Accounting Research Center, Booth School of Business, University of Chicago

The Role of Accounting Data in Performance Evaluation, Budgetary Participation, and

Organizational Effectiveness
Author(s): Peter Brownell
Source: Journal of Accounting Research, Vol. 20, No. 1 (Spring, 1982), pp. 12-27
Published by: Wiley on behalf of Accounting Research Center, Booth School of Business, University
of Chicago
Stable URL: http://www.jstor.org/stable/2490760 .
Accessed: 17/06/2014 23:33

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley and Accounting Research Center, Booth School of Business, University of Chicago are collaborating
with JSTOR to digitize, preserve and extend access to Journal of Accounting Research.

http://www.jstor.org

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
Journal of Accounting Research
Vol. 20 No. 1 Spring 1982
Printed in U.S.A.

The Role of Accounting Data in

Performance Evaluation,
Budgetary Participation, and
Organizational Effectiveness
PETER BROWNELL*

1. Introduction
The relationship between managerial leadership styles and measur
of organizational effectiveness, such as performance of subordinates,
an unsettled issue in management accounting research. Despite prelim
nary evidence by Argyris [1952], and later by DeCoster and Fertak
[1968], that the effectiveness of budgetary systems is associated wi
supervisory leadership style, subsequent research has produced inconcl
sive findings. Two cases in point are the results of Hopwood [1972] an
Otley [1978]. Hopwood found that a leadership style characterized by
heavy emphasis on budget-related performance was significantly assoc
ated with job-related tension (JRT). He went on to speculate that JR
would be dysfunctional for performance. Otley, attempting to replica
and extend Hopwood's work, explored the relationship between leade
ship style and performance, hypothesizing that the tension-producin
style found by Hopwood would be associated with inferior subordina
performance. Although his results were inconclusive, they suggested th
superior performance levels were associated with a budget-focused lea
ership style.
This paper reports the results of a study undertaken in an attempt
reconcile the results of Hopwood and Otley. In the study, I hypothesize

*
Assistant Professor, Massachusetts Institute of Technology. [Accepted for publicatio
March 1981.]
12
Copyright ?, Institute of Professional Accounting 1

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

that directly observable associations between leadership evaluative sty

and performance should not be expected because the relationship will b
moderated by budgetary participation. The results confirm the hypoth
esis, indicating that a budget-focused leadership style is most effectiv
under conditions of high participation, but is ineffective where partic
pation is low.
In the next section, the operant conditioning (Skinner [1938]) an
balance (Heider [1946]) theoretical paradigms are presented to provid
the foundation for the expectations and hypotheses of the study. This
followed by sections dealing with the hypotheses, the research method
and an analysis of the results. Finally, I conclude with a discussion of th
limitations and implications of the study.

2. Some Theoretical Constructs

The basic principle of operant conditioning concerns the developmen
of stimulus/response (SIR) bonds in association with appropriate rei
forcement conditions. Usually, the reinforcement takes the form of
desired outcome or the avoidance of an undesired outcome. However,
has been shown (e.g., see Maddi [1972]) that secondary reinforcement
or outcomes which, while of no potential value in and of themselves, a
valued because of their past association with desired outcomes, can b
responsible for the development and maintenance of an SIR bond. Th
notion takes a familiar form in the expectancy theory model of motivatio
(see Ronen and Livingstone [1975]) where "first-" and "second-lev
outcomes" are arguments of the function.
The role of reward structure as a secondary reinforcer in the relatio
ship between budgetary participation and performance and job satisfa
tion was studied by Cherrington and Cherrington [1973] (CC). The
reasoned that a reward structure which was based on budget achievemen
would represent appropriate reinforcement only for individuals who a
largely responsible for the determination of budget targets, while a rewa
structure deemphasizing the budget would provide appropriate reinforc
ment only for individuals for whom the budget is largely imposed.
other words, rewards for budget-related performance should be compa
ible with the level of emphasis placed on budgets. In an experiment
task involving 230 undergraduate business students, CC obtained resul
which tended to confirm that performance and reported job satisfactio
were higher in the budget-based reward/high participation and no
budget-based reward/low participation groups. I believe that leadersh
evaluative style will operate in a similar fashion, in that superiors wh
exhibit a budget-constrained style and place primary emphasis in the
evaluation of subordinates on budget achievement, will provide appr
priate reinforcement only for those individuals who are heavily involve
and influential in the budget-setting process.
The theory of operant conditioning essentially provides the basis f

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
14 JOURNAL OF ACCOUNTING RESEARCH, SPRING 1982

the identification of appropriate and inappropriate degrees of emphas

on budget achievement in superior evaluative style. As an extensio
balance theory (Heider [1946]), part of a broader class of theories know
as consistency theories, provides a theoretical foundation for the pred
tion of the consequences of inappropriate degrees of emphasis. T
fundamental thrust of consistency theory is that individuals strive for
balanced or equilibrium cognitive structure, with unbalanced or disequ
librium situations construed as cognitive conflict (Brown [1965, cha
11]).
A subordinate is said to have a balanced cognitive structure if h
attitude toward his superior is (un)favorable and both he and his superi
maintain other attitudinal orientations that (dis)agree completely. Ther
fore, if the subordinate is unfavorably disposed toward the evaluati
style employed by his superior, and (as is reasonably assumed) t
superior is favorably disposed toward it, balance can prevail only if (
the subordinate's attitude toward high performance also conflicts wi
that of his superior (presumably, the superior is favorably dispos
toward high performance), and (b) the subordinate's attitude toward t
superior, himself, is unfavorable. In such circumstances, it is likely th
actual performance and reported satisfaction with job in general (a
with supervision in particular) will suffer compared with the alternati
balanced situation where the subordinate is favorably disposed towa
the superior's evaluative style (and, hence, toward the superior himse
and high performance also).
The two theoretical paradigms of operant conditioning and balan
theory formed the basis for positing the hypotheses in my study. Opera
conditioning suggests that the orientation of the subordinate toward t
evaluative style of his superior is dependent on the level of budgeta
participation in the planning phase. A favorable orientation toward tho
evaluative styles which place a high emphasis on budget results in t
-ontrol phase will occur only under conditions of high participatio
Similarly, a favorable orientation toward a low budget-emphasis sty
will occur only under conditions of low budgetary participation. A
unfavorable orientation thus results either from high budget-emphasi
low participation or low budget-emphasis/high participation.
The predicted consequences of differing orientations toward the sup
rior's evaluative style are based on balance theory. Specifically, bo
performance and reported levels of job satisfaction should be lower wi
inappropriate than with appropriate degrees of emphasis placed by t
superior on budget results.
In summary, the hypotheses in the present study will relate leadersh
evaluative style, budgetary participation, and their interaction with bo
performance and job satisfaction.
3. Hypotheses
The hypotheses can be stated in their null form as follows: Hi: The
is no significant interaction between supervisory evaluative style a

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

budgetary participation affecting performance. H2: There is no signif

cant interaction between supervisory evaluative style and budgeta
participation affecting job satisfaction.

4. Method
A questionnaire was administered to a sample of 48 managers draw
from a large San Francisco Bay Area manufacturing company. Th
managers were selected by the company management on the basis
availability at the time the research was conducted. The average age
the respondents was 43.6 years, and the average tenure in their curre
position was 10.3 years. The managers represented eight separate fun
tional divisions of the organization, each of which was involved in on
phase of the production or distribution activities of the business. A
managers held cost-center responsibility for their respective organiz
tional subunits, but preliminary meetings with plant management ind
cated that technological, regulatory, and environmental characteristic
varied somewhat across cost centers leading to the use of a variety
managerial approaches throughout the organization.' The questionnair
elicited information on four variables: two independent variables (eval
ative style and budgetary participation) and two dependent variabl
(performance and job satisfaction).

The Independent Variables

1. EVALUATIVE STYLE

To measure evaluative style, I employed the eight-dimensional inde

first developed by Hopwood [1971], and later used by Otley.2 Eac
manager was asked to rank-order eight performance criteria according
the relative importance placed on each criterion by supervisors in the
evaluation. Their rankings were categorized according to the four eval
ative styles of Hopwood, based on two of the criteria:3
(i) Budget-Constrained Style (BC)-meeting the budget, but n
concern with costs, ranked among the top three criteria.
(ii) Budget-Profit Style (BP)-both meeting the budget and concer
with costs ranked among the top three criteria.
(iii) Profit-Conscious Style (PC)-concern with costs, but not meetin
the budget ranked among the top three criteria.
(iv) Non-Accounting Style (NA)-neither meeting the budget n
concern with costs ranked among the top three criteria.
Table 1 presents the results of the use of the evaluative style index an
' This is an important consideration since a focus on a single organization can substa
tially reduce the variance of the measured variables across the sample, rendering hypothe
testing extremely difficult.
2 The version of this index used by Otley differed slightly because of the use of pro

center rather than cost-center managers in his study. See Otley [1978, p. 147].
3 The other six criteria were job-effort, concern with quality, ability to handle m

attitude toward work and company, coworker relationships, and collegial cooperation.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
16 PETER BROWNELL

TABLE 1
Total Numbers and Percentages of Respondents Perceiuing Each Eualuative Style
Hopwood, Otley, and Present Study
Style of Evaluation
Sample
BC BP PC NA

Hopwood (N = 167) .............. 20 10 26 44

Otley (N= 39) .................. 13 56 28 3
Present study (N = 38*) .......... 21 24 24 31
* As explainedlater, this was the largestsamplesize actuallyused for hypothesistesting.

TABLE 2
Rotated Factor Loadings from Milani Measure
Item Factor1 Factor2 Communal
Extent of involvement in budget setting 0.824 0.166 0.707
Reasoning given by superior for budget
revisions .0.091 0.961 0.932
Frequency of giving opinions to superior
about the budget .0.736 -0.246 0.602
Degree of influence on final budget 0.888 0.248 0.850
Degree of importance of input to budget 0.860 0.214 0.785
Frequency of being asked for opinions
about the budget by the superior 0.902 0.001 0.814
Eigenvalues .3.562 1.037
Explained variance 60.9% 17.2% 78.1%

provides comparative data from both Hopwood's and Otley's use of

index. Interestingly, the distribution of respondents in my study is m
similar to Hopwood's distribution than to Otley's, consistent with
notion raised by Otley [1978, p. 143 and 1980, p. 415] that type
responsibility center may influence evaluative styles chosen by superio
For purposes of analysis, the four evaluative styles were collapsed i
two classes, a high budget-emphasis (BC and BP) and a low budg
emphasis (PC and NA ).5

4 A chi-square test indicates no significant difference between the distribution in H

wood's study and mine (X2 = 5.72, df = 3, not significant). By contrast, my distributio
significantly different from Otley's (X2 15.62, df = 3, p < 0.01). Apparently, profit-ce
managers (Otley's sample) are evaluated much more on the basis of accounting informa
than are cost-center managers (Hopwood's sample and the sample in this study), as refle
in the dominance of the nonaccounting style in Hopwood's and my samples, of only tr
importance in Otley's sample.
5 Following the collapsing of the four categories into two, the difference between

distribution of respondents in Hopwood's study and that in the present study rem
insignificant (X2 = 2.44, df = 1, not significant). The difference between the distributio
Otley's study and that in the present study is smaller (and not quite significant atp '
after collapsing the categories (X2 = 3.77, df = 1, p < 0.06).

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

TABLE 3
Descriptive Statistics for Milani and Mahoney Measures

Std. Theoretical Range Actual Range

Meastlle Meall D
Dev'n Minimum Maximum Minimum Maxinum
Milani 25.79 7.10 6 42 13 39
Mahoney
Dimensions:
Overall . . 6.84 0.95 1 9 5 8
Planning . 6.58 1.48 1 9 2 9
Investigating 6.50 1.66 1 9 2 8
Coordinating 6.84 1.37 1 9 3 9
Evaluating . 6.45 1.46 1 9 3 9
Supervising . 7.53 1.01 1 9 5 9
Staffing .. 6.26 1.62 1 9 3 9
Negotiating 4.84 2.10 1 9 1 9
Representing 3.95 2.44 1 9 1 8

2. BUDGETARY PARTICIPATION

Budgetary participation was measured using both the Hofstede [196

and Milani [1975] approaches in order to obtain some cross-validatio
They correlated at +0.74. Since the multiitem structure of the Mila
measure permits a reliability assessment, it was used for the hypothes
testing. To assess its reliability, the Milani-measure responses were fact
analyzed. As a multiitem additive scale, one would hope that all six item
in the measure load on a single factor. The analysis led to the emergen
of two eigenvalues greater than unity, therefore extracting two comp
nents. The rotated factor loadings are presented in table 2, and the resul
reveal that with the exception of item 2, all items in the measure load
a single factor. In order to provide comparability of my results with tho
of Milani, I constructed participation scores using the responses to all s
items in the measure, as he did. The upper section of table 3 presen
descriptive statistics for the Milani measure.

The Dependent Variables

1. PERFORMANCE

A self-rating version of the Mahoney et al. [1963; 1965] nine-dimen

sional performance measure was used to gather data on the performanc
This approach was chosen because a self-rating rather than superio
rating version of the measure overcomes the problem of "halo error
which seems to occur with superior's ratings (Thornton [1968]; Nealy an
Owen [1970]). Also, the nine-dimensional structure of the measure clear
captures the multidimensional nature of performance without introducin

("5"Haloerror" is the tendency to evalute "globally" or, in other words, to evaluate

only one cognitive dimension. A high intercorrelation among separate dimensions is e
dence of halo error.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
18 PETER BROWNELL

the problem of excessive dimensionality. Kavanagh et al. [1971], fo

example, obtained disappointing results on a discriminant validity test o
a twenty-dimension performance rating scale employed by them. Inde
pendent assessments of reliability and validity of the Mahoney instru
ment have provided supportive evidence of the measure's sound devel
opment (Penfield [1974]; Heneman [1974]).
The nine-dimensional structure of the Mahoney measure includes
single overall performance rating, together with ratings on eight subdi
mensions.7 As a preliminary test of the consistency of the ratings,
regressed the overall rating on the ratings of the eight separate dimen
sions. The regression explained 60.8 percent of the variance in the overal
ratings. This result is consistent with Mahoney's early work, where it wa
found that approximately 55 percent of the functions critical to effective
managerial performance were common to the 452 managerial assignments
in 13 companies studied, while approximately 45 percent were job specific
(Mahoney et al. [1963, pp. 106-7]). The lower section of table 3 presents
descriptive statistics for the Mahoney measure.
2. JOB SATISFACTION

Robinson et al. [1969] cataloged no less than 16 different instruments

employed in the literature to measure job satisfaction. The number of a
hoc measures used in single studies defies estimation. However, the us
of reliability and validity criteria in the selection substantially reduce
the number of feasible alternatives. For my study, I considered both the
Job Descriptive Index (JDI) (Smith et al. [1969]) and the Minnesota
Satisfaction Questionnaire (MSQ) (Weiss et al. [1967]), each of which
has been carefully developed and extensively tested in the field (Gille
and Schwab [1975, p. 313]). I chose the MSQ because of the availability
of substantial norm group data and the fact that it has outperformed th
JDI in tests of discriminant validity (Dunham et al. [1977]). On th
question of reliability, Weiss et al. provide Hoyt analysis of varianc
reliability coefficients both for an entire sample of 27 norm groups an
for a specific group of managers. A comparison of the reliability coeffi
cients from my results and those of the other two sets of data is presented
in table 4. While the coefficients obtained from my data are satisfactory
per se, they are generally lower than the coefficients obtained from th
two comparison groups reported by Weiss et al.
Table 5 presents descriptive statistics for the present sample on each
of the 20 subscales of the MSQ, together with the overall index of job
satisfaction.

5. Results
Completed questionnaires were collected from all 48 managers. Tw
sets of responses were omitted due to improper completion of both th

7The eight subdimensions are planning, investigating, coordinating, evaluating, supe

vising, staffing, negotiating, and representing.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

TABLE 4
Hoyt Analysis of Variance Reliability Coefficients for Three Samples of MSQ
Respondents

Subscale Present Other 27 Norm

Study Managers Groups

Ability utilization .......... .72 .92 .91

Achievement .............. .76 .73 .84
Activity ................... .60 .81 .86
Advancement .............. .68 .96 .93
Authority ................. .70 .91 .85
Company policies .......... .71 .87 .90
Compensation ............. .74 .95 .91
Coworkers ................. .50 .67 .85
Creativity ............. .... .71 .87 .87
Independence .............. .74 .73 .85
Moral values .............. .66 .77 .81
Recognition ............... .80 .96 .93
Responsibility ............. .77 .85 .78
Security ................... .66 .78 .80
Social service .............. .64 .89 .89
Social status ............... .63 .76 .79
Supervision-human ....... .81 .90 .89
Supervision-technical ..... .80 .71 .86
Variety ............... ... .70 .85 .86
Working conditions ......... .77 .94 .89
GENERAL SATISFACTION ..... .72 .85 .88

Mahoney measure and the MSQ (leaving N = 46 for H1 and H2), and a
additional six were omitted due to evidence of response-set bias
completion of the MSQ only (leaving N = 40 for H2). A disturbing eig
sets of responses were also omitted due to improper completion of th
ranking called for on the Hopwood evaluative style measure, reducin
the final sample sizes for testing of H1 and H2 to 38 and 32, respectively

PERFORMANCE H1
The following equation was developed for purposes of testing H1:
Y=31 +/82X+f3Z+ f4XZ (

where Y is performance
X is budgetary participation
Z is a binary variable,
Z f -1 for high budget-emphasis
Z
+1 for low budget-emphasis, and
XZ is the interaction between participation an
budget-emphasis.
Deviation scores from the overall mean were substituted for the ra
scores on the Milani measure. The effect of the subtraction procedu
was to code low participation scores positively (smaller raw scores we

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
20 PETER BROWNELL

TABLE 5
Descriptive Statistics froomMSQ Administration
Subscale Mean Std. Dev'n Minimum Maximum
Ability utilization ........... 20.03 2.76 13 25
Achievement ............... 20.84 2.80 12 25
Activity ................... 20.53 2.98 10 25
Advancement .............. 18.34 3.64 10 25
Authority .................. 19.91 2.22 16 25
Company policies ........... 17.88 3.41 7 23
Compensation . ............. 18.16 3.47 12 25
Coworkers ................. 19.47 2.92 11 25
Creativity ................. . 20.25 3.45 10 25
Independence .............. 17.91 2.13 15 22
Moral values ............... 20.63 2.76 16 25
Recognition ................ 17.97 4.05 9 25
Responsibility .............. 20.88 2.31 15 25
Security ................... 21.19 2.68 17 25
Social service .............. 19.09 2.79 12 23
Social status ............... 17.22 2.03 14 22
Supervision-human ........ 18.59 4.20 10 25
Supervision-technical ...... 19.16 3.52 11 25
Variety .................... 1.9.72 2.61 11 25
Working conditions ......... 19.03 3.30 10 25
GENERAL SATISFACTION ..... 76.50 7.25 58 93
The theoretical range for the 20 subscales is 5-25, while for general satisfaction it is 20-100.

subtracted from the mean) and to code high participation scores ne

tively (larger raw scores subtracted from the mean). This provided
clearer basis for predicting the sign of 84, the coefficient for the interact
term. Recall that the hypothesis is that the low participation/low budg
emphasis and high participation/high budget-emphasis combinations
both expected to be associated with high performance. The product ter
XZ, would be positively signed in both of these cases. The other t
combinations, low participation/high budget-emphasis and high part
pation/low budget-emphasis were expected to be associated with low
performance, with the product term, XZ, negatively signed in each ca
Hence, the basic question of the significance of the interaction could
tested by determining whether ABwas significantly different from zero
either direction. For an interaction of the predicted form, however
significant, positive ,4 was actually expected.
The results of this regression are presented in table 6. ,4 is significan
positive, indicating rejection of the null hypothesis in the expec
direction. 8L3, the coefficient for evaluative style, is less significant (p
0.10), but still suggestive of the tendency for higher performance to
associated with a deemphasis on the budget in supervisory evaluat
8 To insure a proper specification of the interaction term in equation (1), X and Z w

tested for independence. Using a chi-square test of the sign of the participation score (+
low participation, - for high participation) in each of the two budget-emphasis conditi
the null hypothesis of no relationship could not be rejected at the 0.05 level (X2 = 3.0
= 1, not significant).

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

TABLE 6
Results of Regression: Hypothesis Test Hi
Coefficient Value Standard Error t Probability

JBI 6.71 0.14 48.92 <0.01

/2 -0.09 0.02 -4.32 <0.01
/33 0.19 0.14 1.40 <0.10
34 0.04 0.02 2.22 <0.025
R2 = 0.38
df= 34

style. This result is consistent with Hopwood [1972], but contradictory

Otley [1978], whose results suggested that better performance tended
be associated with a budget-constrained style.9 Participation (8B2) is high
significant (p < 0.01), indicating, as is often suggested in the literatur
that higher participation is associated with higher performance. Howeve
this is in contrast to Milani's results, which indicated only very we
support of a positive relationship between participation and performanc
using the same participation measure. While the interaction of parti
pation and evaluative style is important, the results here indicate th
participation alone exerts a powerful direct influence on performance.
To explore the nature of the interaction further, I employed t
Johnson-Neyman technique,10 which establishes a region of values f
the continuous independent variable (budgetary participation) with
which no significant differences in the dependent variable could
ascribed to the categorical variable (budget-emphasis in evaluative style
The test makes use of the slopes and intercepts derived from tw
regressions of performance on the recoded budgetary participation scor
one regression for each of the two categories of evaluative style.11 T
derived range of values for this test was 2.73 to -48.66 (recall th
following their transformation, the participation scores used in equati
(1) have a mean of zero with high participation negatively signed and lo
participation positively signed). Considering the standard deviation
participation (ax = 7.10), these results indicate that for scores larger th
X + 0.38 ax (i.e., low participation), performance is significantly high
under conditions of a deemphasis on the budget in evaluative style. T
intersection of the two curves occurs at X = -4.36 (X - 0.61ax), suggesti
that for larger negative scores than this (i.e., high participation), perfor
ance is superior under conditions of a heavy emphasis on the budget
evaluative style. However, the difference between performance leve
under the two evaluative styles does not become significant until X
' It should be noted, however, that the performance measure used in this study (s
ratings) differs from that used by Otley (formal evaluations obtained from the participa
company).
1See, for example, Kerlinger and Pedhazur [1973, pp. 256-58].
" It is not necessary to perform these two regressions since the slopes and intercepts
each can be derived from the coefficients produced from the multiple regression in equat
(1), given Z= -1 or +1.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
22 PETER BROWNELL

e r fo rmrancne

- 0.04.1 5

I - {- Elm Lou buoate-et - a

a id I' ( PC and d NA)

* --0.1296

6 51

High budaet-emphea
: * (BC and EN)

-48.66 -4. 36 C 2.73

High Participation Participation Low Participation
Mean

FIG. 1.-Results of Johnson-Neyman test.

-48.66 (X-6.85ux).12 These results are depicted in figure 1. The hatch

area in figure 1 represents the range of participation scores within wh
no significant differences in performance can be attributed to evaluat
style. Following Kerlinger and Pedhazur [1973, p. 248], the presence o
point of intersection within the research range of interest (i.e., not
removed from the mean score for participation) suggests a disordi
interaction, implying that the "preferred" (in terms of performan
evaluative style switches at a point which lies within this range.
What is suggested by these results is that performance differen
resulting from the use of alternative evaluative styles can be bet
explained when budgetary participation is introduced as a moderat
factor. In particular, the undesirable consequences of an excessive e
phasis on budgets in evaluative style, suggested by Hopwood [1972],
ameliorated by a suitably high level of budgetary participation, a res
which may tend to explain why Otley [1978] obtained disappoint
results when studying the direct effects of evaluative style on perfor
ance.

FURTHER ANALYSIS

In order to explore the effects of the evaluative style/participati

interaction on the various subdimensions of performance, regressi
similar to equation (1) were performed for each of the eight subsca
from the Mahoney measure. The results of these eight regressions
summarized in table 7.
Some interesting results emerge. Note, in particular, the strong effe

12 This value of -48.66 translates into a raw score for participation of 74.45, whic

somewhat of a fictitious measure given that the feasible range for the measurem
instrument was 6-42.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION 2
TABLE 7
Results of Regressions Using Performance Subscales' t-Statistics

Dimension 8 83 8 R2
t t
Planning ............. -2.21** -0.04 0.74 0.15
Investigating ......... -2.28** 2.46*** 2.56*** 0.27
Coordinating ......... -0.98 0.67 0.63 0.04
Evaluating ........... -1.47 0.30 1.28 0.09
Supervising .......... -0.99 0.68 0.96 0.05
Staffing .............. -1.72* 1.10 0.23 0.09
Negotiating .......... -2.92*** 0.00 --0.27 0.24
Representing ......... -3.05*** -1.03 -0.26 0.31
p < 0.05.
**p < 0.025.
P < 0.01.

TABLE 8
Results of Regression: Hypothesis Test H2
Coefficient Value Standard Error t Probability

JBI 76.48 1.44 53.09 <0.01

/2 -0.19 0.21 -0.91 N.S.
f3 -0.79 1.44 -0.55 N.S.
/4 0.05 0.21 0.21 N.S.
R= 0.06
df= 28

of all three variables (evaluative style, participation, and their interaction

on the subscale of "investigating." This subscale provides the stronge
support for Hopwood's contention that a budget-constrained style ca
directly result in adverse consequences. It is also true, however, th
participation directly affects performance on this dimension, as does th
participation/evaluative style interaction. The strong direct effects
participation on performance in the areas of "negotiating," "planning
and "representing" are worthy of note also. These results are intuitivel
appealing.
JOB SATISFACTION H2

Equation (1) was again employed as the basis for testing H2, th
hypothesis of interaction affecting job satisfaction. The results of th
regression are presented in table 8.
The results shown in table 8 do not permit rejection of H2. None of th
independent variables has significant coefficients. However, the resul
of using the 20 subscales of the MSQ for further testing produce som
interesting relationships. Table 9 presents the results of 20 regressions
the form of equation (1), each of which employs one of the 20 subscale
as the dependent variable.
The significant coefficients for participation for "ability utilization
"achievement," "creativity," "recognition," "responsibility," and "soci

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
24 PETER BROWNELL

TABLE 9
Results of Regressions Using Satisfaction Subscales' t-Statistics

Dimension 2 8.:3 4 R2
t t t
Ability utilization -2.64** 0.81 1.82* 0.25
Achievement . . -1.70* 0.21 0.50 0.11
Activity . . -1.10 -0.42 -0.70 0.09
Advancement -0.96 0.02 0.09 0.04
Authority -1.01 0.02 -0.99 0.08
Company policy & practice -0.42 -1.31 1.26 0.14
Compensation ..-0.28 -0.31 -1.33 0.07
Coworkers -0.11 - 1.81* 0.23 0.13
Creativity -2.50*** 1.20 1.07 0.20
Independence . .-0.60 -0.07 0.70 0.03
Moral values . .-1.22 -0.26 -0.38 0.08
Recognition . . -1.96* -0.78 0.15 0.20
Responsibility . . -1.75* 0.63 0.77 0.11
Security .......................... -1.17 -0.63 1.26 0.13
Social service . . -1.26 -0.11 1.14 0.10
Social status . . -2.07** -0.49 0.66 0.19
Supervision-human relations -0.83 -0.60 0.65 0.07
Supervision-technical . -1.58 -0.58 -0.03 0.13
Variety -1.65 0.03 -0.20 0.11
Working conditions 0.24 -1.90* 1.40 0.18
p < 0.05.
*p < 0.025.
***p < 0.01.

status" are not surprising, since high participation should be associate

with higher reported satisfaction with these individual job facets. Th
strongest results for evaluative style occur with "working conditions" an
"coworkers." The latter is particularly interesting since it suggests th
a budget-constrained style of evaluation is associated with higher r
ported satisfaction with coworkers. This result seems inconsistent with
conclusion of Argyris [1952], who suggested that one means of relievin
the pressure associated with a heavy emphasis on budgets in evaluatio
was to blame other workers, with the consequence of increased intrad
partmental strife. While only one of the interaction coefficients is signi
cant ("ability utilization"), all but six are in the predicted direction.
sign test (Siegel [1956]) shows that the probability of this result
approximately 0.06, suggesting a general tendency for the interactio
terms to be as predicted.

DISCUSSION
The results of this study indicate that the impact of supervisor
evaluative style on performance is moderated by budgetary participatio
which, in turn, exerts a substantial positive influence on performanc
Since no conclusive results emerged regarding job satifaction, this d
cussion is confined to a consideration of the performance-related result

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

The fact that evaluative style exerted no direct influence on perfor

ance by itself was not surprising. Otley [1978] pointed to several facto
which might explain this result, including the type of responsibility cent
involved (profit- vs. cost-centers), the degree of staff supportiveness, a
the specific environmental conditions.
My results suggest that a budgetary participation is a moderati
influence which is consistent with Otley's last factor. Hayes [1977] h
indicated that accounting information is less appropriate as a foc
element in organizational control with increasing exposure of the org
nization, or subunits of it, to the environment. Hopwood [1973, p. 1
made a similar argument, suggesting that as environmental exposu
increases, those behaviors necessary to accomplish organizational obje
tives are decreasingly captured by the formal accounting system a
excessive reliance on accounting information can be dysfunctional in su
cases.'3 In these circumstances, participation in the budgeting proce
may help to ensure that the formal accounting system captures on
organizationally functional information, legitimizing a heavier emphas
on budget results.
Like all other studies, this one has its limitations and weaknesse
several of which deserve mention. First, caution must be exercised
avoid overly strong statements of causation, in particular, the directi
of causation. While most studies implicitly view style of leadership as
independent variable with effects on such dependent variables as pe
formance, a reverse linkage might be conceivable. That is, it is feasib
that a high-performance organizational unit will permit one type
leadership style, while a unit experiencing substandard performan
might command a different style. Indeed, Goodstadt and Kipnis [197
have shown that effective work groups will elicit generally supporti
behavior from superiors while less effective groups will generate clo
supervision.
Second, the measurement instruments used here were crude. Measur
ments of budgetary participation are still in need of refinement. T
factor structure of the Milani measure (see table 2) suggests a multid
mensional construct whose properties are relatively unknown. Hopwood
measure of evaluative style also has problems, as pointed out by Kah
[1972].
Third, the present sample was drawn nonrandomly from a sing
corporation, thus limiting the generalizability of the results.
With these limitations in mind, however, one can say that the resu
of this study cast doubt on the direct effects of leadership evaluative sty
on measures of organizational effectiveness. Instead, one finds that bud
etary participation has a significant moderating effect on any su
relationship.
" Otley [1978,
pp. 124-25] provides a diagrammatic representation of this argument.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
26 PETER BROWNELL

REFERENCES
ARGYRIS,C. The Impact of Budgets on People. New York: Controllership Foundati
1952.
BROWN,R. Social Psychology. New York: The Free Press, 1965.
CHERRINGTON, D. J., ANDJ. 0. CHERRINGTON. "Appropriate Reinforcement Contingenc
in the Budgeting Process." Empirical Research in Accounting: Selected Studies, 19
Supplement to Journal of Accounting Research 11: 225-53.
DECOSTER,D. T., ANDJ. P. FERTAKIS."Budget-Induced Pressure and its Relationship
Supervisory Behavior." Journal of Accounting Research (Autumn 1968): 237-46.
DUNHAM,R. B., F. J. SMITH,AND R. S. BLACKBURN. "Validation of the Index of Orga
zational Reactions with the J.D.I., the M.S.Q. and Faces Scale." Academy of Manageme
Journal (September 1977): 420-32.
GILLET,B., AND D. P. SCHWAB."Convergent and Discriminant Validities of Correspond
Job Descriptive Index and Minnesota Satisfaction Questionnaire Scales." Journal
Applied Psychology (February 1975): 313-17.
GOODSTADT, B., AND D. KIPNIS."Situational Influences on the Use of Power." Journal
Applied Psychology (June 1970): 201-7.
HAYES,D. "The Contingency Theory of Managerial Accounting." The Accounting Revi
(January 1977): 22-39.
HEIDER, F. "Attitudes and Cognitive Organization." Journal of Psychology (January 19
107-12.
HENEMAN, H. G. "Comparisons of Self and Superior Ratings of Managerial Performanc
Journal of Applied Psychology (March 1974): 638-42.
HOFSTEDE, G. H. The Game of Budget Control. Assen: Van Gorcum, 1967.
HOPWOOD, A. G. "An Accounting System and Managerial Behaviour." Ph.D. dissertati
University of Chicago, 1971.
."An Empirical Study of the Role of Accounting Data in Performance Evaluatio
Empirical Research in Accounting: Selected Studies, 1972. Supplement to Journal
Accounting Research 10: 156-82.
. An Accounting System and Managerial Behaviour. London: Saxon House, 197
KAHN, R. L. "Discussion of Hopwood's 'An Empirical Study of the Role of Account
Data in Performance Evaluation'." Empirical Research in Accounting: Selected Studi
1972. Supplement to Journal of Accounting Research 10: 183-86.
KAVANAGH, M. J., A. C. MACKINNEY, AND L. WOLINS."Issues in Managerial Performan
Multitrait-Multimethod Analyses of Ratings." Psychological Bulletin (January 19
34-49.
KERLINGER, F. N., AND E. J. PEDHAZUR. Multiple Regression in Behavioral Resear
New York: Holt, Rinehart & Winston, 1973.
MADDI, S. R. Personality Theories: A Comparative Analysis. Homewood, Ill.: The Dor
Press, 1972.
MAHONEY, T. A., T. H. JERDEE, AND S. J. CARROLL.Development of Managerial Perfor
ance: A Research Approach. Cincinnati: South-Western Publishing Co., 1963.
."The Jobs of Management." Industrial Relations (February 1965): 97-110.
MILANI,K. W. "The Relationship of Participation in Budget-Setting to Industrial Sup
visor Performance and Attitudes: A Field Study." The Accounting Review (April 19
274-84.
NEALY, S. M., AND T. W. OWEN. "A Multitrait Multimethod Analysis of Predictors a
Criteria of Nursing Performance." Organizational Behavior and Human Performan
(July 1970): 348-65.
OTLEY, D. T. "Budget Use and Managerial Performance." Journal of Accounting Resear
(Spring 1978): 122-49.
. "The Contingency Theory of Management Accounting: Achievement and Pr
nosis." Accounting, Organizations and Society (December 1980): 413-28.
PENFIELD, R. V. "Time Allocation Patterns and Effectiveness of Managers." Personn

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions
 ACCOUNTING DATA IN PERFORMANCE EVALUATION

Psychology (Summer 1974): 245-55.

ROBINSON, J. P., R. ATHANASIOU, AND K. B. HEAD. Measures of Occupational Attitud
and Occupational Characteristics. Ann Arbor: Survey Research Center, Institute
Social Research, University of Michigan, 1969.
RONEN, J., AND J. L. LIVINGSTONE. An Expectancy Theory Approach to the Motivation
Impact of Budgets." The Accounting Review (October 1975): 671-85.
SIEGEL, S. Non-Parametric Statistics for the Behavioral Sciences. New York: McGra
Hill, 1956.
SKINNER, B. F. The Behavior of Organisms. New York: Appleton-Century-Crofts, 1938
SMITH, P. C., L. M. KENDALL, AND C. L. HULIN. The Measurement of Satisfaction in Wo
and Retirement. Chicago: Rand McNally, 1969.
THORNTON, G. C. "The Relationship Between Supervisory- and Self-Appraisals of Exec
tive Performance." Personnel Psychology (Winter 1968): 441-56.
WEISS, D. J., ET AL. Manual for the Minnesota Satisfaction Questionnaire. Minneso
Studies in Vocational Rehabilitation, vol. 22. Minneapolis, 1967.

This content downloaded from 80.198.48.48 on Tue, 17 Jun 2014 23:33:22 PM

All use subject to JSTOR Terms and Conditions