You are on page 1of 16

Accounting, Organizations an&Society, Vol. 10, No. 3, pp. 287-302, 1985. 0361-3682/85 $3.00+.

00
Printed in Great Britain @ 1985 Pergamon Press Ltd.

A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION


AND INTENDED TURNOVER IN THE LARGE CPA FIRM*

MARL4 LOMBARD1 BULLEN


School ofAccountancy, Georgia State University

and

ERIC G. FLAMHOLTZ
Graduate School of Management, University of Cal@ornia, Los Angeles

Abstracl

This research reports on the development and test of a model of job satisfaction and intended turnover in
the large CPA firm environment. On an overall basis, the model was generally supported. Satisfaction
showed strong significance in a multiple regression equation predicting the probability of turnover. In addi-
tion, eleven dimensions of job satisfaction developed through factor analysis, as well as a “career prepara-
tion” variable, correlated positively with a measure of overall job satisfaction and negatively with a measure
of the probability of turnover.

This research reports on the theoretical de- chology literature that satisfaction is inversely
velopment and empirical test of a model of job related to turnover (Brayfield & Crockett, 1955;
satisfaction and intended turnover in the large Herzberg et al., 1957; March & Simon, 1958;
CPA firm environment. The empirical test Vroom, 1964; Porter & Steers, 1973; Lawler,
studies one large metropolitan office of a “Big 1973). In the accounting area, several resear-
Eight” CPA firm in the United States. However, chers, including the authors of this paper, have
due to the similarities in large CPA firms, the shown that aspects of satisfaction are related to
study may have implications for other large firms turnover (Sorensen, 1967; Capin, 1969; Fetyko,
as well. 1972; Loeb & Gannon, 1972; White & Hellriegel,
A key objective of this research was to 1973; Lammers, 1975; lstvan & Wollman, 1976;
uncover the causes of dissatisfaction in order to Rhode et al., 1976, 1977; Faircloth & Carrell,
reduce them and, in turn, to control turnover to 1978; Dillard & Ferris, 1979; Tyra, 1980; Bullen,
the optimal extent. In general, although some 1982). Other studies which have considered
turnover is expected and considered necessary aspects of job satisfaction in the CPA firm envi-
by large CPA firms, the uncontrolled high rate of ronment have included Strawser et al., 1969;
turnover which these firms experience appears Ferris, 1977; Amernic et al., 1979; Benke &
to be dysfunctional. Rhode, 1980; Albrecht et al., 198 1; and Gaertner
It is generally accepted in the industrial psy- & Ruhe, 1981.

l The authors wish to acknowledge Professor Roger H. Hermanson of Georgia State University, Professors David M. Boje.
James E. Bruno. Louis E. Davis. and Paul I. Rosenthal. all of the Universiry of California, Los Angeles, and two anonymous
reviewers for their helpful comments on an earlier version of this paper.

287
288 MARIA LOMBARD1 BULLBN and ERIC G. FLAMHOLTZ

Within the accounting area, but not specifi- era1 investments involving lower productivity
cially the large CPA firm environment, a recent and greater supervision required during training
study of accountants (Aranya et al., 1982) from periods. Whereas some of these costs may be
a variety of firm settings did not find an associa- absorbed by clients, it is doubtful that this would
tion between job satisfaction and turnover. be true for all of the costs, especially considering
These results, which apparently conflict with the increased competition for clients in the CPA
prior findings, emphasize the need for research profession. In addition, there are possible costs
concentrating on the specific large firm environ- resulting from negative goodwill of dissatisfied
ment. migrating employees who may move to client or
Within the accounting literature, high rates of prospective client positions.
turnover are commonplace, with most accoun- Intended turnover is likely to lead to turnover
tants leaving in the first few years. For example, which in turn leads to the above described costs
Grossman (1967) as reviewed in Leathers to either the CPA firms themselves or to their
( 1971) reported that 15% of new hires left after clients. Thus, research which attempts to find
one year, 45% after three years, 62% after five ways to decrease turnover and intended tur-
years and 84% after ten years. In studying 7500 nover is of economic significance to CPA firms
employees in a 2 1 year period in one large firm, and to their clients. Since owner-partners have
Capin ( 1969) found that 15% of starting invested and are planning future investments in
employees had left after one year, 62% after human resources with the expectation of realiz-
three years, and 82% after ten years. A partner in ing a good return, it seems reasonable that own-
charge of a major office of a Big Eight firm ers would seek ways to protect their human cap-
(Steele, 1976) provided statistics for his firm for ital investments.
the past ten years. These showed that 19% of
newly hired accountants had left after one year,
50% after three years, and 80% after five years. THE MODEL
Relatively few leave after the fifth year. Tyra
( 1980) reported that 32% of the professional The model providing the basic framework for
stalYfrom one Big Eight firm left by the end of the study in this research is presented in Fig. 1.
second year, 66% by the end of the fourth year According to this model, three major areas
and 78% by the end of the sixth year. impact on the individual firm member’s assess-
The particular CPA firm investigated in this ment of his or her job satisfaction: ( 1) satisfac-
study had experienced employee shortages at tion with various “dimensions” of the work envi-
the semi-senior, senior and supervisor levels. ronment, (2) perception of the job’s success in
Whereas the office experienced an overall tur- preparing the individual for his or her career
nover rate in their professional stti of approxi- goals, and (3) perception of the availability to
mately 34% in the year prior to this study, the the individual of extraorganizational job
partners desired to reduce turnover to the low options.
20% range. The satisfaction “dimensions” shown in the
High turnover rates result in certain costs to model were developed from the broad
the CPA firm. Fetyko ( 1972) estimates an expen- categories of (a) work itself, (b) working condi-
diture of $15,000 per newly recruited accoun- tions, (c) organizational environment, (d) finan-
tant for the first year for selecting, recruiting, cial rewards, and (e) promotion opportunity.
training and salary. Other authors (Sorensen et This preliminary classification was formulated as
al., 1973) report that the replacement cost of an a result of a literature review and prior personal
entry-level staff accountant has been estimated CPA experience. Next, from preliminary inter-
to exceed 510,000. The costs associated with views at the research site, these broad categories
turnover include the specific, direct costs of were broken down into a larger set of 30 prelimi-
selection and training, as well as the more gen- nary satisfaction dimension variables (Table 1).
A THEOFtETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 289

Perceptron of Perceptjon af
job’5 success avarlabilky of
Communlcbtron in preparmg extraorgonrzat~onal
Appreciation/porticipotion for cower Options
Supervision/evaluation gwts
Staffmg
Counseltng/lroswg
Financial
Promotion
I
* :
1 ,Amq! 1
Overah assessment _ LevelT L,
of
c---- ---I * ) Probabjlity of turnover
zz,z?:;, I.,\
DL.,.“, (Intent to leave)
job sahsfactjon Other factors J§ L

* Tested by Hypotheses I-3


t Tested by Hypotheses 4 -7

) Although beyond the scope of thus paper, differences by these vanobkes In the other model vorlabtes
have teen tested, and a number of drfferences found. See Butlen (1982).

5 Other factors which may affect job sotjsfactron and turnover that have not been tested by the
researchers Include performance, ability, personality, ond certorn other personal charactenstlcs,
geogrophicat locatron and general economic condttjons.

Fig. 1. Model of the components of job satishction and the probability of turnover.

These were used as a guide in constructing a the job’s success in preparing for career goals,
questionnaire “tailor-made” for the large CPA perception of availability of extraorganizational
firm environment. Subsequently, factor analysis options, overall assessment of satisfaction, and
was used to reduce the 70 satisfaction dimen- probability of turnover) are likely to exhibit dif-
sions questionnaire items based on the 30 pre- ferences due to the individual’s ( 1) area special-
liminary variables to 11 “factors” or scales. ity or department in the firm, (2) level in the
These became the actual variables used in the firm, (3) career goal preferences, and (4)
analysis: ( 1) higher needs/work itself, (2) client demographic characteristrcs (i.e., length with
range/assignments, ( 3) pressures, (4) office sup- firm, education and sex). “Other” factors
port, (5) communication, (6) appreciatiompar- include performance record, ability for public
ticipation, (7) supervision/evaluation, (8) staf- accounting work, personality and certain other
fing, (9) counselingtraining,,( 10) financial, and personal characteristics, geographical location,
( 11) promotion. and general economic conditions. Although
The model also postulates that the individual reporting on these differences is beyond the
CPA firm member’s probability of turnover (i.e., scope of this study, items ( 1) through (4) above
intent to leave) is affected by four factors: ( 1) have been tested in Bullen ( 1982 ).
overall assessment of satisfaction, (2) satisfac- A listing of variables tested in this study
tion with the dimensions of the working envi- appears in Table 2.
ronment, (3) perception of the job’s success in
preparing for career goals, and (4) perception of
the availability of extraorganizational options. METHODOLOGY
Thus, the model maintains that the latter three
items affect the probability of turnover both Subjects
directly and through their indirect effect on The subjects were from one large metropoli-
overall satisfaction. tan office of a U.S. “Big Eight” firm. However, due
The model specifies that its five major compo- to the similarities in large CPA firms, this study
nents (satisfaction dimensions, perception of may also have implications for other large firms.
290 MARL4 LOMBARD1 BULLEN and ERIC G. FLAMHOLTZ

.
TABLE 1. Listing of preliminary satisfaction dimension such as identifying and acting upon potentially
variables controllable sources of job satisfaction and dis-
satisfaction.
a Work itserf
1. Skill variety
Data for the study were obtained through a
2. Autonomy questionnaire distributed to nonpartner firm
3. Authority members. The number of responses was 125,
4. Self-esteem ’ representing a response rate of 76%.
5. Growth opportunity
6. Interaction with high level clients
7. Work identity
Questionnaire
8. Worksignificance The questionnaire was developed specifically
9. Client range for measuring the variables under study in the
large CPA firm environment. An important part
b Working conditions
of the study involved the participation of 17 firm
10. SecretariaUadministrative support
I 1. Office physical environment
members in a preliminary interview process for
12. Multiple Assignments the purpose of identifying relevant items for
13. Time factors inclusion in the questionnaire, reviewing the
14. Travel requirements completed instrument for reasonableness, and
making predictions regarding the hypotheses.
C Organizationaienvironment
IS. Communication
A literature review was also undertaken to
16. Recognition identify possible sources of satisfaction or dis-
17. Appreciation satisfaction to be included in the measuring
18. Participation instrument, as was prior personal experience in
19. Bureaucracy
the Big Eight CPA environment.
20. Relationship with co-workers
The format of the questionnaire was a five
2 1. Supervision
22. Evaluation point Likert-type scale. Seventy questionnaire
23. Training items were constructed to measure the satisfac-
24. Counseling tion dimensions. Overall satisfaction and proba-
25. StatBng
bility of turnover were each measured by two
d Financial rewards TABLE 2. Listing of variables
26. Salary
27. Fringe benefits Satisfaction dimensions variables (from factor analysis)
28 Salary equity 1. Higher needs/work itself
2. Pressures
e mmotion opportunityljo6 security
3. Communication
29. Promotion opportunity 4. Promotion
30. Job security 5. Supervision/evaluation
6. Office support
7. Staffing
8. Appreciation/participation
The CPA firm under study estimated that an 9. Client range/assignments
10. Financial
average of $5,000 per recruit was expended for
I I. Counseling/training
the more identifiable costs of recruiters’ salaries
and travel, staff time spent recruiting and formal Other variables in model
training. The firm indicated that certain factors 12. Overall satisfaction: overall assessment of satisfaction
significantly affecting turnover were beyond 13. Probability of turnover: intent to migrate
14. Career preparation: perception of job’s success in pre-
their control (i.e., the economy’s influence on
paring for future career goals
the availability of outside options). However, 15. Availability of options: perception of availability of
they realized the possibility of, and expressed extraorganizational job opportunities
interest in reducing turnover through means
A THEOBETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 291

questions; career preparation and availability of information on the degree to which the applica-
options by one questionnaire item each. The ble variables were interrelated.
complete instrument can be found in Bullen
(1982).
RESULTS
Statement of hypotheses
The hypotheses investigated were: Table 4 shows the descriptive statistics, fre-
quencies and reliabilities of the satisfaction
1. Prediction of overall satisfaction dimensions and other variables. The lowest
Hl: The satisfaction dimensions cannot be used to pre- mean (“ 1 = Very dissatisfied”; “5 = Very satis-
dict overall satisfaction. fied”) for the satisfaction dimensions was office
H2: Career preparation cannot be used to predict overall
support (2.72); while the highest was higher
satisfaction.
H3: Availability of options cannot be used to predict needs/work itself (3.77). Satisfaction dimen-
overaIl satisfaction. sions are discussed further in connection with
the results of the hypothesis testing.
2. Prediction of probability of turnover The mean for the overall satisfaction scale
H4: The satisfaction dimensions cannot be used to pre- revealed that most people scored in the middle
dict probability of turnover.
or on the positive side. The implication is that a
H5: Overall satisfaction cannot be used to predict proba-
bility of turnover. major immediate overhaul of the entire working
H6: Career preparation cannot be used to predict proba- environment is not needed. Rather, particularly
bility of turnover. for those members giving responses in the mid-
H?: Availability of options cannot be used to predict dle ranges, satisfaction is likely to increase with
probability of turnover.
changes in those dimensions of the working
environment which are identified as sources of
Analysis of data dissatisfaction.
Table 3 shows the specific statistical For probability of turnover, the mean indi-
methodology used in testing the hypotheses. cated that, on an overall basis, respondents per-
Hypotheses 1 through 7 were tested using muhi- ceived slightly more of an intention to stay with
ple regression analysis (using SPSS), while Pear- than to leave the firm, at least in the relatively
son Product Moment correlations provided short run. However, further examination of the

TABLE 3. Statistical analysis used to test hypotheses

Hypothesis Independent
tested variable Dependent variable Type of analysis

a. Related to researchquestion I: prediction of and relationship to overall satisfaction


Hl Satisfaction Overall satisfaction 1. Multiple regression 2. Correlation analysis
dimensions
H2 Career preparation Overall satisfaction 1. Multiple regression 2. Correlation analysis
H3 Availability of Overall satisfaction 1. Multiple regression 2. Correlation analysis
options

b. Related to researchquestion 2:prediction of and relationship toprobability of turnover


H4 Satisfaction Probability of turnover 1. Multiple regression 2. Correlation analysis
dimensions
H5 Overall satisfaction Probability of turnover 1. Multiple regression 2. Correlation analysis
Hb Career preparation Probability of turnover 1. Multiple regression 2. Correlation analysis
H7 Availability of
options
292 MARIA LOMBARD1 BULLEN and ERIC G. FL4MHOLTZ

TABLE 4. Descriptive statistics, frequencies and reliabilities of constructed satisfaction scales and other variables

Average scale
Number item
of items score
in Relia- Range (Mean f
Variable scale bility* Meant S.D. Low High no. items)

Satisfactionscalesi
1. Higher needs/ 12 0.91 45.26 7.96 20 60 3.77
work itself
2. Pressures 0.84 25.65 5.18 10 39 3.21
3. Communication 0.88 29.76 6.83 10 44 3.31
4. Promotion 0.83 21.70 4.88 8 32 3.10
5. Supervision/ 0.84 23.16 4.66 9 32 3.31
evaluation
6. Office support 5 0.80 13.59 3.75 5 23 2.72
7. StatTing 6 0.85 17.38 4.70 6 27 2.90
8. Appreciation/ 6 0.87 18.86 4.96 7 28 3.14
participation
9. Client range/ 5 0.74 18.30 3.11 8 24 3.66
assignments
10. Financial 4 0.71 11.80 2.86 4 19 2.95
11. Counselin@aining 4 0.74 12.39 3.20 5 20 3.10

Other variables
Overall satisfaction5 2 0.81” 4.79 1.79 2 10 2.40$
Probability of tumover(/ 2 0.71tt 5.56 2.71 2 10 2.78
Career preparation7 1 - 3.8 1 0.96 1 5 3.81
Availability of options7 1 - 4.26 0.78 2 5 4.26

l Cronbach’s “coefficient alpha”.


t The means should not be compared down the column as the scales are comprised of different numbers of items.
$ The scale range is “1” = very dissatisfied; “5” = very satisfied.
5 For the overall satisfaction scale, unlike the other scales, a “1” represents the most satisfied/positive response and a “5”
the most dissatisfied/negative response. Thus a “2.4” average on the overall satisfaction scale would be a “3.6” if the
scaling had been similar to the other scales.
/IA “1” = highest probability of turnover, “5” = lowest.
fl The scale range is “ 1” = not at all; “5” = to a very great extent.
** Pearson Correlation of two questionnaire items in scale.
tt Pearson Correlation of two questionnaire items in scale.

frequencies to the questionnaire responses many people perceived that their job was still
revealed that about 80% did intend to leave preparing them for their future career goals.
within five years. Since turnover is a problem Firm members showed that they valued their
particularly within the semi-senior through public accounting experience and considered it
beginning supervisor levels (two to five year a stepping stone to their future careers, within or
ranges), improvements made to control the outside the firm.
level of turnover by identifying potentially con- As was expected, for availability of options,
trollable sources of job satisfaction and dissatis- most of the respondents answered on the posi-
faction are of potential benefit. tive side, indicating that they perceived a high
The mean for career preparation suggests that availability of extraorganizational job oppor-
A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 293

tunities in the particular large city in which the TABLE 5. Prediction of overall satisfaction from facet
research firm is located. The implication is that satisfaction scales, career preparation and availability
of options
the probability of turnover is likely to be higher
and overall satisfaction possibly lower than if Unstand- Stand-
there had been a low availability perceived. Variable ardized ardized F

Prediction of overall satisfaction: hypotheses Satisfaction scales


1. Higher needs/work itself -0.09 -0.38 11.96t
1-3 -0.17 4.15’
2. Pressures -0.59
In the multiple regression analysis, overall 0.13 0.05 0.15NS
3. Communication
satisfaction was the “dependent” variable pre- 4. Promotion -0.43 -0.12 1.58NS
dicted by the 11 satisfaction dimensions vari- 5. Supervision/evaluation 0.26 0.07 0.53NS
ables, career preparation and availability of 6. Office support -0.39 -0.01 O.OINS
7. Staffing -0.38 -0.10 1.46NS
options.
8. Appreciation/participation -0.16 -0.04 0.15NS
As the results in Table 5 indicate, the F-ratio is -0.20 -0.03 0.15NS
9. Client Range/assignments
significant at p < 0.001. The R’, which is the 10. Financial 0.28 0.05 0.32NS
proportion of variation explained by the vari- 11. Counseling/training -0.25 -0.04 0.30NS
ables included in the regression equation, was
Other variables
0.64. The other measure of the accuracy of use-
Career preparation -0.63 -0.34 24.55t
fulness of the prediction equation, the standard Availability of options 0.85 0.04 0.38NS
errors of the estimate, is 1.40. These measures
indicate a reasonable level of accuracy. l Significant atpC0.05.
The results show that only two of the satisfac- t Sign&cant at p<O.OOl.
tion dimensions - i.e., higher needs/work itself NS = Not significant atpc0.05.
and pressures - and the career preparation var- R2 = 0.64.
iable, are statistically sign&ant (at the 0.05 Standard error = 1.40.
level) in predicting overall satisfaction. How- F( 13,110) = 15.04t.
ever, according to the basic model of this study, N = 124.
it was expected that all the satisfaction dimen-
sions would have an impact on overall satisfac- shown in Table 6. (Since the scales for the satis-
tion. It should be noted that the reason why faction dimensions and overall satisfaction run
some of the other satisfaction dimensions were in opposite directions, the correlations show
not signticant may be explained by their corre- negative signs on Table 6 even though the sub-
lation with the dimensions that were significant. stantive meaning is a positive relationship.) As
As explained below, these correlations indicate indicated in Table 6, the correlations of higher
that there does appear to be some multicol- needs/work itself with a number of the other
linearity, which may have some effect on the dimensions are high - particularly with com-
multiple regression equation. However, this is munication (0.75), client range/assignments
not considered a serious problem in the theoret- (0.68), appreciation/participation (0.66), staf-
ical development of the satisfaction model in fing (0.60), and supervision/evaluation (0.57).
this study. For example, proving a strict direc- Table 6 shows that pressures was correlated
tional relationship is not the only approach in fairly highly (0.62) with supervision/evaluation.
providing information regarding the relation- The point to be made is that most of the variabil-
ship of the satisfaction dimensions and other var- ity in overall satisfaction was “explained” by the
iables with overall satisfaction. Correlation higher needs/work itself and pressures dimen-
analysis can also suggest which variables are sions. Since these two were related to the other
likely to have most impact on overall satisfac- dimensions, there was not much variability left
tion. The intercorrelations between the satisfac- to be explained by them. In other words, if
tion dimensions and the other variables are higher needs/work itself and pressures had been
TABLE 6. Intercorrelations among facet satisfaction scales and other variables

Variable
satisfactiondimensions$ 1 2 3 4 5 6 7 8 9 10 11 OS. P.T. C.P. A

1. Higher needs/work itself 1.00


2. Pressures 0.45 1.00
3. Communication 0.75 0.55 1.00
4. Promotion 0.43 0.46 0.60 1.oo
5. Supervision/evaluation 0.57 0.62 0.60 0.53 1.oo
6. Office support 0.17t 0.26 0.28 0.37 0.35 1.00
7. StafBng 0.60 0.51 0.59 0.45 0.6 1 0.34 1.00
8. Appreciation$articipation 0.66 0.52 0.83 0.68 0.55 0.24 0.53 1.00
9. Client range/assignments 0.68 0.52 0.52 0.4 1 0.57 0.30 0.57 0.44 1.00
10. Financial 0.30 0.49 0.50 0.62 0.47 0.38 0.46 0.49 0.34 1.00
11. Counselling/training 0.43 0.25 0.58 0.36 0.50 0.46 0.42 0.48 0.35 0.32 1.oo

Other variables:
Overall satisfaction11 -0.70 -0.49 -0.58 -0.45 -0.52 -0.24 -0.54 -0.55 -0.60 -0.32 -0.37 1.00
Probability of turnover)) 0.51 0.35 0.46 0.28 0.39 0.17t 0.31 0.48 0.38 0.23 0.25 -0.61 1.00 n
0.15’ 0.24 -0.60 0.47 1.00 w
Careerpreparation” 0.47 0.25 0.33 0.22 0.38 0.15t 0.29 0.31 0.48
Availabilityofoptions” 0.12’ O.lO$ 0.09$ -0.08$ O.lO$ -0.03$ 0.11s 0.11s 0.07$ O.Ol# -0.01* -0.07$ 0.02$ 0.14$ 1.00 E
m
- 3
NOTE: With the exception of the following items, all of the above correlations are significant atjKO.01. 2
l Significant atp<O.lO. # The scale range is “ 1” -very dissatisfied; “5” -very satisfied
7 Significant atpC0.05. ((The scale range is ” 1” -most satisfied, positive response; “5” -most dissatisfied, negative response. Note that this is
the opposite of the direction of the satisfaction dimensions scales; thus the correlations of the satisfaction
$ Not significant atpt0.10. dimensions with overall satisfaction show negative signs even though they are positively correlated.
1 In the scale range a “ 1” represents the highest probability of turnover; and a “5” the lowest
l* Thescale range is”1”-not at all; “5”-toaverygreat extent-i.e., these “run thesameway”as the
satisfaction dimensions scales.
A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 295

left out of the regression equation, it is conceiva- ence they are getting, are all important in deter-
ble that these other dimensions would have mining these individuals’ overall job satisfaction.
shown significance. In fact, the above items appear to be even more
In addition to multiple regression analysis, important than, for example, monetary compen-
examination of the correlations of the individual sation in the sense that they are more highly cor-
satisfaction dimensions with overall satisfaction related with overall satisfaction.
provided useful information. As shown in Table The correlation analysis shows that career
6, the satisfaction dimensions having the highest preparation is one of the most positively corre-
correlation with overall satisfaction were higher lated variables (0.60) with overall satisfaction.
needs/work itself (0.70) client range/assign- The results of the multiple regression indicate
ments (0.60) communication (0.58) apprecia- that, as predicted, career preparation signifi-
tion/participation (0.55), staffing (0.54) super- cantly predicts overall satisfaction (at the 0.00 1
vision/evaluation (0.52) pressures (0.49) and level). Thus, the firm members’ perceptions of
promotion (0.45). Thus, these variables are the job’s success in preparing them for their
likely to be most related to overall satisfaction. future career goals can be considered an impor-
On the other hand, the dimensions with the tant factor in determining how satisfied they are
lowest correlations were office support (0.24) with their positions on an overall basis.
financial (0.32) and counseling/training (0.37). Availability of options was not significant.
Upon first review, it seems somewhat surprising Most people, including the more satisfied and
that office support and financial were the least less satisfied members, seemed to feel that there
correlated with overall satisfaction. Office sup- was a good availability of employment oppor-
port, with an average item scale mean of 2.72 tunities outside the firm. Thus, there was not
(“1” = very dissatisfied; “5” = very satisfied) much variability in this variable in the subjects
showed the lowest mean of the 11 satisfaction (from one large city) in this study. However, it is
dimensions and financial (2.95) was the third still possible that, in other situations, availability
lowest. Additionally, complaints regarding them of options would show more of a relationship to
were made in comments both in the preliminary overall satisfaction.
interviews and in an open-ended section of the
questionnaire. Prediction of probability of turnover: bypotb-
The explanation appears to be that, while eses 4-7
some people may be annoyed with these items, This last set of hypotheses tests the major
they are not as important in determining their relationships in the model forming the theoreti-
overall satisfaction as are the other more highly cal basis of this study - i.e., that the probability
correlated dimensions, and career preparation. of turnover is predicted by the satisfaction
Overall, people were on the positive side of the dimensions (H4), overall satisfaction (H5),
satisfaction scale; thus the dimensions on the career preparation (HG), and availability of
positive side are likely to show higher correla- options (H7).
tion with overall satisfaction. For example, the Table 7 shows the results of the multiple
highest correlation with overall satisfaction is regression analysis using probability of turnover
higher needs/work itself As Table 4 indicates, as the dependent variable. Again, the F-ratio is
this scale was rated the highest in satisfacion. significant at the 0.001 level. However the R2
As expected, in determining their “overall (0.45) is lower, and the standard error (2.13)
satisfaction,” the CPA firm members are factor- higher than for the equation predicting overall
ing in both negative and positive factors, but satisfaction. Thus, this multiple regression equa-
somewhat emphasizing the positive. Apparently, tion is somewhat weaker, although it is consi-
factors related to their organizational environ- dered adequate for the purpose of testing the
ment, the satisfaction of their higher needs, the hypotheses and drawing some conclusions.
work they are doing, and the clients and experi- Examination of the significance levels of the
296 ,MABIALOMBARD1 BULLEN and ERIC G. FLAMHOLTZ

TABLE 7. Prediction of probability of turnover from facet coefficients did not show significance. Examina-
satisfaction scales, overall satisfaction, career preparation tion of the intercorrelations in Table 6 reveals
and availability of options
that overall satisfaction showed the highest
Unstand- Stand-
negative correlation (0.61) with probability of
Variable ardized ardized F turnover. The second highest negative correla-
tion (0.5 1) is with higher needs/work itself.
Satisfaction scales However, as indicated in the previous discussion
1. Higher needs/work itself 0.29 0.09 0.36NS of the hypotheses predicting overall satisfaction,
2. Pressures 0.84 0.02 0.02NS higher needs/work itself was also most signihc-
3. Communication 0.16 0.04 0.06NS
4. Promotion
ant in predicting overall satisfaction @ < 0.00 1)
-0.11 -0.20 3.07.
5. Supervision/evaluation 0.63 0.11 0.89NS and most correlated to it (0.70). So it is quite
6. Office support 0.58 0.08 0.85NS possible that, had overall satisfaction been left
7. StafBng -0.70 -0.12 1.35NS out of the regression equation, higher needs/
8. Appreciation/participation 0.17 0.30 4.04t work itself would have been significant in pre-
9. Client range/assignments -0.64 -0.07 0.42NS
10. Financial
dicting the probability of turnover. The same
0.33 0.04 0.13NS
11. Counseling/training -0.90 -0.10 1.13NS situation appears also applicable to some of the
other satisfaction dimensions which showed the
Other variables higher positive correlations with overall satisfac-
Overall satisfaction -0.62 -0.41 11.95$ tion and the higher negative correlations with
Career preparation 0.47 0.17 3.16’
probability of turnover. For example, for higher
Availability of options -0.31 -0.09 1.46N.S
needs/work itself the two sets of correlations for
overall satisfaction (positive) and probability of
l Significant atp40.10. turnover (negative) were 0.70 and 0.5 1 respec-
t Significant at p<O.OS. tively; for pressures 0.49 and 0.35; for communi-
$ Significant at pCO.00 1. cation 0.58 and 0.46; for staffing 0.54 and 0.31;
NS = Not significant at p<O. 10. and for client range/assignments 0.60 and 0.38.
R’ = 0.45. Thus, it is likely that all of these dimensions, in
Standard error = 2.13. addition to promotion and appreciation/partici-
F( 14,109) = 6.44$ pation, may be useful in predicting probability of
N= 124. turnover.
Again, the dimensions which correlated least
regression equation coefficients for the satisfac- with overall satisfaction also correlated least
tion dimensions reveals that appreciatiotipar- with probability of turnover. These were offtce
ticipation is significant at the 0.05 level. HOW- support (0.24 and 0.17) financial (0.32 and
ever the coefficient showing the highest signift- 0.23) and counseling/training (0.37 and 0.25).
cance (0.001) was overall satisfaction. It was From this data it appears that these three dimen-
predicted that overall satisfaction would sions are not as influential in the decision to mig
strongly impact on the probability of turnover. It rate as are some of the other variables.
was also expected that the individual satisfaction At first, it seems somewhat surprising that the
dimensions would have an effect. However, financial dimension was not correlated more
most of them did not show significant coeffi- highly with probability of turnover. In the pre-
cients. The explanation may be similar to that liminary interviews financial factors were cited a
provided in the last section on overall satisfac- number of times as reasons for people leaving.
tion. In this case, it is likely that, because so However, the idea was also expressed that
much of the variation in probability of turnover people value the experience they receive in the
was accounted for by overall satisfaction, and firm and are willing to forgo dollars for it.
since the satisfaction dimensions are correlated Nevertheless, after a certain period with the
with overall satisfaction, their partial regression firm, they may feel that continuation in public
A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 297

accounting is no longer in their best interests the two satisfaction dimensions, higher needs/
when a position comes along offering “more work itselfand pressures, were significant in pre-
money for less work.” The number of years that dicting overall satisfaction. Overall satisfaction
a higher value is placed on the experience factor, and the satisfaction dimension, appreciation/
rather than on working conditions and compen- participation, were significant in predicting
sation, varies with the individual. The point is probability of turnover. However, availability of
that many people consider their public account- options was not significant in the multiple
ing experience a training period, and as such an regression. Concerning the satisfaction dimen-
opportunity to build their own human capital. sions, all were positively correlated with overall
They are willing to postpone leaving for higher satisfaction and negatively correlated with prob-
salary until the right time, even if they are dis- ability of turnover (Table 6). However, most of
satisfied with their compensation. The basic the regression coefficients in the multiple
model in this study proposes what was regression equation were not significant.
suggested by these comments - i.e., that per-
ception of the job’s success in preparing for
career goals influences the probability of tur- TABLE 8. Summary of results of hypothesis testing

nover. From the interviews, and the study’s


Null Sign.
results, it appears that this variable is more hypothesis level’
important in influencing probability of turnover
than are some of the individual satisfaction 1. Prediction of owall satisfaction
dimensions. H 1 Overall satisfaction cannot be
predicted by the satisfaction
However, in the multiple regression equation,
dimensions. Rejected Variousi
career preparation was not signscant at the 0.05 H2 Overall satisfaction cannot be
level in predicting the probability of turnover. predicted by career
Nevertheless its correlation (0.47) with the preparation. Rejected O.OOl$
probability of turnover is one of the higher cor- H3 Overall satisfaction cannot be
predicted by availability of
relations, as is indicated in Table 6. Only overall
options. Retained 0.25
satisfaction (0.6 1), higher needs/work itself
(0.5 1 ), and appreciation/participation (0.48) 2. Prediction ofprobability of turnover
are higher. Thus we can conclude that, as pre- H4 Probability of turnover cannot
dicted, career preparation is one of the major be predicted by the satisfaction
dimensions. Rejected Various5
factors entering into the turnover decision.
H5 Probability of turnover cannot
Again, availability of options was not signifi- be predicted by overall
cant, most probably for the type of reasons dis- satisfaction. Rejected 0.001
cussed in the last section on overall satisfaction. H6 Probability of turnover cannot
be predicted by career Between
preparation. Retained 0.05 & 0.10
H7 Probability of turnover cannot
be predicted by availability Between
Summay of results of hypothesis testing ofoptions. Retained 0.108~0.25
Table 8 shows the significance levels of the
statistical analyses used in the hypothesis test- l A signlticance level of 0.05 was necessary to reject the nuIl
hypothesis.
ing; and whether the hypothesis was considered
“retained” or “rejected.” In order to reject the t Further explanation is provided in Table 5 and the text of
this paper.
null hypothesis and accept the alternative as
$ The SPSScomputer program for regression does not print
being true, a 0.05 level of significance was consi-
the actual sign&axe levels; thus the F table approximation
dered necessary. shown is considered appropriate.
On an overall basis, the model was generally 5 Further explanation is provided in Table 7 and the text of
supported. For example, career preparation, and tis paper.
298 MARL4 LOMBARD1 BULLEN and ERIC G. FLAMHOLTZ

IMPLICATIONS AND DIRECTIONS evaluation, pressures and promotion. The satis-


FOR FUTURE RESEARCH faction dimensions with the highest (negative)
correlation with probability of turnover were
higher needs/work itself, appreciationparticipa-
Of particular signitlcance for theory develop- tion, communication, supervision/evaluation,
ment in large CPA lirm job satisfaction and tur- client range/assignments, pressures and staffing.
nover is the models inclusion of both the career From the above it appears as if these factors
preparation and career goal choice variables. To are particularly important to firm members’
review, the model proposes that a CPA firm overall feeling of job satisfaction and their deci-
member’s perception of the job’s success in pre- sions to migrate or stay with the firm. Addition-
paring for career goals impacts on the indi- ally, it is felt that more work is needed to ascer-
viduals overall job satisfaction, as well as on his tain the relationship of these variables to job
or her decision to leave or to stay with the firm. satisfaction and turnover; and thus that future
The results of the hypothesis testing showed that research should incorporate study of them.
these propositions of the model were generally Availability of options was not significant in
supported. Additionally, the model suggests that testing the model; although, as suggested earlier,
there will be differences in the individuals job the reason may be that most people perceived a
satisfaction and turnover decision as a result of high availability of outside job opportunities in
differences in the individuals future career the large city in which the particular firm
goals. This proposition has also been supported. studied was located. It is expected that this vari-
Thus, forming an integral part of the indi- able would show more of an impact on overall
vidual’s overall assessment of satisfaction and satisfaction and probability of turnover in
probability of turnover are the individual’s research utilizing a wider range of subjects from
career goals and perceptions of how these goals varying geographical areas and different time
are being met in his or her CPA firm position. It periods. Future research of a longitudinal nature
is suggested that aspects of career be included in is needed to fully study the impact of this vari-
the framework guiding future research studies of able.
job satisfaction and turnover in the large CPA In conclusion, there is a need for future work
firm environment. The results obtained in the in the general area of personnel aspects of the
hypotheses testing of this study can be used as a public accounting profession and the specihc
basis in making predictions in future such areas of job satisfaction, turnover, and careers in
hypotheses. the large CPA firm environment. This is espe-
It is also felt that the identification of the satis- cially so as prior work in the area has not been
faction dimensions provides implications for extensive.
theory and research by specifying variables Especially of interest to this research is the
which might be included in future studies of job replication of the current study using a represen-
satisfaction and turnover. These factors should tative sample from all of the large firms. This
be considered because they are likely to be future work is necessary in order to support the
sources of job satisfaction and dissatisfaction. As generalizeability of the results to this larger envi-
shown in the study’s hypothesis testing, satisfac- ronment. Such a study would contribute further
tion on each of the 11 satisfaction dimensions by taking a longitudinal approach in examining
was shown to be positively related to overall actual turnover data at later points in time after
satisfaction and inversely related to probability collection of the original data. In this way the
of turnover. models ability to predict both intended and
The highest correlations (positive) with over- actual turnover could be tested. The findings of
all satisfaction were higher needs/work itself, such future studies are likely to be of value to
client range/assignments, communication, app- CPA firm owner-partners, employees, prospec-
reciation/participation, staffing, supervision/ tive employees and university educators.
A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 299

BIBLIOGRAPHY

Albrecht, W. S., Brown, S. W. & Field, D. R., Toward Increased Job Satisfaction of Practicing CPAs, Journal
ofAccountancy (August 1981) pp. 61-66.
Amemic, J, H., Aranya, N. & PolJock, J,, Is there a Generally Accepted Accountant? CA Magazine (October
1979) pp. 34-42.
Aranya, N., La&man, R. & Amernic, J., Accountants’Job Satisfaction: A Path Analysis, Accounting. Organ-
izations and Society ( 1982) pp. 20 l-2 11.
Bet&e, R. L. & Rhode, J. G., The Job Satisfaction of Higher Level Employees in Large Certified Public
Accounting Firms, Accounting Organizations andsociety ( 1980) pp. 187-201.
Brayfield, A. H. & Crockett, W. H., Employee Attitudes and Employee Performance, PsychologicalBulletin
( 1955) pp. 39-24.
Bullen, M. L., An Empirical Study ofSources ofJob Satisfaction in the Big Eight CPA Firm (Ph.D. dissertation,
University of California, Los Angeles, 1982).
Capin, R. S.. How to Cope With the StaRMan Shortage. TbePructicul Accountant ( 1969).
DiJJard, J. F. & Ferris, K R., Sources of Professional Staff Turnover in Public Accounting Firms: Some Further
Evidence,Accounting OrganirationsandSociety (1979)~~. 179-186.
Faircloth, A. & Carrell, M., Evaluating Personnel Procedures in CPA Fhms -A Study, Journal ofAccount-
ancy (October 1978).
Ferris, K R., Perceived Uncertainty and Job Satisfaction in the Accounting Environment, Accounting,
Organizations and Society ( 1977) pp. 2328.
Fetyko, D. F., Retention of Staff Personnel in Michigan CPA Firms, The Michigan CPA (May-June 1972)
pp. 21-25.
Gaertner, J. F. & Ruhe, K R., Job-Related Stress in Public Accounting, Journal ofAccountancy (June 198 1)
pp. 68-74.
Herzberg. F., Mausner, B., Peterson. R. 0. & Capwell, D. F., Job Attitudes: Review of Research and Opinion
(Pittsburgh: Psychological Service of Pittsburgh, 1957).
Istvan, D. F. & Wollman, J. B., Turnover in CPA Firms, The CPA Joumaf uuly 1976) pp. 2 l-25.
Iammers, L.. An Empirical Study of the Validity and Reliability of the Job Descriptive Index for the Measure-
ment of Accountants’ Job Satisfaction (Ph.D. dissertation, University of Winsconsin-Madison. 1975 ).
Lawler, E. E.,Motivution in Work Organizations (Monterey, CA: Brooks/Cole, 1973).
Leathers, P. E., Staff Retention in Public Accounting Firms, The Journul ofAccountancy (January 1971)
pp. 87-90.
Loeb, S. E. & Gannon, M. J., Need Satisfaction and Staff Retention in large Certified Public Accounting Firms,
The CPA Journal (April 1972) pp. 327-329.
March, J. G. & Simon, H. A.. Organizations (New York: John Wiley, 1959).
Nie, N., Hull, C., Jenkins, J.. Steinbrenner. K & Brent, D., Statistical Puckugefor the Social Sciences SPAS
(New York: McGraw-Hill, 1975).
Porter, L. W. & Steers, R. M., Organizational, Work and Personnel Factors in Employee Turnover and
Absenteeism, PslychoiogicaI Bulletin ( 1973) pp. 15 l-l 76.
Rhode. J. G., Sorensen, J. E. & Lawler, E. E., An Analysis of Personal Characteristics Related to Professional
StaffTurnover in Public Accounting Firms, Decision Sciences (October 1976) pp. 771-880.
Rhode, J G.. Sorensen, J. E. & Lawler. E. E., Sources of Professional Staff Turnover in Public Accounting
Firms Revealed by the Exit Interview, Accounting. Organizations and Society ( 1977) pp. 165-J 75.
Sorensen, J. E., Professional and Bureaucratic Organization in the Public Accounting Firm, Accounting
Revieu~ (July 196’) pp. 553565.
Sorensen, J. E.. Rhode, J. G. 8r Bawler. E. E., The Generation Gap in Public Accounting, The Journal of
Accountunq (December 1973) pp. 42-50.
Steele. C. G., Discussant’s Response to Sorensen, J. E., Sorensen, T., Rhode, J. & Bawler, E., A Behavioral Study
of Staff Retention in the Public Accounting Profession, Symposium on Auditing Research 1974 (Urbana-
Champaign: University of Illinois, 1976).
Strawser. R. H., Ivancevich. J. M. & Lyon. H. L., A Note on the Job Satisfaction of Accountants in Iarge and
Small CPA Firms, Journal ofAccounting Research (Autumn 1969) pp. 339-345.
Tyra. A.. Staff Turnover in CPA Firms, The Woman CPA (January 1980) pp. 13-28.
Vroom. V. H., U’oork and Moth&on (New York: John Wiley, 1964).
White. G. E. 8i Hellriegel. D.. Attitudes of CPAs Related to Professional Turnover, Journal ofAccountancy
(June 1973) pp. 86-88.
300 MARIA LOMBARD1 BULLEN and ERIC G. FLAMHOLTZ

APPENDIX A. FACTOR ANALYSIS OF FACET SATISFACTION ITEMS


ORTHOGONAL ROTATION: VARIMAX’

Factors
1 2 3 4 5 6 7 8 9 10 11 ct

Factor 1: Higher needs/work itself


S2Q6 Amount of authority you have 0.71 0.74
S2Q3 Opportunity for freedom &flexibility 0.68 0.6 1
S2Q 10 Extent position uses skills & abilities 0.68 0.63
SlQ5 Extent work is significant 0.65 0.56
S2Q2 Opportunity to gain experience & develop skills 0.55 0.45 0.58
S2Ql Opportunity to see total picture of project 0.54 0.56
S4Q5 Recognition 81 respect from firm members 0.52 0.41 0.76
S2Q9 Sense of accomplishment & self-esteem 0.51 0.61
S4Q6 Extent treated as professional 0.47 0.6 1
S2Q4 Interaction with high-level clients 0.46 0.41
S4Q 14 Association with professional colleagues 0.45 0.53
S2Q8 Opportunity for variety in assignments 0.41 0.50 0.56

Factor 2: Pressures
53410 Tie deadlines 0.84 0.75
S3Ql Number hours worked 0.65 0.64
S3Q9 Distances commute in town 0.62 0.46
S3Q7 Number ofpeople report to at same time 0.53 0.52
S3Qll Number of assignments at same time 0.48 0.42 0.59
S3Q12 Start new job before finishing last 0.48 0.52
S3Q6 Engagement time budgets 0.43 0.48
S5Q5 Timeliness of review 0.38 0.38 0.59

Factor 3: Communication
S4Q3 Exchange with higher levels in Department 0.64 0.73
S4Q7 Exchange with own level in Department 0.64 0.54
S4Q2 Extent feel part of a group 0.56 0.48
S4Q 16 Exchange with other offices 0.48 0.58
S6Q5 Counseling partner familiar with work 0.46 0.54 0.66
S4Q4 Partners’ interest in opinions 0.45 0.40 0.64
SQ6 Manner raises % bonuses determined 0.43 0.64
s4Q5 Recognition&respect from firm members 0.52 0.41 0.46 0.76
S4Q 15 Opportunity to make suggestions 0.39 0.69

Factor 4: Promotion
S7Q 10 Opportunity for advancement 0.74 0.66
S7Q9 Manner promotions determined 0.70 0.66
S7Q6 Availability of upper level positions 0.66 0.59
S7Q3 Extent know when will be promoted 0.55 0.55
S6Ql Feedback received on performance 0.42 0.53 0.68
S6Q2 Knowing what people expect 0.40 0.57
S7Q2 Salary, including bonus 0.38 0.47 0.61
A THEORETICAL AND EMPIRICAL INVESTIGATION OF JOB SATISFACTION 301

APPENDIX A continued

Factors
123456789 10 11 ct

Factor 5: Supervision/evaluution
S5Ql Quality of supervision 0.63 0.66
S5Q3 Consistency of supervision 0.61 0.67
SGQlO Extent fairly evaluated 0.55 0.63
sQ9 Extent evaluation forms completed 0.49 0.42
S5Q2 Training in relation to work assigned 0.45 0.57
S5Q9 Nature of reviewers’ points 0.42 0.43
S5Q5 Timliness of review 0.38 0.38 0.59

Factor6: Office support


S3Q5 Typing services 0.75 0.67
S3Q4 Copying services 0.72 0.58
S3Q3 SecretariaVAdministrativesupport 0.71 0.60
S5Q7 Local office training programs 0.58 0.40 0.53
S7Q8 Fringe benefits (ins., pension, etc.) 0.49 0.46

Factor 7: Staffing
S6Q3 Extent of Rumor Mill in assignments 0.70 0.67
S5Q 12 Input in choosing assignments 0.59 0.68
S6Q4 Extent ofRumor Mill in evaluation 0.57 0.57
S5Q8 Equity in job assignments 0.54 0.53
S5Q 10 System of stalfing assignments 0.47 0.55
S5Q4 Advance notice of assignments 0.46 0.53

Factor 8: Appreciation/participation
S4Q 12 Partners’ expression of appreciation 0.75 0.74
S4Q 13 Consistency of support & appreciation 0.60 0.63
SQl Feedback received on performance 0.42 0.53 0.68
S4Q 15 Opportunity to make suggestions 0.39 0.46 0.69
S7Q4 Feeling of job security 0.46 0.52
S4Q4 Partners’ interest in opinions 0.45 0.40 0.64

Factor 9: Clientrange/assignments
S2Q7 Client range (size, type, industry) 0.66 0.62
S2Q7 Opportunity for variety in assignments 0.41 0.50 0.56
S2Q2 Gain experience & develop skills 0.55 0.45 0.58
S3QI 1 No. ofassignments at same time 0.48 0.42 0.59
S4Ql Extent ofprocedures, policies, rules 0.40 0.46

Factor 10: Financial


S7Q7 MBA pay diiferential 0.55 0.39
s7Q2 Salary, including bonus 0.38 0.47 0.61
S7Q5 Equity in salaries in level 0.45 0.38
S4Q 10 Channels of communication understood 0.4 1 0.47

Factor 11: Counseling/training


S6Q5 Counseling partner familiar with work 0.46 0.54 0.66
S6Q8 Guidance from counseling partner 0.52 0.53
S5Ql1 National Firm trainingprograms 0.49 0.33
SSQ- Local office training programs 0.58 0.40 0.53
302 MARlA LOMBARD1 BULLEN and ERIC G. FLAMHOLTZ

APPENDIX A continued

Factors
1 2 3 4 5 6 7 8 9 10 11 ct

Eigenvalues 19.6 3.8 2.7 2.6 2.0 1.8 1.5 1.5 1.2 1.1 1.1

Percent ofVariance 50.3 9.8 7.0 6.8 5.2 4.6 3.9 3.8 3.1 2.9 2.7

l The cutoff point for inclusion is a factor loading of at least 0.38. Items are shown in descending order of magnitude for
each factor. Factors are shown in the order from the computer analysis.
The questionnaire section and number are indicated in the far left column.

Ct = Communality.

You might also like