You are on page 1of 25

The Context of Organization Structures

Author(s): D. S. Pugh, D. J. Hickson, C. R. Hinings, C. Turner


Source: Administrative Science Quarterly, Vol. 14, No. 1 (Mar., 1969), pp. 91-114
Published by: Johnson Graduate School of Management, Cornell University
Stable URL: http://www.jstor.org/stable/2391366
Accessed: 14/09/2009 13:24

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=cjohn.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

Johnson Graduate School of Management, Cornell University is collaborating with JSTOR to digitize, preserve
and extend access to Administrative Science Quarterly.

http://www.jstor.org
D. S. Pugh, D. J. Hickson, C. R. Hinings, and C. Turner

The Context of OrganizationStructures


Aspects of organizational context that have been held to be relevant to organiza-
tional structure were examined. Seven primary concepts of organizational context,
viz.: origin and history, ownershipand control, size, charter, technology, location and
dependence on other organizations, were analyzed and operationally defined scales
constructed. These were used as independent variables in a multivariate'regression
analysis to predict three underlying dimensions of organization structure previously
established. The size of the correlations obtained on a sample of 46 organizations in
the English Midlands (0.75 with structuring of activities using size and technology
as predictors; 0.75 with concentration of authority using dependence and location as
predictors; 0.57 with line control of workflow, using the operating variability scale
of charter as a predictor) indicates that these aspects of context are salient for struc-
ture. The frameworkof contextual and structural variables is seen as making possible
processual studies on a much more rigorous comparative basis than before.'

The structure of an organization is closely causes of differences between structures, and


related to the context within which it functions, large size has even been considered as charac-
and much of the variation in organizationstruc- teristic of bureaucratic structure (Presthus,
tures might be explained by contextual factors. 1958). Others argue for the pre-emptive im-
Many such factors, including size, technology, portance of the technology of production or
organizational charter or social function, and service in determiningstructureand functioning
interdependencewith other organizations, have (Dubin, 1958; Perrow, 1967; Woodward, 1965;
been suggested as being of primary importance Trist et al., 1963). Parsons (1956) and Selznick
in influencing the structure and functioning of (1949) have attempted to show in some detail
an organization. that the structure and functioning of the orga-
There have been few attempts, however, to nization follow from its social function, goals,
relate these factors in a comparativesystematic or "charter." Eisenstadt (1959) emphasizedthe
way to the characteristic aspects of structure, importance of the dependence of the organiza-
for such studies would require a multivariate tion on its social setting, particularly its de-
factorial approach in both context and struc- pendence on external resources and power, in
ture. The limitations of a unitary approach to influencing structural characteristicsand activi-
organizational structure have been elaborated ties. Clearly all of these contextual factors, as
elsewhere (Hinings et al.; 1967), but its defi- well as others, are relevant; but without a multi-
ciencies in the study of contextual factors are variate approach, it is not possible to assess
no less clear. Theorists in this area seem to their relative importance.
have proceededon the assumption that one par- A previous paper described the conceptual
ticular contextual feature is the major determi- frameworkupon which the present multivariate
nant of structure, with the implication that analysis is based (Pugh et al., 1963), and a
they considered the others less important. subsequent paper its empirical development
Many writers from Weber onwards have men- (Pugh et al., 1968). It is not a model of orga-
tioned size as being one of the most important nization in an environment,but a separation of
variables of structureand of organizationalper-
1
This work was conducted when the authors were formance from other variables commonly hy-
members of the Industrial Administration Research pothesized to be related to them, which are
Unit, the University of Aston in Birmingham, England.
Research conducted by that Unit is jointly supported
called "contextual" in the sense that they can
by the Social Science Research Council and the Uni- be regardedas a setting within which structure
versity. is developed. Table 1 summarizes the frame-
91
92 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 1. CONCEPTUAL SCHEME FOR EMPIRICAL STUDY OF WORK ORGANIZATIONS.

Contextual variables Structural variablest


Origin and history Structuring of activities
Ownership and control Functional specialization
SizeFucinlseilzto
Charter Role specialization
Technology Standardization (overall)
Location Formalization (overall)
Resources Concentration of authority
Dependence
Centralization of decision making
Activity variables* Autonomy of the organization
Identification Standardization of procedures for
(charter, image) selection and advancement
Perpetuation Line control of workflow
(thoughtways, Subordinate ratio
finance,
personnel services) Formalization of role performance
recording
Workflowreodn
(production Percentage of workflow superordinates
distribution) Relative size of supportive component

(direction Percentage of clerks


motivation, Percentage of nonworkflow personnel
evaluation, Vertical span (height)
communication) Perf ormance variables
Homeostasis Efficiency
(fusion, (fitabiiy
leadership, (profitability,
problem solving, productivity,
legitimization)
legitimization) market standing)
~~Adaptability
Morale
*Bakke (1959).
t Pugh et al. (1968).

work and also includes a classification of activi- does not in itself test hypotheses about processes
ties useful in the analysis of organization func- (e.g. how changes in size interact with varia-
tioning (Bakke, 1959). tions in structuring of activities), but it affords
The design of the study reported in the pres- a basis for generating such hypotheses.
ent paper treats the contextual variables as
independent and the structural variables as de- SAMPLE AND METHODS
pendent. The structural variables are (i) struc- Data were collected on fifty-two work orga-
turing of activities; that is, the degree to which nizations, forty-six of which were a random
the intended behavior of employees is overtly sample stratified by size and product or pur-
defined by task specialization, standard rou- pose. The sample and methods have been de-
tines, and formal paper work; (ii) concentra- scribed in detail in a previous paper, (Pugh et
tion of authority; that is, the degree to which al., 1968). For scaling purposes, data on the
authority for decisions rests in controlling units whole group were used, but for correlational
outside the organization and is centralized at analyses relating scales to each other, and for
the higher hierarchical levels within it; and prediction analyses relating contextual variables
(iii) line control of workflow; that is, the de- to structural ones, only data on the sample of
gree to which control is exercised by line per- forty-six organizationswere used. None of the
sonnel instead of through impersonal proce- data was attitudinal.
dures. The eight contextual variables were The data were analyzed under the heading
translated into operationaldefinitions and scales of the conceptual scheme. To define the varia-
were constructed for each of them. These were bles operationally, scales were constructed that
then used in a multivariate regression analysis measured the degree of a particular character-
to predict the structural dimensions found. istic. The scales varied widely. Inkson et al.
This factorial study using cross-sectionaldata (1967) discussed the variety of scaling proce-
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 93
dures used. Some were simple dichotomies intercorrelations. The methodological implica-
(such as impersonality of origin) or counts tions of this analysis are discussed in Levy and
(such as numberof operating sites); some were Pugh (1969).
ordered category scales, locating an organiza- The study of contextual aspects of organiza-
tion at one point along a postulated dimension tions will inevitably produce a much more het-
(such as closeness of link with customers of erogeneous set of scales than the comparable
clients). Some were stable, orderedscales estab- study of the structural aspects; for the scales
lished by linking together a large number or are selected, not from a common conceptual
items exhibiting the characteristic on the basis base, but for their postulated links with struc-
of cumulative scaling procedures,such as work- ture. One of the objectives of using the multi-
flow rigidity, an aspect of technology. Some variate approach described here would be to
were summary scales extracted by principal- test the relationship between disparate aspects
components analysis to summarize a whole of context, and to attempt a conceptual clarifi-
dimension, such as operating variability, an cation of those aspects demonstrated to be
aspect of charter. In this way, forty primary salient in relation to organizational structure.
scales of context were constructed and then It was not possible to investigate the variable
reduced to fourteen empirically distinct ele- "resources" adequately. For human and idea-
ments, which are listed in Table 2 together with tional resources, the wide-ranging interviews
their correlationswith the main structural vari- within a comparatively short time span made
ables as defined in Table 1. Table 3 gives their it impossible to obtain adequate data. Material

TABLE 2. ELEMENTS OF ORGANIZATIONCONTEXT.

Product-moment correlation with structural factors


Structuring Concentration Line control
Elements of context of activities of authority of workflow
Origin and history (3)*
Impersonality of
origin -0.04 0.64 0.36
Age 0.09 -0.38 -0.02
Historical changes 0.17 -0.45 -0.03
Ownership and control (7)
Public accountability -0.10 0.64 0.47
Concentration of
ownership with control -0.15 -0.29 -0.21
Size (3)
Size of organizationt 0.69 -0.10 -0.15
Size of parent
organizationt 0.39 0.39 -0.07
Charter (7)
Operating variability 0.15 -0.22 -0.57
Operating diversity 0.26 -0.30 -0.04
Technology (6)
Workflow integration 0.34 -0.30 -0.46
Labor costs -0.25 0.43 0.32
Location (1)
Number of operating
sites -0.26 0.39 0.39
Dependence (10)
Dependence -0.05 0.66 0.13
Recognition of trade
unions 0.51 0.08 -0.35
* Numbers in parentheses indicate number of primary scales.
t Logarithm of number of employees.
94 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 3. INTERCORRELATIONS OF CONTEXTUAL VARIABLES (PRODUCT-MOMENT


COEFFICIENTS, N = 46).

,o >1
- .

o ~ d+0 0 0C 0 0 4-
Ipsni of "a
Age- -0.200
-0 C)
0C) cd~ 9 C- d
Historical changes j~C
-.4- 0.5 0 Cd E

Age -0.20 -~~~~~~~~~~~W 0 C


Public
Simersonality
aconablt
of ignizaiont
or
~ ~ 0.66-
07
0.0
0 -02
02 0.0 -0.21
~
Size~ of paren ~~~~~~=
oranztn 0.4 -0.1 -0.1 0.5 -05 04
Historiain changesility
Concerntaltyo of ownrship -0.26-0.24-0.16
0.50
-0.34 -4 -
Opublticg
accountaity -0.6 0.00 -0.25 -0 00 0.

with control* -0.40 -0.03 0.02 -0.50 -


Size of organizationt 0.07 0.16 0.29 0.00 -0.21 -
Size of parent organizationt 0.45 -0.12 -0.10 0.51 -0.55 0.43 -

Operatingvariability -0.26 -0.24 -0.16, -0.34 0.29 -0.24 -0.19 -

Operatingdiversity -0.23 0.00 0.13 -0.14 0.00 0.26 -0.10 0.02 -

Workflow integration -0.24 0.07 0.05 -0.35 0.10 0.07 -0.09 0.57 0.33 -
Labor Costs 0.41 -0.24 -0.31 0.34 -0.09 -0.28 0.08 -0.27 0.01 -0.50
No. of operating sites 0.14 -0.07 -0.08 0.34 -0.20 0.14 0.16 -0.56 -0.05 -0.58 0.16
Dependence 0.53 -0.32 -0.38 0.53 -0.50 -0.17 0.63 0.05 -0.19 -0.05 0.26 0.05 -
Recognition of trade unions 0.04 -0.04 -0.11 0.17 -0.21 0.36 0.37 0.19 0.01 0.20 -0.15 -0.12 0.22
* N = 42 for all correlations with this variable.
t Logarithm of number of employees.

and capital resources were found to reduce to sonally founded organizations might be ex-
aspects of size, and the relative disposition of pected to have a higher level of structuring of
these resources (e.g. capital versus,labor) was activities, whereas personally founded organiza-
found to be better regarded as an aspect of tions would have a higher degree of concentra-
technology. tion of authority. The data on the present
sample, however, show no relationship between
CONTEXTUAL VARIABLES impersonalityof origin and structuringof activ-
ities (r = -0.04), but a strong relationship be-
Origin and History tween impersonality of origin and concentra-
An organizationmay have grown from a one- tion of authority (r = 0.64). (With N = 46, all
man business over a long period of time, or it correlations 0.29 and above are at or beyond
may have been set up as a branch of an already the ninety-five percent level of confidence.)
existing organization and so develop! rapidly. To a considerable extent this relationship is
During its development it may have undergone due to the fact that government-owned, and
many or few radical changes in purpose, owner- therefore impersonally founded, organizations
ship, and other contextual aspects.. An ade-
tend to be highly centralized. Such organiza-
quate study of the impact of these factors on
tions tend to be line controlled in their work-
organizationalstructure must be conducted on
a comparative longitudinal basis (Chandler, flow, thus contributing to the relationship
1962); but even in a cross-sectionalstudy such (r = 0.36) between impersonality of origin and
as this, it is possible to define and make opera- line control of workflow. The lack of relation-
tional three aspects of this concept. ship with structuring of activities, which is
Impersonality of origin. This variable distin- common to all three scales of this dimension,
guishes between entrepreneurial organizations, underlines the need to examine present con-
personally founded, and bureaucratic ones textual aspects in relation to this factor rather
founded by an existing organization. Imper- than historical ones.
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 95
Age. The age of the organization was taken ferent ownership and control patterns. Two
from the time at which the field work was car- aspects of this concept, public accountability
ried out. The range in the sample varied from and the relationship of the ownership to the
an established metal goods manufacturingorga- management of the organization were investi-
nization, founded over 170 years previously, in gated. For wholly owned subsidiary companies,
1794, to a government inspection department, branch factories, local governmentdepartments,
which began activities in the area as a separate etc. this form of analysis had to be applied to
operating unit 29 years previously. No clear the parent institution exercising owning rights,
relationship was found between age and im- in some cases through more than one intermedi-
personality of origin (r = -0.20). Stinchcombe ate institution (e.g. committees,of the corpora-
(1965) has argued that no relationship should tion, area boards, parent operating companies,
be expected between the age of an organization which were themselves owned by holding com-
and its structure but rather between the struc- panies, etc.). The ultimate owning unit is re-
ture of an organization and the date that its ferred to as the "parent organization."
industry was founded. The present data sup- Public accountability. This was a three-point
port this conclusion in that no relationship is category scale concerned with the degree to
found between age and structuring of activities which the parent organization, (which could, of
(r = 0.09) or line control of workflow (r =0.02). course,be the organizationalunit itself, as it was
Age was related to concentration of authority in eight cases) was subject to public scrutiny
(r = -0.38), older organizations having a tend- in the conduct of its affairs. Least publicly ac-
ency to be more decentralizedand to have more countable would be a company not quoted on
autonomy. the stock exchange; next, organizations that
Historical changes. The organizationsin this raised money publicly by having equity capital
sample did not have adequate historical infor- quoted on the stock exchange, also public co-
mation on the extent of contextual changes for operative societies; and most publicly account-
use in a cross-sectionalinvestigation; but it was able were the departments of the local and cen-
possible to obtain limited information as to tral government. On the basis of the classical
whether particular changes had occurred, and literature on bureaucracy as a societal phe-
thus to develop a scale for the types of con- nomenon, it might be hypothesized that orga-
textual changes that had occurred, namely nizations with the greatest exposure to public
whether at least one change had occurred (i) in accountability would have a higher degree of
the location of the organization, (ii) in the structuring of activities, and a greater concen-
product or service range offered, and (iii) in tration of authority. The data on the present
the pattern of ownership. Item analysis carried sample show relationships more complicated
out using the Brogden-Clemens coefficient than this, however.
(Brogden, 1949) gave a mean item-analysis First, it must be emphasized that although
value of 0.85, suggesting that it was possible to this sample included eight government depart-
produce a scale of historical changes by sum- ments, all the organizations had a nonadminis-
ming the items. The organizations were dis- trative purpose, which could be identified as a
tributed along the scale from no changes to all workflow, Pugh, et al. (1968) Table 1. This is
three types of changes. As expected, there was not surprising in this provincial sample, since
a strong correlation of this scale with age (r = purely administrative units of the requisite size
0.51), older organizations tending to have ex- (i.e., employing more than 250 people) are few
periencedmore types of change. There was also outside the capital. The relationships between
a strong relationship, perhaps mediated by age, public accountability and structure must be in-
between historical changes and concentration terpreted in the light of this particular sample.
of authority (r = -0.45), such changes being as- No relationship was found between public
sociated with dispersionof authority. accountability and structuringof activities (r =
-0.10). This structuring factor applies to the
Ownership and Control
workflow as well as administrative activities of
The differences in structure between a de- the organization, and it appears that govern-
partment of the government and a private ment organizationswith a workflow are not dif-
business will be due to some extent to the dif- ferentiated from nongovernment organizations
96 ADMINISTRATIVE SCIENCE QUARTERLY

on this basis. On the other hand there was a management of the organization. For the four
positive relationship between public account- foreign-ownedorganizations in the sample, this
ability and concentration of authority (r= information was not available in England; the
0.63) standardizationof proceduresfor selection analysis was therefore based on n = 42 (Table
and advancement (r = 0.56) and line control 4). A sixth scale was developed for interlocking
of workflow (r = 0.47). These all point to cen- directorships; that is, the percentage of direc-
tralized but line-controlled government work- tors who held other directorships outside the
flow organizations (Pugh et al., 1969). The owning group. The intercorrelation matrix of
scale of standardizationwas a bipolar one, and these six variables suggested that factor analysis
a high score meant that the organization stan- would be helpful in summarizing an extensive
dardized its procedures for personnel selection analysis of ownership (Table 5). A principal-
and advancement,and also that it did not stan- components analysis was thus applied to the
dardize its procedures for workflow. The rela- matrix, and a large first factor accounting for 56
tionship between public accountability and this percent of the variance was extracted,which was
standardization scale suggests that the govern- heavily loaded on all variables except interlock-
ment workflow organizations standardize their ing directorshipsand was thereforetermed "con-
personnel procedures, but rely on professional centration of ownershipwith control."
line superordinates for workflow control. As would be expected, there was a negative
Relationship of ownership to management. relationship between public accountability and
The concepts of Sargent Florence (1961) were concentration of ownership with control (r =
found most fruitful in studying this aspect of -0.51); the more publicly accountable the own-
ownership and control, but the method used ership, the less concentratedit was, with central
was the selection of variables for a correlational and local government ownership epitomizing
approach,rather than classification on the basis diffuse ownership by the voting public.
of percentages. Florence studied the relation- The discussion about the effects of differing
ships of shareholders,directors, and executives. patterns of personal ownershipon organizations
Where these groups were completely separate and society originated with Marx, and has since
there was full separation of ownership, control, polarized into what Dahrendorf (1959) has
and management; where they were the same, called the, "radical" and "conservative" posi-
then ownership, control and management coa- tions. It is generally agreed that there has been
lesced. Between these two extremes, the scales a progressive dispersion of share ownership fol-
were designed in the present study to measure lowing the rise of the corporation, but there is
the degree of separation. Company records and little agreement, or systematic evidence, on the
public records were examined and five scales effects of this. The radicals (Burnham, 1962;
developed for the patterns of shareholding and Berle and Means, 1937) argue that present
the relationshipsbetween the ownershipand the ownership patterns have produced a shift in

TABLE 4. OWNERSHIP AND CONTROL(N = 42).

Scale number and title Range % Mean S.D.

12.01 Concentration of voteholdings (Percentage of


equity owned by top twenty shareholders) 0-100 38.47 32.37
12.03 Voteholdings of individuals (Percentage of indi-
viduals among top twenty shareholders) 0-100 17.19 26.89
12.04 Directors among top twenty voteholders (Per-
centage of directors among top twenty sharehold-
ers) 0-100 20.69 29.39
12.05 Directors' voteholdings (Percentage of equity
owned by all directors combined) 0- 99.9 9.40 19.61
12.06 Percentage of directors who are executives 0-100 46.11 32.73
12.09 Interlocking directorships (Percentage of direc-
tors with other directorships beyond owning
organization) 0-100 45.22 33.73
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 97

TABLE 5. OWNERSHIP AND CONTROL: INTERCORRELATION MATRIX (PRODUCT-


MOMENT COEFFICIENTS (N = 42).

Directors
Concentration Voteholdings among Percentage of
of of top twenty Directors' directors who Interlocking
voteholdings individuals voteholders voteholdings are executives directorships

Concentration of
volteholdings
Voteholdings of
individuals 0.62
Directors among top
twenty voteholders 0.54 0.87
Directors'
voteholdings 0.55 0.90 0.78
Percentage of directors
who are executives 0.26 0.30 0.37 0.20
Interlocking
directorships 0.32 0.03 0.04 0.09 0.33

control away from the entrepreneurto managers, fect, there seemed to be grounds for not pro-
who become important because of their control ceeding with them in a multivariate analysis.
over the means of production and the organiza-
Size
tion of men, materials, and equipment. The re-
sult then of dispersion of ownership is likely to There has. been much work relating size to
be dispersion of authority. However, the con- group and individual variables, such as morale
servatives (Mills, 1956; Aaronovitch, 1961) and job satisfaction, with not very consistent
argue that the dispersion of capital ownership results (Porter and Lawler, 1965). With few
makes possible the concentration of economic exceptions, empirical studies relating size to
power in fewer hands, because of the inability of variables of organization structure have con-
the mass of shareholders to act, resulting in a fined themselves to those broad aspects of the
concentration of authority. role structure which are here termed "configu-
The results obtained with this sample sup- ration" (Starbuck, 1965). Hall and. Tittle,
port neither of these positions. The correlation (1966), using a Guttman scale of the overall
given in Table 2 of concentration of ownership degree of perceived bureaucratizationobtained
with control with concentration of authority by combining scores on six dimensions of We-
(r = -0.29) might suggest that concentration berian characteristicsof bureaucracyin a study
of ownership is associated with dispersion of of twenty-five different work organizations,
authority; but it must be rememberedthat this found a small relation between their measure-
correlation is obtained for the whole sample, ment of perceived bureaucratizationand orga-
which includesgovernment-ownedorganizations, nization size (r = 0.252 at the 6 percent level
whereas the discussion of the effects of owner- of confidence).
ship patterns has been concerned entirely with In this,study the aspects of size studied were
private ownership. When the government orga- number of employees, net assets utilized, and
nizations were extracted from the sample, the number of employees in the parent organiza-
correlation disappeared(r =-0.08 for N = 34). tions.
No relationshipswere found between the struc- Number of employees and net assets. It was
ture of an organization and the ownership pat- intended that the sample be taken from the
tern of its parent organization. This lack of population of work organizations in the region
relationship is quite striking, particularly in employing more than 25-0people, but the sam-
view of the extent of the correlation found with ple ranges from an insurance company employ-
other contextual variables. Since ownership ing 241 people to a vehicle manufacturingcom-
and control seemed to have its impact through pany employing 25,052 (mean 3,370; standard
the degree of public accountability, and the deviation 5,313). In view of this distribution,
other variables did not have an additional ef- it was felt that a better estimate of the correla-
98 ADMINISTRATIVE SCIENCE QUARTERLY

tion between size and other variables would be the factor may obscure particular relationships
obtained by taking the logarithm of the num- with the source variables which it summarizes.
ber of employees (mean 3.12; standard devia- For some purposes therefore, it may be inter-
tion 0.57). esting to examine particular relationships. The
"Net assets employed by the organization" lack of relationship between size and concen-
was also used, because financial size might ex- tration of authority, for example, summarizes
pose some interesting relationships with orga- (and therefore conceals) two small but distinct
nization structure that would not appear when relationships with two of the component vari-
only personnel size was considered. The sample ables. There is no relationship between size
ranged from under ?100,000-an estimate for and autonomy (r= 0.09), but there is a nega-
the government inspection agency whose equip- tive relationshipbetween size and centralization
ment was provided by its clients-to a confec- (r = -0.39), and a positive one between size
tionery manufacturing firm with ?38 million. and standardization of procedures for selection
The attempt to differentiate between these two and advancement (r = 0.31). The relationship
aspects of size proved unsuccessful, however, as with centralization has clear implications for
the high correlation between them (r = 0.78) the concept of bureaucracy. Centralizationcor-
shows. Taking the logarithm of the two vari- relates negatively with all scales of structuring
ables raised the correlation (r = 0.81). For of activities except one: the more specialized,
this sample, therefore, a large organizationwas standardized, and formalized the organization,
big both in number of employees and in finan- the less it is centralized. Therefore on the basis
cial assets. The logarithm of the number of of these scales, there can be no unitary bureauc-
employees was therefore taken to represent racy, for an organization that develops spe-
both these aspects of size. cialist offices and associated routines is decen-
The correlationbetween the logarithm of size tralized. Perhaps when the responsibilities of
and structuring of activities (r = 0.69) lends specialized roles are narrowly defined, and
strong support to descriptive studies of the ef- activities are regulated by standardized proce-
fects of size on bureaucratization. (This corre- dures and are formalized in records, then au-
lation may be compared with that between thority can safely be decentralized. Pugh, et al.
actual size and structuringof activities, r = 0.56, (1969) discuss the interrelationshipof the struc-
to demonstrate the effects of the logarithmic tural variables in particular types of organiza-
transformation.) Larger organizations tend to tion.
have more specialization, more standardization Size of parent organization. This is the num-
and more formalization than smaller organiza- ber of employees of any larger organization to
tions. The lack of relationship between size which the unit belongs. The literature on
and the remaining structural dimensions, i.e., bureaucracy often implies that it is the size of
concentrationof authority (r = -0.10) and line the larger parent organization that influences
control of workflow (r = -0.15) was equally the structure of the sub-unit. The important
striking. This clear differential relationship of factor about a small government agency may
organization size to the various structural di- not be its own size, but that of the large minis-
mensions underlines the necessity of a multi- try of state of which it is a part. Similarly, the
variate approach to context and structure if structureof a subsidiary company may be more
oversimplificationsare to be avoided. related to the size of its holding company. The
Indeed, closer examination of the relation- number of employees in the parent organiza-
ship of size to the main structural variables tions ranged from 460 to 358,000 employees.
underlying the dimension of concentration of The size of the parent organization correlated
authority (Pugh, et al., 1968: Table 4) points positively (after logarithmic transformation)
up a limitation in the present approach, which with structuring (r= 0.39) and concentration
seeks to establish basic dimensions by means of of authority (r = 0.39) but not with line con-
factor analysis. As was explained in that paper, trol of workflow (r = -0.07). The classical
the structural factors represent an attempt to concept of bureaucracy would lead to the hy-
summarize a large amount of data on a large pothesis that the size of the parent organization
number of variables to make possible empiri- would be highly correlated with structuring of
cally based comparisons. But the cost is that activities and concentrationof authority, there-
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 99

TABLE 6. CHARTER.

Distribution
N = 46 Score Scale number and title
Scale No. 14.02
Multiplicity of outputs
19 1 Single output with standard variations
8 2 Single output with variations to customer specification
19 3 Two or more outputs
Scale No. 14.03
Type of output
14 1 Service (nonmanufacturing)
32 2 Manufacturing (new physical outputs in solid, liquid
or gaseous form)
Scale No. 14.04
Type of output
16 1 Consumer (outputs disposed of to the general public
or individuals)
7 2 Consumer and producer
23 3 Producer (outputs disposed of to other organizations
which use them for, or as part of, other outputs)
Scale No. 14.06
Customer orientation
11 1 Standard output(s)
7 2 Standard output(s) with standard modifications
6 3 Standard output(s) with modification to customer
specification
22 4 Output to customer specification
Scale No. 14.07
Self-image
24 1 Image emphasizes qualities of the organization itself
6 2 Image emphasizes both the organization and the
output
16 3 Image emphasizes qualities of the output of the
organization
Scale No. 14.08
Policy on outputs multiplicity
5 1 Contracting the range of outputs
26 2 Maintaining the range
15 3 Expanding the range
Scale No. 14.09
Ideology: client selection
28 1 No selection, any clients supplied
14 2 Some selection of clients
4 3 Clients specified by parent organization

fore the support from this sample was relatively Charter


modest. The correlation with structuring (r = Scales. Institutional analysts have demon-
0.39) is much lower than the correlation of strated the importance of the charter of an
organization size and structuring (r = 0.69). organization; that is, its social function, goals,
The impact of the size of an organization is ideology, and value systems, in influencing
thus considerably greater than the size of the structure and functioning (Parsons, 1965;
parent organization on specialization, standard- Selznick, 1949). To transform concepts.which
ization, formalization, etc. But a relationship had been treated only descriptively into a quan-
with concentration of authority is not found titative form that would make them comparable
with organization size (r = -0.10). Thus large to other contextual aspects, seven ordered cate-
groups have a small but definite tendency to gory scales were devised. Four of them char-
have more centralized subunits with less auton- acterized the purpose or goal of the organiza-
omy. This relationship would be partly due to tion in terms of its "output," the term being
the government-ownedorganizations,inevitably taken as equally applicable to products or ser-
part of large groups, which were at the concen- vices: (i) multiplicity of outputs-ranging from
trated end of this factor. a single standard output to two or more out-
100 ADMINISTRATIVE SCIENCE QUARTERLY

puts; (ii) type of output-a manufacturing- Operating variability. This factor, account-
service dichotomy, (iii) consumer or producer ing for 30 percent of the variance was highly
outputs or a mixture of both, and (iv) customer loaded on the variables.,consumer or producer
orientation of outputs-ranging from com- outputs, customer orientation of outputs, and
pletely standard outputs to outputs designed type of output. It was therefore conceptualized
entirely to customer or client specification. as being concerned with manufacturing non-
Three scales were devised for ideological aspects standard producer goods as against providing
of charter: (v) self-image-whether the ideology standard consumer service. The manufacturing
of the organizationas indicated by slogans used producer end of the scale was linked with an
and image sought emphasized the qualities of organizational emphasis on self-image, whereas
its outputs; (vi) policy on multiple outputs- the consumer service end emphasized outputs.
whether the policy was to expand, maintain, or The scale was therefore constructed by a
contract its range of outputs; and (vii) client weighted summing of the scores on all these
selection-whether any, some, or no selectivity variables (the weighting being necessary to
was shown in the range of customers or clients equate the standard deviations) and then stan-
served by the organization. Table 6 gives the dardizing the sums to a mean of 50 and a stan-
details of the seven scales and Table 7 the inter- dard deviation of 15. This produced the range

TABLE 7. INTERCORRELATION MATRIX (PRODUCT-MOMENT COEFFICIENTS, N = 46).

Multi- Cus-
plicity Service- tomer Expansion-
of manufac- Consumer- orien- Self- Client contraction
outputs turing producer tation image selection of range

Multiplicity of outputs
Service-manufacturing 0.15
Consumer-producer 0.05 0.37 -
Customer orientation 0.38 0.18 0.59
Self -image -0.05 -0.17 -0.33 -0.13 -
Client selection -0.14 -0.02 0.28 -0.04 -0.18
Expansion-Contraction of range 0.07 -0.14 -0.09 0.07 0.10 -0.09

correlation matrix between them. This sug- of scores, on the scale given in Table 9. Th'e
gested that factor analysis would be helpful in lower scores distinguished organizations giving
summarizing the data, and a principal compo- only a standard service (e.g., teaching, trans-
nents analysis applied to the matrix gave the port, retailing), from organizations (with high
results shown in Table 8. scores) producing nonstandard producer out-
puts to customer specification (metal goods
firm, engineering repair unit, packaging manu-
TABLE 8. CHARTER: PRINCIPAL-COMPONENTS
facturer, etc.), with those organizations having
ANALYSIS.
a standard output range in the middle.
Factor loadings Operating diversity. This factor of charter,
accounting for 20 percent of the variance, em-
Operating Operating
Scales variability* diversityt phasized multiplicity of outputs, policy on
whether to expand the range of kinds of out-
Consumer-producer output 0.85 0.16 puts, client selection, and self-image. The more
Customer orientation
of outputs 0.74 -0.41 diversely operating organizations were a glass
Type of output (service- manufacturer, a metal manufacturer, and a
manufacturing) 0.5 7 0.00 brewery; the more restricted were a motor
Self-image -0.52 -0.34
Multiplicity of outputs 0.37 -0.66 component manufacturer,a domestic appliance
Client selection 0.23 0.66 manufacturer,and a scientific inspection agency.
Expansion-contraction
of range -0.15 -0.48 Eisenstadt (1959), Parsons (1956), Selznick
* Percentage of variance = 30%
(1949, 1957), Wilson (1962), and Clark (1956)
t Percentage of variance = 20% have discussed the effects of the goals of an
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 101

TABLE 9. OPERATING VARIABILITY.

Number of
organizations
N =46 Score Type of organization

1 48 Component manufacturer
6 45 Two metal goods manufacturers
Component manufacturer
Abrasives manufacturer
Packaging manufacturer
Glassmanufacturer
2 43 Printer
Repairs for government department
4 42 Two component manufacturers
Motor component manufacturer
Metal motor component manufacturer
2 41 Vehicle manufacturer
Engineering tool manufacturer
1 40 Component manufacturer
3 37 Civil engineering firm
Carriage manufacturer
Metal goods manufacturer
3 36 Vehicle manufacturer
Confectionery manufacturer
Local authority water department
2 35 Motor-tire manufacturer
Commercial vehicle manufacturer
4 34 Motor component manufacturer
Non ferrous metal manufacturer
Research division
Food manufacturer
3 33 Engineering component manufacturer
Domestic appliances manufacturer
Local authoritycivil engineeringdepartment
1 32 Component manufacturer
2 31 Government inspection department
Toy manufacturer
3 30 Brewery
Insurance company
Food manufacturer
1 27 Local authority transport department
6 25 Local authority baths department
Co-operative chain of retail stores
Chain of retail stores
Savings bank
Chain of shoe repair stores
Department store
1 23 Omnibus company
1 21 Local authority education department

organization on its structure, but there has cruitment and means of selection. Clark (1956)
been almost no detailed empirical work on the as well as Thompson and Bates (1957) empha-
actual relationshipbetween goals and structure. sized both the marginality and the degree of
Selznick (1949) showed how the goal of democ- concretenessof the goal as a determinantof the
racy led to decentralization in the TVA, and direction of organizational adaptation. Blau
also suggested that the role structure of an and Scott (1962) made one of the few attempts
organization is the institutional embodiment to classify organizationsby their goals, suggest-
of its purpose. Wilson (1962) suggested a re- ing that internal democracy goes with mutual
lationship between goals and methods of re- benefit goals, efficiency with business goals, a
102 ADMINISTRATIVE SCIENCE QUARTERLY

professional structure with service goals and of the organization, even if the physical tech-
bureaucratic structure with commonweal goals. niques involve only pen, ink, and paper. The
Scales of organizational charter were related concept covers both the pattern of operations
to structure, and operating variability was and the equipment used, and all the scales de-
shown to be strongly associated with line con- veloped are applicable to service as well as to
trol of workflow (r = -0.57). Thus the more an manufacturing organizations. Five scales of
organization is concerned with manufacturing related aspects of technology were developed.
nonstandard producer goods, the more it re- Thompson and Bates (1957) defined the
lies upon impersonal control of workflow; the "adaptability" of the technology as "the extent
more it is providing a standard consumer ser- to which the appropriatemechanics, knowledge,
vice, the more it uses line control of its work- skills and raw materials can be used for other
flow through the supervisory hierarchy. Orga- products" and, it may be added, services. An
nizations showing operating diversity, however, attempt to operationalize some aspects of this
tended to be more structured in activities definition is given in Table 10, which shows a
(r = 0.26) and more dispersed in authority scale of workflow rigidity. This consists of
(r = -0.30). eight biserial items concerned with the adapt-
ability in the patterns of operations; for exam-
Technology ple, whether the equipment was predominantly
Scales. Technology has come to be consid- multi-purpose or single-purpose, whether re-
ered increasingly important as a determinantof routing of work was possible, etc. Since this
organizational structure and functioning, al- was a scale of composite items, item analysis
though comparative empirical studies of its was used to test the scaleability. The mean
effects on structureare few, mainly case studies item-analysis value of 0.84 indicates that it is
on the effects on the operator's job and atti- legitimate to add the scores on these items to
tudes (Walker, 1962). Thompson and Bates form a workflow rigidity score for an organiza-
(1957), however, compared a hospital, a uni- tion.
versity, a manufacturing organization, and a Two other scales of technology utilized the
mine for the effects of their technologies on concepts outlined by Amber and Amber (196.2).
the setting of objectives, the managementof re- They postulated that "the more human attri-
sources, and the execution of policy. The main butes performed by a machine, the higher its
work on the classification of technology in re- automaticity" and compiled a scale of automa-
lation to organization structure has been that ticity together with clear operational defini-
of Woodward (1965). She related mainly "con- tions, which could be applied to any piece of
figuration" aspects of the structure of manu- equipment from a pencil to a computer, and
facturing organization (e.g., number of levels of which categorized each into one of six classes.
authority, width of spans of control) to a classi- The two scales based on these concepts were:
fication of their production systems according the automaticity mode, i.e., the level of auto-
to the "controlability and predictability" of maticity of the bulk of the equipment of the
the process. organization; and the automaticity range; i.e.,
In the present study the need to develop the highest-scoringpiece of equipment an orga-
suitable measurementsof overall organizational nization used, since every organization also
technology made the level of generality achieved scored the lowest possible by using hand tools
by the Woodward classification desirable; but and manual machines.
the need to develop concepts of technology that The fourth scale, interdependence of work-
applied to all the organizations in the sample flow segments, was a scale of the degree of link-
precluded the direct adoption of that scale. A age between the segments of an organization;
full account of the development of scales of a segment being defined as those parts into
technology and their relationship to organiza- which the workflow hierarchy was divided at
tion structure is given in Hickson, et al. (1969). the first point of division beneath the chief
Only the scales included in the present analysis executive. The three points on the scale were:
are describedhere. (i) segments duplicated in different locations,
Technology is here defined as the sequence all having the same final outputs; (ii) segments
of physical techniques used upon the workflow having different final outputs, which are not
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 103

TABLE 10. SCALE OF WORKFLOW RIGIDITY.

Number of Item
organizations analysis
Item (N = 52)+ value*
No waiting time possible
(versus waiting time) 8 0.82
Single-purpose equipment
(versus multi-purpose) 13 0.78
Production or service line
(versus no set line) 42 1.00
No buffer stocks and no delays possible
(versus buffer stocks and delays) 8 0.71
Single-source input
(versus multisource input) 12 0.67
No rerouting of work possible
(versus rerouting possible) 15 0.80
Breakdown stops all workflow immediately
(versus not all workflow stops) 6 0.97
Breakdown stops some or all workflow immediately
(versus no workflow stops) 35 0.95
* Mean item analysis value= 0.84.
+ Since this is a test of internal consistency and scaleability, the whole group of 52 organizations was used.
(D. S. Pugh et al., 1968.)

inputs of other segments; (iii) segments having with public accountability (r =-0.35), largely
outputs which become inputs of other segments. because the government-ownedorganizationsin
The final scale, specificity of criteria of quality the sample were predominantly service and
evaluation, was a first attempt to classify the therefore at the diverse end of the workflow
precision with which the output was compared integration scale. The correlations between
to an acceptable standard. The three points workflow integration and operating variability
on the scale were: (i) personal evaluation only; (r = 0.57) and diversity (r = 0.33) reflect the
(ii) partial measurements,of some aspect(s) of close relationship between the ends of the orga-
the output(s); (iii) measurements used over nization and the means it employs to attain
virtually the whole output, to compare against them.
precise specification (the "blueprint"concept). Workflow integration showed modest but
Correlations. As expected, these measures distinct correlations with all the three struc-
tend to be highly intercorrelated. A principal tural factors, the only contextual variable to do
components analysis extracted a large first so, as can be seen from Table 2. The relation-
factor accounting for 58 percent of the vari- ships of technology are therefore much more
ance, with loadings of over 0.6 on all scales, general than is the case with size, for example,
and of over 0.8 on three of them. A scale of which has a greater but more specific effect.
workflow integration was therefore constructed The positive correlation between workflow in-
by summing the scores on the componentscales. tegration and structuringof activities (r = 0.34)
Among organizations scoring high, with very would be expected since highly integrated and
integrated, automated, and rather rigid tech- thereforemore rigid technologieswould be asso-
nologies, were an automobile factory, a food ciated with a greater structuring of activities
manufacturer, and a swimming baths depart- and procedures. Similarly, the correlationwith
ment. Among those scoring low, with diverse, concentration of authority (r = -0.30) sug-
nonautomated, flexible technologies, were retail gests that because of the increasing control re-
stores, an education department,and a building sulting directly from the workflow itself in an
firm. integrated technology, decisions tend to become
There were no clear relationships between more routine and can be decentralized. But the
workflow integration and the variables of size, fact that the correlations are not higher than
origin and history, or concentration of owner- this emphasizes that structuringmay be related
ship with control and negative relationship to other contextual factors, such as size. The
104 ADMINISTRATIVE SCIENCE QUARTERLY

relationship of technology to line control of scale of technology (r= -0.58), and with public
workflow, however, was very clear (r = 0.46); accountability (r = 0.34), this last correlation
the more integrated the technology, the more reflecting the predominantlyservice function of
the reliance on impersonal control. It must be the group of government-ownedorganizations.
emphasized, however, that these relationships This pattern of inter-relationshipsamong the
were found on the whole sample of manufac- contextual variables led to the expectation of
turing and service organizations. When manu- relationships between number of operating sites
facturing organizations only were considered, and the structural dimensions which would be
some of the relationships showed considerable congruent with those of operating variability
change (Hickson et al., 1969). and workflow integration. The correlations of
Labor costs. This is a second related, but number of operating sites with structuring of
conceptually distinct, aspect of the technology activities (r = -0.26), concentration of author-
of the workflow and is expressed as a percent- ity (r = 0.39) and line control of workflow
age of total costs. The range in the sample was (r = 0.39) confirm the relationshipswith charter
from 5 to 70 percent, with engineering organi- and technology, and suggest a charter-technol-
zations scoring low and public services high. ogy-location nexus of interrelated contextual
The scale correlated with workflow integration variables having a combined effect on structure.
(r =-0.50), high integration being associated
with reduced labor costs. Its correlations with Dependence
the structural factors are comparablewith those The dependence of an organization reflects
for technology (after adjusting the signs). its relationships with other organizations in its
social environment, such as suppliers, custom-
Location -ers, competitors, labor unions, management
The geographical, cultural, and community organizations,and political and social organiza-
setting can influence the organization markedly tions.
(Blau and Scott, 1962). This study controls Dependence on parent organization. The
for some of these effects in a gross way, for all most important relationship would be the de-
organizationsof the sample were located in the pendence of the organizationon its parent orga-
same large industrial conurbation,and the com- nization. The relative size of the organization
munity and its influence on the organizations in relation to the parent organizationwas calcu-
located there were taken as given (Duncan lated as a percentage of the number of em-
et al., 1963). Compared with the national ployees. This ranged from under one percent
distribution, the sample was overrepresentedin in two cases-a branch factory of the central
the engineering and metal industries, and un- government, and a small subsidiary company
representedin mining, shipbuilding, oil refining of one of the largest British private corpora-
and other industries. Because of the location, tions in the country-to 100 percent in eight
however, regional cultural differences of the independent organizations. The distribution
sort found by Thomas (1959) as to role con- was Poisson in form with a mean and standard
ceptions, were avoided. deviation of 37 percent. The next scale was a
One aspect of location which discriminated four-point category scale concerned with the
between organizations in the sample, was num- status of the organization in relation to the
ber of operating sites. The range formed a parent organization: (i) principal units (8 orga-
Poisson distribution, with 47 percent of the nizations) where the organization was inde-
sample having one site; but six organizations pendent of any larger group although it might
had over a hundred sites, and two over a thou- itself have had subsidiaries or branches; (ii)
sand. This distribution did not appear to be a subsidiary units (18 organizations) which, al-
function of size (r = 0.14) but of the operating though part of a larger group, had their own
variability aspect of charter(r = -0.56). Manu- legal identity with, for example, their own
facturing organizations were concentrated in boards of directors; (iii) head branch units (4
a small number of sites (the largest number organizations)which did not have separate legal
being nine), whereas services range across the identity although they were the major operating
scale. The number of operating sites was,there- components of the parent organization and the
fore correlated with the workflow integration head office of the parent organization was on
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 105

TABLE 11. DEPENDENCE.

Distribution
N = 46 Score Scale number and title
Scale No. 18.07
Relative Size*
Range = 0-100 Mean - 37.4 S.D. - 37.3
Scale No. 12.10
Status of Organization Unit
16 1 Branch
4 2 Head branch (headquarters on same location)
18 3 Subsidiary (legal identity)
8 4 Principal unit
Scale No. 12.11
Organizational representation on policy-making bodiest
19 1 Organization not represented on top policy-making body
Organization represented on local policy-making body
4 2 but not on top policy-making body
23 3 Organization represented on policy-making body
Scale No. 18.06
Number of specializations contracted out+
Range = 1-16 Mean 7.2 S.D. 4.0
Scale No. 18.17
Vertical integrations
Range = 1-16 Mean 7.7 S.D. 3.5
Scale No. 18.03
Integration with suppliers
4 1 No ownership ties and single orders
7 2 No ownership and single contracts or tenders
8 3 No ownership and short-term contracts, schedule and call-off
6 4 No ownership and yearly contracts, standing orders
7 5 Ownership and contractual ties
14 6 Ownership and tied supply
Scale No. 18.05
Response in outputs volume to customer influence
12 1 Outputs for stock
5 2 Outputs for stock and to customer order
21 3 Outputs to customer order
2 4 Outputs to customer order and to schedule and call-off
6 5 Outputs to schedule and call-off
Scale No. 18.08
Integration with customers: type of link with customers
24 1 Single orders
9 2 Regular contracts
10 3 Long-term contracts (over two years)
3 4 Ownership
Scale No. 18.09
Integration with customers: dependence of organization
on its largest customer
30 1 Minor outlet (less than 10% of output)
10 2 Medium outlet (over 10% of output)
3 3 Major outlet (over 50% of output)
3 4 Sole outlet
Scale No. 18.10
Integration with customers: dependence of largest customer
on organization
11 1 Minor supplier (less than 10% of particular item)
5 2 Medium supplier (over 10% of particular item)
21 3 Major supplier (over 50% of particular item)
9 4 Sole supplier with exclusive franchise
* Size of unit as a percentage of size of parent organization.
t Internal and parent organizations.
$ The specializations are those of functional specialization (D. S. Pugh et al., 1968: Appendix A). Scores
are out of a possible 16.
? This scale is formed by the total of the scores on the 18 items representing the following five scales: 18.03,
18.05, 18.08, 18.09, 18.10.
106 ADMINISTRATIVE SCIENCE QUARTERLY

the same site; (iv) branch units (16 organiza- Dependence on other organizations. The sup-
tions) operating parts of a parent organization pliers and customers or clients of the organiza-
which did not satisfy the preceding criteria. tion must also be considered. The operating
The third aspect of the relation between the function of the organizationcan be regarded as
organization and the parent organization was being the processing of inputs, and outputs be-
given by the degree of organizationalrepresen- tween supplier and client, and the degree to
tcation on policy-making bodies. This three- which the organization is integrated into the
point scale ranged from the organization being processual chain by links at either end can be
represented on the policy-making body of the measured. Five category scales were developed
parent organization (e.g. board of directors, to elucidate this concept (with details given in
city council), through the organization being Table 11). They were concernedwith the inte-
represented on an intermediate policy-making gration with suppliers and clients, and response
body (e.g. board of directors of an operating in the output volume to client influence, etc.
company but not of the ultimate owning hold- To establish a single dimension measuring the
ing company, committee of the city council), degree to which the organizationwas integrated
to the organization having no representativeon into this system, the five scales were trans-
any policy-making body of the parent organiza- formed into biserial form. Item analysis was
tion. As would be expected, these three vari- carried out on the 18-item scale generated and
ables were highly correlated (Tables 11, 12). yielded a mean item analysis value of 0.70,

TABLE 12. DEPENDENCE: INTERCORRELATION MATRIX (PRODUCT-MOMENT


COEFFICIENT N - 46).

Special-
Status of Organizational izations
Relative organiza- representationon contracted Vertical Trade
size tion unit policy-makingbody out integration unions
Relative size
Status of
organizationunit 0.68
Organizationalrepresentation
on policy-makingbody 0.50 0.65
Specializations
contractedout -0.60 -0.51 -0.52 -
Vertical
integration -0.40 -0.34 -0.36 0.45 -

Trade unions -0.09 -0.16 -0.25 0.19 0.28

A related variable was the number of spe- which seemed to justify the addition of the
cializations contracted out by the organization. items into a total scale, vertical integration. At
In many cases these would be available as ser- one extreme was a confectionery manufacturer
vices of the parent organizationto the organiza- and an engineering components firm supplying
tion, although account was also taken of the goods from stock with a large number of cus-
various specialist services (e.g., consultants) tomers after obtaining their supplies from a
used outside the parent organization. The spe- large variety of sources; at the other extreme
cializations were as defined in the structural were organizations (vehicle components, civil
scale of functional specialization (Pugh et al., engineering, scientific research) obtaining their
1968: Appendix A), and ranged from one spe- resources from a small number of suppliers and
cialization contracted out (two engineering supplying their product or service to a small
works, a printer, and a builder) to no less than number of clients (often the owning group
fifteen of the sixteen specializations contracted only) who had a marked effect upon their work-
out (an abrasives manufacturerand a packaging flow scheduling.
manufacturer) with a mean 7.2 and standard For trade unions, a scale of five orderedcate-
deviation 4.0. gories was developed of the extent to which
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 107

unions were accepted as relevant to the activi- applied to the parent organization) than with
ties of the organization. The scale was (i) no operating variability or workflow integration
recognition given; ('ii) only partial recognition (variables applied to the operations of the in-
given (i.e., discussions for certain purposes, but dividual organizations themselves). This pat-
not negotiations); (iii) full recognition given tern lends support to the view that these mea-
to negotiate on wages and conditions of service sures are tapping aspects of the dependence of
on behalf of their members; (iv) full recogni- the organization,particularly its dependence on
tion given plus facilities for union meetings to external resources and power as in Eisenstadt's
be held regularly on the time and premises of (1959) formulation. The one exception was the
the organization; (v) as in the preceding plus variable of recognition of trade unions, which
the recognition of a works convenor to act on had its largest contextual correlationwith orga-
behalf of all unions with members in the orga- nization size, and is therefore concerned with a
nization. Organizations in the sample were different aspect of interdependence. Imperson-
located in all the categories, with the modal ality of origin (from origin and history) and
position being full recognition; but five orga- public accountability (from ownershipand con-
nizations did not recognize unions, and eleven trol) show the same pattern of higher correla-
gave the maximum recognition including a tion with the parent organization than with the
works convenor. unit, indicating that impersonally founded
Examination of the intercorrelationsbetween organizations are likely to be more dependent
these six variables of dependence (Table 12) on their founding organizations; and that more
and of their correlations with other important publicly accountable organizations are more
aspects of context (Table 13) shows consider- likely to be dependent on outside power with
ably higher correlations with size of parent government-ownedorganizations being the ex-
organization than with size of organization, treme case.
and considerably higher correlations with con- These relationships suggested the applica-
centration of ownershipwith control (a variable tion of factor analysis to a correlation matrix

TABLE 13. DEPENDENCE.

Size of Line
Concentration parent Workflow Size of Structur- Concentra- control
of ownership organi- Operating integra- organi- ing of tion of of
with control Size zation variability tion zation activities authority workflow

Status of
organization
unit 0.45 0.11 -0.27 0.01 0.05 0.17 0.13 -0.63 -0.07
Organizational
representation
on policy-
making bodies 0.41 0.15 -0.19 -0.01 0.19 0.20 0.14 -0.63 -0.18
Number of
specializations
contracted out -0.32 0.14 0.40 0.11 0.09 0.01 0.18 0.53 0.00
Relative Size 0.47 0.06 -0.38 -0.08 -0.03 0.16 0.03 -0.40 -0.13
Vertical
integration -0.15 -0.01 0.39 0.21 -0.12 -0.06 0.06 0.29 -0.04
Trade unions* -0.21 0.26 0.25 0.19 0.21 0.36 0.51 0.08 -0.35
Impersonality
of origin -0.40 0.13 0.36 -0.27 -0.25 0.07 -0.04 0.64 0.36
Public
accountability
of parent
organization -0.51 0.04 0.45 -0.35 -0.35 0.01 -0.10 0.64 0.47
Dependence -0.49 -0.06 0.37 0.05 -0.05 -0.17 -0.05 0.66 0.13
* This variable was not included in the scale of dependence.
108 ADMINISTRATIVE SCIENCE QUARTERLY

containing the seven variables. A principal- the level in the organization which has the
components analysis applied to the matrix pro- necessary authority to take particular decisions,
duced a large first factor dependence2account- (Pugh et al., 1968: 76); the higher the neces-
ing for 55 percent of the variance, which was sary level, the greater the centralization. No
heavily loaded on all seven scales (on six of the account was taken of the degree of participa-
seven, the loadings were above 0.7; the remain- tion or consultation in decision-making as in
ing loading on vertical integration was 0.58). Hage and Aiken's (1967) formulation of the
The scores for dependencewere obtained by an concept. These were regarded as aspects for
algebraic weighted sum of the scores on the study at the group level of analysis. Neither
four most highly loaded component scales, the is it possible for such a statement as the fol-
weightings being obtained by a multiple re- lowing to hold: "The decisions were centralized
gression analysis of the component scales on on the foreman since neither the superintendent
the factor. A high score characterizedorganiza- nor the departmental manager had the neces-
tions with a high degree of dependence, which sary experience." In the present formulation
tended to be impersonally founded, publicly this would be regarded as relative decentraliza-
accountable, vertically integrated, with a large tion. Autonomy was measured by the propor-
number of specializations contracted out, small tion of decisions that could be taken within the
in size relative to their parent organization, low organization as distinct from those which had
in status, and not represented at the policy- to be taken at the level above it. Thus inde-
making level in the parent organization (e.g., pendent organizations of necessity had more
branch units in packaging, civil engineering, autonomy, since there was no level above the
and food manufacture, a central government chief executive, and the correlation between
repair department, and a local government dependence and this component of concentra-
baths department). Organizationswith low de- tion of authority was r -0.72. The relation
pendence were independent organizations char- of centralization (which is concerned with the
acterized by personal foundation, low public whole range of levels in the hierarchy) with
accountability, little vertical integration, few dependence is less, but still high (r= 0.57).
specializations contracted out, and where the Dependent organizations also have a distinct
parent organization was the organization itself tendency to standardize the procedures for
(e.g., a printing firm, the very old metal goods selection and advancement (r = 0.40), a major
firm, a chain of shoe repair stores, and an engi- component of concentration of authority. So
neering componentmanufacturer). dependent units have the apparatus of recruit-
The correlationof dependencewith the struc- ment routines, selection panels, formal estab-
tural factors was focused largely on concentra- lishment figures, etc. of their parent organiza-
tion of authority (r = 0.66), in every case, for tions.
dependence and its component scales the cor-
Relation between Structure and Context
relation being much greater than with the other
factors, as Table 13 shows. Indeed, apart from In this investigation of the relationship of
the correlations with impersonality of origin organization structure to aspects of the context
and public accountability, none of the other in which the organization functions, the use of
correlations reached the 5 percent level of con- scaling and factor analytic techniques has made
fidence. Dependent organizations have-a more possible the condensation of data and reorga-
centralized authority structure and less auton- nization of concepts and has established eight
omy in decision making; independent organiza- distinctive scales of elements of context. These
tions have more autonomy and decentralize de- scales, shown in Table 14 together with their
cisions down the hierarchy. correlations with the structural dimensions, de-
The relationships between dependence and note the variables that are salient among those
the component scales of concentration of au- which have been thought to affect structure.
thority vary. Centralization, as defined and Relationships between structure and age, size,
measured in this study, is concerned only with charter (operating variability, operating diver-
2
sity), technology (workflow integration) loca-
We are grateful to our colleague Diana C. Pheysey
for suggesting this formulation and for much valuable
tion (numberof operating sites) and dependence
critical comment on an earlier draft of this paper. on other organizationsare exposed by the corre-
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 109

TABLE 14. SALIENT ELEMENTS OF CONTEXT (PRODUCT-MOMENT CORRELATIONS


WITH STRUCTURAL FACTORS).*

Structuring Concentration Line control


Elements of context of activities of authority of workflow
Age -0.38
Size of organizationt 0.69
Size of parent organizationt 0.39 0.39
Operating variability - o0.57
Operating diversity -0.30
Workflow integration 0.34 -0.30 -0.46
Number of operating sites - 0.39 0.39
Dependence 0.66
* With N - 46, correlations of 0.29 are at the 5% level of confidence, and correlations of 0.38 are at the 1%
level of confidence.
t Logarithm of number of employees.

lations. At the same time the correlationsraise contribution to the prediction than workflow
questions about the relationshipbetween owner- integration, since it has a strong correlation
ship pattern and administrative structure. with the first predictor (r = 0.43); whereas the
technology measure is not correlated with
THE MULTIVARIATE PREDICTION organization size (r -0.08). This is in fact
OF STRUCTURE FROM CONTEXT the case as shown in the first section of Table
From inspection of Table 14 and of the inter- 15, which gives the multiple prediction analyses
correlation matrix in Table 3, certain elements for the three structural factors.
of context can now be identified. The variables Table 15 shows for each predictor variable,
in Table 14 are now used as independent vari- the single correlation with the criterion, the
ables in a prediction analysis of the structural multiple correlation obtained by adding this
dimensions. The pattern of these correlations, predictor to the preceding ones, the F ratio
that is, that where they are high they are corresponding to the increase obtained on the
specific, and where they are low, they are dif- addition of this predictor, the degrees of free-
fused, indicates that the predictions should be dom correspondingto the F ratio when N = 46,
attempted on a multivariate basis. In this case and the level of confidence at which the in-
consideration had to be given to choosing not crease due to this predictor can be quoted. It
only predictors with high correlations with the will be seen from the first section of Table 15
criterion, but also having low intercorrelations that the correlation 0.69, between size and
among themselves. If high intercorrelations structuringof activities, is increased to a multi-
among the predictors were allowed, then, since ple correlation of 0.75 when workflow integra-
the high correlations with the criterion would tion is added as a predictor. But the multiple
be aspects of the same relationship,the multiple correlation shows no noticeable increase when
correlation would not be increased to any ex- size of parent organization is added as a third
tent. If the intercorrelationsbetween the pre- predictor; that is, its predictive power has al-
dictors were low, then each would make its ready been tapped by the two previous vari-
distinct contribution to the multiple correlation. ables.
These problems can be illustrated from the It must be emphasized that this procedure
attempt to obtain a multiple prediction of struc- assesses only the predictive power of the con-
turing of activities from the three contextual textual variables, not their relative importance
variables correlatedwith it (Table 14). Size is in any more general sense. It cannot be con-
clearly the first predictor, with a correlation of cluded that the relationship of size of parent
r 0.69, and the question is whether taking organization to structuring of activities is less
account of size of parent organization and importantthan that of workflowintegration,be-
workflow integration will increase predictive cause it adds less to the multiple correlation. In-
accuracy. In spite of its greater correlation deed the original higher correlation shows that
with the criterion, the size of the parent orga- this is not the case. Because of the interaction of
nization would be expected to make a smaller the variables, the effects of organizational size
110 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 15. MULTIPLE PREDICTION ANALYSIS OF STRUCTURAL FACTORS.

Contextual predictors Single Multiple Degrees of Level of


of structural factors correlation correlation F ratio freedom confidence

Structuring of activities
Size 0.69 0.69 39.6 1: 44 > 99%
Workflow integration 0.34 0.75 8.2 1: 43 > 99%
Size of parent organization 0.39 0.76 1.9 1: 42 NS
Concentration of authority
Dependence 0.66 0.66 34.2 1: 44 > 99%
Location (number of
operating sites) 0.39 0.75 12.5 1: 43 > 99%
Age of organization -0.38 0.77 2.5 1: 42 NS
Operating diversity -0.30 0.78 3.0 1: 41 NS
Workflow integration -0.30 0.78 0.0 1: 40 NS
Size of parent organization 0.39 0.79 0.4 1: 39 NS
Line control of workflow
Operating variability -0.57 0.57 20.7 1: 44 > 99%
Workflow integration -0.46 0.59 1.7 1: 43 NS
Number of sites 0.39 0.59 0.1 1: 42 NS

and size of parent organization are confounded, single correlation of operating variability with
in this study, as the correlation between them the criterion.
shows. A full examination of their relative ef- The size of the multiple correlationsobtained
fects would require a sample in which they with the first two factors, each 0.75 with two
were not correlated, as is the case with the predictors, together with the small number of
technology measure. predictors needed, strongly supports the view
The same argument applies to the multiple that in relation to organization structure as de-
prediction of concentrationof authority (Table fined and measured in this study, salient ele-
15). Here again there is a clear first predictor, ments of context have been identified. Thus a
dependence,with a correlationof 0.66 but then knowledge of the score of an organization on a
a choice of intercorrelatedvariables. The selec- small number of contextual variables makes it
tion was made in order to get as high a multiple possible to predict within relatively close limits,
correlation as possible with as few predictors its structural profile. Given information about
as possible, but the fact that the later predictors how many employees an organization has, and
add nothing to the multiple correlationdoes not an outline of its technology in terms of how
mean that they have no impact, only that pre- integrated and automated the work process is,
dictive power has been exhausted by previous its structuring of activities can be estimated
related variables. The existence of the charter- within fairly close limits. Since in turn the
technology-location nexus referred to above is score of the organization on structuring of
supported by the fact that when any one of activities summarizes an extensive description
these variables is used as a predictor, the re- of broad aspects of bureaucratization,the orga-
maining two do not add to the multiple corre- nization is thereby concisely portrayed in terms
lation. Table 15 shows the multiple correlation of this and similar concepts. Likewise, knowing
of 0.75 obtained by using the location measure the dependence of an organization on other
together with dependence as predictors. When organizations and its geographical dispersion
the technology scale of workflow integration is over sites tells a great deal about the likely con-
substituted as the second predictor,the multiple centration of authority in its structure. Size,
correlation is 0.71; when the operating diver- technology, dependence and location (number
sity scale of charter is used, the multiple corre- of sites) are critical in the prediction of the two
lation is 0.70. major dimensions (structuring of activities,
The prediction of line control of workflow concentration of authority) of the structures
shows this same phenomenon, where the addi- of work organizations.
tion of predictors, because of their interrela- Multiple predictions of the order of magni-
tionships, does not improve on the original tude obtained are as high as can be expected
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 11

TABLE 16. MULTIPLE REGRESSION ON STRUCTURAL FACTORS.

Subsamples
Whole - "Robust"
Structural factors sample 1 2 weightings

Structuring of activities
Weightings of predictors
Size 0.67 0.72 0.60 2
Workflow
Integration 0.29 0.14 0.43 1
Multiple correlation 0.75 0.73 0.79 0.74
Concentration of authority
Weightings of predictors
Dependence 0.64 0.50 0.77 2
Location 0.36 0.40 0.33 1
Multiple correlation 0.75 0.66 0.84 0.75

with this level of analysis. Higher values would tions Starting from a framework as outlined
imply that there were no important deviant in the conceptual scheme summarizedin Table
cases, and that differences as to policies and 1, aspects of the context and structure of the
proceduresamong the members of an organiza- organizationwere sampled in order to establish
tion have no effect on its structure. And this is scales which discriminatedamong organizations
obviously not so. The multiple predictions dis- in a large number of aspects. From this sam-
cussed here are applicable only to this sample. pling 103 primary scales of structure and con-
When the regressionequations obtained are ap- text were developed as a basis for the analysis
plied to another similar sample for prediction- of the interrelationshipsamong them.
purposes, there is likely to be a reduction in the By scaling and factor analytic techniques,
multiple correlations. The extent of this reduc- these were then summarizedto form three basic
tion can be strictly gauged only by investigat- dimensions of structure and eight salient ele-
ing another similar sample of organizations. ments of context (Table 14). The analogy
This cross-validation study is at present being with the psychological test constructor who
undertaken, but a first attempt to estimate the samples behavior in order to establish dimen-
likely amount of reduction was made by split- sions of personality is clear, and the same
ting the sample into two subsamples of 23 limitations apply. Thus while a claim can be
organizations, each stratified in the same way made for the internal consistency and scale-
as the whole sample. Table 16 gives the multi- ability of these measures,no claim can be made
ple regressions on structuring of activities and as to the comprehensivenesswith which they
concentrationof authority for the whole sample cover the field. This is particularly clear in the
and for the two subsamples separately. The attempt to elucidate aspects of context, a con-
multiple correlations and the weightings are of cept which, although in some respects narrower
the same order of magnitude. A "robust" pre- than that of environment, is still very wide.
diction on the basis of simple weightings was Emphasis was thereforeplaced on those aspects
also calculated. These correlations should be of context that had been held to be relevant to
less subject to shrinkage. The stability of the structure on the basis of previous writings.
correlation of 0.57 between operating variabil- The size of the multiple correlations obtained
ity and line control of workflow is indicated by indicates that at least some of the salient
correlationson the two subsamples of 0.50 and aspects of context were tapped.
0.65. The predicability of the structural dimen-
sions from contextual elements serves as exter-
SUMMARY AND DISCUSSION
nal validating evidence for the structural con-
This study has demonstratedthe possibilities cepts themselves. It has now been shown that
of a multivariate approach to the analysis of besides being internally consistent and scaleable,
the relationships between the structure of an as previously demonstrated, they can also be
organization and the context in which it func- related in a meaningful way to external refer-
112 ADMINISTRATIVE SCIENCE QUARTERLY

ents. Indeed the size of the correlations in- way. It can be suggested that a policy of spe-
evitably raises the question of causal implica- cializing roles and standardizing procedures,
tions. It is tempting to argue that these clear that is, of structuring, would require more
relationships are causal-in particular, that people, that is, growth in size. Concentration
size', dependence, and the charter-technology- of decisions in the hands of an owning group is
location nexus largely determine structure. likely to result in more economic integration
It can be hypothesized that size causes struc- among the subsidiaries concerned, that is, more
turing through its effect on intervening varia- dependence; while the production control, in-
bles such as the frequency of decisions and spection, and work-study procedures of staff
social control. An increased scale of operation control might raise the level of workflow inte-
increases the frequency of recurrentevents and gration in the technology.
the repetition of decisions, which are then stan- But a cross-sectional study such as this can
dardized and formalized (Haas and Collen, only establish relationships. Causes should be
1963). Once the numberof positions and people inferred from a theory that generates a dynamic
grows beyond control by personal interaction, model about changes over time. The contribu-
the organization must be more explicitly struc- tion of the present study is to establish a frame-
tured. In so far as structuringincludes the con- work of operationally defined and empirically
cept of bureaucracy, Weber's observation that validated concepts, which will enable processual
"the increasing bureaucraticorganization of all and dynamic studies to be carried out on a
genuine mass parties offers the most striking much more rigorousand comparativebasis than
example of the role of sheer quantity as a lever- has been done previously. The framework is
age for the bureaucratizationof a social struc- also seen as a means of controlling for organiza-
ture" is pertinent (Gerth and Mills, 1948). tional factors when individual and group level
Dependence causes concentration of author- variables are being studied. Such studies must
ity at the apex of publicly owned organizations now be conducted with reference not only to
because pressure for public accountability re- differences in size, but also in dependence,
quires the approval of central committees for operating function, workflow integration, etc.,
many decisions. The similar position of small and with reference to the demonstrated rela-
units in large privately owned groups is demon- tionship between these aspects of context and
strated by the effect that a merger may have organizationstructure.
upon authority. After a merger, a manager of
the smaller unit "may no longer be able to take
a certain decision and act upon it independently. D. S. Pugh is director of research and reader
He may have to refer matters to people who in organizationalbehavior at the London Busi-
were complete strangers to him a few months ness School, England. D. J. Hickson is visiting
earlier" (Stewart et al-., 1963). professor of organizational behavior, and C. R.
Integrated technology may be hypothesized Hinings is associate professor in the Organiza-
to cause an organization to move towards the tional Behaviour Research Unit of the Univer-
impersonal control end of the line-control di- sity of Alberta, Edmonton, Canada. C. Turner
mension. Line control is adequate in shops or is lecturer of sociology at the University of East
in municipal schools or building maintenance Anglia.
gangs, where the technology of the tasks is not REFERENCES
Aaronovitch, S.
mechanized and each line supervisor and pri-
1961 The Ruling Class. London: Lawrence and
mary work group is independent of all the Wishart.
others. But as workflow integration reaches the Amber, G. H., and P. S. Amber
production line or automated stages, where 1962 Anatomy of Automation. Englewood
large numbersof tasks are interdependent,more Cliffs, N.J.: Prentice-Hall.
control is needed than can be exercised by the Bakke, E. W.
line of commandalone. Udy (1965) summarizes 1959 "Concepts of the social organization." In
M. Haire (ed.), Modern Organization
this in his proposition, "The more complex the Theory. New York: Wiley.
technology . . ., the greater the emphasis on Berle, A. A., and G. Means
administration." 1937 The Modern Corporation and Private
The causal argument need not run only one Property. New York: Macmillan.
Pugh et al: CONTEXT OF ORGANIZATION STRUCTURES 113
Blau, P., and W. R. Scott Inkson, J. H. K., R. L. Payne, and D. S. Pugh
1962 Formal Organizations. San Francisco: 1967 "Extending the occupational environment:
Chandler- the measurement of organizations." Oc-
Brogden, H. E. cupational Psychology, 41: 33-47.
1949 "A new coefficient: application to biserial Levy, P., and D. S. Pugh
correlation and to estimation of selective 1969 "Scaling and multivariate analyses in the
efficiency." Psychometrika, 14: 169-182. study of organizational variables." Sociol-
Burnham, J. ogy, 3 (to appear).
1962 The Managerial Revolution. London: Mills, C. Wright
Penguin. 1956 The Power Elite. London: Oxford Uni-
Chandler, A. D. versity.
1962 Strategy and Structure. Cambridge,Mass.: Parsons, T.
Massachusetts Institute of Technology. 1956 "Suggestions for a sociological approach
Clarke, B. R. to the theory of organizations, I and II."
1956 "Organizational adaptation and precar- Administrative Science Quarterly, 1 (June
ious values." American Sociological Re- and September): 63-85, 225-239.
view, 21: 327-336. Perrow, C.
Dahrendorf, R. 1967 "A framework for the comparative anal-
1959 Class and Conflict in Industrial Society. ysis of organizations." American Socio-
Stanford, California: Stanford University. logical Review, 32 (April): 194-208.
Dubin, R. Porter, L. W., and E. E. Lawler III
1958 The World of Work. Englewood Cliffs, 1965 "Properties of organization structure in
N.J.: Prentice-Hall. relation to job attitudes and job behavior."
Duncan, 0. D., W. R. Scott, S. Lieberson, B. Dun- Psychological Bulletin, 64: 25-51.
can and H. Winsborough Presthus, R. V.
1963 Metropolis and Region. Baltimore, Md.: 1958 "Towards a theory of organizational be-
Johns Hopkins University. havior." Administrative Science Quarterly,
Eisenstadt, S. N. 3: 48-72.
1959 "Bureaucracy, bureaucratization and de- Pugh, D. S., D. J. Hickson, and C. R. Hinings
bureaucratization." Administrative Sci- 1969 "An empirical taxonomy of work organi-
ence Quarterly, 4: 302-320. zation structures." Administrative Science
Florence, P. S. Quarterly, 14 (to appear).
1961 Ownership, Control and Success of Large Pugh, D. S.. D. J. Hickson, C. R. Hinings, K. M.
Companies. London: Sweet and Maxwell. Macdonald, C. Turner, and T. Lupton
Gerth, H. H., and C. Wright Mills (eds.) 1963 "A conceptual scheme for organizational
1948 From Max Weber: Essays in Sociology. analysis.." Administrative Science Quar-
London: Routledge and Kegan Paul. terly, 8: 289-315.
Haas, E., and L. Collen Pugh, D. S., D. J. Hickson, C. R. Hinings, and C.
1963 "Administrative practices in university de- Turner
partments." Administrative Science Quar- 1968 "Dimensions of organization structure."
terly, 8: 44-60. Administrative Science Quarterly, 13: 65-
Hage, J., and M. Aiken 105.
1967 "Relationship of centralization to other Selznick, P.
structural properties." Administrative Sci- 1949 T.V.A. and the Grass Roots. Berkeley,
ence Quarterly, 12 (June): 72-92. Calif.: University of California.
Hall, R. H., and C. R. Tittle 1957 Leadership in Administration. Evanston,
1966 "A note on bureaucracy and its corre- Ill.: Row and Peterson.
lates." American Journal of Sociology, Starbuck, W. H.
72: 267-272. 1965 "Organizationalgrowth and development."
Hickson, D. J., D. S. Pugh, and D. C. Pheysey In J. G. March (ed.), Handbook of Orga-
1969 "Operations technology and formal orga- nizations. Chicago: Rand McNally.
nization: an empirical reappraisal." Ad- Stewart, R., P. Wingate, and R. Smith
ministrative Science Quarterly (to ap- 1963 Mergers: The Impact on Managers. Lon-
pear). don: Acton Society Trust.
Hinings, C. R., D. S. Pugh, D. J. Hickson, and C. Stinchcombe, A. L.
Turner 1965 "Social structure and organization." In
1967 "An approach to the study of bureauc- J. G. March (ed.), Handbook of Organiza-
racy." Sociology, 1 (January): 62-72. tions. Chicago: Rand McNally.
114 ADMINISTRATIVE SCIENCE QUARTERLY

Thomas, E. J. tions." In J. G. March (ed.), Handbook of


1959 "Role conceptions and organizationalsize." Organizations. Chicago: Rand McNally.
American Sociological Review, 24: 30-37. Walker, C. R.
Thompson, J. D., and F. E. Bates 1962 Modern Technology and Civilization. New
1957 "Technology, organization and administra- York: McGraw-Hill.
Wilson, B. R.
tion." Administrative Science Quarterly,
1962 "Analytical studies of social institutions."
2: 323-343.
In A. T. Welford et al., Society: Prob-
Trist, E. L., G. W. Higgin, H. Murray, and A. B. lems and Methods of Study. London:
Pollock Routledge and Kegan Paul.
1963 OrganizationalChoice. London: Tavistock. Woodward, J.
Udy, S. H. 1965 Industrial Organization,Theory and Prac-
1965 "The comparative analysis of organiza- tice. London: Oxford University.

You might also like