central role of the learning process in socialization. Findings indicated that the relationship between socialization tactics
and outcomes, job satisfaction and organizational commitment, was fully
mediated by the level of newcomer learning (Cooper-Thomas & Anderson,
186
JARON HARVEY ET AL.
2002). As we have discussed the learning process and socialization experience are interrelated; this is because individuals need to learn
in order to be socialized. However, whereas learning is about using
information acquired through different experiences and observations to
inform future behaviors (Ormand, 1999), socialization focuses on employees
becoming more familiar with the organization, their job, and those with
whom they interact (Van Maanen & Schein, 1979). Hence, the two processes
have different outcomes. Individuals learn based on things that they observe
or experience, with the purpose of the learning being that they know how to
behave in a given situation (Bandura, 1986; Ormand, 1999). The socialization experience focuses on integrating workers into their jobs by helping them
to learn the ropes (Ashforth et al., 2007). We further suggest that the way
individuals learn plays an important role in the socialization experience. We
next focus on three different modes of learning to help us better understand
how individuals make sense of their new roles in organizational settings.
How the Wheels Turn: Types of Learning

Because the goal of the socialization experience is to change from an outsider
into an insider, through becoming familiar with job roles and organizational
culture, employees who seek to accomplish this will seek to draw information
from both their experiences and observations. Central to our paper are the
three different modes of learning that help individuals to process
information. In a reformulation of organizational learning types, necessitated by the organizational learning jungle (Huysman, 2000, p. 81), Visser
(2007) offered a typology of three modes of learning used in organizational
settings. These are planned, meta-, and deutero-learning. By considering each
of these learning processes, we can gain a better understanding of
socialization experiences.
Planned Learning
Planned learning occurs when organizations create and maintain a system,
routine, procedure, or structure with the specic purpose of inducing
organizational members to learn (Schon, 1971, 1975). Planned learning is
the type of learning that organizations rely on to create better employees.
When an individual enters an organization, there are certain processes
and rituals that he or she experiences. Many individuals spend their rst
few hours as newcomers in an organization, reading forms and lling out
required paperwork. There may be hours or days of specied training about
187
how their job is properly preformed. The processes of orientation and other
forms of training are part of a broader effort to inuence and teach
employees what it takes to become an insider in the organization (Van
Maanen & Schein, 1979). By utilizing these types of systems organizations
require employees to engage in planned learning.
The most important element of planned learning is that it inuences
employee behavior (Visser, 2007). The incorporation of planned learning
leads employees to alter their behavior in some way to better fulll job
roles or t into a work group. Planned learning is most likely to occur
when organizations take an institutionalized approach to socialization. For
example, both sequential and xed tactics of socialization can facilitate to
planned learning. Employees are given explicit information about the
activities they will participate in when sequential tactics are employed and
xed tactics provide precise information about what will occur during each
stage of the socialization process (Ashforth & Saks, 1996; Van Maanen &
Schein, 1979). Additionally, planned learning is the ideal way for gaining
referent information. This is because referent information lends itself to
communication through an established routine or system that helps workers
to know explicitly what is expected of them in their jobs. Other implications
for the use of planned learning during a socialization experience are that an
organization can adjust and adapt the socialization process based on the
outcomes, such as employee role mastery, to obtain the desired results.
Planned learning has the potential to induce meta-learning among employees
because of the creation and maintenance of learning systems (Visser, 2007).
Because of this, planned learning is a critical component for organizations
that desire to inuence employees during the socialization process and we
suggest the following:
Guideline 6. Researchers should theorize about the circumstances under
which planned learning will be most effective during the socialization
process. In contemplating these circumstances, they should focus on the
desired outcomes of the socialization process.
Meta-Learning
The next mode of learning that underlies socialization experiences is metalearning. When a specic course of action fails to produce the desired
results, individuals will try to learn what went wrong. This discrepancy,
between the desired and actual results, is where the learning process begins
(Visser, 2007). The process of meta-learning occurs when an individual
begins to process the inconsistency between expectations and actual
188
JARON HARVEY ET AL.
consequences, searching for reasons why it may have occurred. This

learning process results in two levels of inquiry, typically referred to as
single- and double-loop learning (Argyris, 2003; Argyris & Schon, 1978).
The initial level of learning in the process of meta-learning, single-loop,
involves reection to better understand how to detect errors, and thoughts
about how to effectively correct those errors (Argyris & Schon, 1974; Visser,
2007). For example, if an individual nds himself or herself in the out-group
due to consistent tardiness, there may be reection on how this could have
been anticipated. Upon reection, the individual may remember that a
supervisor and other coworkers made several comments about the
importance of being timely in this organization. Initially these comments
were simply brushed off, because the individual did not consider his or her
level of tardiness to be relevant. In retrospect, the individual sees that his or
her coworkers were making subtly important suggestions about how to t
in. Moving forward this individual plans to be aware that even seemingly
inconsequential remarks may have important implications. From this
example, we can see that the individual focused on how the error could have
been detected and how similar errors can be avoided in the future; this is the
purpose for rst level of learning during meta-learning. Thus, such learning
is essential as individuals work to understand their job and the culture
around them during socialization experiences.
The second level of learning concerns seeking ways to improve the
understanding of the norms and values of the situation (Argyris, 2003;
Crossan, 2003). Continuing the previous example, the employee may engage
in double-loop learning by starting a conversation about why being on time
is so important, and perhaps if possible, some consideration of extime by
the organization. An individualized approach to socialization (Jones, 1986) is
likely to lead employees to engage in high levels of meta-learning. Because an
individualized approach is unstructured and unique for each employee, they
will need to reect on both their experiences and observations in order to
learn what they need to do to become fully enmeshed in the organization. In
sum then, meta-learning describes how individuals learn from the differences
between the actual and expected outcomes of the socialization process. Metalearning will help individuals detect errors and aid them when looking for
ways to improve the understanding of the norms and values of a situation.
Meta-learning provides valuable insight into the learning process by
describing how employees alter both their thoughts and behavior. It is a
conscious process that involves employees stopping to think about the
intended consequences of their actions (Argyris & Schon, 1996). As
employees gain more information about different job roles, group processes,
189
or organizational attributes, they will enact that knowledge through their

behavior. When those behaviors do not result in the anticipated results, the
employee will stop and think about what has happened. Why has it not
worked? or What did I miss? are a few of the questions an individual
may ask. After some consideration, workers may make adjustments, begin
experimenting, and looking for ways to improve (Visser, 2007). If an
organization uses divestiture tactics to strip away the personal characteristics of employees (Van Maanen & Schein, 1979), the employees could use
meta-learning to process what this means for them as individuals and decide
what types of behaviors they need to change. This type of learning may be
especially benecial when workers are striving to master job-related
content or negotiate group interactions as they endeavor to be accepted in
the organization. As discrepancies between workers expectations and the
outcomes occur, they will reect on this appraisal information. By engaging
in meta-learning with appraisal information, employees will be able to
discern what types of changes they need to take in order to correct their
performance so they can obtain their desired outcomes. Thus, meta-learning
will be critical as employees seek to better understand their jobs, the
organizational culture, and how to succeed in their new position. Thus we
suggest the following guideline for future research.
Guideline 7. Researchers should theorize about how meta-learning can be
better integrated into the socialization process. Researchers may want
to focus on ways to add specic moments for meta-learning into existing
planned-learning processes.
Deutero-Learning
The context that surrounds an employee provides a wealth of information
for employees as they attempt to learn about their job and place in the
organization. Deutero-learning draws information from context-specic
events and nonverbal cues (Visser, 2007). At an unconscious level,
individuals are constantly learning and absorbing knowledge about the
organization, acceptable forms of behavior, expectations, requirements,
and other nuances of organizational life. Deutero-learning is based on two
main principles: all systems are capable of adaptation, in which learning is
inherent, and learning occurs through an interaction (Bateson, 1963, 1972).
First, all systems are capable of adaptation that requires learning. All
systems, be they biological, social, ecological, or organizational, change over
time. This change occurs as the system adapts to the existing environment.
Inherent in this adaptation is learning about what works and what does
190
JARON HARVEY ET AL.
not; with those who learn and adapt continuing their existence. The second
principle of deutero-learning is that learning occurs through an interaction.
Individuals do not learn in isolation, there must be an interaction with
something or someone in order for learning to occur (Bateson, 1972; Visser,
2007). Employees learn by witnessing or experiencing event A in the context
of X. The interaction of A and X creates information, which then facilitates
learning. The results of this learning will be extended to similar situations
where it will be maintained, or altered depending on the results of the new
interaction and the patterns of behavior that develop from it (Bateson, 1972).
Three main factors characterize deutero-learning. First, it is a continuous
process of behavioral communication that is mostly unconscious (Haley,
1963; Watzlawick, Bavelas, & Jackson, 1967) and can be both outward
and incidental in nature. Interactions occur continuously as individuals are
around one another and communicate with each other. Simple messages
provide learning experiences by creating the context of a situation and
through the elicitation of a response and nonverbal cues. We sometimes
know that we are learning and at other times we do not know we are
learning. The second feature of deutero-learning is that it cannot be overtly
controlled. Attempts to control a deutero-learning situation create a new
learning process, because attempts to control result in a new context, which
then creates a new interaction and from this interaction employees are able to
extract new information (Visser, 2007). The nal characteristic of deuterolearning is that it does not always lead to the improvement in the individual
or organization; individuals sometimes learn counterproductive behaviors.
While this type of learning can certainly occur in a fashion that leads to the
betterment of the individual or the organization, research has indicated that
certain types of situations may result in negative outcomes for both the
individual and the organization (Bateson, 1972; Dopson & Neumann, 1998;
Haley, 1963; Smith, 1976; Watzlawick, Bavelas, & Jackson, 1967).
Because deutero-learning is constantly occurring, it is inevitable that
workers having a socialization experience will engage in this learning
process. For example, in the socialization setting, workers may be told
that certain procedures are important; however, the tone, expression, and
posture of the person communicating this message may convey that this
procedure is actually unimportant. Individuals responsible for the socialization of employees may unconsciously or consciously communicate nonverbal
messages that contradict the verbal message being communicated. For
employees who are socialized through serial tactics, which provide an
experienced organizational member as a role model (Cable & Parsons, 2001),
deutero-learning helps them to process this experience by using information
191
from both their experience and the context of the situation. Additionally,
relational information, which aids individuals in knowing where they stand
in their relationships with other individuals (Miller & Jablin, 1991), is most
likely to be acquired through deutero-learning. The workers relationship
status will most likely be communicated through subtle tones, posture, and
expressions of those they interact with; which will provide information for
individuals to process through deutero-learning.
Deutero-learning may also be particularly important for individuals
who are worried about appearing incompetent because they ask too many
questions (Ostroff & Kozlowski, 1992) or for socialization experiences that
occur later in careers. Employees who are promoted, returning from an
international assignment, or adapting to a change in leadership may rely on
deutero-learning to glean important details about how they can become
organizational insiders once again. Because this type of learning is not
always a conscious effort on the part of the learner (Bateson, 1972), it is
likely to continue throughout individuals careers as they experience various
boundary crossings and move closer to being insiders. In many cases,
deutero-learning helps employees to read the information between the lines
as an organizational insider should, and for this reason we suggest the
following guideline:
Guideline 8. Researchers should theorize about ways in which deuterolearning may inuence the socialization process. They may consider how
the environment around the employee will inuence opportunities to
engage in deutero-learning.
Each of these different modes of learning plays an important role during
the socialization experience. These modes of learning are the culmination of
the learning process that happens during socialization. As individuals cross
boundaries, they are more motivated to learn and have more sources of
information. These three modes of learning are the processes by which
employees take the information they have gained and use it to adjust and
adapt their thoughts and behaviors.
SOCIALIZATION AND LEARNING:

THE NEVER-ENDING PROCESS
Organizational life is in constant ux. The environment around a rm
changes, innovations provide new opportunities, technology alters existing
192
JARON HARVEY ET AL.
jobs, and employees must cope with these changes the best they can. In
addition to the outside pressures, inuencing workers, there are personal
choices that alter career paths and create new and different situations in
organizational life. Employees may apply for promotions, be transferred to
different positions, experience leadership changes, or switch organizations
all together. As individuals work through these different changes and
adjust to new roles, jobs, and coworkers they continue to have socialization experiences. This happens because they are making every effort to
become, or regain their status as, organizational insiders. In this paper,
we have sought to understand how the learning process, which underlies
socialization works.
According to Fig. 1, there are several important elements to learning
during the socialization experience. We can see that workers in an
organization experience boundary crossings, or changes in their job, role,
or people around them. Boundary crossings, which result in increases to
employees uncertainty and anxiety, then increase employees motivation to
learn. Additionally, boundary crossing also inuences both the sources and
types of information to which workers have access. However, tenure in a
certain position or with the organization is likely to lower employees motivation to learn, because the more familiar individuals are with the people
around them and their situation the less uncertain they will feel because of it
(Berger & Calabrese, 1975). For this reason, the longer an individual is
tenured in an organization the less uncertainty he or she will face when
crossing boundaries. On the other hand, tenure in an organization, and the
associated familiarity with other employees and the processes and practices
of the organization, will increase the number of sources from which workers
can acquire information. The increased motivation to learn, that results from
boundary crossings, drives workers to engage in learning, and the newly
acquired sources that provide a variety of information types, all of which
Fig. 1.
The learning process during a socialization experience.
193
feed into the planned, meta-, and deutero- modes of learning. The processing
of information through these three modes of learning is the nal step in the
learning process, and leads employees to change what they think and how
they behave.
In taking a long-term view of the socialization process, we suggest that
individuals will have multiple socialization experiences, and a learning
view of socialization is applicable to an entire career rather than for
organizational entry only. The most common focus of socialization
research is the point of organizational entry when a newcomer enters
and adjusts to an organization (Ashforth et al., 2007; Louis, 1980). This
creates a constricted focus on a very limited period of time and a very
select group of individuals. Ashforth et al. (2007) called this narrow focus
into question when they asked researchers to challenge some of the
traditional assumptions of socialization research. There appears to be a
need to expand the time period surrounding socialization research, so as to
recognize different times in an employees career and the inuences of
different groups of employees. Using a learning perspective, we contend
that socialization has an overlooked temporal aspect. Individuals will
learn throughout their careers, and socialization experiences can occur at
different times during the course of a career; for instance they may happen
after promotions, transfers, and organizational changes. Thus, the
socialization experience does not occur a single point in time, but rather
the multiple boundary crossings that occur during the course of an
individuals career create many opportunities for this experience. These
crossings trigger the learning process so employees can regain their insider
status. When this happens, employees are motivated to learn and they
begin to draw from all their sources of information to learn what it is that
they need to do next.
Conclusions and Directions for Future Research

The learning processes, which underlie the socialization experience, have
been an implicit part of socialization from the beginning of research in this
area (Van Maanen & Schein, 1979). In this paper, we have endeavored
to propose how the learning process occurs during socialization, and what
some of the important elements of this process are. By examining this
learning process we have sought to better understand how boundary
crossings trigger the learning process, how it unfolds through motivation to
learn and the acquisition of information, and how three modes of learning
194
JARON HARVEY ET AL.
lead to the changes in thoughts and behaviors that indicate that learning has
occurred (Ormand, 1999). This paper, and the uncovering of the learning
processes that occur during socialization, suggests a series of important
directions for future research.
The rst direction is in clearly linking socialization tactics to different
types of information and modes of learning. As we have suggested, different
types of tactics and approaches to socialization lend themselves to providing
different types of information and engaging different modes of learning.
Some tactics are more likely to be linked with certain types of information.
Employees in organizations that have an individualized approach to
socialization, which separates employees from each other (Cable & Parsons,
2001), are less likely to receive relational information (Miller & Jablin, 1991)
about their relationships with other employees and the organization. When
considering modes of learning, an institutionalized approach to socialization
(Jones, 1986) is likely to have a stronger association with planned learning
than it is with meta-learning. Future research should seek to explore these
relationships to help organizations understand what types of information
workers will receive, and how their employees will learn from different
socialization tactics.
Second, by extending the temporal window of the socialization process
to include the span of an entire career and focusing instead on boundary
crossings, we can see how individuals will have multiple socialization
experiences within a single organization. During an individuals career span,
individuals will experience myriad socialization events and may be more
susceptible to different learning modes during these multiple socialization
events. The way individuals learn during these socialization experiences also
has important implications for understanding why certain outcomes occur.
The model we propose integrates learning types into the socialization
process, and we posit there are several outcomes of the socialization process
are the result of these learning processes. With each boundary crossing
employees are motivated to learn, acquire different types of information
from sources, and learn from that the way they need to behave so they can
reestablish themselves as insiders.
Third, in future work researchers need to address issues of measurement
for these different modes of learning. For both planned and meta-learning,
prior work that has focused on role ambiguity (e.g., Gruman et al., 2006;
Holder, 1996) may be particularly helpful. Because reducing role ambiguity
is an important outcome during socialization, and socialization programs
are designed to reduce ambiguity, this may provide a starting point for
scholars. In developing measures of planned and meta-learning researchers
195
may focus on how much an employee gains from organizational programs

and how much they learn by stopping and thinking about the implications
of their personal experiences. Gauging planned learning will help organizations know how effective their socialization programs are. In measuring
modes of learning, deutero-learning presents the greatest difculty, because
it often happens unconsciously. For guidance in this area researchers can
look to the work of McClelland (1961). The needs that McClelland
described (e.g., Need for Power, Need for Achievement) were unobservable,
and as such, measurement was problematic. To help capture the needs in
question, ambiguous meaning thematic apperception tests where used
(Holmstrom, Silber, & Karp, 1990). While such tests may not fully capture
deutero-learning they may provide a starting point for developing a
measurement of this mode of learning by providing some insight into the
unconscious motivations, which drive our performance and are deemed
important by an individual.
As further research provides a more complete picture of the learning
process that underlies the socialization experiences of employees, there will
be important implications for organizations. By knowing how employees
learn, what sets off the learning process, and what modes of learning are
active during socialization, organizations may be able rene their programs
to help workers become more fully enmeshed in the organization. By
knowing that boundary crossings initiate the learning process, organizations
can be prepared to provide the information necessary to aid employees
learning processes as they adapt to a new position or organization. Helping
employees learn the ropes will not only lead to more satised workers, but
may also lead to a more effective organization.
REFERENCES
Allen, D. G. (2006). Do organizational socialization tactics inuence newcomer embeddedness
and turnover? Journal of Management, 32, 237256.
Argyris, C. (2003). A life full of learning. Organization Studies, 27, 11781192.
Argyris, C., & Schon, D. A. (1974). Theory in practice: Increasing professional effectiveness.
Oxford: Oxford University Press.
Argyris, C., & Schon, D. A. (1978). Organizational learning: A theory of action perspective.
Reading, MA: Addison-Wesley.
Argyris, C., & Schon, D. A. (1996). Organizational learning II: Theory, method, and practice.
Reading, MA: Addison-Wesley.
Ashford, S. J., & Black, J. S. (1996). Proactivity during organizational entry: The role of desire
for control. Journal of Applied Psychology, 81, 199214.
196
JARON HARVEY ET AL.
Ashforth, B. E., & Saks, A. M. (1996). Socialization tactics: Longitudinal effects on newcomer
adjustment. Academy of Management Journal, 39, 149178.
Ashforth, B. E., & Saks, A. M. (2000). Personal control in organizations: A longitudinal
investigation with newcomers. Human Relations, 53, 311339.
Ashforth, B. E., Sluss, D. M., & Harrison, S. H. (2007). Socialization in organizational
contexts. In: G. P. Hodgkinson & J. K. Ford (Eds), International review of industrial and
organizational psychology (Vol. 22). Chichester, NY: Wiley.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.
Bateson, G. (1963). Exchange of information about patterns of human behavior. In:
W. S. Fields & W. Abbott (Eds), Information storage and neural control (pp. 173186).
Springeld, IL: Thomas Books.
Bateson, G. (1972). Steps to an ecology of mind. San Francisco: Chandler.
Bauer, T. N., Bodner, T., Erdogan, B., Truxillo, D. M., & Tucker, J. S. (2007). Newcomer
adjustment during organizational socialization: A meta-analytic review of antecedents,
outcomes, and methods. Journal of Applied Psychology, 92, 707721.
Bauer, T. N., & Green, S. G. (1994). Effect of newcomer involvement in work-related activities:
A longitudinal study of socialization. Journal of Applied Psychology, 79, 211223.
Bauer, T. N., & Green, S. G. (1998). Testing the combined effects of newcomer information seeking and manager behavior on socialization. Journal of Applied Psychology, 83,
7283.
Bauer, T. N., Morrison, E. W., & Callister, R. R. (1998). Organizational socialization: A review
and directions for future research. In: G. R. Ferris (Ed.), Research in personnel and
human resources Management (Vol. 16, pp. 149214). Samford, CT: JAI Press.
Berger, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the
development of interpersonal relationships. In: H. Giles & R. N. St. Clair (Eds),
Language and social psychology. Baltimore: University Park Press.
Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and beyond:
Toward a developmental theory of interpersonal communication. Human Communication
Research, 1, 99112.
Black, J. S., Mendenhall, M., & Oddou, G. (1991). Toward a comprehensive model of
international adjustment: An integration of multiple theoretical perspectives. Academy
of Management Review, 16, 291317.
Bolino, M. C. (2007). Expatriate assignments and intra-organizational career success:
Implications for individuals and organizations. Journal of International Business Studies,
38, 819835.
Brockner, J., Konovsky, M., Cooper-Schneider, R., Folger, R., Martin, C., & Lies, R. J. (1994).
Interactive effects of procedural justice and outcome negativity on victims and survivors
of job. Academy of Management Journal, 37, 397409.
Burke, R. J., & Bolf, C. (1986). Learning within organizations: Sources and content.
Psychological Reports, 59, 11871196.
Cable, D. M., & Parsons, C. K. (2001). Socialization tactics and person-organization t.
Personnel Psychology, 54, 123.
Carlson, D. S., Bozeman, D. P., Kacmar, K. M., Wright, P. M., & McMahan, G. C. (2000).
Training motivation in organizations: An analysis of individual-level antecedents.
Journal of Managerial Issues, 12, 271287.
197
Chan, D., & Schmitt, N. (2000). Interindividual differences in intraindividual changes in

proactivity during organizational entry: A latent growth modeling approach to
understanding newcomer adaptation. Journal of Applied Psychology, 85, 190210.
Chen, G. (2005). Newcomer adaptation in teams: Multilevel antecedents and outcomes.
Cooper-Thomas, H., & Anderson, N. (2002). Newcomer adjustment: The relationship between
organizational socialization tactics, information acquisition and attitudes. Journal of
Occupational and Organizational Psychology, 75, 423437.
Crossan, M. M. (2003). Altering theories of learning and action: An interview with Chris
Argyris. Academy of Management Executive, 17, 4046.
Davis, T. R. V., & Luthans, F. (1980). A social learning approach to organizational behavior.
Academy of Management Review, 5, 281290.
De Vos, A., Buyens, D., & Schalk, R. (2003). Psychological contract development during
organizational socialization: Adaptation to reality and the role of reciprocity. Journal of
Organizational Behavior, 24, 537559.
Dopson, S., & Neumann, J. E. (1998). Uncertainty, Contrariness and the double-bind: Middle
managers reactions to changing contracts. British Journal of Management, 9, S53S70.
Fedor, D. B., Rensvold, R. B., & Adams, S. M. (1992). An investigation of factors expected to
affect feedback seeking: A longitudinal eld study. Personnel Psychology, 45, 779805.
Feldman, D. C. (1976). Contingency theory of socialization. Administrative Science Quarterly,
21, 433452.
Feldman, D. C. (1981). The multiple socialization of organizational members. Academy of
Finkelstein, L. M., Kulas, J. T., & Dages, K. D. (2003). Age differences in proactive
newcomer socialization strategies in two populations. Journal of Business and
Fisher, C. D. (1986). Organizational socialization. In: K. M. Rowland & G. R. Ferris (Eds),
Research in personnel and human resource management (Vol. 4, pp. 101146). Greenwich,
CT: JAI Press.
Godshalk, V. M., & Sosik, J. J. (2003). Aiming for career success: The role of learning goal
orientation in mentoring relationships. Journal of Vocational Behavior, 63, 417437.
Graen, G. (1976). Role-making processes within complex organizations. In: M. D. Dunnette
(Ed.), Handbook of industrial and organizational psychology (pp. 12011245). Chicago:
Rand McNally College Publishing Company.
Greenberger, D. B., Strasser, S., & Lee, S. (1988). Personal control as a mediator between
perceptions of supervisory behaviors and employee reactions. Academy of Management
Journal, 31, 405417.
Gruman, J. A., Saks, A. M., & Zweig, D. I. (2006). Organizational socialization tactics and
newcomer proactive behaviors: An integrative study. Journal of Vocational Behavior,
69, 90104.
Hackman, J. R., & Oldham, G. R. (1976). Motivation through the design of work: Test of a
theory. Organizational Behavior and Human Performance, 16, 250279.
Haley, J. (1963). Strategies of psychotherapy. New York: Grune & Stratton.
Hanser, L. M., & Muchinsky, P. M. (1978). Work as an informational environment.
Organizational Behavior and Human Performance, 21, 4760.
Holder, T. (1996). Women in nontraditional occupations: Information-seeking during
organizational entry. Journal of Business Communication, 33, 926.
198
JARON HARVEY ET AL.
Holmstrom, R. W., Silber, D. E., & Karp, S. A. (1990). Development of the apperceptive
personality test. Journal of Personality Assessment, 54, 252264.
Hom, P. W., Roberson, L., & Ellis, A. D. (2008). Challenging conventional wisdom about who
quits: Revelations from corporate America. Journal of Applied Psychology, 93, 134.
Huysman, M. (2000). Rethinking organizational learning: Analyzing learning processes of
information systems designers. Accounting, Management and Information, 10, 8189.
Ibarra, H., & Andrews, S. B. (1993). Power, social inuence, and sense making: Effects of
network centrality and proximity on employee perceptions. Administrative Science
Quarterly, 38, 277303.
Jex, S. M., & Britt, T. W. (2008). Organizational psychology: A scientist-practitioner approach
(2nd ed.). Hoboken, NJ: Wiley.
Jones, G. R. (1986). Socialization tactics, self-efcacy, and newcomers adjustment to
organizations. Academy of Management Journal, 29, 262279.
Kammeyer-Mueller, J., & Wanberg, C. R. (2003). Unwrapping the organizational entry
process: Disentangling multiple antecedents and their pathways to adjustment. Journal
of Applied Psychology, 88, 779794.
Kim, T.-Y., Cable, D. M., & Kim, S.-P. (2005). Socialization tactics, employee proactivity, and
person-organization t. Journal of Applied Psychology, 90, 232241.
Klein, H. J., & Weaver, N. A. (2000). The effectiveness of an organizational-level orientation
training program in the socialization of new hires. Personnel Psychology, 53, 4766.
Kramer, M. W. (1993). Communication and uncertainty reduction during job transfers:
Leaving and joining processes. Communication Monographs, 60, 178198.
Kramer, M. W. (1994). Uncertainty reduction during job transitions: An exploratory study of
communication experiences of new comers and transferees. Management Communication
Quarterly, 7, 384412.
Kramer, M. W., & Noland, T. L. (1999). Communication during job promotions: A case of
ongoing assimilation. Journal of Applied Communication Research, 27, 335355.
Lester, R. E. (1987). Organizational culture, uncertainty reduction, and the socialization of new
organizational members. In: S. Thomas (Ed.), Culture and communication: Methodology,
behavior, artifacts, and institutions. Norwood, NJ: Ablex.
Louis, M. R. (1980). Surprise and sense making: What newcomers experience in entering
unfamiliar organizational settings. Administrative Science Quarterly, 25, 226251.
Major, D. A., Kozlowski, S. W. J., Chao, G. T., & Gardner, P. D. (1995). A longitudinal
investigation of newcomer expectations, early socialization outcomes, and the moderating
effects of role development factors. Journal of Applied Psychology, 80, 418431.
Major, D. A., Turner, J. E., & Fletcher, T. D. (2006). Linking proactive personality and the Big
Five to motivation to learn and development activity. Journal of Applied Psychology, 91,
927935.
Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational
trust. Academy of Management Review, 20, 709734.
McClelland, D. C. (1961). The achieving society. New York: Free Press.
Mignery, J. T., Rubin, R. B., & Gorden, W. I. (1995). Organizational entry: An investigation of
newcomer communication behavior and uncertainty. Communication Research, 22, 5485.
Miller, V. D., & Jablin, F. M. (1991). Information seeking during organizational entry: Inuences,
tactics and a model of the process. Academy of Management Review, 16, 92120.
Morrison, E. W. (1993a). Longitudinal study of the effects of information seeking on newcomer
socialization. Journal of Applied Psychology, 78, 173183.
199
Morrison, E. W. (1993b). Newcomer information seeking: Exploring types, modes, sources, and
outcomes. Academy of Management Journal, 36, 557589.
Morrison, E. W. (1995). Information usefulness and acquisition during organizational
encounter. Management Communication Quarterly, 9, 131155.
Morrison, R. F., & Brantner, T. M. (1992). What enhances or inhibits learning a new job?
A basic career issue. Journal of Applied Psychology, 77, 926940.
Ormand, J. E. (1999). Human learning (3rd ed.). Uppersaddle River, NJ: Prentice-Hall.
Ostroff, C., & Kozlowski, S. W. J. (1992). Organizational socialization as a learning process:
The role of information acquisition. Personnel Psychology, 45, 849874.
Ostroff, C., & Kozlowski, S. W. J. (1993). The role of mentoring in the information gathering
processes of newcomers during early organizational socialization. Journal of Vocational
Behavior, 42, 170183.
OHara, K. B., Beehr, T. A., & Colarelli, S. M. (1994). Organizational centrality: A third
dimension of intraorganizational career movement. Journal of Applied Behavioral
Science, 30, 198216.
Payne, S. C., Culbertson, S. S., Boswell, W. R., & Barger, E. J. (2008). Newcomer psychological
contracts and employee socialization activities: Does perceived balance in obligations
matter? Journal of Vocational Behavior, 73, 465472.
Saks, A. M., & Ashforth, B. E. (1996). Proactive socialization and behavioral self-management.
Saks, A. M., & Ashforth, B. E. (1997). Socialization tactics and newcomer information
acquisition. International Journal of Selection and Assessment, 5, 4861.
Saks, A. M., Uggerslev, K. L., & Fassina, N. E. (2007). Socialization tactics and newcomer
adjustment: A meta-analytic review and test of a model. Journal of Vocational Behavior,
70, 413446.
Salancik, G., & Pfeffer, J. (1978). A social information-processing approach to job attitudes and
task design. Administrative Science Quarterly, 23, 224253.
Schein, E. H. (1971). The individual, the organization, and the career: A conceptual scheme.
Journal of Applied Behavioral Science, 7, 401426.
Schein, E. H. (1978). Career dynamics. Reading, MA: Addison-Wesley.
Schon, D. A. (1971). Beyond the stable state. New York: Random House.
Schon, D. A. (1975). Deutero-learning in organizations: Learning for increased effectiveness.
Organizational Dynamics, 4, 216.
Simpson, I. H. (1967). Patterns of socialization into professions: The case of student nurses.
Sociological Inquiry, 37, 4754.
Skinner, B. F. (1969). Contingencies of reinforcement. New York: Appleton-Century-Crofts.
Smith, E. K. (1976). Effect of the double-bind communication on the anxiety levels of normals.
Journal of Abnormal Psychology, 85, 356363.
Teboul, J. C. B. (1995). Determinants of new hire information-seeking during organizational
encounter. Western Journal of Communication, 59, 305325.
Van Maanen, J. (1975). Police socialization: Longitudinal examination of job attitudes in an
urban police department. Administrative Science Quarterly, 20, 207228.
Van Maanen, J. (1976). Breaking in: Socialization to work. In: R. Dubin (Ed.), Handbook of
work, organization, and society (pp. 67130). Chicago: Rand McNally College Publishing
Company.
Van Maanen, J., & Schein, E. H. (1979). Toward a theory of organizational socialization. In:
B. Staw (Ed.), Research in organizational behavior (Vol. 1). Greenwich, CT: JAI Press.
200
JARON HARVEY ET AL.
Visser, M. (2003). Gregory Bateson on deutero-learning and double bind: A brief conceptual
history. Journal of the History of the Behavioral Sciences, 39, 269278.
Visser, M. (2007). Deutero-learning in organizations: A review and a reformulation. Academy of
Watzlawick, P., Bavelas, J. B., & Jackson, D. D. (1967). Pragmatiscs of human communication:
A study of interactional patterns, pathologies and paradoxes. New York: Norton.
Zalesny, M. D., & Ford, J. K. (1990). Extending the social information processing perspective:
New links to attitudes, behaviors, and perceptions. Organizational Behavior and Human
Decision Processes, 47, 205246.
COMPARING APPLES AND

ORANGES: TOWARD A TYPOLOGY
FOR ASSESSING E-LEARNING
EFFECTIVENESS
N. Sharon Hill and Karen Wouters
ABSTRACT
E-learning programs exist in a wide variety of formats. Without a
framework for distinguishing between different e-learning programs, it is
a challenge for researchers to compare their effectiveness or identify
characteristics of e-learning that contribute to learning effectiveness.
Based on general theories of learning, we develop a typology that
compares e-learning programs in terms of the nature of the learning
interactions they provide for learners in three dimensions: degree of
interaction, learner control of interactions, and informational value of
interactions. The typology dimensions apply to learnerinstructor,
learnerlearner, and learnerinstructional material interactions. We also
discuss important theoretical implications of the typology. First, we show
the utility of the typology for comparing the effectiveness of different
e-learning programs. Second, we apply the typology dimensions to
develop a theoretical framework for e-learning research that provides a
foundation for examining factors that inuence learning effectiveness in
an e-learning program. The framework identies e-learning program
characteristics, learner characteristics, and contextual factors that impact
ISSN: 0742-7301/doi:10.1108/S0742-7301(2010)0000029008
201
202
N. SHARON HILL AND KAREN WOUTERS
learning effectiveness in different e-learning environments. It also shows

how the typology dimensions align with learning goals to inuence
learning effectiveness.
Organizations are using e-learning extensively for employee development

(DeRouin, Fritzsche, & Salas, 2005). E-learning, which we dene as the use
of information and communication technologies (ICT) to deliver instruction
to learners, offers many potential benets to both organizations and
employees. The benets to the organization include savings on travel and
accommodation for attending training, the ability to offer training to more
people, greater consistency in training delivery, and improved tracking of
course completion and testing (Noe, 2005; Welsh, Wanberg, Brown, &
Simmering, 2003). For employees, the benets include greater exibility over
when and where training is completed, the potential to complete course
segments for just-in-time training needs, and exposure to a wider offering
of courses (Noe, 2005; Welsh et al., 2003).
The growth in e-learning has been characterized as a revolution in training
and development (Galagan & Drucker, 2000; Noe, 2005). Out of this
revolution, a large number of different e-learning formats have emerged. As
described by Welsh et al. (2003) in their review of the e-learning literature,
e-learning applications can range from asynchronous e-learning, in which
learners work completely at their own pace, to synchronous e-learning that is
live and requires that learners be at their computers at the same time as
the instructor and other learners participating in the e-learning program.
In addition, a wide range of technologies are used to support e-learning,
resulting in a large variety of delivery formats. These range from static,
primarily text-based content to more sophisticated programs that integrate
graphics, animation, video, and audio. To add to this complexity, many
organizations use different combinations of e-learning formats and classroom instruction to form blended learning.
As a result of the wide variety of e-learning and blended formats,
comparing the effectiveness of e-learning to traditional classroom instruction, or the effectiveness of different e-learning and blended learning
programs is akin to comparing apples and oranges. Given this, Smith and
Dillon (1999) argued that to facilitate empirical research in this area,
researchers should rst articulate the dimensions along which e-learning
programs vary. However, to our knowledge, there is currently no
comprehensive typology for comparing e-learning programs. Therefore, the
primary purpose of this chapter is to address this shortcoming in the extant
Comparing Apples and Oranges
203
literature by delineating dimensions of a typology that can be used to

characterize different types of e-learning programs. Our typology is
grounded in theories of traditional learning that emphasize the critical role
of interactions between the learner and instructor, the learner and other
learners, and the learner and instructional material for facilitating learning.
We refer to these interactions as learning interactions and argue that
e-learning programs can be differentiated by the extent to which they vary in
terms of three key learning interaction characteristics: degree of interaction,
learner control of interactions, and informational value of interactions.
These characteristics form the three dimensions of our typology.
Another purpose of this chapter is to highlight important theoretical
contributions of the typology and its potential to advance e-learning
research. First, we discuss the utility of the typology for comparing the
effectiveness of different e-learning, blended and classroom instruction
formats. We show how the typology can help to explain current equivocal
research ndings in this area as well as stimulate new research directions.
Second, we apply the typology dimensions to develop a theoretical
framework for e-learning research that examines factors that inuence
learning effectiveness in an e-learning program. In this framework, the
dimensions of the typology act as mediating variables that link characteristics of e-learning programs to learning effectiveness. We describe how the
typology can be used to explain existing relationships found in the literature
between the characteristics of e-learning programs and learning effectiveness
as well as to identify new e-learning program characteristics that the typology
suggests are worthy of attention. The framework also highlights learner
characteristics and contextual factors that moderate the relationship between
the typology dimensions and learning effectiveness. We extend research in
this area by showing how the typology can be used to move beyond the
focus in the existing literature on individual characteristics and contextual
factors that are important for e-learning, in general, to also identify those
that promote e-learning effectiveness in different types of e-learning environments. Finally, although several scholars agree that for effective learning to
occur, the characteristics of an e-learning program should align with the
learning goals of the program, there is no clear agreement on the form this
alignment should take. We describe how the typology can be used to
reconcile and extend different perspectives in this area.
The rest of the chapter proceeds as follows. We rst dene e-learning in
order to set the boundaries for the scope of learning programs addressed by
the typology. We then review existing typologies that distinguish between
different e-learning programs, and discuss their limitations relative to the
204
typology we develop in this chapter. Next, we review the major theories of

learning related to more traditional instruction that form the theoretical
foundation for the typology. Drawing on these theories of learning and
e-learning research, we develop the typology, which consists of three
dimensions (degree of interaction, learner control of interactions, and
informational value of interactions). We discuss each dimension of the
typology in turn, including its theoretical basis and relationship to
e-learning effectiveness. Having described the typology, we turn our
attention to its theoretical contributions, including as a foundation for
comparing the effectiveness of different e-learning programs and as a central
mechanism in a theoretical framework for identifying factors that inuence
e-learning effectiveness.
In summary, our goal is to make a signicant contribution to the
e-learning literature by developing a typology for distinguishing between
different e-learning programs. The development of the typology relies on
the integration of several literature streams, including the training and
educational literature as well as the information systems, and organizational
behavior literature.
E-LEARNING DEFINITION
We dene e-learning as the use of ICTs (e.g., internet, intranet, CD-ROM,
interactive TV, teleconferencing, computer-conferencing, and chat) to
deliver instruction to learners. Consistent with several existing denitions
in the literature, our denition focuses on technology-mediated learning as
the primary dening characteristic of e-learning. In other words, content is
transmitted to the learner using technology, and technology is also used to
facilitate communication between the learner and any other participants in
the learning program (i.e., instructor and other learners). We also include
in the e-learning denition all possible ICTs that can be used to deliver
instruction. Some researchers have dened e-learning more narrowly,
focusing on the use of computer-based technologies only (e.g., Lowe &
Holton, 2005; Welsh et al., 2003). Our denition is consistent with others
(e.g., Kaplan-Leiserson, 2002; DeRouin et al., 2005) who have offered a
broader denition that incorporates the use of computer-based technologies
(e.g., internet, intranet/extranet, and CD-ROM) as well as other ICTs
(e.g., audio, video, and TV). We believe this broader denition is essential
for developing a more comprehensive framework that applies to the wide
range of e-learning applications currently found in organizations.
205
Our denition also includes both technology-mediated learning where

learner, instructor, and fellow learners are spatially separated from each
other as well as technology-mediated self-study programs where there are no
instructor and other learners available. Several existing denitions make
specic assumptions regarding the presence of a human instructor and that
instructors spatial separation from the learner (e.g., Klein, Noe, & Wang,
2006); however, consistent with others (e.g., Bell & Kozlowski, 2002;
DeRouin et al., 2005; Welsh et al., 2003), our denition is not limited to
e-learning programs where there is a human instructor or other learners.
Between the two extremes of traditional classroom instruction and fully
technology-mediated e-learning are different forms of blended learning.
Blended learning has been dened as training that combines traditional
classroom sessions with e-learning and self-study (Kovaleski, 2004, p. 35).
Depending on the extent to which learning is technology-mediated, the
blended learning program will be closer to the classroom instruction or
the e-learning end of the continuum we have described. Given the need for
e-learning researchers to compare e-learning to blended and classroom
instruction, the typology we develop can be applied to learning programs
that fall at all positions on the classroom instructione-learning continuum.
REVIEW OF EXISTING E-LEARNING TYPOLOGIES

We performed a search to identify existing typologies that have been used
to distinguish between different e-learning programs. We used an electronic
search of EBSCO (Academic Search Premier, Business Search Premier,
Education Research Complete, Eric, PsychArticles, Psychology and
Behavioral Science Collection, and PsychInfo) and web searches performed
with Google Scholar to locate e-learning typologies in the literature from
1985 to the present. We then supplemented these electronic searches with
manual searches of books on e-learning. Search terms included e-learning,
online, web-based, distributed, distance, technology-mediated, and computer-based learning/training/instruction. Our review of the literature revealed
only a few existing typologies. In this section, we briey review these existing
typologies, and discuss their limitations relative to the typology we develop
in this chapter.
A rst set of existing typologies distinguishes between different types of
e-learning based on the delivery medium. Specically, Davidson-Shivers
and Rasmussen (2006), De Volder (1996), and Taylor (2001) describe
different generations of distance learning. Distance learning refers to the use
206
of different technologies to deliver instruction to learners who are spatially

separated from the instructor and other learners (Keegan, 1986). The rst
generation of distance learning is the print-based model (i.e., correspondence education supported by distance instruction through written
messages), which is not consistent with our e-learning denition. The
remaining generations use ICT to deliver instruction to learners; hence, they
fall within our denition of e-learning. The second generation of distance
learning uses radio, telephone, and television in a nonintegrated form. The
third generation uses computers and digital technologies to unite instructors
and learners and deliver the content of the course. The delivery medium in
the fourth and last generation of distance learning is the internet.
A second set of typologies differentiates between e-learning programs
based on specic characteristics of the technologies used to deliver
instruction and the spatial separation of the instructor and learners. For
example, Aggarwal (2000) and Brewer, De Jonghe, and Stout (2001)
developed a typology based on two dimensions: time (synchronous versus
asynchronous) and place (same versus different). Traditional classroom
instruction, where instructor and learner share the same space at the same
time and communicate in real time (synchronous), is at one extreme. At the
other extreme, instructor and learners are at different locations and
communicate with a time lag (asynchronous). Hedberg, Brown, and Arrighi
(1997) added a third dimension, group size (individual versus group),
to these two dimensions. Some delivery media are targeted to an individual
learner, whereas other media allow learners to work in group (e.g.,
groupware tools). Dillemans, Lowyck, Van der Perre, Claeys, and Elen
(1998) and Proost (1998) identied three additional technology dimensions
that can be used to differentiate e-learning programs: type of interaction
(humanhuman versus humancomputer), information modality (the ability
to transmit verbal and/or nonverbal cues) and linearity (the extent to which
the learner can navigate through the material).
Despite the intuitive appeal of focusing on the delivery media or
underlying technology characteristics to compare e-learning programs, such
a focus provides little information about the learning experience created
for the learner when these features are combined in the context of different
instructional designs. For example, the use of more synchronous technologies, such as videoconferencing, provides the potential for high levels of
interaction between the instructor and learners; however, depending on the
instructional design, such technologies might be used for one-way lecturing
by the instructor (low level of interactivity) or for a highly interactive
exchange between the instructor and learners. In other words, the amount of
207
interaction that actually occurs depends on the instructional design and

the motivation of the instructor and other learners to participate in the
discussion.
We also found two typologies that do not focus on the technology used,
but on aspects of the programs instructional design. Horton and Horton
(2003) made a distinction between instructor- and learner-led e-learning,
referring to the difference in control over the learning process. Clark and
Mayer (2003) and Holmes and Gardner (2006) distinguished between types
of e-learning based on the learning theory underlying the programs design.
(We include a more detailed discussion of learning theories in a later section
of this chapter.) E-learning programs based on behavioral learning theory
(e.g., drill and practice type programs) are characterized by demonstrations or examples and frequent practice with corrective feedback. A second
type of e-learning based on information processing theory aims to present
the information in a way that enhances the internal processes of acquiring,
understanding, and retaining knowledge. A third type of e-learning draws
from the constructivist learning theory and focuses on making sense of the
presented material guided by the instructor and/or in collaboration with
other learners.
By focusing primarily on the design philosophy underlying an e-learning
program, and not the technology features, these last two typologies do not
account for the fact that the use of different media and/or technologies
can create very different learning experiences in the context of a particular
design. For example, different technologies facilitate different levels of
interaction with other learners in a design based on constructivist
learning theory.
To address the shortcomings of these existing typologies, the typology we
develop focuses on the learning experience created for the learner, rather
than the specic features of the e-learning program in terms of technology
used or instructional design philosophy. In other words, we argue that it is
ultimately the type of learning experience created through the combination
of technology and instructional design choices that will inuence learning
effectiveness, and not the specic program features per se. Related to our
approach, there has been an ongoing debate in the educational literature as
to whether it is the use of a particular delivery technology or the instructional method that improves learning (Anderson & Elloumi, 2004; Carter,
1996; Jonassen, Campbell, & Davidson, 1994; Richey, 2000). Instructional
methods are dened as any way to shape information that compensates
for or supplants the cognitive processes necessary for achievement or
motivation (Clark, 2001, p. 208). Two main protagonists in this discussion
208
have been Clark (1983) and Kozma (1991). Clark (1983) argued that any
medium, appropriately applied, can be used to provide quality instruction,
while Kozma (1991) argued that media attributes alone inuence learning,
and the effectiveness of a medium to provide quality instruction depends on
how much of the learners cognitive work is performed by the medium. Two
recent meta-analyses (Bernard et al., 2004; Sitzmann, Kraiger, Stewart &
Wisher, 2006) conrmed Clarks position by showing that the instructional
methods built into the program tend to take precedence over the technology
used to deliver instruction. Bernard et al. (2004) found that there were
instances in which a distance learning group outperformed the traditional
classroom instruction group, whereas in other instances the opposite
occurred. The authors concluded that it is the instructional methods, such as
the feedback provided and the degree of learner engagement, independent of
the medium, that determine learning effectiveness. Consistent with this,
Sitzmann et al.s (2006) ndings indicated that web-based instruction and
classroom instruction were equally effective for teaching declarative
knowledge when the same instructional methods (e.g., the same level of
interaction with the learner and the same level of learner control) were used.
Consistent with the dominant position in the distance education literature
(Anderson & Elloumi, 2004; Carter, 1996; Jonassen et al., 1994), we argue
that both technological features and instructional methods shape the
learning experience. Further, these two factors combine with characteristics
of the instructor and other learners in the program to create different
learning experiences for learners. These learning experiences vary in the
nature of the learning interactions that occur, including interactions between
the learner and instructor, the learner and other learners, and the learner
and instructional material. Our focus on learning interactions is based on
both general and distance learning theories that predict that effective
learning is facilitated by learning interactions. Next, we review these theories
that provide the foundation for our typology.
THEORETICAL FOUNDATION
As noted by Salas, Kosarzycki, Burke, Fiore, and Stone (2002), there is
currently no theory that predicts e-learning effectiveness. However, general
theories of learning provide a useful starting point for developing an
e-learning typology. Furthermore, since researchers frequently seek to
compare e-learning with traditional classroom instruction, a typology that
is based on more general learning theories should apply to the full range of
209
instructional approaches, including classroom, blended, and e-learning.

The major learning theories that have emerged over the past century include
the behaviorist, cognitive, constructivist, and social learning theories
(Leidner & Jarvenpaa, 1995; Merriam & Caffarella, 1999; Salas et al.,
2002). Although, these theories differ in their major assumptions and can
be distinguished from each other in many different ways, for our current
purpose, a meaningful way to compare the theories is to consider how they
characterize the nature of interactions required for effective learning. In all
learning environments, interactions are the means by which information is
transmitted and knowledge is constructed.
Proponents of behavioral theories, such as Skinner (1974) and Thorndike
(1932), consider learning the uncritical absorption of objective knowledge and
modication of behavior. The instructor and instructional material facilitate
learning by providing the environmental stimuli for behavioral change to
occur, with the learner as a passive receiver of this stimuli (Leidner &
Jarvenpaa, 1995). In other words, there is a one-way transfer of knowledge
from the instructor and instructional material to the learner, with very little
interactivity between the learner and these components of the program.
In contrast to behavioral theories, cognitive learning theories place
more emphasis on the learner as an active participant in the learning
process. According to these theories, learning is a process in which the
learner constructs knowledge through continuous interaction with the
other components of the learning program instructor, other learners, and
instructional material (Salas et al., 2002). Cognitive constructivist theories,
such as Mezirows (1991) theory of transformational learning, consider
learning a process of meaning construction that takes place in interaction
with the instructional material. From a social constructivist perspective
(Wenger, 1998), learning is fundamentally a social activity taking place
through interaction with other learners and the instructor. Finally, social
learning theories, such as Banduras (1986) social cognitive theory, posit
that individuals learn from observing others, and hence, interactions with
others are key.
Distance learning researchers have also built their theories around the
central theme of interactions in the learning process. As mentioned earlier,
the term distance learning has been used as a more general term to describe
both e-learning and other forms of learning that rely on alternative types
of technology (e.g., correspondence by mail), not just ICTs, to deliver
instruction to learners who are spatially separated from the instructor and
other learners (Keegan, 1986). For example, Garrison (1989) maintained
that any educational transaction is based on seeking understanding and
210
knowledge through dialogue and debate; therefore, it requires two-way

communication between teacher and learner. In a learning setting where
learner and instructor are separated from each other, and communicate
using technology, this process is compromised. In addition, Keegan (1986)
argued that in distance learning the interactions between instructor and
learner that facilitate learning have to be articially recreated in order for
the instruction to be effective. Finally, Moore (1991, 1993) argued that the
reliance on technology-mediated communication creates a psychological
and communication gap in the interactions that occur between the learner
and instructor, the learner and other learners, and the learner and
instructional material. According to Moore, the success of distance learning
is determined by the extent to which this gap can be reduced.
In line with these theories of general learning and distance learning, we
build our typology around the central theme of interactions that occur in the
learning process between the learner and the different components of an
e-learning program (i.e., instructor, other learners, and instructional
material). We refer to these interactions as learning interactions. Consistent
with distance learning theories, we propose that the use of technology to
mediate learning can signicantly change the amount and nature of these
interactions; hence, one way to distinguish between different e-learning
programs is to compare the nature of the learning interactions they create
for learners. Based on the theories reviewed above, we identify three types of
learning interactions: learnerinstructor interactions, learnerlearner interactions, and learnerinstructional material interactions. Learnerinstructor
interactions provide content, motivation, feedback, and dialogue between
instructor and learner. Learnerlearner interactions involve the exchange of
information, ideas, and dialog between students about the course content.
Finally, learnerinstructional material interactions involve the process by
which students obtain intellectual information from instructional material
(Chen, 2001). Having discussed its theoretical foundation, we turn now to a
more detailed description of each dimension of the typology.
THREE-DIMENSIONAL TYPOLOGY
OF LEARNING INTERACTIONS
Based on the theories reviewed in the previous section, we argue that
the range of different e-learning programs used in organizations create
signicant variation in the types of learning interactions they produce
211
for learners. This variation can be described in three dimensions: degree

of interaction, learner control of interactions, and informational value of
interactions. Together these learning interaction dimensions form a
typology for distinguishing between different e-learning programs. In this
section, we describe each learning interaction dimension, and also discuss its
theoretical basis. From a theoretical perspective, each dimension should
facilitate greater learning effectiveness. Learning effectiveness refers to the
benets to learners that result from participating in the e-learning program,
for example, increased knowledge and skills, or new behaviors (Noe, 2005).
However, as we discuss in a later section, there are other factors
(e.g., learner characteristics) that moderate the relationship between each
dimension and learning effectiveness.
Degree of Interaction
This dimension describes the extent to which the e-learning program
provides the learner with opportunities for interaction (i.e., the amount of
interactivity) with the instructor, the other learners, and the instructional
material. The importance of this dimension is shown by our discussion in the
previous section of learning theories (e.g., social cognitive theory,
constructivist learning theories) that emphasize the critical role of interactions for effective learning. Degree of learnerinstructor and degree of
learnerlearner interaction refer respectively to the extent to which a learner
is able to interact with an instructor and with other learners in the program
(Moore, 1991, 1993). Interactions with an instructor, in which the learner
receives direction and guidance, have been shown to increase learning
effectiveness (Lemak, Shin, Reed, & Montgomery, 2005; Najjar, 1996).
Similarly, interactions with other learners, for example, through technologymediated group learning, have also been positively associated with learning
effectiveness (Arbaugh, 2005; Lou, Abrami, & Apollonia, 2001). Finally,
learning programs also vary in the degree of interaction built into the
instructional material. For instance, degree of interaction with the
instructional material is high when the learner has to solve a problem
through question and answer and is regularly prompted with test questions.
Research shows that increased interaction with the instructional material
provides the learner with more practice and feedback and leads to more
time spent on task, which positively impacts learning (for a review, see
Brown, 2001). In summary, learnerinstructor, learnerlearner, and learner
instructional material interactions should facilitate the learning process.
212
Learner Control of Interactions

This dimension describes the extent to which a learner has control over the
interactions available in the e-learning program, such that the learner is able
to tailor the instruction to his or her individual needs. E-learning programs
in which the sequencing of the content is structured in advance (e.g.,
traditional lecture via videoconferencing) are situated at the lower end of
this dimension. At the higher end of the learner control dimension are
e-learning programs that allow learners to assess their own learning needs
and adjust the program accordingly (e.g., web-based training providing
hyperlinks to different knowledge sources).
In the existing literature, learner control most typically refers to control
of interaction with the instructional material (e.g., DeRouin et al., 2005).
In this regard, several types of learner control and their impact on learning
effectiveness have been discussed, including learner control of sequence,
pacing, content, context, method of presentation, task difculty, and
incentives (DeRouin et al., 2005). However, we propose that it is also
important to consider learner control of interactions with the instructor and
other learners. This is consistent with the broad denition of learner control
provided in the literature as any instructional strategy in which learners
assume some form of control (DeRouin et al., 2005). For example, Wydra
(1980) dened learner control as a mode of instruction in which one or
more key instructional decisions are delegated to the learner (p. 3). Reeves
(1993) dened learner control as the degree to which an individual is given
control over various instructional features during a lesson or training
program. New forms of technologies allow learners to control interactions
with the instructor and other learners. For example, learner control of
interactions with the instructor is high when the e-learning course provides
the opportunity to chat with a content expert whenever the learner desires.
This exibility allows the type (i.e., instructor versus other learners) and
amount of interaction to be tailored to meet learners needs (Keller, 1983;
Milheim & Martin, 1991).
Several different theoretical perspectives have been used to explain the
positive impact of learner control on learning effectiveness (Milheim &
Martin, 1991). From a motivational perspective, learner control should
inuence motivation to learn, which has been shown to increase learning
effectiveness (Klein et al., 2006). This occurs through several mechanisms
(Milheim & Martin, 1991). First, learners needs are more likely to be
satised when training is made more relevant to a learner by providing the
learner with greater control. Second, consistent with adult learning theory
213
(Knowles, 1990), learner control also satises a learners need to be selfdirected. Finally, from an expectancy theory (Vroom, 1964) perspective,
learner control increases a learners expectancy of success because the
learner is made to feel that learning outcomes are in his or her hands.
According to Milheim and Martin (1991), two additional theoretical
perspectives for understanding the impact of learner control on learning
effectiveness are attribution theory (Kelley, 1967) and information processing theory (Gagne, 1985). Attribution theory states that learners seek to
understand and explain why an event has occurred, and the explanation
they construct inuences future action. By inuencing a learners perception
of the locus, stability, and controllability of learning, learner control can
increase the learners expectation of success, which will in turn produce
higher levels of effort. Finally, information processing theory emphasizes
the internal processes that occur when training content is processed and
retained. Learner control facilitates the process through which information
is coded for long-term memory by allowing the learner to organize information in a way that makes it personally meaningful. In summary, learner
control of interactions should positively impact the learning process.
Informational Value of Interactions

Informational value is the extent to which, within the different interactions,
communication or data are transmitted that are valuable for learning
(Kirkman & Mathieu, 2005). In addition to considering how much
opportunity for interaction an e-learning program offers, and how much
control the learner has over those interactions, it is also important to
recognize that not all learning interactions are created equal in terms of their
ability to transfer information that is valuable for learning. Media richness
theory (Daft & Lengel, 1986), a theory from communications research,
provides a foundation for this dimension. A basic tenet of this theory is
that communication media differ in terms of richness or their ability to
clarify ambiguity and facilitate understanding of communication messages.
It further proposes that communication media can be placed on a
continuum of richness. For example, videoconferencing is richer than email
because it allows for both nonverbal and verbal communication. Email
communication lacks the body language and other nonverbal cues that help
to clarify the meaning of messages. In general, according to these theoretical
perspectives, less rich communication media are viewed as less effective for
communication. Applying these theoretical arguments to e-learning suggests
214
that e-learning consisting of static text will transmit fewer cues to aid
understanding of learning than e-learning programs in which multimedia is
used to create a richer learning experience.
Based on the construct of media richness, Kirkman and Mathieu
(2005) dened a new construct: informational value. They argued that
media richness was dened as a characteristic of the information carrying
capacity of the communication media; however, the construct of richness
also applies to other types of information exchange, beyond direct
communications between one individual and another. For example, an
individual might post a presentation or engineering drawing to a website to
be viewed by others. Informational value is a more general term that applies
to all types of information exchanges, not just direct communications
between individuals. Therefore, in an e-learning context, informational
value is relevant to learning interactions involving direct communication
with an instructor and other learners as well as the presentation of
information to learners via the instructional material. This is consistent with
our denition of e-learning as the use of both the information and
communication technologies to deliver instruction to learners.
Kirkman and Mathieus (2005) construct of informational value focused
on the media used for communication and information exchange. In an
e-learning context, we view informational value as a characteristic of
the interaction that is shaped, not only by the media characteristics but
also by the extent to which that media is used by instructors and other
learners to provide cues that enhance learning. For example, an interaction
in which the instructor uses a richer communication media in a way that is
very expressive and provides a lot of nonverbal cues will create an
interaction of higher informational value than one in which an instructor
uses the same media with a lower level of expressiveness. We discuss this
idea further in a later section in which we discuss the antecedents of
informational value. Based on the theoretical perspectives above, interactions that are higher in informational value should facilitate greater
understanding on the part of the learner, and therefore enhance learning
effectiveness.
Taken together, the three learning interaction dimensions we have dened
form a typology for distinguishing between different e-learning programs.
The typology has important theoretical implications for research that
(1) compares the effectiveness of different e-learning, blended learning, and
classroom instruction and (2) identies factors that inuence e-learning
effectiveness. We discuss these two theoretical implications in the sections
that follow.
215
COMPARING THE EFFECTIVENESS OF

DIFFERENT E-LEARNING PROGRAMS
The typology we have developed provides a means to compare the effectiveness
of different e-learning programs. In this section, we discuss the utility of our
typology for both deepening our understanding of existing research ndings in
this area and identifying important new research directions. We argue that the
equivocal results found when comparing the effectiveness of e-learning to
classroom instruction (for reviews, see DeRouin et al., 2005; Welsh et al., 2003)
is in large part due to the lack of specicity regarding the characteristics of the
targeted e-learning programs. In this regard, the dimensions in our typology
can shed light on these equivocal ndings.
As discussed in our review of existing typologies in an earlier section,
existing research has already recognized that when comparing the effectiveness of e-learning, blended, and classroom instruction, it is necessary to move
beyond the technology per se and focus on how the technology is used within
the context of the programs instructional strategies (Bernard et al., 2004;
Clark, 1983; Sitzmann et al., 2006). Our typology builds on this idea by
providing a means to compare e-learning programs that takes into account
all components of an e-learning program, including the technology, the
instructional design, and the other participants in the learning program
(instructor and other learners). Our typology consists of three dimensions
along which learning programs can differ. Rather than being tied to a
particular technology or instructional design feature, these dimensions
describe the learning experience created for the learner in terms of the nature
of the interactions in which the learner engages while moving through the
program. Further, because the framework is independent of any particular
technology or instructional design strategy, it can be applied to learning
programs that consist entirely of classroom instruction as well as those that
are entirely technology-mediated. Hence, it can be used to facilitate
comparisons between learning programs that fall at any position along the
continuum between these two extremes.
The utility of the typology for comparing the effectiveness of different
learning formats is demonstrated by research that has compared learning
programs where all characteristics of the learning programs were the same,
with the exception of differences related to one of the learning interaction
dimensions in the typology. For example, Zhang (2005) conducted an
experiment that compared levels of student satisfaction and performance for
a course that was delivered using three different instructional methods:
classroom lecture, a videotape of that lecture, and a videotape of the lecture
216
in which students were allowed to control the sequencing of content and the
presentation format. Zhang found the highest level of student satisfaction
and performance in the e-learning condition with learner control. In
contrast, Tutty and Klein (2008) found that students who were paired in
learning dyads performed better for an individual learning assignment
in the face-to-face condition compared to a computer-mediated program.
Using follow-up surveys of the students and observation, the research team
explained this result by the fact that students in the face-to-face condition
had higher levels of interaction (degree of interaction) and also were able to
use nonverbal and contextual cues to more effectively share information
with their partners (informational value).
The research ndings above demonstrate the importance of clearly
isolating where the learning programs that are being compared fall along
the dimensions of our typology. The importance of this approach is also
illustrated by considering research that has compared blended learning with
classroom instruction (for a review, see Klein et al., 2006). Researchers have
explained their ndings that blended learning is more effective by pointing to
the fact that blended learning incorporates the advantages of both classroom
and e-learning (Klein et al., 2006; Sitzmann et al., 2006). For example,
according to Sitzmann et al. (2006), The instructional advantage of blended
learning may be due to incorporating the benets of personal interaction
typically found in CI [classroom instruction] and self study between
instructional meetings using the Web (p. 646). These benets of blended
learning can be more precisely understood by comparing blended learning
to classroom instruction using our typology. On the one hand, because
blended learning programs involve some classroom instruction, they have
the potential to provide a high level of interaction and informational value
of learning interactions compared to pure e-learning. On the other hand, the
e-learning component frequently allows the learner more control over the
pace and depth of learning than traditional classroom learning. As a next
step, researchers could use our typology to more precisely identify the types
of blended learning programs that are most effective. This is important
because within the continuum of blended learning, many different
combinations of classroom instruction and e-learning are possible.
A THEORETICAL FRAMEWORK
FOR E-LEARNING RESEARCH
Another important theoretical contribution of the typology is its application
to develop a theoretical framework for research that identies factors that
217
E-learning Program Characteristics
TYPOLOGY
Instructional design (e.g., learning model,

including requirements for interaction and
number of participants)
Learning Interaction
Dimensions
Technology (e.g., time dependency,

perceived ease of use)
Instructor (e.g., personality, teaching
style) and other learners (e.g., cognitive
style, technical expertise)
Instructional design (e.g., learning model,
accessibility of instructor and other
learners)
Technology (e.g., linearity, time
dependency)
Instructor (e.g., teaching style) and other
learners (e.g., cognitive style, technical
expertise)
Instructional design (e.g., presentation of
information, one-way versus two-way
communication, human-human versus
human-computer interaction)
Technology (e.g., information modality,
human-human versus human-computer
interaction, one-way versus two-way
communication, time dependency)
Degree of interaction
Instructor
Other learners
Instructional
materials
Learner Characteristics
Contextual Factors
Learning Goals
Learner control of
interactions
Instructor
Other learners
Instructional
materials
Learning
Effectiveness
Informational value of
interactions
Instructor
Other learners
Instructional
materials
Instructor and other learners (e.g.,

immediacy behaviors)
Fig. 1.
Theoretical Framework for E-learning Research.
inuence e-learning effectiveness (see Fig. 1). In this framework, the learning
interaction dimensions of the typology are central mediating mechanisms
that explain the relationship between characteristics of an e-learning
program and learning effectiveness. This theoretical framework facilitates
research related to factors that inuence e-learning effectiveness in four
categories: e-learning program characteristics, learner characteristics,
contextual factors, and learning goals. Next, we describe each component
of this framework and its relationship to the dimensions of our typology.
We also develop propositions to guide e-learning research related to each
component of the framework.
E-learning Program Characteristics

Fig. 1 shows the characteristics of the e-learning program as antecedents of
the typology dimensions. As shown in Fig. 1, we categorize these e-learning
program characteristics according to the four potential components of an
e-learning program: technology, instructional design, instructor, and other
learners. With regard to the rst two components, we argued earlier,
218
following the dominant position taken in the design-technology discussion

in the educational literature (Anderson & Elloumi, 2004; Carter, 1996;
Jonassen et al., 1994), that both the technology used and instructional
design impact learning interactions. For example, interaction with fellow
learners is not possible without technology features that provide human
human interaction (e.g., chat). However, by the same token, the degree
of interaction will primarily be determined by design decisions, such as
those related to the learning model (behavioral versus constructivist), the
number of participants (individual versus group assignments), and the
frequency of interactions.
The design and technology decisions provide the potential for certain
types of interactions; however, the nature of the learning interactions in an
e-learning program will also be inuenced by how the instructor and other
learners choose to apply this potential in their actual interactions. For
example, an instructor who uses the opportunity to interact with learners via
videoconference to provide an animated and interactive presentation will
provide higher interactivity and informational value than an instructor that
delivers a monotone, one-way delivery. Similarly, even though the program
uses technology that allows learners to communicate with each other, and
the design allows for learner interaction, the extent to which the instructor
promotes and facilitates interaction between learners can play an important
role in determining how much interaction actually occurs. Based on this,
researchers have described the instructor as a major force in shaping the
nature of the interactions that occur in an e-learning program (Webster &
Hackley, 1997).
Below, we describe the e-learning characteristics that are antecedents
to the typology dimensions in our theoretical framework. While e-learning
programs differ in terms of many different characteristics, we believe that
the typology helps to identify characteristics of e-learning that are likely
to be most impactful for learning. Our goal here is not to be exhaustive in
listing all possible e-learning characteristics, but to highlight some that are
likely to most inuence our typology dimensions.
Antecedents of Degree of Interaction
Instructional Design. The degree of interaction with the instructor, the
other learners, and the instructional material is primarily determined by the
design decisions made when developing the e-learning program (Moore,
1993). Decisions regarding learning model, including requirements for
interaction and number of participants, will set boundary conditions for
the extent to which interaction can occur within the e-learning program.
219
First, with regard to the requirements for interaction, the cognitive

constructivist view of learning (Dewey, 1938; Mezirow, 1991) emphasizes
that the learner takes responsibility for constructing meaning actively
through dialogue with oneself and others. An instructional design based
on this perspective will pay more attention to building in opportunities
for interaction, such as live question/answer, discussion (Garrison, 1993)
and collaborative problem-based learning opportunities, instead of
individual assignments (Gorsky & Caspia, 2005). This is in contrast to a
behaviorist orientation to learning (Skinner, 1974; Thorndike, 1932), which
underscores the role of reinforcement by the external environment to ensure
learning, and translates into a program in which the learner does not play
an active role in the learning process (Merriam & Caffarella, 1999).
An example of a design characteristic that facilitates a high degree of
interaction is the requirement for the learner to participate in group
activities or submit questions on a discussion board. This is in contrast to elearning programs that may have built-in opportunities for interaction, but
do not make these interactions a requirement, or a prerecorded lecture that
does not allow for interaction at all (Anderson & Garrison, 1995; Gorsky &
Caspia, 2005).
Second, design decisions related to group size can inuence the extent to
which learners interact with fellow learners and with the instructor (Chen &
Willits, 1998). E-learning programs vary in the number of participants who
can interact simultaneously. Some programs do not allow interaction with
other participants or are very limited in this regard (e.g., CD-ROM); others
allow large groups to participate (e.g., web-based application). The
e-learning program needs to allow at least two participants to interact with
each other, in order for interactions to take place. However, it is less clear
how large the group should be in order to optimize the level of interactivity
with fellow learners. Both the training and team literatures have shown that
group size has an effect on the extent to which groups interact (Arbaugh,
2005; Cohen & Bailey, 1997). Larger groups may be less effective because
of a decrease in involvement or participation by the individual members
(Hare, 1981; McGrath, 1984). Also, the degree of interaction with the
instructor will diminish in larger groups (Gorsky & Caspia, 2005).
Technology. First, the time dependency of the programs technology will
determine the degree of interaction with the instructor and other learners.
Time dependency captures the distinction between synchronous versus
asynchronous technologies. Synchronous technologies allow for real-time
interaction among learners and instructor, even if they are in different places
220
(e.g., tele- or videoconferencing system). Synchronous technologies increase

the likelihood of interactions with others (Lauzon & Moore, 1992; Proost,
1998). The more synchronous technologies are used, the more learners will
interact with the instructor and fellow learners as they participate in the
program at the same time. With asynchronous technologies, learners and
instructors are time independent and learners participate in the training
program at their convenience. As a result, asynchronous technologies
typically have a lower level of interactivity.
Second, following the technology acceptance model (Davis, 1989), we
expect perceived ease of use of the technologies to positively inuence
learners attitude toward the technologies. As a result, they will be more
willing to use the technologies to interact with the instructor, the other
learners, and the instructional material. In line with this, research has shown
that the more computer software, email, world wide web, etc. are perceived
as easy to use, the more individuals make use of these technologies
(Arbaugh, 2005).
Instructor and Other Learners. Several researchers have discussed
characteristics of the instructor and other learners that have an impact on
the degree to which they interact with a learner in the learning process
(Gorsky & Caspia, 2005). Instructors personality traits and teaching style,
for instance, play a critical role in creating and maintaining learner
instructor interaction (Moore, 1993). Chans (2002) study revealed that
instructors characterized by a high degree of extraversion were more
likely to interact with their students in a distance learning environment.
Webster and Hackley (1997) found that a more interactive teaching style
(encouraging learners to interact) was positively related to learners
involvement and participation in the learning process.
We further expect that characteristics of the other learners who are
participating in the program will inuence the extent to which they are
open to and seek out opportunities for interaction with others. Two such
characteristics are cognitive style and expertise in using e-learning
technologies. These individual characteristics mirror those described later
in this chapter as learner characteristics that facilitate learning in an elearning environment with a high degree of interactivity. Here, it is
important to note that certain cognitive styles have been associated with
preference for collaborative learning methods (for reviews, see Riding &
Cheema, 1991; Smith, Murphy, & Mahoney, 2003), and expertise in using elearning technologies has been positively associated with increased
participation in e-learning (Martins & Kellermanns, 2004). Hence, when
221
other learners have more collaborative learning styles and greater technical
expertise, they will tend to seek out more interactions with a learner,
resulting in a higher degree of learnerlearner interaction.
Proposition 1. The degree of interaction in an e-learning program is
inuenced by characteristics of the instructional design (e.g., learning
model, including requirements for interaction and number of participants), characteristics of the technology (e.g., time dependency and
perceived ease of use), and characteristics of the instructor (e.g.,
personality, teaching style) and other learners (e.g., cognitive style and
technical expertise).
Antecedents of Learner Control of Interactions
Instructional Design. Instructional design decisions will also establish
boundary conditions for the impact that technology as well as
characteristics of the instructor and other learners have on the degree of
learner control of interactions. First, we expect the learning model
underlying the e-learning program to have an effect on learner control. A
constructivist orientation toward learning (Dewey, 1938; Mezirow, 1991)
not only recognizes the critical importance of frequent interactions for
learning to occur but also that these interactions take place in an
individualized fashion while the learner is drawing his or her own lessons
from experience (Merriam & Caffarella, 1999). Compared to a behaviorist
approach toward learning, the constructivist orientation is much more likely
to translate into a design that allows and encourages learners to exercise
control over their interactions with the instructor, other learners, and
instructional material. In contrast, the behaviorist approach will more likely
translate into a program- or instructor-led e-learning program.
Second, following Chen (2001) and Gorsky, Caspia, and Tuvi-Arad
(2004), we expect the accessibility of the instructor to determine the degree
of control the learner has over learnerinstructor interactions, with
signicantly lower levels of learner control when the instructor is available
one hour per month versus seven days a week. Similarly, the accessibility
of other learners has a signicant impact on the extent to which learners
can chose to interact with each other (Gorsky, Caspia, & Trumper, 2004).
For example, providing the learner with other learners contact details
(e.g., email addresses) and providing access to other students seven days
per week leads to more learner control of learnerlearner interactions than
giving learners the opportunity to interact at limited times during the
program (Gorsky & Caspia, 2005).
222
Technology. Within the constraints of design decisions that are made to

enhance learner control, technology features will further impact this
dimension. First, we expect that the linearity (linear versus nonlinear) and
the time dependency (synchronous versus asynchronous) of the technologies
used will inuence the learner control of leanerinstructional material
interactions. Linearity describes technologies that only allow linear
predened learning paths versus those that enable learners to customize
their path through the program. With the exception of conferencing
systems (e.g., tele- and videoconferencing systems), most computer networkbased technologies (e.g., web-based applications) allow a nonlinear
presentation of the content, and thus provide the exibility to adapt
the learning path to individual learning needs and preferences (Proost,
1998).
Whereas synchronous technologies increase the likelihood of interactions
with the instructor and fellow learners (as discussed above), they decrease
the likelihood of individualized interactions with the program. The
more synchronous technologies are used, the less learners will be able to
control their own learning process (Hedberg et al., 1997; Proost, 1998).
In particular, when synchronous technologies are used for content delivery
(and not mere communication), it is more likely that the instructor will
make the key decisions regarding the content, sequence, and pace of
learning. In contrast, asynchronous technologies provide the opportunity
for learning to be time independent and for the learner to participate at his
or her own pace. In summary, nonlinear and asynchronous technologies will
increase the potential that learners have to control their interactions with the
instructor, the other learners, and the instructional material.
Instructor and Other Learners. The control that learners have over
interactions with the instructor, the other learners, and the instructional
material will also be determined by characteristics of the instructor and
other learners. First, as argued by Anderson and Elloumi (2004), instructors
differ in their teaching style and skills for responding to diverse learner
needs. Even given the constraints of the programs instructional design,
instructors vary in the extent to which they take a more instructor-led versus
student-led approach to instruction. In a more instructor-led approach, an
active instructor teaches a passive learner and decides where to build in
opportunities for interaction with others. In addition, the instructor
determines the pace, sequence, and methods of instruction (Anderson &
Elloumi, 2004). This kind of learning reects the goals of the instructor and
ignores the learners needs (Moore, 1973). In contrast, learner control of the
223
different learning interactions will increase when the instructor, within the
constraints of the programs design, takes a learner-centered (i.e.,
supporting individualized learning activities), community-centered (i.e.,
encouraging collaborative and individual interactions in many formats), and
content-centered (i.e., providing direct access to vast libraries of content)
approach (Anderson & Elloumi, 2004; Whipp & Chiarelli, 2004).
Learner characteristics that will have an impact on the extent to which
learners can control their interactions with each other are cognitive style
and expertise in using e-learning technologies. A learners control over
interactions with other learners requires that these other learners be willing
to engage in interactions that suit the learners needs. As discussed above,
we expect cognitive styles that relate to preference for learning collaboratively with others and expertise in using e-learning technologies to have a
positive effect on willingness to engage in learning interactions. Thus, when
other learners in an e-learning program have these characteristics, this
should result in a learner having more control over interactions with these
learners.
Proposition 2. Learner control of the interactions in an e-learning
program is inuenced by characteristics of the instructional design (e.g.,
learning model and accessibility of the instructor and other learners),
characteristics of the technology (e.g., linearity and time dependency), and
characteristics of the instructor (e.g., teaching style) and other learners
(e.g., cognitive style and technical expertise).
Antecedents of Informational Value of Interactions
Instructional Design. The informational value of the interactions with the
instructor, the other learners, and the instructional material will be primarily
determined by the programs instructional design. The informational value
of the learnerinstructional material interactions will depend on the design
decisions related to how content is presented. Whereas text is the simplest
way to present information, additional features such as still images, graphics,
images in motion, and sound can be added to increase the informational
value (Bell & Kozlowski, 2006). In addition, the informational value of the
learnerinstructor and learnerlearner interactions will be determined by
the type of communication allowed by the design. Specically, a program
design that allows for two-way communication will increase the
informational value of interactions with the instructor and other learners.
Two-way communication allows the learner to receive feedback and
seek clarication. This, in turn, enhances the informational value of the
224
interaction (Bell & Kozlowski, 2006). In contrast, if the e-learning program is

designed to have only one-way communication, then informational value will
be diminished (Bell & Kozlowski, 2006).
Further, an e-learning program can be designed to allow for interactions
that are humanhuman or humancomputer. Communicating with other
people or working on assignments with other learners (i.e., humanhuman
interactions) will result in higher informational value. This is because
such communication allows for detailed and immediate feedback and the
exchange of nonverbal cues that enhance the meaning of the communication. When learners only have the option to ask questions to an instructor
proxy or e-tutor (i.e., humancomputer interactions: Maruping & Agarwal,
2004; Proost, 1998), informational value is lower.
Technology. The technology used to implement design decisions will also

inuence the informational value of the interactions. For example, the
design might allow for humanhuman interactions; however, there are a
range of technologies (e.g., email, chat, and phone) that can facilitate such
interaction. Four characteristics of the technology that are relevant to
informational value are information modality (single versus multiple cues),
humanhuman versus humancomputer interaction, one-way versus twoway communication, and time dependency. Information modality refers to
the extent to which the technology allows for the communication of multiple
cues and language variety (Proost, 1998). This dimension ranges from
having the potential to convey text only (single cue) to having the potential
to convey text in combination with graphics, pictures, sound, and nonverbal
information (multiple cues). Humanhuman versus humancomputer
interaction (Proost, 1998) refers to whether humans communicate with
humans, mediated by technology (e.g., in a virtual classroom or
chat session) or humans communicate directly with the computer (e.g.,
CD-ROM). Finally, technologies such as CD-ROM, video tape, audio
tape, and DVD only allow one-way communication, whereas technologies
such as interactive media and videoconferencing support two-way
communication. In line with media richness theory (Daft & Lengel, 1986),
we expect the information modality and time dependency of the
technologies used to have a signicant impact on the informational value
of the delivery and communication processes in an e-learning program.
For example, a medium that only supports text-based cues and is
asynchronous provides the learner with less rich information than
synchronous multimedia (Bell & Kozlowski, 2006). Also, as argued above,
225
whether the technology supports one-way versus two-way communication

or humanhuman versus humancomputer interaction will also impact the
interactions informational value.
Instructor and Other Learners. A relevant concept with respect to the
antecedents inuencing the informational value of learnerinstructor and
learnerlearner interactions is immediacy behaviors. This concept was
rst introduced by Mehrabian (1972) who dened it as behaviors that
increase mutual sensory stimulation between two people and includes both
verbal and nonverbal behaviors. Examples of nonverbal immediacy
behaviors are eye contact, facial expressions, gestures, and body position,
whereas verbal immediacy involves behaviors such as using personal
examples, using humor, providing and inviting feedback, and addressing
students by name (Gorham, 1988). We expect immediacy behaviors to
inuence the informational value of the learnerinstructor and learner
learner interactions.
Proposition 3. The informational value of the interactions in an e-learning
program is inuenced by characteristics of the instructional design (e.g.,
presentation of information, one-way versus two-way communication,
and humanhuman versus humancomputer interaction), characteristics
of the technology (e.g., information modality, one-way versus two-way
communication, humanhuman versus humancomputer interaction, and
time dependency), and characteristics of the instructor and other learners
(e.g., immediacy behaviors).
So far, we have considered how our typology can be used to identify
characteristics of e-learning programs that inuence learning effectiveness
through their inuence on the typology dimensions. We now turn our
attention to how the typology dimensions inuence learning effectiveness.
As discussed earlier in our description of each learning interaction
dimension, the dimensions are expected to positively inuence learning
effectiveness. However, it is important to note that the typology describes
the potential for learning interactions that is available to the learner.
As shown in Fig. 1, a number of factors can inuence the extent to
which this potential translates into effective learning. These factors are
shown as moderators in Fig. 1 and include learner characteristics,
contextual factors, and learning goals. Next, we describe each of these
moderators and discuss how our typology can help to advance research
related to each one.
226
Learner Characteristics
A good match between the characteristics of the learner and the learning
interaction opportunities offered by the e-learning program will enhance
learning effectiveness, both directly and through motivation to learn
(Colquitt, LePine, & Noe, 2000). Motivation to learn describes a learners
desire to learn the content of the training program and has been shown to
inuence learning effectiveness (Colquitt et al., 2000). Motivation to learn
may be particularly important for e-learning because compared to more
traditional classroom instruction, e-learning requires learners to take greater
responsibility for their own learning. Further, since many e-learning programs
are completed over a longer period of time, there is a greater need for learners
to demonstrate persistence to complete an e-learning program.
With regard to the direct inuence, learners whose characteristics make
them more able to cope with demands of the e-learning environment will
perform better in that environment. Also, such learners tend to spend more
time in the e-learning program, which allows for greater exploration of the
program content and more comprehensive practice, leading to greater
learning effectiveness (Brown, 2001; Wang & Newlin, 2000). With regard to
the inuence through motivation to learn, several theoretical perspectives
suggest that a match between individual characteristics and the e-learning
program characteristics will inuence motivation to learn (for a review, see
Colquitt et al., 2000). For example, social cognitive theory (Bandura, 1986)
proposes that ones believe about ones ability to execute a task will increase
motivation to perform that task. Learners whose characteristics are better
matched to the program will be more condent of their ability to learn
using the e-learning program, and hence more motivated to learn. Similarly,
based on the technology acceptance model (Davis, 1989), a match in learner
characteristics is likely to inuence perceived ease of use of the e-learning
program, which according to this perspective, will inuence motivation to
use the program.
In their review of distance learning research, Salas et al. (2002) noted that
there is a growing body of research to show that individual characteristics
predict distance learning outcomes. However, they cautioned that although
some individual characteristics will be important for all distance learning
environments, research has yet to identify the learner characteristics that
are important in specic DL [distance learning] environments (p. 145).
We agree that a shortcoming of most of the existing research in this area is
that researchers fail to specify the type of e-learning for which a particular
individual characteristic will be most important. Categorizing an e-learning
227
program according to the learning interaction dimensions we have dened

helps to identify which learner characteristics are important for e-learning in
general, and which will be important in specic e-learning environments.
For example, individual characteristics related to a learners ability to
monitor his or her own learning process and make adjustments where
needed will be most important for learning in an e-learning program
characterized by the potential for a high degree of learner control. Hence,
this individual characteristic varies in importance depending on where an
e-learning program falls along this learning interaction dimension.
In this section, we provide examples of learner characteristics that
are likely to be important for all e-learning environments and those that
align with specic dimensions of our typology. As with our discussion of
e-learning characteristics, our purpose is not to provide an exhaustive list
of learner characteristics, but to illustrate the application of the typology
by focusing on some that are particularly germane to e-learning and to the
different typology dimensions.
Salas et al.s (2002) review summarized learning characteristics that are
likely to facilitate learning effectiveness in most e-learning environments,
regardless of where they fall on the dimensions of our typology. These
include cognitive ability, learning self-efcacy, prior knowledge and
experience, and mastery goal orientation. In addition, since a dening
characteristic of e-learning is the use of technology to deliver instruction,
several researchers have focused on technical expertise as an important
individual characteristic in an e-learning environment, (Brown, 2001;
Hill, Smith, & Mann, 1987; Lowe & Holton, 2005; Martins & Kellermanns,
2004; Thompson & Lynch, 2003). Since e-learning interactions are mediated
using technology, poor technology expertise is likely to impact the learners
ability to take advantage of any learning interactions that are available in
the e-learning program. Learners who lack technical expertise might not
explore the full degree of interactivity or learner control options available in
the program and may also be less receptive to the use of technology that
offers higher informational value, since this technology typically has more
complex technological features.
Another individual characteristic that is important across all learning
interaction dimensions is self-regulation. Self-regulation consists of strategies that individuals use in response to difcult or anticipated difculties
in goal-directed action to guide their own behavior over time and across
changing circumstances (Kanfer & Heggestad, 1997). According to Kanfer
and Heggestad (1997) individuals vary in the extent to which they have
the skills to apply self-regulation strategies to deal with obstacles to goal
228
attainment. Individuals who are more self-regulating have a greater

tendency to engage in motivation control strategies (e.g., self-goal setting
and monitoring of progress toward goals and self-reinforcement) and
emotion control strategies that protect on-task attention and prevent
distracting emotional states (Kanfer & Heggestad, 1997). The use of such
self-regulation strategies has been shown to lead to improved performance
and more successful goal attainment (e.g., Creed, King, Hood, & McKenzie,
2009; Porath & Bateman, 2006).
Self-regulation behaviors should ensure that learners are able to
successfully integrate e-learning into their everyday activities and maintain
a high level of self-motivation to engage in e-learning. This will better
position a learner to take full advantage of the degree of interactions
available in the e-learning program. For example, in an e-learning program
in which learners communicate in an online discussion forum, a learner with
poor self-regulation skills may nd less time to participate in and monitor
the discussion. As a result, the learner will be less engaged and derive less
benet from this aspect of the program. For similar reasons, learners who
are more effective at self-regulation will also be able to make better use of
the learner control available in an e-learning program. Finally, with regard
to informational value, programs that are low in informational value may
be less engaging and require additional effort on the part of the learner to
derive meaning from the learning interactions. Learners that are more
motivated to persist with learning in the face of such obstacles are more
likely to expend the greater effort required to learn in such an environment.
In line with these arguments, researchers have pointed to self-regulation
during the learning process as a critical success factor for learners in an
e-learning environment (for reviews, see Salas et al., 2002; Smith, 2005;
Smith et al., 2003).
In addition to learner characteristics that are important for all e-learning
environments, we also identify learner characteristics that align more
specically with each typology dimension.
Learner Characteristics and Degree of Interaction
With regard to the degree of interaction dimension, individual differences in
desire for interaction with others during the learning process are likely to be
important. Differences in desire for interaction might stem from differences
in how information is processed during learning or differences in the need
for social interaction during the learning process. First, from a learning style
perspective, learners differ in the extent to which interaction with others
facilitates their learning. For example, Sternbergs (1997) theory of thinking
229
styles proposes that people differ in the cognitive mechanisms they use to
organize and govern tasks. Externals prefer to work through problems
together with others and prefer approaches, such as group brainstorming,
where solutions evolve through collaborative problem solving (cf. Workman,
Kahnweiler, & Bommer, 2003). In contrast, Sternberg dened an internal
thinking style that describes people who prefer working alone. Internals nd
it disruptive to their concentration to interact with others while problem
solving or analyzing information. Given their preference for joint problem
solving and collaborative information processing, externals are likely to
learn less effectively in a learning environment with fewer opportunities for
interaction with others (i.e., low degree of interaction).
Research related to learning styles has also identied cognitive style
dimensions that encompass preference for collaborative learning methods
(for reviews, see Riding & Cheema, 1991; Smith et al., 2003). For example,
the Wholist cognitive style dimension identied by Riding and Cheema
(1991) describes learners who prefer to learn in groups and to interact
frequently with other learners, as well as the instructor, as opposed to
learners who respond better to more independent and more individualized
approaches. Underlying this cognitive style is a tendency to be sociable and
socially dependent. Related to this, Salas et al. (2002) suggested that
researchers should strive to better understand how differences in social needs
of learners affect learning effectiveness in a distance learning environment.
Personality researchers have identied a number of individual differences
related to individuals need for social interaction, for example, Cheek and
Buss (1981) sociability and Hills (1987) interpersonal orientation. In
summary, we argue that learners who have a greater desire for interaction
with others during the learning process will benet more from programs that
have a high degree of learnerinstructor and learnerlearner interactions.
Learner Characteristics and Learner Control of Interactions
An important learner characteristic for success in e-learning programs that
provide a high degree of learner control is metacognition (for a review, see
DeRouin et al., 2005; Salas et al., 2002). Metacognition is concerned with
how the learner navigates and guides him- or herself within the training
program and is a measure of the degree to which learners reect on their
own learning process (Flavell, 1979). Individuals with greater metacognitive
skills are better able to monitor their progress, determine when they are
having problems, and adjust their learning accordingly (Ford, Smith,
Weissbein, Gully, & Salas, 1998; Schmidt & Ford, 2003). These types of
learners make better decisions about learning strategies and where to direct
230
their attention (Salas et al., 2002). As a result, metacognition has been

identied as an important characteristic for learners in an environment
characterized by a high degree of learner control (for reviews, see Lowe &
Holton, 2005; Salas et al., 2002; Smith, 2005; Smith et al., 2003).
Learner Characteristics and Informational Value of Interactions
Learner characteristics that align with the informational value dimension
of our typology are those related to learner preferences for how information
is presented and for certain modes of communication. These learner
characteristics have received little attention in the existing literature
compared to learner characteristics that align with the other two
dimensions. However, research suggests that these differences do exist, and
we believe they are worthy of further research attention. For example,
research related to learning preferences and readiness for online learning has
identied a learning preference related to comfort with communicating
electronically with other learners and an instructor (for a review, see Smith,
2005). Smith (2000) analyzed learning preferences of 1,252 vocational
students and identied a difference related to comfort with learning using
text or listening as opposed to nontextual interaction, including direct
experience, observation, and practice. Learners who have a preference for
richer communication modes will require higher levels of informational
value to facilitate their learning.
In addition, cross-cultural research has identied differences in communication styles that could be relevant to this dimension (Hall, 1976). In lowcontext cultures, individuals communicate predominantly through explicit
statements in text and speech, whereas in high-context cultures individuals
rely more heavily on contextual and nonverbal cues during communication
both communicate and interpret the meaning of messages. Extending this
line of research, communication researchers have identied an individual
difference that aligns with these differences in communication style and that
describes the extent to which individuals are indirect in their communication
style, relying less on the content of the spoken or written word and more
on the nonverbal aspects (Holtgraves, 1997). Learners who are more indirect
in their communication style are likely to nd it more difcult to learn in
e-learning programs with lower levels of informational value, which will
negatively impact learning effectiveness.
Proposition 4. Some learner characteristics will moderate the relationship
between all learning interaction dimensions and learning effectiveness
(e.g., cognitive ability, learning self-efcacy, prior knowledge and experience,
231
mastery goal orientation, technical expertise, and self-regulation skills).

Other learner characteristics will moderate the relationship between
specic learning interaction dimensions and learning effectiveness.
For example, desire for interaction with others during the learning
process will interact with degree of interaction, metacognition will interact
with learner control, and preference for richer communication modes
and a more indirect communication style will interact with informational
value.
Contextual Factors
Since most e-learning research has taken place in educational, rather than
workplace settings (DeRouin et al., 2005), contextual factors related to
e-learning have received relatively little research attention. Yet, the limited
research in this area (e.g., Klein et al., 2006), as well as the practitioner
literature (Frankola, 2001; Hequet & Johnson, 2003; Moshinskie, 2001;
American Society for Training and Development/The Masie Center, 2001),
suggests that contextual factors can have a signicant impact on whether
learners engage in and complete e-learning programs. In Fig. 1, we include
contextual factors as important moderators of the relationships between the
typology dimensions and learning effectiveness. We argue that learning
effectiveness will be enhanced when characteristics of the learning context
align with the learning interaction characteristics of the e-learning program.
Our argument for this relationship is based on a similar rationale to that
presented for the moderating role of learner characteristics. First, contextual
factors can act as enablers or present barriers to engage in e-learning (Klein
et al., 2006; Martins & Kellermanns, 2004). In addition, the alignment of
contextual factors with learning interaction characteristics can inuence
learning effectiveness through motivation to learn. First, such an alignment
can inuence learners perception of how easy it is to engage in e-learning.
As discussed above, this perception has been linked theoretically to
motivation to learn (e.g., social cognitive theory, Bandura, 1986; technology
acceptance model, Davis, 1989; and expectancy theory, Vroom, 1964).
Second, this alignment can also increase motivation to learn by inuencing
perceptions of the utility of participating in e-learning. For example,
expectancy theory (Vroom, 1964) suggests that when employees perceive
that there are benets associated with participating in e-learning, their
motivation to engage in e-learning will increase.
Traditional training research has focused on contextual factors such
as climate for transfer and opportunity to perform (Colquitt et al., 2000;
232
Noe, 2005) that inuence learning outcomes. These factors will also be
important in an e-learning environment; however, as with our discussion of
learner characteristics, we focus here on contextual factors that are likely to
be particularly germane to e-learning. Two contextual factors that we
propose will interact with all the typology dimensions to inuence learning
effectiveness is the extent to which a learner has easy access to appropriate
technology tools and technology support. E-learning programs that offer
more interactivity, greater learner control, and increased informational
value will typically require the use of technologies with more complex
technical requirements. Hence, it is important for learners to have access to
appropriate technologies and adequate support in using those technologies.
Access to the technology required to effectively use an e-learning program
and availability of technical support are related to a learners use of the
system (Martins & Kellermanns, 2004). In addition, as noted above, the
extent to which learners have the opportunity to engage in e-learning
without distractions will inuence the extent to which a high degree of
interaction, high learner control, and high-informational value translates
into learning effectiveness.
DeRouin, Fritzsche, and Salas (2004) reviewed several contextual factors,
based on theories of motivation, which are relevant for increasing a learners
willingness to engage with an e-learning program. These include supervisors
expressing condence in learners, providing them with encouragement
and valued rewards for participation in e-learning, working with learners
to set difcult but attainable goals for learner performance as a result of
participating in an e-learning program, and monitoring learner progress
against those goals.
Based on the arguments above, we propose that it is possible to
differentiate different work contexts in terms of their climate for e-learning.
We dene climate for e-learning as learners shared perceptions about
characteristics of the work environment that facilitate e-learning in general,
regardless of the type of e-learning program. Our denition is based on the
denition of climate in the organizational literature. Climate has been dened
as employees shared perception of the events, practices, and procedures as
well as the kind of behaviors that get rewarded, supported, and expected in
a particular organization (Schneider, 1990). The relationship between climate
and behavior has been demonstrated at different levels of analysis for
different elements of the work setting (e.g., Colquitt, Noe, & Jackson, 2002;
Hofmann, Morgeson, & Gerras, 2003; Simons & Roberson, 2003).
It is possible to conceptualize and measure climate for e-learning at
multiple levels of analysis. For example, organizations with strong climates
233
for e-learning will integrate e-learning into the organizations performance

management system, promote the value of e-learning training, provide
technology to support different types of e-learning formats and technological
support, provide workspaces free from distraction for completing
e-learning, and monitor e-learning participation. In addition, at the workgroup level, the existing literature has shown that managers play an important
role in reducing barriers to engage in e-learning and have considerable
discretion in how organizational policies related to e-learning participation are
implemented at the workgroup level (Hequet & Johnson, 2003; American
Society for Training and Development/The Masie Center, 2001).
In addition to factors that create a general climate for e-learning, relevant
contextual factors that specically align with each typology dimension are
described below.
Contextual Factors and Degree of Interaction
An important contextual factor that aligns with the degree of interaction
dimension is the extent to which the work context provides opportunities
for interaction to compensate for the lack of interaction available in an
e-learning program. Without a high degree of interaction in an e-learning
program, learners lack social support and feedback, which can lead to
feelings of isolation, uncertainty, and anxiety. This in turn can negatively
impact learning effectiveness (Benbunan-Fich & Hiltz, 2003). Interactions
with managers or peers during the learning process can compensate for lack
of interactivity in an e-learning program. For example, Moshinskie (2001)
suggests that managers provide learners with opportunities to practice
and get feedback on what they are learning in the e-learning program.
In addition, the organization can facilitate employees forming peer-learning
groups with other e-learners in the organization.
Contextual Factors and Learner Control
DeRouin et al. (2004) suggest that a specic aspect of the organizational
climate that will help to facilitate learner control is the extent to which the
organization generally promotes employee participation, empowerment,
and autonomy. In such a climate, employees will be more prepared and
willing to take control of their own learning because they are used to doing
so as a natural part of their job.
Contextual Factors and Informational Value
Although we have already discussed the importance of providing learners
access to technology required to complete e-learning, an aspect of this that is
234
particularly important for the informational value dimension is the extent

to which learners have access to technology that can support highinformational value interactions. For example, although an e-learning
program might offer audio and video capability in addition to text, a learner
may be limited to text-based content because his/her computer equipment
cannot cope with the additional bandwidth required for the other richer
media options. Also, since programs with high informational value will
typically use more complex technologies (e.g., videoconferencing and
audiovisual), the need for more advanced technical support to address any
problems encountered in using these technologies is particularly important
(Martins & Kellermanns, 2004).
Proposition 5. Some contextual factors will moderate the relationship
between all learning interaction dimensions and learning effectiveness
(i.e., climate for e-learning, including opportunity to engage in e-learning
without distraction, access to appropriate technology tools and support,
rewards for e-learning participation, setting e-learning performance goals,
and supervisor encouragement). Other contextual factors will moderate
the relationship between specic learning interaction dimensions and
learning effectiveness. For example, supervisor and peer interaction
during the learning process will interact with degree of interaction,
participative organizational climate will interact with learner control, and
availability of richer technology media and associated technical support
will interact with informational value.
Learning Goals
There is general agreement in the traditional training literature that desired
learning outcomes are more likely to be attained by aligning characteristics
of an e-learning program with the targeted learning goals (e.g. Gagne, 1985;
Noe, 2005). Similarly, scholars in the distance and technology-based
education literature have underscored the importance of such alignment
for learning effectiveness (e.g. Kozma, 1991; Moore, 1993). For example,
Kozma (1991) stated that whether or not a technology-based program
makes a difference in learning depends on how the program corresponds to
the particular learning task, because tasks vary in the demands they place on
the learner. Consistent with this existing research, we include learning goals
as a moderator in the relationship between learning interaction dimensions
and learning effectiveness.
235
Although several scholars agree on the need to align characteristics of an

e-learning program with learning goals, there is less clarity on the exact
nature of this match. Recently Bell and Kozlowski (2006) and Lowe and
Holton (2005) developed theoretical models that address this question. Both
models make a distinction between different types of learning goals, which
range from basic level to advanced knowledge and skills. Lowe and Holton
(2005) based their model on Bloom, Engelhart, Furst, Hill, and Krathwohls
(1956) classication consisting of ve categories: knowledge, comprehension, application, synthesis, and evaluation. Bell and Kozlowski (2006)
distinguished between declarative, procedural, strategic, and adaptive
knowledge and skills. Both models show that different learning goals place
different demands on the e-learning instruction. Lowe and Holton (2005)
discuss the match in terms of the instructional control and support provided
in computer-based training. Bell and Kozlowskis (2006) primary assumption is that as the complexity of the knowledge and skills targeted increases,
the richness of the distributed learning experience in terms of content,
immersion, interactivity, and communication must also increase.
Translated to our model, this suggests that at the most basic level, when
the learner is asked to acquire declarative knowledge, a program with little
interaction, low level of learner control, and with interactions with lower
informational value is sufcient to attain targeted learning goals. First,
a program-controlled learning experience will be more efcient to instruct
specic well-dened concepts and facts (Lowe & Holton, 2005), as at this
level one does not require learners to regulate their own learning process.
Second, as stated by Bell and Kozlowski (2006), a text-based experience with
a low level of interactivity will be most efcient to learn at the lowest
learning goal level. The authors based their statement on empirical evidence
showing that richer interactions with more stimuli interfere with, rather than
stimulate, memorization. Finally, the authors argue that efforts to increase
the interactivity with the instructor and fellow learners, by for instance
providing sophisticated ICTs, will not add much value to the learning
process compared to the investments made.
In contrast, at the most complex level of knowledge and skills, where the
learner is asked to acquire more advanced problem-solving skills, there
needs to be more interaction, with more learner control and higher
informational value. First, an e-learning experience in which the learner has
control over the program will be more effective to learn more advanced
knowledge and skills. This is because this learning goal level requires that
learners play an active role in their own learning process (Bell & Kozlowski,
2006; Kolb, 1984) by adapting existing models, interpreting specic events,
236
etc. Second, more complex learning goals may require higher interactivity
with the instructor and fellow learners in order to exchange information and
ask for feedback (Garrison, 1993; Lowe & Holton, 2005). Finally, the
informational value provided in the interactions is important if more
advanced knowledge and skills are the target. Learners need to acquire
coherent mental models and situational awareness to reach this target,
and this can be facilitated by providing multiple data and contextual cues
(Bell & Kozlowski, 2006).
We believe that our typology is a useful way to integrate existing research
related to matching learning goals to the e-learning experience provided to
the learner. To extend this research, researchers should examine the relative
importance of each learning interaction dimension to different levels of
learning goals. This will allow e-learning program designers to make tradeoffs in the design process. For example, given a more complex learning goal,
is it best to allocate resources to providing greater levels of learner control or
increasing the informational value of the program? We believe that these are
important questions to address in future research in this area.
Proposition 6. E-learning effectiveness will increase when the characteristics of the learning interactions are matched to the level of learning goals.
CONCLUSION
E-learning holds much promise for helping organizations meet their future
training demands. However, to realize its full potential, there needs to be a
clearer understanding of how different e-learning program congurations
inuence learning outcomes. To this end, Smith and Dillon (1999) noted the
importance of clearly articulating the dimensions along which e-learning
programs differ. In response to this, we developed a typology based on
general theories of learning that compares e-learning programs according
to the nature of the learning interactions they provide for learners. The three
dimensions of the typology that describe important characteristics of the
learning interaction dimensions are degree of interaction, learner control
of interactions, and informational value of interactions. These dimensions
apply to interactions that occur between the learner and instructor, other
learners, and instructional material.
We highlighted several important theoretical contributions of the typology.
First, we discussed how the typology can be used in research to compare the
effectiveness of different e-learning, blended learning, and classroom
237
instruction formats. The typology helps to reconcile existing research ndings

as well as stimulate new research questions in this area. We believe that our
typology is the rst to provide a comprehensive foundation for comparing the
range of different e-learning programs that are currently available, including a
comparison of e-learning with classroom and blended instruction. Unlike
existing typologies that are framed in terms of specic technology features of
the e-learning program, our typology is exible enough to apply to e-learning
programs at any point along the continuum from pure classroom instruction
to instruction that is fully technology-mediated.
Second, based on the typology dimensions, we developed a theoretical
framework and related propositions that can be used to identify factors that
inuence success in different types of e-learning environments (i.e., e-learning
programs located at different points along each of the typology dimensions).
The framework identies important characteristics of e-learning programs
that are antecedents to the typology dimensions, and therefore inuence
learning effectiveness through these dimensions. It also identies learner
characteristics and contextual factors that moderate the relationships
between the typology dimensions and learning effectiveness. Finally, it
shows how the typology dimensions can be used to match different levels of
learning goals to the appropriate e-learning format.
Recent reviews of the traditional training and e-learning literature have
commented on the paucity of theory to guide e-learning research (Klein
et al., 2006; Welsh et al., 2003). This is in contrast to the progress that has
been made in both theoretical and empirical research in the traditional
training area (Salas & Cannon-Bowers, 2001). In summary, Salas and
Cannon-Bowers (2001, p. 483) noted that A few researchers have begun to
scratch the surface y of this topic, but a science of distance learning and
training needs to evolve. We hope that our typology and theoretical
framework will help to advance the science of e-learning and stimulate
research that can help organizations realize the full potential of their
e-learning implementations.
REFERENCES
Aggarwal, A. (2000). Web-based learning and teaching technologies: Opportunities and
challenges. Hershey, PA: IDEA Group Publishing.
American Society for Training and Development/The Masie Center. (2001). E-learning: If we
build it, will they come? Available at: http://www.astd.org/NR/rdonlyres/E4C5C3EB2F0B-456D-8136-9B38F52CC5B1/0/84416110pdf.pdf
238
Anderson, T., & Elloumi, F. (2004). Theory and practice of online learning. Athabasca:
Athabasca University.
Anderson, T., & Garrison, D. (1995). Transactional issues in distance education: The impact of
design in audioteleconferencing. The American Journal of Distance Education, 9, 2745.
Arbaugh, J. B. (2005). Is there an optimal design for on-line MBA courses? Academy of
Management Learning & Education, 4, 135149.
Bandura, A. (1986). Social foundations of thought & action: A social cognitive theory.
New Jersey: Prentice-Hall.
Bell, B. S., & Kozlowski, S. W. J. (2002). Adaptive guidance: Enhancing self-regulation, knowledge,
and performance in technology-based training. Personnel Psychology, 55(2), 267306.
Bell, B. S., & Kozlowski, S. W. J. (2006). Aligning training and technology: A theoretical
framework for the design of distributed learning systems. Paper submitted for 21st SIOP
Conference, Dallas.
Benbunan-Fich, R., & Hiltz, S. R. (2003). Mediators of the effectiveness of online courses.
IEEE Transactions on Professional Communication, 46(4), 298312.
Bernard, R. M., Abrami, P. C., Lou, Y., Borokhovski, E., Wade, A., & Wozney, L. (2004).
How does distance education compare with classroom instruction? A meta-analysis of
the empirical literature. Review of Educational Research, 74, 379439.
Bloom, B. S., Engelhart, M. D., Furst, E. J., Hill, W. H., & Krathwohl, D. R. (1956). Taxonomy
of educational objectives: The cognitive domain. New York: Longman.
Brewer, E. W., De Jonghe, J. O., & Stout, V. J. (2001). Moving to online. Thousand Oaks, CA:
Corvin Press Inc.
Brown, K. G. (2001). Using computers to deliver training: Which employees learn and why?
Carter, V. (1996). Do media inuence learning? Revisiting the debate in the context of
distance education. Open Learning: The Journal of Open and Distance Learning, 11(1), 3140.
Chan, B. (2002). A study of the relationship between personality and teaching effectiveness:
Does culture make a difference? International Review of Research in Open and Distance
Learning, 3, 3140.
Cheek, J. M., & Buss, A. H. (1981). Shyness and sociability. Journal of Personality and Social
Psychology, 41(2), 330339.
Chen, Y.-J. (2001). Dimensions of transactional distance in the world wide web environment:
A factor analysis. British Journal of Educational Technology, 32(4), 459470.
Chen, Y.-J., & Willits, F. K. (1998). A path analysis of the concepts in Moores theory of
transactional distance in a videoconferencing learning environment. Journal of Distance
Education, 13(2), 111.
Clark, R. (1983). Reconsidering research on learning from media. Review of Educational
Research, 55, 445459.
Clark, R., & Mayer, R. E. (2003). E-learning and the science of instruction: Proven guidelines for
consumers and designers of multimedia learning. San Francisco: Pfeiffer.
Clark, R. E. (2001). Learning form media: Arguments, analysis and evidence. Greenwich, CT:
Information Age Publishers.
Cohen, S. G., & Bailey, E. B. (1997). What makes teams work: Group effectiveness research
from the shop oor to the executive suite. Journal of Management, 23(3), 239290.
Colquitt, J. A., LePine, J. A., & Noe, R. A. (2000). Toward an integrative theory of training
motivation: A meta-analytic path analysis of 20 years of research. Journal of Applied
Psychology, 85(5), 678707.
239
Colquitt, J. A., Noe, R. A., & Jackson, C. L. (2002). Justice in teams: Antecedents and
consequences of procedural justice climate. Personnel Psychology, 55(1), 83109.
Creed, P. A., King, V., Hood, M., & McKenzie. (2009). Goal orientation, self-regulation
strategies, and job-seeking intensity in unemployed adults. Journal of Applied Psychology,
94(3), 806813.
Daft, R. L., & Lengel, R. H. (1986). Organizational information requirements, media richness
and structural design. Management Science, 32, 554571.
Davidson-Shivers, G. V., & Rasmussen, K. L. (2006). Web-based learning: Design,
implementation and evaluation. Columbus, Ohio: Pearson Merrill Prentice Hall.
Davis, F. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information
technology. MIS Quarterly, 13, 319340.
DeRouin, R. E., Fritzsche, B. A., & Salas, E. (2004). Optimizing e-learning: Researchbased guidelines for learner-controlled training. Human Resource Management, 43(2/3),
147162.
DeRouin, R. E., Fritzsche, B. A., & Salas, E. (2005). E-learning in organizations. Journal of
Management, 31, 920940.
Dewey, J. (1938). Experience and education. West Lafayette, IN: Kappa Delta Pi.
Dillemans, R., Lowyck, J., Van der Perre, G., Claeys, C., & Elen, J. (1998). New technologies for
learning: Contribution of ICT to innovation in education. Leuven, Belgium: Leuven
University Press.
De Volder, M. (1996). From penny post to information super-highway: Open and distant learning
in close up. Leuven/Amersfoort: ACCO.
Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitivedevelopmental inquiry. American Psychologist, 34(10), 906911.
Ford, J. K., Smith, E. M., Weissbein, D. A., Gully, S. M., & Salas, E. (1998). Relationships of
goal orientation, metacognitive activity, and practice strategies with learning outcomes
and transfer. Journal of Applied Psychology, 83(2), 218233.
Frankola, K. (2001). Why online learners drop out. Workforce, 80(10), 53.
Gagne, R. M. (1985). The conditions of learning (4th ed.). New York: Holt, Rinehart & Winston.
Galagan, P. A., & Drucker, P. F. (2000). The e-learning revolution. Training & Development,
54(12), 24.
Garrison, D. R. (1989). Understanding distance education: A framework for the future. London:
Routledge.
Garrison, R. (1993). A cognitive constructivist view of distance education: An analysis of
teaching-learning assumptions. Distance Education, 14(2), 199211.
Gorham, J. (1988). The relationship between verbal teacher immediacy behaviors and student
learning. Communication Education, 37, 4053.
Gorsky, P., & Caspia, A. (2005). Dialogue: A theoretical framework for distance education
instructional systems. British Journal of Educational Technology, 36(2), 137144.
Gorsky, P., Caspia, A., & Trumper, R. (2004). Use of instructional dialogue by university
students in a distance education physics course. Open Learning: The Journal of Open and
Distance Learning, 19(3), 265277.
Gorsky, P., Caspia, A., & Tuvi-Arad, I. (2004). Use of instructional dialogue by university
students in a distance education chemistry course. Journal of Distance Education, 19(1),
119.
Hall, E. T. (1976). Beyond culture. New York: Doubleday.
Hare, A. P. (1981). Group size. American Behavioral Scientist, 24, 695708.
240
Hedberg, J., Brown, C., & Arrighi. (1997). Interactive multimedia and web-based learning:
Similarities and differences. In: B. H. Khan (Ed.), Web-based instruction. New Jersey:
Englewood Cliffs.
Hequet, M., & Johnson, G. (2003). The state of the e-learning market. Training, 40(8), 24.
Hill, C. A. (1987). Afliation motivation: People who need peopley but in different ways.
Journal of Personality and Social Psychology, 52(5), 10081018.
Hill, T., Smith, N. D., & Mann, M. F. (1987). Role of efcacy expectations in predicting
the decision to use advanced technologies: The case of computers. Journal of Applied
Psychology, 73(2), 307313.
Hofmann, D. A., Morgeson, F. P., & Gerras, S. J. (2003). Climate as a moderator of the
relationship between leader-member exchange and content specic citizenship: Safety
climate as an exemplar. Journal of Applied Psychology, 88, 170178.
Holmes, B., & Gardner, J. (2006). E-learning: Concepts and practice. London: Sage Publications.
Holtgraves, T. (1997). Styles of language use: Individual and cultural variability in conversational indirectness. Journal of Personality and Social Psychology, 73(3), 624637.
Horton, W., & Horton, K. (2003). E-learning tools and technologies. Indianapolis, IN: Wiley
Publishing, Inc.
Jonassen, D. H., Campbell, J. P., & Davidson, M. E. (1994). Learning with media:
Restructuring the debate. ETR&D, 42, 3140.
Kanfer, R., & Heggestad, E. D. (1997). Motivational traits and skills: A person centered
approach to work motivation. Research in Organizational Behavior, 19, 156.
Kaplan-Leiserson, E. (2002). E-learning glossary. Available at http://www.learningcircuits.org/
glossary.html
Keegan, D. (1986). The foundations of distance education. Kent: Croom Helm.
Keller, J. M. (1983). Instructional design theories and models: An overview of their current status.
Philadelphia: Lawrence Erlbaum Associates.
Kelley, H. H. (1967). Attribution theory in social psychology. In: D. Levine (Ed.), Nebraska
symposium on motivation (Vol. 15, pp. 192238). Lincoln, NE: University of Nebraska Press.
Kirkman, B. L., & Mathieu, J. E. (2005). The dimensions and antecedents of team virtuality.
Journal of Management, 31(5), 700.
Klein, H. J., Noe, R. A., & Wang, C. (2006). Motivation to learn and course outcomes: The
impact of delivery mode, learning goal orientation, and perceived barriers and enablers.
Knowles, M. (1990). The adult learner. Houston, TX: Gulf Publishing.
Kolb, D. A. (1984). Experiential learning: Experience as the source of learning and development.
New Jersey: Prentice-Hall.
Kovaleski, D. (2004). Blended learning in focus. Corporate Meetings & Incentives, 23, 3536.
Kozma, R. B. (1991). Learning with media. Review of Educational Research, 61, 179211.
Lauzon, A. C., & Moore, G. A. B. (1992). A fourth generation distance education system:
Integrating computer-assisted learning and computer conferencing. In: M. G. Moore
(Ed.), Readings in distance education (Vol. 3, pp. 2637). University Park, PA:
Pennsylvania State University, American Center for the Study of Distance Education.
Leidner, D. E., & Jarvenpaa, S. L. (1995). The use of information technology to
enhance management school education: A theoretical view. MIS Quarterly, 19(3),
265292.
Lemak, D. J., Shin, S. J., Reed, R., & Montgomery, J. C. (2005). Technology, transactional
distance, and instructor effectiveness: An empirical investigation. Academy of Management Learning & Proceedings, 4, 150159.
241
Lou, Y., Abrami, P. C., & Apollonia. (2001). Small group and individual learning with
technology: A meta-analysis. Review of Educational Research, 71, 449521.
Lowe, J. S., & Holton, E. F. I. (2005). A theory of effective computer-based instruction for
adults. Human Resource Development Review, 4(2), 159188.
Martins, L. L., & Kellermanns, F. W. (2004). A model of business school students acceptance of
a web-based course management system. Academy of Management Learning & Education,
3(1), 726.
Maruping, L. M., & Agarwal, R. (2004). Managing team interpersonal processes
through technology: A task-technology t perspective. Journal of Applied Psychology,
89(6), 975990.
McGrath, J. E. (1984). Groups: Interaction and performance. Englewood Cliffs, NJ: Prentice-Hall.
Mehrabian, A. (1972). Nonverbal communication. Chicago: Aldine.
Merriam, S., & Caffarella, R. (1999). Learning in adulthood: A comprehensive guide. San
Francisco: Jossey-Bass.
Mezirow, J. (1991). Transformative dimensions of adult development. San Francisco: Jossey-Bass.
Milheim, W. D., & Martin, B. L. (1991). Theoretical bases for the use of learner control: Three
different perspectives. Journal of Computer-Based Instruction, 18, 5156.
Moore, M. G. (1973). Toward a theory of independent learning and teaching. Journal of Higher
Education, 44, 661679.
Moore, M. G. (1991). Distance education theory. The American Journal of Distance Education,
5, 16.
Moore, M. G. (1993). Theory of transactional distance. In: D. Keegan (Ed.), Theoretical
principles of distance education. London: Routledge.
Moshinskie, J. (2001). How to keep e-learners from escaping. Performance Improvement, 40(6),
2835.
Najjar, L. J. (1996). Multimedia information and learning. Journal of Educational Multimedia
and Hypermedia, 5, 129150.
Noe, R. A. (2005). Employee training & development (3rd ed.). Burr Ridge: Irwin McGraw-Hill.
Porath, C., & Bateman, T. (2006). Self-regulation: From goal orientation to job performance.
Proost, K. (1998). Telematic learning environments: Description and dimensions. In:
SCIENTER (Ed.), Research perspectives on open and distance learning. Bologna, Italy:
SCIENTER. Available at: http://www.scienter.org/; http://icdllit.open.ac.uk/icdlbrowse1.
php?a=00010141
Reeves, T. (1993). Pseudoscience in computer-based instruction: The case of learner control
research. Journal of Computer-Based Instruction, 20(2), 3946.
Richey, R. C. (2000). Reections on the state of educational technology research and development: A response to Kozma. Educational Theory Research & Development, 48, 1618.
Riding, R. J., & Cheema, I. (1991). Cognitive styles: An overview and integration. Educational
Salas, E., & Cannon-Bowers, J. A. (2001). The science of training: A decade of progress. Annual
Review of Psychology, 52(1), 471.
Salas, E., Kosarzycki, M. P., Burke, C. S., Fiore, S. M., & Stone, D. L. (2002). Emerging themes
in distance learning research and practice: Some food for thought. International Journal
of Management Reviews, 4(2), 135153.
Schmidt, A. M., & Ford, J. K. (2003). Learning within a learner control training environment:
The interactive effects of goal orientation and metacognitive instruction on learning
outcomes. Personnel Psychology, 56(2), 405.
242
Schneider, B. (1990). The climate for service: An application of the climate construct. In:
B. Schneider (Ed.), Organizational climate and culture (pp. 383412). San Fransisco:
Jossey-Bass.
Simons, T., & Roberson, Q. (2003). Why managers should care about fairness: The effects of
aggregate justice perceptions on organizational outcomes. Journal of Applied Psychology, 88(3), 432443.
Sitzmann, T., Kraiger, K., Stewart, D., & Wisher, R. (2006). The comparative effectiveness of
web-based and classroom instruction: A meta-analysis. Personnel Psychology, 59, 623664.
Skinner, B. F. (1974). About behaviorism. New York: Random House, Inc.
Smith, P. J. (2000). Preparedness for exible delivery among vocational learners. Distance
Education, 21(1), 2948.
Smith, P. J. (2005). Learning preferences and readiness for online learning. Educational
Psychology, 25(1), 312.
Smith, P. J., Murphy, K. L., & Mahoney, S. E. (2003). Towards identifying factors underlying
readiness for online learning: An exploratory study. Distance Education, 24(1), 5767.
Smith, P. L., & Dillon, C. L. (1999). Comparing distance learning and classroom learning:
Conceptual considerations. American Journal of Distance Education, 13, 107124.
Sternberg, R. J. (1997). Thinking styles. New York: Cambridge University Press.
Taylor. (2001). Fifth generation distance education. Available at www.usq.edu.au/users/taylorj/
conferences.htm
Thompson, L. F., & Lynch, B. J. (2003). Web-based instruction: Who is inclined to resist it and
why? Journal of Educational Computing Research, 29(3), 375385.
Thorndike, E. L. (1932). The fundamentals of learning. New York: Teachers College, Columbia
University.
Tutty, J., & Klein, J. (2008). Computer-mediated instruction: a comparison of online and faceto-face collaboration. Educational Technology Research & Development, 56, 101124.
Vroom. (1964). Work and motivation. New York: Wiley.
Wang, A. Y., & Newlin, M. H. (2000). Characteristics of students who enroll and succeed in
psychology web-based classes. Journal of Educational Psychology, 92(1), 137143.
Webster, J., & Hackley, P. (1997). Teaching effectiveness in technology-mediated distance
learning. Academy of Management Journal, 40(6), 12831309.
Welsh, E. T., Wanberg, C. R., Brown, K. G., & Simmering, M. J. (2003). E-learning: Emerging
uses, empirical results and future directions. International Journal of Training &
Development, 7(4), 245258.
Wenger, E. (1998). Systems thinker. Cambridge, MA: Cambridge University Press.
Whipp, J. L., & Chiarelli, D. (2004). Self-regulation in a web-based course: A case study.
Educational Technology Research & Development, 4, 522.
Workman, M., Kahnweiler, W., & Bommer, W. (2003). The effects of cognitive style and media
richness on commitment to telework and virtual teams. Journal of Vocational Behavior,
63(2), 199219.
Wydra, F. T. (1980). Learner controlled instruction. Englewood Cliffs, NJ: Educational
Technology Publications.
Zhang, D. (2005). Interactive multimedia-based e-learning: A study of effectiveness. The
American Journal of Distance Education, 19(3), 149162.
ABOUT THE AUTHORS

M. Ronald Buckley holds the JC Penney Company Chair of Business
Leadership and is a professor of management and a professor of psychology
in the Michael F. Price College of Business at the University of Oklahoma.
He earned his Ph.D. in industrial/organizational psychology from Auburn
University. His research interests include, among others, work motivation,
racial and gender issues in performance evaluation, business ethics,
interview issues, and organizational socialization. His work has been
published in journals such as the Academy of Management Review,
Personnel Psychology, Journal of Applied Psychology, Organizational
Behavior and Human Decision Processes, and the Journal of Management.
Michael J. Burke is the Lawrence Martin Chair in Business at Tulane
University, and he holds adjunct appointments in Tulanes Department of
Psychology and School of Public Health and Tropical Medicine. He received
his Ph.D. in psychology from Illinois Institute of Technology. He is a past
president of the Society for Industrial and Organizational Psychology and he
served as editor of Personnel Psychology. In 2006, he was awarded the
Decade of Behavior Research Award for his research on workplace safety
from a federation of professional scientic associations. In 2009, Professor
Burke completed a three-year term on the Safety and Occupational Health
Study Section of the National Institute for Occupational Safety and Health.
In the domain of workplace safety, his research interests focus on safety
climate, safety and health training, and safety performance.
Michelle K. Duffy is a professor and the Board of Overseers Professor of
HRIR in the Carlson School of Management at the University of
Minnesota. She received a Ph.D. from the University of Arkansas. Her
research interests include employee well-being, antisocial behaviors,
emotions, and mood.
Beth Florin is Managing Director of Pearl Meyer and Partners and leads the
Survey and Employee Compensation Practice. She has specialized experience
in the design, development, and implementation of broad-based compensation programs and total remuneration compensation surveys. She was a
243
244
ABOUT THE AUTHORS
co-founder of Executive Alliance, a technology industry compensation

consultancy that was acquired by Clark Consulting in 2001. Prior to that,
she was a senior human resource consultant with William M. Mercer,
Incorporateds High Tech Compensation Practice and held human resource
positions at Data General Corporation. She is a graduate of the University of
Florida and holds and M.S. in human resource management and research
methodology from Cornell University. She is a member of the Deans
Advisory Council at the ILR School at Cornell and is a member of the Board
of the Compensation Research Initiative (CRI) at Cornell.
Juliya Golubovich is a student in the Graduate Program in Organizational
Psychology at Michigan State University. She received her B.B.A. in
industrial/organizational psychology from Baruch College, City University
of New York. Her primary research interests are in adverse impact against
minorities in testing, development of alternate test instruments, and noncognitive predictors of performance.
Jonathon R. B. Halbesleben (Ph.D., University of Oklahoma) is the
HealthSouth Chair of Health Care Management and associate professor
in the Culverhouse College of Commerce and Business Administration at
the University of Alabama. His research interests include stress and
burnout, workfamily issues, and health care management. His work has
appeared in such journal outlets as the Journal of Applied Psychology,
Journal of Management, Academy of Management Learning and Education,
and Research in Personnel and Human Resource Management.
Kevin F. Hallock is a professor of labor economics and of human resource
studies and director of the Compensation Research Initiative (CRI) at the
ILR School at Cornell University. He is also a research associate at the
National Bureau of Economic Research (NBER), a senior fellow for
executive compensation, board compensation and board practices at The
Conference Board and a member of the board of directors at WorldatWork.
His current research includes projects on executive and director compensation, the valuation of stock options and the design of compensation systems.
He earned a B.A. in economics from the University of Massachusetts at
Amherst and a Ph.D. in economics from Princeton University.
Jaron Harvey is an assist professor in the Culverhouse College of Commerce
and Business Administration at the University of Alabama. He received
his Ph.D. from the University of Oklahoma. His research interests are in
the exchange relationships that exist between employers and employees.
He is specically interested in what creates these exchange relationships and
About the Authors
245
the exchanges that occur when employees go the extra mile for their
employers.
N. Sharon Hill is an assistant professor at The George Washington
University School of Business. She received her Ph.D. in organizational
behavior and human resources from University of Maryland, College Park.
Dr. Hills research interests include technology-mediated work arrangements (e-learning and virtual teams) and organizational change. These
interests are motivated by her extensive corporate work experience prior to
obtaining her Ph.D. Dr. Hill has presented her research at national and
international conferences. Her work has appeared in Organizational
Behavior and Human Decision Processes and the Journal of Applied
Behavioral Science, and has been recognized as a Best Paper by the
Academy of Management Organizational Development and Change
Division.
Peter W. Hom is a professor of management at Arizona State University
(Tempe, AZ). He received his Ph.D. from the University of Illinois
(Champaign-Urbana) in industrial/organizational psychology. Dr. Hom has
investigated theories of employee turnover in various occupations (Chinese
managers, Swiss bankers, industrial salesmen, retail sales personnel,
National Guardsmen, Mexican factory workers), designed realistic job
previews to reduce reality shock and early quits among new nurses and
accountants, and estimated the economic costs of turnover for mental health
agencies. He has authored scholarly articles in the Academy of Management
Journal, the Journal of Applied Psychology, Organizational Behavior and
Human Decision Processes, and Personnel Psychology. He has authored two
books entitled Employee Turnover and Retaining Valued Employees with
Rodger Griffeth. Dr. Hom serves on the editorial board for the Journal of
Applied Psychology.
Jenny M. Hoobler is an assistant professor of management in the College of
Business Administration at the University of Illinois at Chicago. She
received her Ph.D. from the University of Kentucky. She serves on the
editorial boards of the Journal of Organizational Behavior and Journal of
Management Studies, and on the Executive Committee of the Academy of
Managements Human Resource Management Division. She has published
in a variety of journals including the Academy of Management Journal and
Journal of Applied Psychology. Her research focuses on supervisor
subordinate relationships, gender and diversity, and intersections between
work and non-work domains.
246
ABOUT THE AUTHORS
Frederick T. L. Leong is a professor of psychology at Michigan State

University in the Industrial/Organizational and Clinical Psychology
programs. He is also the director of the Consortium for Multicultural
Psychology Research at MSU. He has authored or co-authored over 130
articles in various psychology journals, 80 book chapters, and also edited or
co-edited 10 books. He is an editor-in-chief of the Encyclopedia of
Counseling (Sage Publications) and editor of the Division 45 Book Series
on Cultural, Racial and Ethnic Psychology. He is the founding editor of the
Asian American Journal of Psychology. Dr. Leong is a fellow of the
American Psychological Association, Association for Psychological Science,
Asian American Psychological Association and the International Academy
for Intercultural Research. His major research interests center around
culture and mental health, cross-cultural psychotherapy (especially with
Asians and Asian Americans), cultural and personality factors related to
career choice, work adjustment, and occupational stress.
Jason D. Shaw is a professor and the Curtis L. Carlson Professor of
Industrial Relations at the University of Minnesota. He received a Ph.D.
from the University of Arkansas. His research interests include organizational turnover, team effectiveness, pay systems, and personality.
Sloane M. Signal is a doctoral student in the A.B. Freeman School of
Business at Tulane University. Before enrolling in the Freeman School, she
served as sequence coordinator and faculty for Advertising and Public
Relations at the Howard University John H. Johnson School of Communications. From 2001 to 2005, Sloane was a member of the Journalism faculty
at the University of Nebraska in Lincoln, where she co-authored The Peer
Review of Teaching Portfolio Project as Scholarship Assessment in Higher
Education: An Advertising Curriculum Example. Her research interests include
communicating across cultures both inside and outside of the United States,
diversity and multiculturalism in the workplace and the role of acculturation,
and the scholarship of teaching and learning.
Bennett J. Tepper is a professor of managerial sciences in the J. Mack
Robinson College of Business at Georgia State University. He received his
Ph.D. in organizational psychology from the University of Miami. He is a
fellow of the American Psychological Association, Society of Industrial and
Organizational Psychology and the Southern Management Association. His
research interests include leadership, employee well-being, and employee
performance contributions. He currently serves on the editorial boards of
About the Authors
247
Academy of Management Journal, Journal of Applied Psychology, Journal of

Organizational Behavior, and the Journal of Management.
Douglas Webber is a Ph.D. student in economics at Cornell University and is
being funded on a National Science Foundation Fellowship. He is a member
of the Cornell Higher Education Research Institute (CHERI) and the
Compensation Research Initiative (CRI) at Cornell. He is a graduate of the
University of Florida. His research interests are in compensation design and
the economics of education.
Anthony Wheeler is an associate professor of human resources management
in the Schmidt Labor Research Center and the College of Business
Administration at the University of Rhode Island. He completed his
undergraduate degree at the University of Maryland, College Park and
earned both his masters and doctoral degrees at the University of
Oklahoma; moreover, he is a certied senior professional in human
resources management (SPHR). His research interests include the inuence
of HRM practices on personenvironment t and include examining issues
related to alternative stafng strategies. This research has lead to the
publication of several scholarly articles in outlets such as Journal of
Management Education, Work & Stress, Leadership Quarterly, Journal of
Occupational and Health Psychology, Issues in Multilevel Research, Research
in Personnel and Human Resources Management, International Journal of
Selection and Assessment, Journal of Business Ethics, Journal of Managerial
Issues, and Journal of Business Logistics
Karen Wouters is a lecturer at the University of Marylands Robert H. Smith
School of Business. Prior to joining the Smith School, she was a research
associate at the Vlerick Management School, Belgium. She received her
Ph.D. in applied economic sciences from Ghent University, Belgium. Her
research interests are primarily in the areas of leadership development,
executive coaching, learning from on-the-job experiences and e-learning.
Dr. Wouters has written articles in the areas of e-learning, vocational
training, and on-the-job learning and has presented her research at national
and international conferences. One of her articles was given the 2002 Highly
Commended Award by Emerald Literati Club. In 2006 and 2008, she
received the Global Forum Best Paper and the Best Paper in
Management Development from the Academy of Management.

HRM

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HRM

Uploaded by

Copyright:

Available Formats

RESEARCH IN PERSONNEL

AND HUMAN RESOURCES

Edited by Kendrith M. Rowland and

Volumes 1120: Edited by Gerald R. Ferris

International Human Resources

International Human Resources

International Human Resources

International Human Resources Management

Edited by Joseph J. Martocchio and

Edited by Joseph J. Martocchio and

Volumes 2327: Edited by Joseph J. Martocchio

Edited by Joseph J. Martocchio and Hui Liao

RESEARCH IN PERSONNEL AND HUMAN RESOURCES

United Kingdom North America Japan

Emerald Group Publishing Limited

WORKPLACE SAFETY: A MULTILEVEL,

COMPARING APPLES AND ORANGES:

ABOUT THE AUTHORS

Price College of Business, Division of

Freeman School of Business, Tulane

Carlson School of Management, University

Pearl Meyer & Partners, Southborough,

Department of Psychology, Michigan State

Department of Management and Marketing,

Cornell University and NBER, ILR School,

Department of Management and Marketing,

School of Business, The George Washington

Department of Management, W.P. Carey

College of Business Administration,

Department of Psychology, Michigan State

Carlson School of Management, University

Freeman School of Business, Tulane

J. Mack Robinson College of Business,

Cornell University, ILR School, Ithaca,

Schmidt Labor Research Center, College of

Robert H. Smith School of Business,

Research in Personnel and Human Resources Management, Volume 29, 147

MICHAEL J. BURKE AND SLOANE M. SIGNAL

An impressive body of literature on workplace safety that spans across

A MULTILEVEL MODEL OF WORKPLACE SAFETY

Workplace Safety: A Multilevel, Interdisciplinary Perspective

theoretical foundation. For our purposes, we begin with a discussion of safety

MICHAEL J. BURKE AND SLOANE M. SIGNAL

Given the literature on job performance, Marchand et al.s (1998) factor

Workplace Safety: A Multilevel, Interdisciplinary Perspective

MICHAEL J. BURKE AND SLOANE M. SIGNAL

the applied psychology literature that identies knowledge and motivation

ANTECEDENTS OF SAFETY COMPLIANCE

Workplace Safety: A Multilevel, Interdisciplinary Perspective

America, and South Africa (Ahasan, Mohiuddin, Vayrynen, Ironkannas, &

National/Regional Level Antecedents

MICHAEL J. BURKE AND SLOANE M. SIGNAL

employees perceptions of workplace safety and accident involvement (i.e.,

Workplace Safety: A Multilevel, Interdisciplinary Perspective

characteristics of the political economy that may signicantly impact the

MICHAEL J. BURKE AND SLOANE M. SIGNAL

by the cultural value of uncertainty avoidance. These arguments form the

Workplace Safety: A Multilevel, Interdisciplinary Perspective

Furthermore, cultural values would be expected to inuence the

MICHAEL J. BURKE AND SLOANE M. SIGNAL

Workplace Safety: A Multilevel, Interdisciplinary Perspective

MICHAEL J. BURKE AND SLOANE M. SIGNAL

Workplace Safety: A Multilevel, Interdisciplinary Perspective

considerable evidence showing that differences between individuals often