Description
Economic theory suggests that multiple financial and non-financial measures (i.e., a strategic
performance measurement system ‘‘SPMS”) be used in compensation contracting to
properly direct employees’ attention and motivate behavior aligned with organizational
goals. Conversely, linking incentives to the SPMS can result in various dysfunctional behaviors,
including game playing by employees, the achievement of unbalanced performance,
and the potential of basing compensation on an incomplete performance measurement
system. Prior literature has investigated the use of subjectivity in compensation contracting
as a means of potentially mitigating these problems; however, subjectivity can introduce
other problems including claims of favoritism and bias. Economic theorists have
recently begun expanding the traditional agency model to include the notion of fairness
or justice. In this study, we obtain data from an organization that uses an SPMS as the basis
for the allocation of bonuses and investigate whether characteristics of the SPMS are associated
with perceived organizational fairness. Specifically, we hypothesize and show that
the extent to which employees perceive that the SPMS reflects a strategic causal model
and the degree to which it is technically valid are positively associated with their perceptions
of organizational justice. We also provide evidence that heightened levels of organizational
justice are the mechanism though which the perceived characteristics of the SPMS
are associated with employee performance. The implication is that firms do not necessarily
need to introduce subjectivity into the incentive contracting system, but can enhance performance
by linking incentive contracts to their SPMS if the system contains characteristics
that enhance employees’ perceptions of justice.
A path model examining the relations among strategic performance
measurement system characteristics, organizational justice,
and extra- and in-role performance
Laurie L. Burney
a
, Christine A. Henle
b
, Sally K. Widener
c,
*
a
Adkerson School of Accountancy, Mississippi State University, P.O. Box EF, Mississippi State, MS 39762, United States
b
College of Business Administration, University of Rhode Island, Ballentine Hall, 7 Lippitt Road, Kingston, RI 02881, United States
c
Jesse H. Jones Graduate School of Management, Rice University, Room 331-MS531, 6100 Main Street, Houston, TX 77005, United States
a b s t r a c t
Economic theory suggests that multiple ?nancial and non-?nancial measures (i.e., a strate-
gic performance measurement system ‘‘SPMS”) be used in compensation contracting to
properly direct employees’ attention and motivate behavior aligned with organizational
goals. Conversely, linking incentives to the SPMS can result in various dysfunctional behav-
iors, including game playing by employees, the achievement of unbalanced performance,
and the potential of basing compensation on an incomplete performance measurement
system. Prior literature has investigated the use of subjectivity in compensation contract-
ing as a means of potentially mitigating these problems; however, subjectivity can intro-
duce other problems including claims of favoritism and bias. Economic theorists have
recently begun expanding the traditional agency model to include the notion of fairness
or justice. In this study, we obtain data from an organization that uses an SPMS as the basis
for the allocation of bonuses and investigate whether characteristics of the SPMS are asso-
ciated with perceived organizational fairness. Speci?cally, we hypothesize and show that
the extent to which employees perceive that the SPMS re?ects a strategic causal model
and the degree to which it is technically valid are positively associated with their percep-
tions of organizational justice. We also provide evidence that heightened levels of organi-
zational justice are the mechanism though which the perceived characteristics of the SPMS
are associated with employee performance. The implication is that ?rms do not necessarily
need to introduce subjectivity into the incentive contracting system, but can enhance per-
formance by linking incentive contracts to their SPMS if the system contains characteristics
that enhance employees’ perceptions of justice.
Ó 2008 Elsevier Ltd. All rights reserved.
Introduction
‘‘An organization that is performing well is one that is
successfully attaining its objectives” (Otley, 1999, p. 364).
Recent trends indicate that many organizations pursue
these objectives by implementing a strategic performance
measurement system (SPMS) in which multiple ?nancial,
operational, and strategic measures are selected to trans-
late and communicate the ?rm’s strategy to its employees
(Ittner, Larcker, & Randall, 2003). An organization’s long-
term success depends on individual employee actions;
likewise, an SPMS provides employees with useful infor-
mation to facilitate behaviors that translate to organiza-
tional success (Otley, 1999; Schiemann & Lingle, 1997).
In this study, we investigate whether employees’ percep-
tions of two characteristics of an SPMS used as the basis
for incentive compensation affect their perception of orga-
nizational justice, or fairness. Investigating employees’
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.11.002
* Corresponding author. Fax: +1 713 348 6331.
E-mail address: [email protected] (S.K. Widener).
Accounting, Organizations and Society 34 (2009) 305–321
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
perceptions of fairness in an organizational context is
important since they are a primary antecedent of impor-
tant work-related outcomes, including employee satisfac-
tion, organizational commitment, organizational
citizenship behavior, and job performance, that ultimately
affect organizational success (Johnson, Selenta, & Lord,
2006). The management accounting literature has begun
studying fairness in the use of ?nancial versus non-?nan-
cial measures (Lau & Sholihin, 2005) and the applicability
to the controllability principle (Giraud, Langevin, & Men-
doza, 2008).
1
These studies support the contention that con-
sideration of fairness issues is relevant. We propose that
perceptions of fairness are especially relevant in the com-
plex SPMS arena.
Organizations use SPMSs to facilitate decision-making
and to in?uence employee behaviors (Sprinkle, 2003).
Agency, expectancy, goal setting, and reinforcement theo-
ries
2
predict a positive relation between incentives and per-
formance (Jenkins et al., 1998; Bonner & Sprinkle, 2002);
however, some researchers suggest that incentives can
cause employees to over invest efforts in one area at the ex-
pense of other areas (Beer & Cannon, 2004). Moreover,
poorly designed incentive schemes can produce undesirable
work efforts (Locke & Latham, 1990). Indeed, empirical re-
sults investigating the effects on performance are mixed
(Jenkins et al., 1998; Bonner & Sprinkle, 2002). One reason
for these mixed results is con?ict regarding the ideal design
of the incentive plan.
A formulaic incentive plan is problematic because any
one measure is incomplete. For example, if a principal con-
tracts with an agent on the basis of pro?t, the agent may
shirk with regards to efforts with long-term payoffs. With
an SPMS incentive plan (henceforth, we mean one in which
an SPMS is used to determine incentives), ?rms can assign
weights to each measure to calculate incentives (Ittner,
Larcker, & Meyer, 2003). Although an SPMS incentive plan
may be an improvement over a single measure, formulaic
system, employees may still engage in game playing to
maximize compensation or the system may result in
unbalanced performance (Ittner, Larcker, & Meyer, 2003;
Kaplan & Norton, 1996). To counteract the negative effects
of a formulaic incentive system, an alternative is to allow
for subjectivity in the evaluation process (e.g., Budde,
2007); however, this option may foster feelings of favorit-
ism or bias (e.g., Moers, 2005).
We investigate another alternative and suggest that
there are system attributes for which users will form per-
ceptions that in turn facilitate more effective behaviors
(Whyte, Bytheway, & Edwards, 1997). Employees behave
based on how they perceive the system; not necessarily
on the actual system design (Robbins, 2003). Therefore, it
is important for research to examine perceptions to under-
stand subsequent employee and organizational outcomes
(Robbins, 2003).
3
Two important characteristics of an SPMS
are that it (1) re?ects a strategic causal model and (2) is
technically valid (Kaplan & Norton, 1996). Strategic align-
ment translates the organization’s strategy to the employ-
ees, shows relations among the measures, and
demonstrates how actions will result in outcomes neces-
sary for achieving desired goals (Kaplan & Norton, 1996).
Technical validity ensures that the measures are accurate,
accessible, understandable, reliable, and timely. In the next
section, we develop hypotheses that employees’ percep-
tions of these two characteristics will foster perceptions
of organizational justice. If employees perceive that they
work in a fair environment, they are less likely to engage
in game playing and more likely to perform at a higher le-
vel. This proposition is consistent with proponents of
incentive plans who suggest that good system design and
implementation coupled with perceptions of fairness can
overcome negative aspects of incentive plans (Beer & Can-
non, 2004). Thus, our primary research question is whether
employees’ perceptions of two attributes of an SPMS
incentive plan are associated with organizational fairness,
and in turn, are positively associated with more effective
employee outcomes in the form of organizational citizen-
ship behaviors (i.e., discretionary, extra-role behaviors
aimed at the organization), which lead to employee perfor-
mance (i.e., required, in-role behaviors, included in the job
description).
We use two organizational justice constructs to capture
the notion of fairness. Distributive justice re?ects percep-
tions of fairness associated with the distribution of out-
comes employees receive. In this study, we de?ne
outcomes as pay, promotions, and recognition. Procedural
justice captures the perceived fairness of procedures used
in the decision-making process. In this study, we adopt a
broad view of organizational decisions, including but not
restricted to decisions related to outcomes.
4
Using a sample of 242 employees, we hypothesize and
?nd that employees’ perceptions of both the degree of tech-
nical validity and the extent to which the SPMS re?ects a
strategic causal model are positively associated with their
perceptions of distributive and procedural justice. We show
that perceptions of procedural justice affect organizational
citizenship behavior, which in turn affects their job perfor-
mance. Finally, we ?nd that perceptions of distributive
justice are associated with perceptions of procedural
justice. Our hypothesized model is illustrated in Fig. 1.
This study is important to both academics and to
designers of SPMS incentive plans. Previous research has
1
See also the application to transfer pricing (Kachelmeier & Towry,
2002) and budgeting (e.g., Lau & Lim, 2002; Libby, 1999, 2001; Lindquist,
1995; Wentzel, 2002).
2
Higher levels of performance are achieved because incentives provide
extrinsic motivation, which lead to greater effort (expectancy theory),
reinforce the performance goals that are linked to incentives (reinforce-
ment theory), and motivate employees to expend effort to achieve dif?cult
goals (goal-setting theory) (Jenkins, Gupta, Mitra, & Shaw, 1998).
3
In the long-run we would expect to see an association between
employee perceptions of the SPMS system design and the actual system
design as employees become more informed about the system and receive
feedback. Thus over time, any gap between perceptions and actual system
design should become smaller (Robbins, 2003).
4
An SPMS incentive plan is a decision-facilitating tool that provides
information to reduce ex ante uncertainty in the organizational decision-
making arena. The SPMS serves as a basis to guide employees in making
decisions regarding where to take actions and exert effort. In addition, since
the SPMS is linked to incentives, it serves as a basis for making allocative
decisions. Thus, we adopt the standard de?nition of procedural justice and
do not con?gure it to capture any one speci?c type of decision.
306 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
explored the use of subjectivity in incentives as a way to
mitigate some of the problems of a formula-based SPMS
incentive plan. But there are also problems inherent to
the use of subjectivity in incentives (Ittner, Larcker, &
Meyer, 2003). Our study suggests that it is important to
understand the attributes that employees governed by
SPMS incentive plans perceive matter because these per-
ceptions affect outcomes such as organizational citizenship
behaviors and performance. Understanding which attri-
butes matter is an important ?rst step towards implement-
ing and communicating a more effective SPMS incentive
plan that may resolve some of the con?ict that currently
exists in motivating employee behaviors. As Otley (1999,
p. 368) states, performance measurement must also be
‘‘necessarily concerned with issues of motivation and em-
ployee behaviour” and that gaining an understanding of
these issues is necessary to appropriately design and
implement control systems. The ?ndings in this study,
which are based on integrating fairness into an econom-
ics-based problem, are consistent with Ittner, Larcker,
and Meyer (2003, p. 754) who state ‘‘the evidence suggests
that psychology-based explanations may be equally or
more relevant than economics-based explanations in
understanding measurement practices in some settings”.
Thus, this paper contributes to the (limited) ?eld-based/
survey research on SPMS incentive systems.
The remainder of this paper is organized in four sec-
tions. In ‘Background and hypothesis development’ section,
we discuss the related literature and develop our research
hypotheses. We describe our research setting, sample, and
development of variables in ‘Sample and research method’
section. In ‘Results’ section, we discuss the statistical tests
that address our research hypotheses and present a discus-
sion of results. Finally, in ‘Conclusions’ section we summa-
rize our results, present the study’s limitations, and discuss
implications of our work.
Background and hypothesis development
Background literature
Financial measures are often criticized as providing
incomplete assessments of behavior and performance. As
a result, ?rms have begun implementing SPMSs that in-
clude multiple ?nancial and non-?nancial measures spe-
ci?cally chosen because they are critical success factors
to the achievement of the ?rm’s business strategy and mis-
sion. The measures are chosen through an organized and
deliberate process to ensure understandability, accuracy,
and reliability of the information. They translate strategy
throughout the organization and thus re?ect a series of
hypothesized cause-and-effect relationships (Frigo & Lit-
man, 2002; Kaplan & Norton, 1996). From this communica-
tion, employees gain an understanding of ‘‘the long-term
strategy, the relations among the various strategic objec-
tives, and the association between the employees’ actions
and the chosen strategic goals” (Ittner & Larcker, 1998, p.
223). Consequently, employees are able to understand
the impact of their actions on performance measures, align
their actions with organizational strategy, and achieve bet-
ter performance (Kaplan & Atkinson, 1998). Webb (2004)
shows that the strength of the causal links is an important
feature of an SPMS since it is positively associated with
goal commitment. He also ?nds that a second SPMS fea-
ture, managers’ beliefs that they can achieve stated goals,
are associated with goal commitment.
Agency theory assumes that employees are work-
averse and self-interested; thus, the primary agency prob-
lem is how organizations can use the performance mea-
surement system to motivate employees and mitigate
organizational control problems (Sprinkle, 2003). The
informativeness principle suggests that additional mea-
sures that are costless should be included in the perfor-
mance measurement system as long as the measure
provides incremental information (Holmstrom, 1979).
Since ?nancial measures are incomplete, the inclusion of
non-?nancial measures is often bene?cial (Feltham & Xie,
1994). Hence, ?rms that utilize their SPMS in incentive
contracting as a means to in?uence behavior should induce
employees to achieve a more balanced effort and higher
levels of performance.
Formulaic incentive systems, though, are often prob-
lematic. One problem unique to SPMS incentive plans in-
cludes the stress placed on employees trying to
determine how to balance constrained time and effort
across multiple objectives. Indeed, Kaplan and Norton
(1996, p. 221) suggest ‘‘caution” be used when linking
measures contained in an SPMS to employee compensa-
tion. To counter that risk, research has investigated with
mixed results whether allowing subjectivity in the con-
tracting process is bene?cial (e.g., Gibbs, Merchant, Van
H3 (+)
H2a (+)
In-Role
Performance
(PERF)
H4 (+) H4 (+)
H1a (+)
H2b (+)
H1b (+)
Distributive
Justice
(DJ)
Procedural
Justice
(PJ)
Degree of
Technical
Validity
(TC)
Extent Reflective
of Strategic Causal
Model (SCM)
Organizational
Citizenship
Behaviors
(OCBO)
Fig. 1. Theoretical model.
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 307
der Stede, & Vargus, 2004). Baker, Gibbons, and Murphy
(1994) analytically demonstrate that in certain situations
including a subjective component in the contracting pro-
cess is complementary to formulaic-based, objective mea-
sures. Rajan and Reichelstein (2006) ?nd that if
subjectivity is included in evaluation, then incentives
should include a discretionary bonus pool. Budde (2007)
shows analytically that incorporating subjectivity into a
multiple measure performance measurement incentive
system can result in a ?rst-best solution. However, other
empirical research ?nds that the use of subjectivity nega-
tively affects employees’ motivation and may foster feel-
ings of favoritism or bias (Ittner, Larcker, & Meyer, 2003;
Prendergast & Topel, 1993). For example, Moers (2005)
?nds that managers provide ratings that are both more le-
nient and compressed into a tighter range when they rely
on both objective and subjective performance measures.
Subjectivity can also be introduced through the weighting
of the multiple measures, if weights are not explicitly sta-
ted. Krishnan, Luft, and Shields (2005) ?nd that the preci-
sion of measures in?uences managers’ relative weighting
and can result in ineffective allocation of employee effort.
Overall, the use of an SPMS incentive plan can generate
mixed results, even if subjectivity is allowed. We suggest
another alternative is to consider whether perceptions of
SPMS characteristics are associated with perceptions of
fairness. That is, instead of incorporating another type
of measure or control mechanism into the evaluation
process, perhaps control can be enhanced if employees’
perceive that SPMS design characteristics are fair. Kahn-
eman, Knetsch, and Thaler (1986, p. S285) state ‘‘The
absence of considerations of fairness and loyalty fromstan-
dard economic theory is one of the most striking contrasts
between this body of theory and other social sciences—and
also between economic theory and lay intuitions about
human behavior”. Following Kahneman et al. (1986),
agency-based theorists have begun incorporating fairness
into agency models and experimental economics, recog-
nizing that individuals are motivated on traits other than
purely self-interest (Fehr, Klein, & Schmidt, 2001; Fehr &
Schmidt, 2000; Rabin, 1993).
Psychology-based literatures including human resource
management and organizational behavior have long inves-
tigated issues of fairness (e.g., Simons & Roberson, 2003).
Moreover, during the last decade, organizational justice
has been ‘‘among the most frequently researched topics”
(Cropanzano & Greenberg, 1997), due in part, to ‘‘the belief
that enhanced fairness perceptions can improve outcomes
relevant to organizations (e.g., organizational commit-
ment, job satisfaction, and performance)” (Colquitt, Con-
lon, Wesson, Porter, & Ng, 2001, p. 428). Much of the
research has traditionally focused on two dimensions of
organizational justice: distributive and procedural. Percep-
tions of distributive justice are high when employees be-
lieve their ratio of inputs (e.g., time, effort, and skill put
into jobs) to outcomes matches or exceeds a referent
other’s ratio (Adams, 1965). Procedures are thought to be
fair when they are made within a process characterized
as consistent, accurate, bias-free, ethical, appealable, and
representative of interested parties affected by the proce-
dures (Leventhal, 1980).
Limitations of prior research
Literature generally holds that using multiple measures
to allocate monetary incentives is associated with better
performance (i.e., Kaplan & Norton, 1996). Agency research
provides some support that the use of multiple measures is
advantageous; however, empirical support is mixed (Itt-
ner, Larcker, & Meyer, 2003) and overall, the relation be-
tween use of an SPMS incentive plan and performance
appears tenuous. There are three primary limitations in
this stream of literature.
First, little is known about which characteristics of
SPMS incentive plans are desirable when the objective is
to increase performance outcomes. Prior work has focused
on issues of subjectivity (Gibbs et al., 2004; Ittner, Larcker,
& Meyer, 2003), characteristics of just and unjust perfor-
mance appraisal systems in general (Taylor, Tracy, Renard,
Harrison, & Carroll, 1995), and involvement in budgeting or
resource allocation systems (Libby, 1999); but little re-
search other than Webb (2004) has investigated speci?c
characteristics of effective SPMSs. Second, much of what
is known about SPMS incentive plans is from tractable ana-
lytical models. Third, the stream of research has not yet
considered mediating variables such as organizational jus-
tice; therefore it does not attempt to explain how an SPMS
may affect performance. Moreover, including more links in
the causal model helps reveal weak links (Luft & Shields,
2003) that may exist between characteristics of an SPMS
and performance. We overcome these limitations by (1)
investigating employee perceptions of two speci?c charac-
teristics of the SPMS—its degree of technical validity and
the extent it re?ects a strategic causal model and (2) inte-
grating the justice and organizational citizenship behavior
literatures as additional links in the SPMS—performance
relation, thus providing evidence on mediating variables.
The discussion that follows adheres to Fig. 1. We inves-
tigate two characteristics of an SPMS incentive plan—the
extent to which it re?ects a strategic causal model and
its degree of technical validity. We ?rst develop the rela-
tion between each of the SPMS characteristics and proce-
dural justice (H1a–b) and distributive justice (H2a–b).
We then fully develop the model by linking justice to both
extra-role and in-role performance (H3–H4).
Development of hypotheses
We begin by hypothesizing that an SPMS incentive plan
is a tool, and to the extent it is linked with strategy and has
a high degree of technical validity, it will in?uence
employees’ perceptions that the organizational decision-
making process is just. That is, employees will perceive
the decision-making process as being procedurally just.
The importance of procedural justice to employees was
originally based on process control theory (Erdogan, 2002),
which suggests that employees need to feel they have con-
trol over and input into procedures used to make decisions
(Thibaut & Walker, 1975). Process control, or voice, is also
desirable because it helps employees feel like they are valu-
able members of a group (Lind & Tyler, 1988). Employees
evaluate procedural justice along dimensions including the
assessment of performance standards (Gilliland & Langdon,
308 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
1998), the establishment of a system for employees to en-
gageingoal setting(Gilliland&Langdon, 1998), andduepro-
cess characteristics (Folger, Konovsky, &Cropanzaon, 1992),
which consist of adequate notice (i.e., publishing, distribut-
ing, andcommunicatingof performance standards andfeed-
back), fair hearing (i.e., meeting between superior and
subordinate to assess performance), and judgment based
on evidence (i.e., application of consistent standards and
allocationof rewards (Erdogan, 2002; Taylor, Masterson, Re-
nard, & Tracy, 1998)). It is generally agreed that the criteria
necessary to establish due process include consistency, bias
suppression, accuracy, correctability, representativeness,
and ethicality (Colquitt et al., 2001; Leventhal, 1980).
A search of the accounting literature reveals little re-
search that directly relates SPMS system characteristics
to organizational justice.
5
Lau and Sholihin (2005) ?nd that
the importance placed on ?nancial and non-?nancial mea-
sures is associated with procedural justice, but do not
examine the speci?c characteristics of the performance
measurement system. With the exception of Giraud et al.
(2008), who invoke the controllability principle to investi-
gate incentive plan characteristics and ?nd that managers
perceive that it is fair to be held accountable for internal
events, but not external events, most of the evidence is
found in the psychology and human resource literatures.
Performance appraisal systems with due process charac-
teristics result in more satis?ed managers and in better
manager–subordinate relationships (Taylor et al., 1998);
and employees have a lower likelihood to leave the organi-
zation (Taylor et al., 1995). Empirical evidence also sup-
ports the relation between procedural justice and each of
the speci?c elements of adequate notice and judgment
based on evidence (Greenberg, 1986; Levy & Williams,
1998). In sum, literature has demonstrated that many of
the criteria necessary to establish due process are posi-
tively associated with behavioral outcomes.
Linking incentives to an SPMS instead of to incomplete
?nancial measures should induce more balanced employee
effort and achieve higher levels of performance. An SPMS
incentive plan provides an organization and its employee
with structure; however, it is a complicated performance
measurement system. It involves multiple ?nancial and
non-?nancial performance measures that employees must
weigh and trade-off in making decisions and taking ac-
tions. Thus, it is necessary that the SPMS incentive plan
is clear, transparent and easily understood by employees.
To the extent the underlying SPMS re?ects a strategic cau-
sal model, it clearly communicates to the employee the ac-
tions he/she can take to be aligned with organizational
objectives and meet stated outcomes. A transparent SPMS
that translates a ?rm’s business strategy is associated with
decreased role ambiguity and con?ict, and increased job-
relevant information (Burney & Widener, 2007), which
helps clarify decision-making processes. Recent research
questions the validity of the cause-and-effect relations in
an SPMS (Malina, Norrekilt, & Selto, 2007); however, the
necessary condition for our hypothesis is only that
employees perceive the relations exist. For example, Far-
rell, Luft, and Shields (2007) ?nd that mental representa-
tion of the causal relations in an SPMS is important to
managerial judgments in decision-making (see also Malina
et al. (2007) and Vera-Munoz, Shackell, and Buehnr
(2007)). To the extent the SPMS contains a high degree of
technical validity it contains timely, understandable, and
reliable information. It takes a complex system and clearly
and easily communicates to employees’ appraisal stan-
dards, organizational objectives, and relevant feedback
that facilitates perceptions of due process characteristics
and helps employees feel in control of their environment.
The SPMS is a tool that facilitates and is used in the pro-
cess of organizational decision-making, An SPMS that is
tightly coupled with strategy and/or has a high degree of
technical validity will: (1) empower employees, (2) help
them feel in control and valued by the organization, and
(3) will contain due process characteristics.
6
Thus, based
on the discussion above, we hypothesize the following:
H1a: The extent to which employees perceive the SPMS
re?ects a strategic causal model is positively associated
with procedural justice.
H1b: The degree to which employees perceive the SPMS
is technically valid is positively associated with procedural
justice.
Distributive justiceis rootedinequitytheorywhichholds
that employees calculate their outputs as a function of their
inputs, compare that ratiotoother employees’ input–output
ratios, and make a judgment regarding the fairness of their
outcome (Adams, 1965). In addition to comparisons of out-
comes to referent individuals, antecedents of distributive
justice include a comparison of the outcome received to
theeffort expended(Greenberg, 1986) andanoutput consis-
tent withthestatedgoals (Deutsch, 1975). Greenberg(1986)
surveyed individuals and found that a characteristic of dis-
tributivejusticeis that theevaluationwas anaccuratere?ec-
tion of the effort expended by the employee. Tyler (1994)
used an experimental setting to demonstrate that employ-
ees’ assessment of their outcome relative to their expecta-
tions was a signi?cant determinant of distributive justice.
We argue that it is likely that employees will perceive
that outcomes (e.g., pay, recognition, and promotions) more
accurately re?ect their efforts when an SPMS that re?ects a
strategic causal model and/or has a high degree of technical
validity is used as the basis for the allocation of incentives.
An SPMS re?ective of a strategic causal model is a formal
written document that provides employees with standards
and goals that dictate the distribution of incentives. This
document translates the ?rm’s strategy to the employees
andcommunicates toeachemployeehowhis/her individual
performance helps the ?rmachieve its objectives. The SPMS
sets expectations for employees and provides them with
stated goals. Employees perceive higher levels of distribu-
5
Johnson, Lowe, and Reckers (2008) examined organizational justice
effects of alternative work arrangements (as opposed to SPMS system
characteristics) and found a signi?cant association.
6
The extent to which an SPMS re?ects a strategic causal model and/or it
has a high degree of technical validity are related as they are both
characteristics of the performance measurement system. However, the
presence of one characteristic does not necessarily mean that the other is
present. For further discussion see Ittner and Larker (2003).
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 309
tive justice when the bonus allocation is consistent with the
implicit internal standards that determine allocation (Col-
quitt, 2001). In addition, Erdogan (2002) proposes that if
employees perceive the information to be used in evalua-
tions is consistent and distinctive, then employees will per-
ceive a higher level of distributive justice. In this context,
consistency is de?ned as looking at employee behavior ‘‘in
the same setting across different occasions” while distinc-
tive information ‘‘allows behaviors across different settings
to be included in the appraisal” (Erdogan, 2002, pp. 568–
569). Atransparent SPMS that clearlycommunicates actions
and aligns behaviors across employees will signal employ-
ees that consistent anddistinctive informationis inuse. This
contention is consistent with Malina et al. (2007) who ?nd
that perceptions of causal linkages inanSPMSare associated
with perceptions of distributive justice.
An SPMS high in technical validity will provide the em-
ployee with information that is perceived to be accurate,
timely, relevant, understandable, and accessible. Employ-
ees perceive that they have access to the information,
understand what it means and how to use it in carrying
out their job responsibilities. With this type of information
forming the basis of the employee’s incentive compensa-
tion, employees will likely perceive that their evaluation
accurately re?ects their effort and their expectations.
In sum, an SPMS that better translates strategy and/or is
technically valid will facilitate an employee’s comparison
of both inputs to outputs, and comparison of outcomes to
others. This comparison will be associated with higher per-
ceptions of distributive justice.
7
Thus, we propose the
following:
H2a: The extent to which employees perceive the SPMS
re?ects a strategic causal model is positively associated
with distributive justice.
H2b: The degree to which employees perceive the SPMS
is technically valid is positively associated with distribu-
tive justice.
Cropanzano and Folger (1991) develop a two compo-
nent model suggesting that if employees perceive resource
allocation as fair (unfair), then assessments of procedural
justice are unnecessary (necessary). Libby (1999) shows
in a participative budgeting setting that employees are
more accepting of unfair resource allocations if procedural
justice (proxied as voice) is coupled with an explanation.
Research concludes that procedural and distributive justice
are often linked; furthermore, that perceptions of proce-
dural justice are often in?uenced by perceptions of distrib-
utive justice (Cropanzano & Ambrose, 2001) with many
contending that if employees believe they receive a fair
outcome, they will in turn believe that the procedures
are fair (Colquitt et al., 2001).
The legal system is based on ‘‘due process” and evalua-
tions of outcomes weigh heavily on its participants; thus, it
has similarities to this literature and offers empirical sup-
port for the relation between distributive and procedural
justice (Lind & Tyler, 1988). Tyler (1994) found that citi-
zens had more favorable perceptions of judicial and police
procedures when their outcomes in traf?c and misde-
meanor courts were favorable. Casper, Tyler, and Fisher
(1988) and Landis and Goodstein (1986) ?nd that felons’
perceptions of distributive justice (i.e., sentencing) in?u-
enced their perceptions of procedural justice. Based on this
discussion, we offer the following hypothesis:
H3: Distributive justice is positively associated with
procedural justice.
Theory suggests that once employees assess procedural
justice, they thenmake evaluations of their ownsatisfaction
with the organization. This evaluation results in behavioral
effects (Lind & Tyler, 1988). Libby (1999) ?nds that when
employees are allowed both voice (e.g., procedural justice)
and explanationinthe budgeting process their performance
increases. Parker and Kohlmeyer (2005) ?nd that employ-
ees’ perceptions of the consistent application of standards
across the decision-making process in audit ?rms in?uence
employees’ turnover intentions. Lau and Sholihin (2005)
?nd that importance placed on performance measures is
associated with procedural justice, which in turn, affects
the employee’s trust inhis/her supervisor andhis/her inten-
tion to leave the organization.
We posit that procedural justice will be associated with
organizational citizenship behaviors (OCBs), which are
behaviors above and beyond the requirements of the job.
Social exchange theory suggests that when employees per-
ceive that either an individual or an entity acts in good
faith towards them, then they reciprocate and return the
favor (Lind & Tyler, 1988). This relationship has strong sup-
port in the literature including Organ (1990), Cohen-Cha-
rash and Spector (2001) who ?nd that procedural justice
is signi?cantly related to OCBs directed towards the orga-
nization (OCBOs), and Colquitt et al. (2001) who ?nd that
procedural justice has a stronger relationship with OCBOs
than with citizenship behaviors directed towards individu-
als (OCBIs).
We argue that OCBOs in turn will result in enhanced
employee job performance. Podsakoff and MacKenzie
(1997) propose several reasons why OCBs enhance em-
ployee performance and effectiveness. OCBs increase pro-
ductivity, direct resources toward more productive tasks,
decrease the need to dedicate resources solely to mainte-
nance functions, enhance coordination between members
of work groups, increase organizations’ ability to attract
and retain good performers, increase the consistency of
performance, and help employees adapt to changes in the
environment. Thus, we hypothesize that OCBs targeting
organizations should lead to higher supervisory ratings of
in-role employee performance. Indeed, empirical research
supports a link between OCBOs and job performance. For
instance, Podsakoff, Ahearne, and MacKenzie (1997) dem-
onstrate a positive link between one type of OCBOs, sports-
manship (tolerance of minor annoyances and
inconveniences that are an inevitable part of work life)
and performance quantity. Additionally, Podsakoff and
MacKenzie (1994) ?nd that sportsmanship and civic virtue
(responsible involvement in governance or other activities
that affect performance, such as the employee attending
meetings) accounts for a signi?cant amount of variance
7
We do not make any claims that causal models are suf?cient conditions
for distributive justice, we argue only that they are necessary. That is, we do
not claim to explain the entire variance in the perceptions of distributive
justice.
310 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
in sales performance. To summarize, the literature sup-
ports the hypothesis that procedural justice affects OCBOs,
which in turn, affect performance. Formally, we hypothe-
size the following:
H4: Procedural justice is positively associated with em-
ployee performance through organizational citizenship
behaviors.
Sample and research method
Description of the setting/performance measurement system
To test our hypotheses, we obtained our data from a
large, southeastern ?nancial services organization that
had implemented an SPMS incentive plan.
8
Speci?cally,
we conducted a survey within one region of the Branch
banking division. This region consisted of 53 branches
ranging in size from one to 25 employees. The unit of anal-
ysis for this study is the employee, appropriate for a test of
theory that relates organizational justice to perceptions of
SPMS characteristics. We surveyed employees and mea-
sured their perceptions of two characteristics of the
SPMS—degree of technical validity and extent to which
the SPMS re?ects a strategic causal model
9
—and organiza-
tional justice. We addressed and overcame one of the lim-
itations of survey research by matching the employee’s
responses to two assessments of performance by their
supervisors (i.e., we constructed an employee-supervisor
dyad). That is, we do not rely on a self-reported measure
of performance.
10
Each employee in the division we studied is evaluated
at two levels. First, on each employee’s hiring anniversary,
supervisors complete a performance evaluation, called a
personal development plan (PDP), which culminates in a
rating of employee performance as: needs immediate
improvement, growth opportunity, performer, very good
performer, and exceptional performer. These PDPs cover
aspects of traditional performance reviews, such as organi-
zational values and core competencies, as well as attain-
ment of objectives. The PDPs serve as the foundation for
annual salary raises. Second, each employee’s performance
is reviewed quantitatively through an incentive matrix
system, which serves as the basis for bonus compensation.
The matrix results are updated monthly and accessible on-
line. Moreover, the Vice-President reviews results with
each branch monthly. Individual bonuses are determined
on a calendar-year basis. Our primary focus is on this latter
bonus process.
During our initial inquiries with the division, the vice-
president reported that the company did not use an SPMS.
However, as our discussions progressed, we discovered
that the incentive matrix system appeared to be a sophis-
ticated SPMS. The incentive matrix system contains multi-
ple measures and is linked to division strategy. We
reviewed the stated objectives of the division and noted
that the strategic objectives are clearly tied to the mea-
sures included in the matrix system.
In 2002 the most recent version of the incentive matrix
system was implemented, resulting in the reliance on the
SPMS. The incentive matrix is structured according to
three fundamental objectives. First, the program is tar-
geted to reward top performers. As such, it is intentionally
designed as a ‘‘tough goal” system. Under this philosophy,
earning a bonus is not inevitable. Second, management is
focused on establishing a balanced approach to perfor-
mance evaluation. The design forces individuals to focus
on the trade-offs inherent among the performance
measures by assigning to each measure a relative weight-
ing, which sum to 100%. The organization selected these
weightings so as to reward balanced performance across
growth, pro?tability, credit quality, and client service
quality. Finally, management encourages teamwork
through inclusion of team-based incentives. Bank stan-
dards have been developed and are used as default values
for the incentive matrixes. However, the default values can
be overridden and tailored to the individual. Thus, a de-
tailed incentive matrix is constructed for each individual
based on his/her job title, branch location, speci?c objec-
tives, and other individual-level characteristics so that
each individual’s incentive matrix includes a mix of default
and unique goals. The details of the incentive matrix sys-
tem were communicated to employees through a 39 page
document.
The incentive matrix uses 8–12 weighted, objective per-
formance indicators across multiple categories and per-
spectives. The weighting serves as the midpoint for the
employee’s ‘‘opportunity”. The employee has an opportu-
nity to earn at three levels: a threshold, a target, and a max-
imum level of performance, which are denoted as goals on
the incentive matrix. These goals are intended to motivate
the employee’s behavior. The year-end performance report
multiplies the performance level for each measure by the
appropriate weight. If the employee achieves the target
on all measures, he or she will earn 100 points. Fewer
points will be awarded for achieving minimumgoals, while
a point total in excess of 100 will be awarded for achieving
goals above the target. Overall performance on the matrix is
calculated as the sumof the individual scores. The incentive
system allows employees to earn up to 50% of their base
salary as a bonus. To illustrate the incentive matrix plan,
consider the following simpli?ed example. Lucy is an area
branch manager. Her incentive matrix plan consists of
two measures: average transactions per teller (ATT) with
a weight of 40 and deposit production (DP) with a weight
8
Due to a con?dentiality agreement, we are unable to disclose the name
of the organization.
9
Anderson and Widener (2007) note that one-?rm ?eld studies may be
an appropriate research method for testing theory if the unit of analysis is
the employee. Furthermore, the variance and variable of interest is in the
perception of the system design. Robbins (2003, p. 145) states, ‘‘individuals
behave in a given manner based not on the way their external environment
actually is but, rather, on what they see or believe it to be. It is the
employee’s perception of a situation that becomes the basis for his or her
behavior. . . Similarly, issues such as fair pay for work performed, the
validity of performance appraisals, and the adequacy of working conditions
are not judged by employees in a way that ensures common perceptions;
nor can we be assured that individuals will interpret conditions about their
jobs in a favorable light. Therefore, to be able to in?uence productivity it is
necessary to assess how workers perceive their jobs”.
10
Nor are we subject to common measure bias on the outcome variable
due to the construction of the employee–supervisor dyad.
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 311
of 60. The opportunity for ATT is 35, 40, and 45 (because the
weight is the midpoint of the employee’s ‘‘opportunity”)
and the opportunity for DP is 50, 60, and 70. Based on Lucy’s
performance during the period, Lucy must score at least an
85 (35 + 50) to earn a bonus. Lucy’s bonus reaches a maxi-
mumof 50% of her salary at 115 points.
11
Thus, the available
bonus will be distributed over the range between 85 and 115
points.
Survey procedures and sample selection
We created two versions of the survey to manipulate
the ordering of questions within the longer sections of
the survey. While the survey only took about 15 min, this
change was undertaken to avoid any potential fatigue is-
sues impacting a particular survey item. We distributed
the survey to 540 employees at 47 branches. Survey pack-
ets and instructions were given to branch managers at a
monthly meeting and managers in turn distributed the
surveys to their subordinates. Employees completed a sur-
vey regarding their perceptions of the characteristics of the
SPMS and organizational justice within their workplace.
Employees also provided information regarding their
demographics and work history. In addition, supervisors
were asked to complete a 5 min survey for each of their
subordinates that evaluated in-role job performance and
organizational citizenship behavior. Surveys were returned
(separately for employees and supervisors) to the
researchers’ university address to ensure con?dentiality.
Employee numbers were used to match the supervisor
and employee surveys. Participants were assured that this
coding system was for matching purposes only and that
they could not be identi?ed individually.
Three hundred and twenty-three employees returned
surveys for a response rate of about 60%. Seventeen sur-
veys were omitted due to missing data. Supervisors did
not complete performance ratings for 64 of the respon-
dents, which reduced the ?nal sample size to 242. Most
participants were female (88%), Caucasian (87% and 10%
African American), and worked full-time (87%). Partici-
pants held a variety of jobs within the bank, but most were
relationship bankers (26%), tellers (23%), senior tellers
(19%), and ?nancial center managers (10%). Sample job
descriptions are provided in Appendix A. On average,
employees were 43 years old and had worked for the bank
for over eight years. Twenty-nine percent held high school
degrees, 34% had some college education, 11% had an asso-
ciate’s degree, 21% had a bachelor’s degree, and 2% held
graduate degrees.
Variable measures
To establish content validity, we thoroughly reviewed
the prior literature. We used established scales when they
existed; otherwise, we relied on the underlying literature
to provide a basis for formulating survey questions.
Once an initial draft of our survey instrument was pre-
pared, colleagues reviewed it for readability and clarity.
After incorporating their suggestions into the survey, we
met with the Vice-President of the division. We discussed
the project to ensure that the questions were appropriate
and worded in a manner that facilitated understanding.
We also reviewed archival documents including the incen-
tive matrix and a binder of documentation about the plan
to ensure that our understanding of the plan contents was
accurate and our language was clear and not misleading.
After reviewing the survey items with him, adjustments
were made to the wording and content to clarify meanings
and to make the items relevant for employees in that par-
ticular division.
We used data from multiple respondents to construct
our variables; thus, reducing the concern for common
measure bias. Speci?cally, we used responses from super-
visors to measure organizational citizenship behavior and
performance, and from employees to measure the four
other variables of interest.
12
As a precaution, we ran a Hart-
man’s one-factor test on the thirty survey questions asked of
the employees. The factor solution yielded ?ve factors with
eigenvalues >1.0. We also ran a Hartman’s one-factor test on
the 14 questions asked of the supervisors. The factor solu-
tion yielded three factors with eigenvalues >1.0. Overall,
the results of the Hartman one-factor tests, along with using
two different respondents, support the absence of signi?-
cant single-source bias (Podsakoff & Organ, 1986). An abbre-
viated version of the survey is found in Appendix B (note
that several questions are reverse coded).
13
Descriptive sta-
tistics for the multi-item variables are reported in Table 1. In
the sub-sections that follow, we brie?y discuss each
variable.
Degree of technical validity (TC)
We measure the degree of technical validity by modify-
ing McGowan’s (1998) technical characteristics of costing
system information scale so it applies to an SPMS. We
asked the respondents to rate the following ?ve questions
using a seven-point scale where 1 = strongly disagree and
7 = strongly agree: the information provided by the incen-
tive matrix is (1) accurate, (2) accessible, (3) reliable, (4)
timely, and (5) understandable. The Cronbach’s a is
0.872, indicating a high degree of internal consistency. To
review the variability of this construct, we created a mea-
sure using the average of the sum of the ?ve questions. The
construct has a mean of 5.27, variance of 1.05, and ranges
from 1.57 to 7.00.
Extent re?ective of strategic causal model (SCM)
To measure SCM, we developed survey items based on
the underlying literature to capture the extent the employ-
ee’s incentive matrix is linked to the ?rm’s strategic objec-
11
The lower and upper limits can be set at different levels depending on
the desired amount of variability around the target for each measure. The
key is that the deviation must be equal on both the upper and lower sides of
the midpoint.
12
Unfortunately, we do not have both supervisor and subordinate ratings
of the items.
13
To ensure con?dentiality, we disguise the identify of the ?rm that
provided the data for this study and replace any reference to the entity with
a generic ‘‘BANK” in the Appendix.
312 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
Table 1
Descriptive statistics.
Latent construct Item
N
Item
mean
Standard
deviation
Minimum Maximum Cronbach’s
a (std.)
Extent re?ective of strategic causal model (SCM) 0.950
My incentive matrix includes measures that are. . .
SL1: . . .chosen to ?t BANK’s strategy (versus using a set of generic measures) 242 5.40 1.122 1 7
SL2: . . .customized for each individual (versus being the same for all individuals
in the bank)
242 4.60 1.645 1 7
SL3: . . .chosen because they relate to BANK’s strategy 242 5.44 1.175 1 7
My incentive matrix helps me understand. . .
SL4: . . .how BANK’s strategy is to be achieved 242 5.23 1.269 1 7
SL5: . . .the ?nancial consequences of my decisions and actions 242 5.29 1.268 1 7
SL6: . . .trade-offs that I encounter among performance measures when I am
making decisions
242 5.15 1.204 1 7
SL9: . . .the cause-and-effect relationship between my actions and BANK’s
performance measurement outcomes
242 5.38 1.196 1 7
The incentive matrix system. . .
TMS1: . . .receives strong support from upper management 242 5.67 1.238 1 7
TMS2: . . .has been closely tied to the competitive strategies of BANK 242 5.51 1.188 2 7
MISC1: . . .impacts the way I perform my job responsibilities 242 5.32 1.394 1 7
MISC2: . . .is an appropriate system to evaluate my individual performance 242 4.80 1.622 1 7
USE2: . . .provides me with information that re?ects BANK’s strategy 242 5.33 1.300 1 7
USE3: . . .helps me make decisions that create long-term value for BANK, as
opposed to focusing solely on short-term results
242 5.29 1.367 1 7
USE5: . . .is useful because it uses multiple measures to evaluate performance 242 5.39 1.245 1 7
Degree of technical validity (TC) 0.872
The information provided by the incentive matrix is. . .
TC1:. . .accurate 242 5.34 1.249 1 7
TC2:. . .accessible 242 5.49 1.338 1 7
TC3:. . .reliable 242 5.26 1.256 1 7
TC4:. . .timely 242 4.99 1.469 1 7
TC5:. . .understandable 242 5.09 1.554 1 7
Distributive justice (DJ) 0.662
DJ1: My outcomes (e.g., pay, recognition, promotions) re?ect the effort I have
put into my work
242 4.72 1.988 1 7
DJ2: My outcomes are appropriate for the work I have completed 242 4.60 1.544 1 7
DJ3: My outcomes re?ect what I have contributed to BANK 242 3.81 1.963 1 7
DJ4: My outcomes are justi?ed, given my performance 242 4.54 1.575 1 7
Procedural justice (PJ) 0.793
PJ1: I am able to express my views and feelings during procedures used to make
decisions at BANK
242 5.01 1.613 1 7
PJ2: I have in?uence over outcomes arrived at by procedures used to make
decisions at BANK
242 4.12 1.599 1 7
PJ3: Procedures used to make decisions at BANK are applied consistently 242 4.81 1.613 1 7
PJ4: Procedures used to make decisions at BANK are free of bias 242 4.16 1.578 1 7
PJ5: Procedures used to make decisions at BANK are based on accurate
information
242 5.37 1.317 1 7
PJ6: I am able to appeal outcomes arrived at by procedures used to make
decisions at BANK
242 4.40 1.360 1 7
PJ7: Procedures used to make decisions at BANK uphold ethical and moral
standards
242 6.07 1.109 1 7
Organizational citizenship behaviors – organization (OCBO) 0.722
OCBO1: Attendance at work is above the norm 242 5.63 1.581 1 7
OCBO2: Gives advance notice when unable to come to work 242 5.95 1.270 1 7
OCBO3: Takes undeserved work breaks 242 5.85 1.478 1 7
OCBO4: Spends a great deal of time with personal phone calls 242 5.46 1.510 1 7
OCBO5: Complains about insigni?cant things at work 242 5.24 1.752 1 7
OCBO6: Conserves and protects organizational property 242 6.08 0.950 2 7
OCBO7: Adheres to informal rules devised to maintain order 242 5.77 1.052 2 7
In-role performance (PERF) 0.899
PERF1: Adequately completes assigned duties 242 5.92 1.151 1 7
PERF2: Ful?lls responsibilities speci?ed in job description 242 5.85 1.220 1 7
PERF3: Performs tasks that are expected of him/her 242 5.90 1.211 1 7
PERF4: Meets formal performance requirements of the job 242 5.85 1.274 2 7
PERF5: Engages in activities that will directly affect his/her performance
evaluation
242 5.01 1.569 1 7
PERF6: Neglects aspects of the job he/she is obligated to perform 242 5.69 1.573 1 7
PERF7: Fails to perform essential duties 242 6.02 1.387 2 7
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 313
tives (e.g., Kaplan & Norton, 1996). We used a 14-item
scale
14
that captures various details of the performance
measurement system, such as the extent to which the per-
formance measures are selected to ?t organizational strat-
egy and whether the measures increase employee
understanding of organizational functioning. The respon-
dents rated each question on a seven-point scale where
1 = strongly disagree and 7 = strongly agree. The Cronbach’s
a is 0.95, which indicates high internal consistency. To re-
view the variability of this construct, we created a measure
using the average of the sum of the 14 questions. The con-
struct has a mean of 5.23, variance of 1.29, and ranges from
1.00 to 7.00.
Organizational justice (PJ and DJ)
We measured justice using a multidimensional scale
developed by Colquitt (2001). Distributive justice (DJ) con-
sists of a four-item scale assessing the degree to which
employees perceive they are fairly rewarded when consid-
ering their effort and level of performance. The procedural
justice (PJ) scale uses Leventhal’s (1980) six procedural
rules to evaluate employees’ fairness perceptions of proce-
dures used by organizations to make decisions. The seven-
item scale assesses the presence of voice, consistency,
accuracy, appeal processes, bias, and ethical treatment to
determine if procedures are fair. The Cronbach’s as of
0.662 and 0.793, respectively, for DJ and PJ, indicate high
internal consistency.
Organizational citizenship behaviors (OCBOs)
Organizational citizenship behaviors are often charac-
terized as ‘‘extra-role” behaviors (Williams & Anderson,
1991). We use a seven-item scale developed by Williams
and Anderson (1991) to measure organizational citizen-
ship behaviors directed at the organization. Questions in-
clude such items as whether the employee complains
about insigni?cant things at work (a reverse-coded item),
conserves and protects organizational property, and ad-
heres to informal rules devised to maintain order. The
OCBO scale has a Cronbach’s a of 0.722, which indicates
high internal consistency.
Performance (PERF)
We use a seven-item scale developed by Williams and
Anderson (1991) to measure performance. This measure
captures ‘‘in-role” behavior, de?ned as ‘‘behaviors that
are recognized by formal reward systems and are part of
the requirements as described in job descriptions” (Wil-
liams & Anderson, 1991, p. 606). We asked supervisors to
rate their employees’ completion of required job tasks.
Speci?cally, we asked questions such as whether the em-
ployee ful?lls responsibilities speci?ed in the job descrip-
tion (see Appendix A for a complete listing). The
Cronbach’s a of 0.899 indicates high internal consistency.
To establish that the measure is acting plausibly, we corre-
late PERF to an objective measure, which is the employee’s
rating received on his/her most recent annual performance
PDP evaluation. The ?rm’s evaluation process assesses per-
formance using a scale as follows: growth opportunity,
performer, very good performer, or exceptional per-
former,
15
which we converted to a four-point scale. The
bi-variate correlation between PERF and the assessment of
performance based on the employee’s most recent evalua-
tion is 0.549 (p < 0.01). We conclude that the survey mea-
sure of PERF is behaving plausibly.
16
Results
The correlation matrix is presented in Table 2 and pro-
vides univariate results. As expected, we see that TC and
SCM are correlated with each of PJ and DJ. We also see that
DJ and PJ are correlated with OCBOs, and OCBOs are corre-
lated with PERF.
17
We use AMOS 4.0 to estimate the structural equation
model shown in Fig. 1. Moreover, we model the default
assumption in SEM, allowing the exogenous variables to
covary. The measurement and structural models are jointly
estimated using the full information maximum likelihood
estimation method.
18
In Table 3, we present the measure-
ment model results. Since latent variables are not directly
measured, we assign each of thema scale by setting the load-
Table 2
Correlation matrix.
SCM TC DJ PJ OCBO
SCM
TC 0.612
**
DJ 0.331
**
0.367
**
PJ 0.499
**
0.585
**
0.511
**
OCBO 0.122 0.146
*
0.138
*
0.146
*
PERF 0.098 0.039 0.102 0.056 0.665
**
Where SCM – extent re?ective of a strategic causal model; TC – degree of
technical validity; DJ – distributive justice; PJ – procedural justice; OCBO
– organizational citizenship behaviors – organization; PERF – in-role
performance.
*
Signi?cant at 0.05.
**
Signi?cant at 0.00.
14
The survey contains 20 questions regarding the incentive matrix
system. We discarded three of the questions that speci?cally ask about
bene?ts to the organization as opposed to the employee. Since we
developed this construct speci?cally for this study, we ran an exploratory
factor analysis on the remaining 17 questions. The factor analysis revealed
two factors with eigenvalues >1, with three questions crossloading on both
factors. We removed those three questions, reran the factor analysis and
noted that the factor solution only contained one factor, which we used in
the ?nal analysis.
15
In describing the PDP earlier, we noted ?ve potential evaluation ratings
while here we list four ratings. The ‘‘needs immediate improvement” option
in the PDP was not given for any of the respondents in our sample.
16
If supervisors engage in leniency and in?ate subordinates’ evaluations,
the overall mean performance level will increase; however, we are not
interested in exploring the mean effect, rather we are interested in
investigating the association between perceptions of justice and employee
performance. The more signi?cant concern in this type of study is common
measurement bias, which we overcame by creating an employee–super-
visor dyad and removing the concern of common measure bias (i.e., the
employee provides the perceptual measure of justice while the employee’s
supervisor rates his/her performance).
17
We reviewed the variance in?ation factors and tolerance levels and
found no evidence of multicollinearity.
18
A review of kurtosis and skewness reveals that all variables are well
below 10.0 and 3.0, respectively, indicating that the data is within tolerable
levels of univariate normality (Kline, 1998).
314 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
ing of one itemfor each latent construct to 1.0 (i.e., otherwise
known as identi?cation) (Kline, 1998). This con?rmatory
analysis yields standardized loadings greater than 0.30
(unstandardized >0.40), all of which are signi?cant
(p-value < 0.01), providing evidence of convergent validity.
We also note no excessive correlations between the
Table 3
Measurement model. This table reports the descriptive statistics along with the results of maximum likelihood estimation of the measurement model for the
latent variables used in Fig. 1. The referent indicators that de?ne the scale of the latent variables are denoted by an asterisk (Ã).
Latent variables N Mean Standard
deviation
Standardized factor
loading
Extent re?ective of strategic causal model (SCM)
My incentive matrix includes measures that are. . .
SL1: . . .chosen to ?t BANK’s strategy (versus using a set of generic measures) 242 5.40 1.122 0.782
SL2: . . .customized for each individual (versus being the same for all individuals in the bank) 242 4.60 1.645 0.659
SL3: . . .chosen because they relate to BANK’s strategy 242 5.44 1.175 0.784
My incentive matrix helps me understand. . .
SL4: . . .how BANK’s strategy is to be achieved 242 5.23 1.269 0.806
SL5: . . .the ?nancial consequences of my decisions and actions 242 5.29 1.268 0.872
SL6: . . .trade-offs that I encounter among performance measures when I am making decisions 242 5.15 1.204 0.810
SL9: . . .the cause-and-effect relationship between my actions and BANK’s performance
measurement outcomes
242 5.38 1.196 0.864
The incentive matrix system. . .
TMS1: . . .receives strong support from upper management 242 5.67 1.238 0.651
TMS2: . . .has been closely tied to the competitive strategies of BANK 242 5.51 1.188 0.719
MISC1: . . .impacts the way I perform my job responsibilities
*
242 5.32 1.394 0.760
MISC2: . . .is an appropriate system to evaluate my individual performance 242 4.80 1.622 0.746
USE2: . . .provides me with information that re?ects BANK’s strategy 242 5.33 1.300 0.732
USE3: . . .helps me make decisions that create long-term value for BANK, as opposed to focusing
solely on short-term results
242 5.29 1.367 0.760
USE5: . . .is useful because it uses multiple measures to evaluate performance 242 5.39 1.245 0.827
Degree of technical validity (TC)
The information provided by the incentive matrix is. . .
TC1:. . .accurate 242 5.34 1.249 0.849
TC2:. . .accessible 242 5.49 1.338 0.733
TC3:. . .reliable 242 5.26 1.256 0.907
TC4:. . .timely 242 4.99 1.469 0.610
TC5:. . .understandable
*
242 5.09 1.554 0.723
Distributive justice (DJ)
DJ1: My outcomes (e.g., pay, recognition, promotions) re?ect the effort I have put into my work 242 4.72 1.988 0.598
DJ2: My outcomes are appropriate for the work I have completed 242 4.60 1.544 0.741
DJ3: My outcomes re?ect what I have contributed to BANK 242 3.81 1.963 0.338
DJ4: My outcomes are justi?ed, given my performance
*
242 4.54 1.575 0.700
Procedural justice (PJ)
PJ1: I am able to express my views and feelings during procedures used to make decisions at
BANK
242 5.01 1.613 0.631
PJ2: I have in?uence over outcomes arrived at by procedures used to make decisions at BANK 242 4.12 1.599 0.566
PJ3: Procedures used to make decisions at BANK are applied consistently 242 4.81 1.613 0.706
PJ4: Procedures used to make decisions at BANK are free of bias 242 4.16 1.578 0.448
PJ5: Procedures used to make decisions at BANK are based on accurate information 242 5.37 1.317 0.759
PJ6: I am able to appeal outcomes arrived at by procedures used to make decisions at BANK 242 4.40 1.360 0.428
PJ7: Procedures used to make decisions at BANK uphold ethical and moral standards
*
242 6.07 1.109 0.663
Organizational citizenship behaviors – organization (OCBO)
OCBO1: Attendance at work is above the norm 242 5.63 1.581 0.521
OCBO2: Gives advance notice when unable to come to work 242 5.95 1.270 0.590
OCBO3: Takes undeserved work breaks 242 5.85 1.478 0.386
OCBO4: Spends a great deal of time with personal phone calls 242 5.46 1.510 0.465
OCBO5: Complains about insigni?cant things at work 242 5.24 1.752 0.544
OCBO6: Conserves and protects organizational property 242 6.08 0.950 0.615
OCBO7: Adheres to informal rules devised to maintain order
*
242 5.77 1.052 0.604
In-role performance (PERF)
PERF1: Adequately completes assigned duties 242 5.92 1.151 0.880
PERF2: Ful?lls responsibilities speci?ed in job description 242 5.85 1.220 0.939
PERF3: Performs tasks that are expected of him/her 242 5.90 1.211 0.933
PERF4: Meets formal performance requirements of the job 242 5.85 1.274 0.858
PERF5: Engages in activities that will directly affect his/her performance evaluation 242 5.01 1.569 0.311
PERF6: Neglects aspects of the job he/she is obligated to perform 242 5.69 1.573 0.705
PERF7: Fails to perform essential duties
*
242 6.02 1.387 0.809
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 315
constructs and a review of the modi?cation indices reveals
no signi?cant crossloadings, evidence of discriminant
validity.
19
In Table 4, we report the results of the structural
model. We evaluate the model using the Chi-square test
statistic (1753.403), the Chi-square divided by the model
degrees of freedom (1.961), the comparative ?t index
(0.976), the root mean square error of approximation
(0.063) and Hoelter’s critical N (137). The model appears
to ?t the data reasonably well.
20
The path coef?cients reported in Table 4 provide evi-
dence on the hypotheses. We ?nd support for our hypoth-
esized model as depicted in Fig. 2. H1a and H2a predict
that the extent to which an SPMS re?ects a strategic causal
model is associated with higher levels of both perceived
procedural justice and distributive justice. We ?nd that
both of these hypotheses are supported (p < 0.05 and
p < 0.01, respectively). H1b and H2b predict that the de-
gree to which an SPMS is technically valid is associated
with higher levels of perceived procedural justice and dis-
tributive justice. We ?nd that both of these hypotheses are
supported (p < 0.01 and p < 0.05, respectively). These re-
sults show that employees’ perceptions of the design char-
acteristics of an SPMS incentive plan enhance perceptions
of organizational justice.
H3 predicts that distributive justice is associated with
procedural justice and the results show that H3 is sup-
ported (p < 0.01). We also ?nd support for H4, which pre-
dicts that procedural justice affects performance through
organizational citizenship behaviors (PJ to OCBO, p < 0.05;
OCBO to PERF, p < 0.01). To provide further evidence on
this mediation hypothesis, we also perform the steps sug-
gested by Baron and Kenny (1986). We ?rst estimate the
direct path from PJ to PERF and ?nd that it is not signi?-
cant. Second, we estimate the direct path from PJ to OCBO
and ?nd a signi?cant relation (standardized b = 0.192,
p < 0.05, two-tailed). Finally, we estimate the direct effects
of both PJ and OCBO on PERF. We ?nd that PJ is insigni?-
cant, while OCBO is signi?cant even after controlling for
PJ (standardized b = 0.802, p < 0.01, two-tailed). According
to Kenny (2006), the necessary steps to show mediation
are steps 2 and 3.
21
Overall, we ?nd evidence that supports our hypothe-
sized model shown in Fig. 1. However, common to cross-
sectional survey research, we rely on theory to establish
the expected directions of the paths between constructs.
Therefore, in the next section we test and provide evidence
on possible reverse causation. We also discuss this issue
further in the conclusion.
Alternative models
Although the model presented in Table 4 is well ?tting,
there is no assurance that it is the only model. The purpose
of this section is to compare the base model to alternative
models to rule out alternative model speci?cations (Kline,
1998). The alternative model speci?cations in Table 5 build
the model by adding one or more paths. When building a
model, a signi?cant X
2
difference test indicates that the
alternative model is better ?tting than the base model
(Kline, 1998).
Alternative model 1 adds the direct paths from each of
the two SPMS characteristics—TC and SCM—to PERF. The
results show that the path coef?cients on the additional
two paths and the X
2
difference test are insigni?cant,
which indicates that the alternative model is not better ?t-
ting than the original model, and, thus we reject alterna-
tive model 1. In alternative model 2, we investigate an
additional direct path from DJ to OCBO. This path does have
some support in the literature, although the conclusion is
that the effect of PJ on OCBOs is much stronger than that
of DJ on OCBOs (Colquitt et al., 2001). The results show that
the path coef?cient from DJ to OCBO is insigni?cant and
the model is not better ?tting. Thus, we reject alternative
model 2.
Table 4
Structural model. This table reports the results of maximum likelihood
estimation of the structural equation model depicted in Fig. 1. Each cell
contains the standardized coef?cient.
Dependent
variable (R
2
)
Independent
variable
Hypothesis Hypothesized
direction
Coef?cient
Base model
DJ (0.274) SCM H2a + 0.312
***
PJ (0.652) SCM H1a + 0.179
**
DJ (++) TC H2b + 0.257
**
PJ (++) TC H1b + 0.365
***
PJ (++) DJ H3 + 0.422
***
OCBO (0.028) PJ H4 + 0.169
**
PERF (++) OCBO H4 + 0.793
***
Model ?t statistics
X
2
1753.403
***
DF 894
CMINDF 1.961
CFI 0.976
RMSEA 0.063
Hoelter’s
critical N
137 (0.01)
++ R
2
is reported on another row since this dependent variable is repeated
in the table.
ÃÃÃ, ÃÃ, and à indicate the signi?cance of the p-value at 0.85. Moreover, we
did perform additional robustness checks by running factor analyses with
both TC and SCM indicators; and one with both PJ and DJ indicators. In both
analyses, the factor solution revealed two factors, all indicators loaded
appropriately, and there were no crossloadings.
20
An insigni?cant Chi-square (Joreskog, 1969), a CMINDF ratio less than 5
(Wheaton, Muthen, Alwin, & Summers, 1977), a CFI close to 1 (Arbuckle &
Wothke, 1999; Bentler, 1990), and an RMSEA of less than 0.08 (Browne and
Cudeck, 1993) indicate good ?t. Although we report the Chi-square test
statistic, recent evidence shows that the Chi-square test statistic is sensitive
to sample size and may be misleading (Byrne, 2001).
21
In our study, the direct link between PJ and PERF is weak (step 1) and
thus the effect of PJ on PERF can only be seen through the effect that PJ has
on OCBO. This ?nding is also consistent with Luft and Shields (2003) who
state that including appropriate paths in causal links can help reveal effects.
We also perform the Sobell’s test for mediation and ?nd that OCBO is a
signi?cant mediator in the relation between PJ and PERF (p < 0.01)
(Preacher & Hayes, 2004).
316 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
Alternative models 3–6 consider the possibility that
PERF drives the employees’ perceptions of justice and the
two characteristics of the SPMS. For example, some argue
that higher ratings are associated with higher perceived
PJ and DJ (e.g., Erdogan, 2002; Stoffey & Reilly, 1997); how-
ever, it is unclear whether it is a measurement issue in that
the scale did not capture both favorable and unfavorable
inequity or whether the relation holds (Erdogan, 2002).
Nevertheless, to rule out the plausible alternative of re-
verse causality, we allow the employee’s performance to
‘‘in?uence” all four of the employee’s perceptions. That is,
in each of the four alternative models, we allow for the
possibility of ‘‘feedback” loops or bidirectional causality.
The results show that the path coef?cients on the addi-
tional paths are not signi?cant and the alternative models
are not better ?tting than the base model, thus we reject
the alternative models.
In sum, after testing six alternative models we conclude
that our base model is the best-?tting model.
22
Validity tests for statistical robustness
To ensure that the SEM model results are robust, we
perform two validity tests. First, an untabulated con?rma-
tory factor analysis on OCBO revealed two dimensions. We
run the SEM model using only the ?rst dimension of OCBO
(items 1, 2, 6, and 7) and ?nd that our statistical inferences
are unchanged. Second, although earlier statistical checks
Table 5
Comparison of base model to alternative models. This table presents the maximum likelihood estimation of the structural equation model in Table 4 to three
alternative model speci?cations. The ?rst alternative builds the model by adding two additional paths. The next 5 alternatives build the model by adding one
additional path. A signi?cant X
2
difference test indicates that the alternative model is better ?tting than the base model. Each cell contains the standardized
coef?cient.
Dependent variable Independent variable Base model Alt. 1 Alt. 2 Alt. 3 Alt. 4 Alt. 5 Alt. 6
DJ SCM 0.312
***
0.314
***
0.309
***
0.311
***
0.311
***
0.352
***
0.353
***
PJ SCM 0.179
**
0.178
**
0.183
**
0.182
**
0.176
**
0.245
***
0.247
***
DJ TC 0.257
**
0.256
**
0.260
***
0.255
**
0.258
**
0.264
***
0.263
***
PJ TC 0.365
***
0.367
***
0.363
***
0.366
***
0.368
***
0.373
***
0.370
***
PJ DJ 0.422
***
0.423
***
0.419
***
0.417
***
0.431
***
0.449
***
0.449
***
OCBO PJ 0.169
**
0.197
**
0.091 0.151
*
0.218
**
0.157
*
0.153
***
PERF OCBO 0.793
***
0.811
***
0.794
***
0.792
***
0.798
***
0.792
***
0.792
***
Alternative model speci?cation
PERF SCM – 0.026 – – – – –
PERF TC – À0.112 – – – – –
OCBO DJ – – 0.103 – – – –
DJ Perf – – – 0.045 – – –
PJ Perf – – – – À0.072 – –
TEC Perf – – – – – À0.014 –
SPMS Perf – – – – – – À0.003
Model ?t
X
2
1753.403 1750.227 1752.789 1753.037 1752.010 1875.666 1875.698
p-Value 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Df 894 892 893 893 893 893 893
CMINDF 1.961 1.962 1.963 1.963 1.962 2.098 2.098
CFI 0.976 0.976 0.976 0.976 0.976 0.972 0.972
RMSEA 0.063 0.063 0.063 0.063 0.063 0.068 0.068
++ R
2
is reported on another row since this dependent variable is repeated in the table.
ÃÃÃ, ÃÃ, and à indicate the signi?cance of the p-value at
Economic theory suggests that multiple financial and non-financial measures (i.e., a strategic
performance measurement system ‘‘SPMS”) be used in compensation contracting to
properly direct employees’ attention and motivate behavior aligned with organizational
goals. Conversely, linking incentives to the SPMS can result in various dysfunctional behaviors,
including game playing by employees, the achievement of unbalanced performance,
and the potential of basing compensation on an incomplete performance measurement
system. Prior literature has investigated the use of subjectivity in compensation contracting
as a means of potentially mitigating these problems; however, subjectivity can introduce
other problems including claims of favoritism and bias. Economic theorists have
recently begun expanding the traditional agency model to include the notion of fairness
or justice. In this study, we obtain data from an organization that uses an SPMS as the basis
for the allocation of bonuses and investigate whether characteristics of the SPMS are associated
with perceived organizational fairness. Specifically, we hypothesize and show that
the extent to which employees perceive that the SPMS reflects a strategic causal model
and the degree to which it is technically valid are positively associated with their perceptions
of organizational justice. We also provide evidence that heightened levels of organizational
justice are the mechanism though which the perceived characteristics of the SPMS
are associated with employee performance. The implication is that firms do not necessarily
need to introduce subjectivity into the incentive contracting system, but can enhance performance
by linking incentive contracts to their SPMS if the system contains characteristics
that enhance employees’ perceptions of justice.
A path model examining the relations among strategic performance
measurement system characteristics, organizational justice,
and extra- and in-role performance
Laurie L. Burney
a
, Christine A. Henle
b
, Sally K. Widener
c,
*
a
Adkerson School of Accountancy, Mississippi State University, P.O. Box EF, Mississippi State, MS 39762, United States
b
College of Business Administration, University of Rhode Island, Ballentine Hall, 7 Lippitt Road, Kingston, RI 02881, United States
c
Jesse H. Jones Graduate School of Management, Rice University, Room 331-MS531, 6100 Main Street, Houston, TX 77005, United States
a b s t r a c t
Economic theory suggests that multiple ?nancial and non-?nancial measures (i.e., a strate-
gic performance measurement system ‘‘SPMS”) be used in compensation contracting to
properly direct employees’ attention and motivate behavior aligned with organizational
goals. Conversely, linking incentives to the SPMS can result in various dysfunctional behav-
iors, including game playing by employees, the achievement of unbalanced performance,
and the potential of basing compensation on an incomplete performance measurement
system. Prior literature has investigated the use of subjectivity in compensation contract-
ing as a means of potentially mitigating these problems; however, subjectivity can intro-
duce other problems including claims of favoritism and bias. Economic theorists have
recently begun expanding the traditional agency model to include the notion of fairness
or justice. In this study, we obtain data from an organization that uses an SPMS as the basis
for the allocation of bonuses and investigate whether characteristics of the SPMS are asso-
ciated with perceived organizational fairness. Speci?cally, we hypothesize and show that
the extent to which employees perceive that the SPMS re?ects a strategic causal model
and the degree to which it is technically valid are positively associated with their percep-
tions of organizational justice. We also provide evidence that heightened levels of organi-
zational justice are the mechanism though which the perceived characteristics of the SPMS
are associated with employee performance. The implication is that ?rms do not necessarily
need to introduce subjectivity into the incentive contracting system, but can enhance per-
formance by linking incentive contracts to their SPMS if the system contains characteristics
that enhance employees’ perceptions of justice.
Ó 2008 Elsevier Ltd. All rights reserved.
Introduction
‘‘An organization that is performing well is one that is
successfully attaining its objectives” (Otley, 1999, p. 364).
Recent trends indicate that many organizations pursue
these objectives by implementing a strategic performance
measurement system (SPMS) in which multiple ?nancial,
operational, and strategic measures are selected to trans-
late and communicate the ?rm’s strategy to its employees
(Ittner, Larcker, & Randall, 2003). An organization’s long-
term success depends on individual employee actions;
likewise, an SPMS provides employees with useful infor-
mation to facilitate behaviors that translate to organiza-
tional success (Otley, 1999; Schiemann & Lingle, 1997).
In this study, we investigate whether employees’ percep-
tions of two characteristics of an SPMS used as the basis
for incentive compensation affect their perception of orga-
nizational justice, or fairness. Investigating employees’
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.11.002
* Corresponding author. Fax: +1 713 348 6331.
E-mail address: [email protected] (S.K. Widener).
Accounting, Organizations and Society 34 (2009) 305–321
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
perceptions of fairness in an organizational context is
important since they are a primary antecedent of impor-
tant work-related outcomes, including employee satisfac-
tion, organizational commitment, organizational
citizenship behavior, and job performance, that ultimately
affect organizational success (Johnson, Selenta, & Lord,
2006). The management accounting literature has begun
studying fairness in the use of ?nancial versus non-?nan-
cial measures (Lau & Sholihin, 2005) and the applicability
to the controllability principle (Giraud, Langevin, & Men-
doza, 2008).
1
These studies support the contention that con-
sideration of fairness issues is relevant. We propose that
perceptions of fairness are especially relevant in the com-
plex SPMS arena.
Organizations use SPMSs to facilitate decision-making
and to in?uence employee behaviors (Sprinkle, 2003).
Agency, expectancy, goal setting, and reinforcement theo-
ries
2
predict a positive relation between incentives and per-
formance (Jenkins et al., 1998; Bonner & Sprinkle, 2002);
however, some researchers suggest that incentives can
cause employees to over invest efforts in one area at the ex-
pense of other areas (Beer & Cannon, 2004). Moreover,
poorly designed incentive schemes can produce undesirable
work efforts (Locke & Latham, 1990). Indeed, empirical re-
sults investigating the effects on performance are mixed
(Jenkins et al., 1998; Bonner & Sprinkle, 2002). One reason
for these mixed results is con?ict regarding the ideal design
of the incentive plan.
A formulaic incentive plan is problematic because any
one measure is incomplete. For example, if a principal con-
tracts with an agent on the basis of pro?t, the agent may
shirk with regards to efforts with long-term payoffs. With
an SPMS incentive plan (henceforth, we mean one in which
an SPMS is used to determine incentives), ?rms can assign
weights to each measure to calculate incentives (Ittner,
Larcker, & Meyer, 2003). Although an SPMS incentive plan
may be an improvement over a single measure, formulaic
system, employees may still engage in game playing to
maximize compensation or the system may result in
unbalanced performance (Ittner, Larcker, & Meyer, 2003;
Kaplan & Norton, 1996). To counteract the negative effects
of a formulaic incentive system, an alternative is to allow
for subjectivity in the evaluation process (e.g., Budde,
2007); however, this option may foster feelings of favorit-
ism or bias (e.g., Moers, 2005).
We investigate another alternative and suggest that
there are system attributes for which users will form per-
ceptions that in turn facilitate more effective behaviors
(Whyte, Bytheway, & Edwards, 1997). Employees behave
based on how they perceive the system; not necessarily
on the actual system design (Robbins, 2003). Therefore, it
is important for research to examine perceptions to under-
stand subsequent employee and organizational outcomes
(Robbins, 2003).
3
Two important characteristics of an SPMS
are that it (1) re?ects a strategic causal model and (2) is
technically valid (Kaplan & Norton, 1996). Strategic align-
ment translates the organization’s strategy to the employ-
ees, shows relations among the measures, and
demonstrates how actions will result in outcomes neces-
sary for achieving desired goals (Kaplan & Norton, 1996).
Technical validity ensures that the measures are accurate,
accessible, understandable, reliable, and timely. In the next
section, we develop hypotheses that employees’ percep-
tions of these two characteristics will foster perceptions
of organizational justice. If employees perceive that they
work in a fair environment, they are less likely to engage
in game playing and more likely to perform at a higher le-
vel. This proposition is consistent with proponents of
incentive plans who suggest that good system design and
implementation coupled with perceptions of fairness can
overcome negative aspects of incentive plans (Beer & Can-
non, 2004). Thus, our primary research question is whether
employees’ perceptions of two attributes of an SPMS
incentive plan are associated with organizational fairness,
and in turn, are positively associated with more effective
employee outcomes in the form of organizational citizen-
ship behaviors (i.e., discretionary, extra-role behaviors
aimed at the organization), which lead to employee perfor-
mance (i.e., required, in-role behaviors, included in the job
description).
We use two organizational justice constructs to capture
the notion of fairness. Distributive justice re?ects percep-
tions of fairness associated with the distribution of out-
comes employees receive. In this study, we de?ne
outcomes as pay, promotions, and recognition. Procedural
justice captures the perceived fairness of procedures used
in the decision-making process. In this study, we adopt a
broad view of organizational decisions, including but not
restricted to decisions related to outcomes.
4
Using a sample of 242 employees, we hypothesize and
?nd that employees’ perceptions of both the degree of tech-
nical validity and the extent to which the SPMS re?ects a
strategic causal model are positively associated with their
perceptions of distributive and procedural justice. We show
that perceptions of procedural justice affect organizational
citizenship behavior, which in turn affects their job perfor-
mance. Finally, we ?nd that perceptions of distributive
justice are associated with perceptions of procedural
justice. Our hypothesized model is illustrated in Fig. 1.
This study is important to both academics and to
designers of SPMS incentive plans. Previous research has
1
See also the application to transfer pricing (Kachelmeier & Towry,
2002) and budgeting (e.g., Lau & Lim, 2002; Libby, 1999, 2001; Lindquist,
1995; Wentzel, 2002).
2
Higher levels of performance are achieved because incentives provide
extrinsic motivation, which lead to greater effort (expectancy theory),
reinforce the performance goals that are linked to incentives (reinforce-
ment theory), and motivate employees to expend effort to achieve dif?cult
goals (goal-setting theory) (Jenkins, Gupta, Mitra, & Shaw, 1998).
3
In the long-run we would expect to see an association between
employee perceptions of the SPMS system design and the actual system
design as employees become more informed about the system and receive
feedback. Thus over time, any gap between perceptions and actual system
design should become smaller (Robbins, 2003).
4
An SPMS incentive plan is a decision-facilitating tool that provides
information to reduce ex ante uncertainty in the organizational decision-
making arena. The SPMS serves as a basis to guide employees in making
decisions regarding where to take actions and exert effort. In addition, since
the SPMS is linked to incentives, it serves as a basis for making allocative
decisions. Thus, we adopt the standard de?nition of procedural justice and
do not con?gure it to capture any one speci?c type of decision.
306 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
explored the use of subjectivity in incentives as a way to
mitigate some of the problems of a formula-based SPMS
incentive plan. But there are also problems inherent to
the use of subjectivity in incentives (Ittner, Larcker, &
Meyer, 2003). Our study suggests that it is important to
understand the attributes that employees governed by
SPMS incentive plans perceive matter because these per-
ceptions affect outcomes such as organizational citizenship
behaviors and performance. Understanding which attri-
butes matter is an important ?rst step towards implement-
ing and communicating a more effective SPMS incentive
plan that may resolve some of the con?ict that currently
exists in motivating employee behaviors. As Otley (1999,
p. 368) states, performance measurement must also be
‘‘necessarily concerned with issues of motivation and em-
ployee behaviour” and that gaining an understanding of
these issues is necessary to appropriately design and
implement control systems. The ?ndings in this study,
which are based on integrating fairness into an econom-
ics-based problem, are consistent with Ittner, Larcker,
and Meyer (2003, p. 754) who state ‘‘the evidence suggests
that psychology-based explanations may be equally or
more relevant than economics-based explanations in
understanding measurement practices in some settings”.
Thus, this paper contributes to the (limited) ?eld-based/
survey research on SPMS incentive systems.
The remainder of this paper is organized in four sec-
tions. In ‘Background and hypothesis development’ section,
we discuss the related literature and develop our research
hypotheses. We describe our research setting, sample, and
development of variables in ‘Sample and research method’
section. In ‘Results’ section, we discuss the statistical tests
that address our research hypotheses and present a discus-
sion of results. Finally, in ‘Conclusions’ section we summa-
rize our results, present the study’s limitations, and discuss
implications of our work.
Background and hypothesis development
Background literature
Financial measures are often criticized as providing
incomplete assessments of behavior and performance. As
a result, ?rms have begun implementing SPMSs that in-
clude multiple ?nancial and non-?nancial measures spe-
ci?cally chosen because they are critical success factors
to the achievement of the ?rm’s business strategy and mis-
sion. The measures are chosen through an organized and
deliberate process to ensure understandability, accuracy,
and reliability of the information. They translate strategy
throughout the organization and thus re?ect a series of
hypothesized cause-and-effect relationships (Frigo & Lit-
man, 2002; Kaplan & Norton, 1996). From this communica-
tion, employees gain an understanding of ‘‘the long-term
strategy, the relations among the various strategic objec-
tives, and the association between the employees’ actions
and the chosen strategic goals” (Ittner & Larcker, 1998, p.
223). Consequently, employees are able to understand
the impact of their actions on performance measures, align
their actions with organizational strategy, and achieve bet-
ter performance (Kaplan & Atkinson, 1998). Webb (2004)
shows that the strength of the causal links is an important
feature of an SPMS since it is positively associated with
goal commitment. He also ?nds that a second SPMS fea-
ture, managers’ beliefs that they can achieve stated goals,
are associated with goal commitment.
Agency theory assumes that employees are work-
averse and self-interested; thus, the primary agency prob-
lem is how organizations can use the performance mea-
surement system to motivate employees and mitigate
organizational control problems (Sprinkle, 2003). The
informativeness principle suggests that additional mea-
sures that are costless should be included in the perfor-
mance measurement system as long as the measure
provides incremental information (Holmstrom, 1979).
Since ?nancial measures are incomplete, the inclusion of
non-?nancial measures is often bene?cial (Feltham & Xie,
1994). Hence, ?rms that utilize their SPMS in incentive
contracting as a means to in?uence behavior should induce
employees to achieve a more balanced effort and higher
levels of performance.
Formulaic incentive systems, though, are often prob-
lematic. One problem unique to SPMS incentive plans in-
cludes the stress placed on employees trying to
determine how to balance constrained time and effort
across multiple objectives. Indeed, Kaplan and Norton
(1996, p. 221) suggest ‘‘caution” be used when linking
measures contained in an SPMS to employee compensa-
tion. To counter that risk, research has investigated with
mixed results whether allowing subjectivity in the con-
tracting process is bene?cial (e.g., Gibbs, Merchant, Van
H3 (+)
H2a (+)
In-Role
Performance
(PERF)
H4 (+) H4 (+)
H1a (+)
H2b (+)
H1b (+)
Distributive
Justice
(DJ)
Procedural
Justice
(PJ)
Degree of
Technical
Validity
(TC)
Extent Reflective
of Strategic Causal
Model (SCM)
Organizational
Citizenship
Behaviors
(OCBO)
Fig. 1. Theoretical model.
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 307
der Stede, & Vargus, 2004). Baker, Gibbons, and Murphy
(1994) analytically demonstrate that in certain situations
including a subjective component in the contracting pro-
cess is complementary to formulaic-based, objective mea-
sures. Rajan and Reichelstein (2006) ?nd that if
subjectivity is included in evaluation, then incentives
should include a discretionary bonus pool. Budde (2007)
shows analytically that incorporating subjectivity into a
multiple measure performance measurement incentive
system can result in a ?rst-best solution. However, other
empirical research ?nds that the use of subjectivity nega-
tively affects employees’ motivation and may foster feel-
ings of favoritism or bias (Ittner, Larcker, & Meyer, 2003;
Prendergast & Topel, 1993). For example, Moers (2005)
?nds that managers provide ratings that are both more le-
nient and compressed into a tighter range when they rely
on both objective and subjective performance measures.
Subjectivity can also be introduced through the weighting
of the multiple measures, if weights are not explicitly sta-
ted. Krishnan, Luft, and Shields (2005) ?nd that the preci-
sion of measures in?uences managers’ relative weighting
and can result in ineffective allocation of employee effort.
Overall, the use of an SPMS incentive plan can generate
mixed results, even if subjectivity is allowed. We suggest
another alternative is to consider whether perceptions of
SPMS characteristics are associated with perceptions of
fairness. That is, instead of incorporating another type
of measure or control mechanism into the evaluation
process, perhaps control can be enhanced if employees’
perceive that SPMS design characteristics are fair. Kahn-
eman, Knetsch, and Thaler (1986, p. S285) state ‘‘The
absence of considerations of fairness and loyalty fromstan-
dard economic theory is one of the most striking contrasts
between this body of theory and other social sciences—and
also between economic theory and lay intuitions about
human behavior”. Following Kahneman et al. (1986),
agency-based theorists have begun incorporating fairness
into agency models and experimental economics, recog-
nizing that individuals are motivated on traits other than
purely self-interest (Fehr, Klein, & Schmidt, 2001; Fehr &
Schmidt, 2000; Rabin, 1993).
Psychology-based literatures including human resource
management and organizational behavior have long inves-
tigated issues of fairness (e.g., Simons & Roberson, 2003).
Moreover, during the last decade, organizational justice
has been ‘‘among the most frequently researched topics”
(Cropanzano & Greenberg, 1997), due in part, to ‘‘the belief
that enhanced fairness perceptions can improve outcomes
relevant to organizations (e.g., organizational commit-
ment, job satisfaction, and performance)” (Colquitt, Con-
lon, Wesson, Porter, & Ng, 2001, p. 428). Much of the
research has traditionally focused on two dimensions of
organizational justice: distributive and procedural. Percep-
tions of distributive justice are high when employees be-
lieve their ratio of inputs (e.g., time, effort, and skill put
into jobs) to outcomes matches or exceeds a referent
other’s ratio (Adams, 1965). Procedures are thought to be
fair when they are made within a process characterized
as consistent, accurate, bias-free, ethical, appealable, and
representative of interested parties affected by the proce-
dures (Leventhal, 1980).
Limitations of prior research
Literature generally holds that using multiple measures
to allocate monetary incentives is associated with better
performance (i.e., Kaplan & Norton, 1996). Agency research
provides some support that the use of multiple measures is
advantageous; however, empirical support is mixed (Itt-
ner, Larcker, & Meyer, 2003) and overall, the relation be-
tween use of an SPMS incentive plan and performance
appears tenuous. There are three primary limitations in
this stream of literature.
First, little is known about which characteristics of
SPMS incentive plans are desirable when the objective is
to increase performance outcomes. Prior work has focused
on issues of subjectivity (Gibbs et al., 2004; Ittner, Larcker,
& Meyer, 2003), characteristics of just and unjust perfor-
mance appraisal systems in general (Taylor, Tracy, Renard,
Harrison, & Carroll, 1995), and involvement in budgeting or
resource allocation systems (Libby, 1999); but little re-
search other than Webb (2004) has investigated speci?c
characteristics of effective SPMSs. Second, much of what
is known about SPMS incentive plans is from tractable ana-
lytical models. Third, the stream of research has not yet
considered mediating variables such as organizational jus-
tice; therefore it does not attempt to explain how an SPMS
may affect performance. Moreover, including more links in
the causal model helps reveal weak links (Luft & Shields,
2003) that may exist between characteristics of an SPMS
and performance. We overcome these limitations by (1)
investigating employee perceptions of two speci?c charac-
teristics of the SPMS—its degree of technical validity and
the extent it re?ects a strategic causal model and (2) inte-
grating the justice and organizational citizenship behavior
literatures as additional links in the SPMS—performance
relation, thus providing evidence on mediating variables.
The discussion that follows adheres to Fig. 1. We inves-
tigate two characteristics of an SPMS incentive plan—the
extent to which it re?ects a strategic causal model and
its degree of technical validity. We ?rst develop the rela-
tion between each of the SPMS characteristics and proce-
dural justice (H1a–b) and distributive justice (H2a–b).
We then fully develop the model by linking justice to both
extra-role and in-role performance (H3–H4).
Development of hypotheses
We begin by hypothesizing that an SPMS incentive plan
is a tool, and to the extent it is linked with strategy and has
a high degree of technical validity, it will in?uence
employees’ perceptions that the organizational decision-
making process is just. That is, employees will perceive
the decision-making process as being procedurally just.
The importance of procedural justice to employees was
originally based on process control theory (Erdogan, 2002),
which suggests that employees need to feel they have con-
trol over and input into procedures used to make decisions
(Thibaut & Walker, 1975). Process control, or voice, is also
desirable because it helps employees feel like they are valu-
able members of a group (Lind & Tyler, 1988). Employees
evaluate procedural justice along dimensions including the
assessment of performance standards (Gilliland & Langdon,
308 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
1998), the establishment of a system for employees to en-
gageingoal setting(Gilliland&Langdon, 1998), andduepro-
cess characteristics (Folger, Konovsky, &Cropanzaon, 1992),
which consist of adequate notice (i.e., publishing, distribut-
ing, andcommunicatingof performance standards andfeed-
back), fair hearing (i.e., meeting between superior and
subordinate to assess performance), and judgment based
on evidence (i.e., application of consistent standards and
allocationof rewards (Erdogan, 2002; Taylor, Masterson, Re-
nard, & Tracy, 1998)). It is generally agreed that the criteria
necessary to establish due process include consistency, bias
suppression, accuracy, correctability, representativeness,
and ethicality (Colquitt et al., 2001; Leventhal, 1980).
A search of the accounting literature reveals little re-
search that directly relates SPMS system characteristics
to organizational justice.
5
Lau and Sholihin (2005) ?nd that
the importance placed on ?nancial and non-?nancial mea-
sures is associated with procedural justice, but do not
examine the speci?c characteristics of the performance
measurement system. With the exception of Giraud et al.
(2008), who invoke the controllability principle to investi-
gate incentive plan characteristics and ?nd that managers
perceive that it is fair to be held accountable for internal
events, but not external events, most of the evidence is
found in the psychology and human resource literatures.
Performance appraisal systems with due process charac-
teristics result in more satis?ed managers and in better
manager–subordinate relationships (Taylor et al., 1998);
and employees have a lower likelihood to leave the organi-
zation (Taylor et al., 1995). Empirical evidence also sup-
ports the relation between procedural justice and each of
the speci?c elements of adequate notice and judgment
based on evidence (Greenberg, 1986; Levy & Williams,
1998). In sum, literature has demonstrated that many of
the criteria necessary to establish due process are posi-
tively associated with behavioral outcomes.
Linking incentives to an SPMS instead of to incomplete
?nancial measures should induce more balanced employee
effort and achieve higher levels of performance. An SPMS
incentive plan provides an organization and its employee
with structure; however, it is a complicated performance
measurement system. It involves multiple ?nancial and
non-?nancial performance measures that employees must
weigh and trade-off in making decisions and taking ac-
tions. Thus, it is necessary that the SPMS incentive plan
is clear, transparent and easily understood by employees.
To the extent the underlying SPMS re?ects a strategic cau-
sal model, it clearly communicates to the employee the ac-
tions he/she can take to be aligned with organizational
objectives and meet stated outcomes. A transparent SPMS
that translates a ?rm’s business strategy is associated with
decreased role ambiguity and con?ict, and increased job-
relevant information (Burney & Widener, 2007), which
helps clarify decision-making processes. Recent research
questions the validity of the cause-and-effect relations in
an SPMS (Malina, Norrekilt, & Selto, 2007); however, the
necessary condition for our hypothesis is only that
employees perceive the relations exist. For example, Far-
rell, Luft, and Shields (2007) ?nd that mental representa-
tion of the causal relations in an SPMS is important to
managerial judgments in decision-making (see also Malina
et al. (2007) and Vera-Munoz, Shackell, and Buehnr
(2007)). To the extent the SPMS contains a high degree of
technical validity it contains timely, understandable, and
reliable information. It takes a complex system and clearly
and easily communicates to employees’ appraisal stan-
dards, organizational objectives, and relevant feedback
that facilitates perceptions of due process characteristics
and helps employees feel in control of their environment.
The SPMS is a tool that facilitates and is used in the pro-
cess of organizational decision-making, An SPMS that is
tightly coupled with strategy and/or has a high degree of
technical validity will: (1) empower employees, (2) help
them feel in control and valued by the organization, and
(3) will contain due process characteristics.
6
Thus, based
on the discussion above, we hypothesize the following:
H1a: The extent to which employees perceive the SPMS
re?ects a strategic causal model is positively associated
with procedural justice.
H1b: The degree to which employees perceive the SPMS
is technically valid is positively associated with procedural
justice.
Distributive justiceis rootedinequitytheorywhichholds
that employees calculate their outputs as a function of their
inputs, compare that ratiotoother employees’ input–output
ratios, and make a judgment regarding the fairness of their
outcome (Adams, 1965). In addition to comparisons of out-
comes to referent individuals, antecedents of distributive
justice include a comparison of the outcome received to
theeffort expended(Greenberg, 1986) andanoutput consis-
tent withthestatedgoals (Deutsch, 1975). Greenberg(1986)
surveyed individuals and found that a characteristic of dis-
tributivejusticeis that theevaluationwas anaccuratere?ec-
tion of the effort expended by the employee. Tyler (1994)
used an experimental setting to demonstrate that employ-
ees’ assessment of their outcome relative to their expecta-
tions was a signi?cant determinant of distributive justice.
We argue that it is likely that employees will perceive
that outcomes (e.g., pay, recognition, and promotions) more
accurately re?ect their efforts when an SPMS that re?ects a
strategic causal model and/or has a high degree of technical
validity is used as the basis for the allocation of incentives.
An SPMS re?ective of a strategic causal model is a formal
written document that provides employees with standards
and goals that dictate the distribution of incentives. This
document translates the ?rm’s strategy to the employees
andcommunicates toeachemployeehowhis/her individual
performance helps the ?rmachieve its objectives. The SPMS
sets expectations for employees and provides them with
stated goals. Employees perceive higher levels of distribu-
5
Johnson, Lowe, and Reckers (2008) examined organizational justice
effects of alternative work arrangements (as opposed to SPMS system
characteristics) and found a signi?cant association.
6
The extent to which an SPMS re?ects a strategic causal model and/or it
has a high degree of technical validity are related as they are both
characteristics of the performance measurement system. However, the
presence of one characteristic does not necessarily mean that the other is
present. For further discussion see Ittner and Larker (2003).
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 309
tive justice when the bonus allocation is consistent with the
implicit internal standards that determine allocation (Col-
quitt, 2001). In addition, Erdogan (2002) proposes that if
employees perceive the information to be used in evalua-
tions is consistent and distinctive, then employees will per-
ceive a higher level of distributive justice. In this context,
consistency is de?ned as looking at employee behavior ‘‘in
the same setting across different occasions” while distinc-
tive information ‘‘allows behaviors across different settings
to be included in the appraisal” (Erdogan, 2002, pp. 568–
569). Atransparent SPMS that clearlycommunicates actions
and aligns behaviors across employees will signal employ-
ees that consistent anddistinctive informationis inuse. This
contention is consistent with Malina et al. (2007) who ?nd
that perceptions of causal linkages inanSPMSare associated
with perceptions of distributive justice.
An SPMS high in technical validity will provide the em-
ployee with information that is perceived to be accurate,
timely, relevant, understandable, and accessible. Employ-
ees perceive that they have access to the information,
understand what it means and how to use it in carrying
out their job responsibilities. With this type of information
forming the basis of the employee’s incentive compensa-
tion, employees will likely perceive that their evaluation
accurately re?ects their effort and their expectations.
In sum, an SPMS that better translates strategy and/or is
technically valid will facilitate an employee’s comparison
of both inputs to outputs, and comparison of outcomes to
others. This comparison will be associated with higher per-
ceptions of distributive justice.
7
Thus, we propose the
following:
H2a: The extent to which employees perceive the SPMS
re?ects a strategic causal model is positively associated
with distributive justice.
H2b: The degree to which employees perceive the SPMS
is technically valid is positively associated with distribu-
tive justice.
Cropanzano and Folger (1991) develop a two compo-
nent model suggesting that if employees perceive resource
allocation as fair (unfair), then assessments of procedural
justice are unnecessary (necessary). Libby (1999) shows
in a participative budgeting setting that employees are
more accepting of unfair resource allocations if procedural
justice (proxied as voice) is coupled with an explanation.
Research concludes that procedural and distributive justice
are often linked; furthermore, that perceptions of proce-
dural justice are often in?uenced by perceptions of distrib-
utive justice (Cropanzano & Ambrose, 2001) with many
contending that if employees believe they receive a fair
outcome, they will in turn believe that the procedures
are fair (Colquitt et al., 2001).
The legal system is based on ‘‘due process” and evalua-
tions of outcomes weigh heavily on its participants; thus, it
has similarities to this literature and offers empirical sup-
port for the relation between distributive and procedural
justice (Lind & Tyler, 1988). Tyler (1994) found that citi-
zens had more favorable perceptions of judicial and police
procedures when their outcomes in traf?c and misde-
meanor courts were favorable. Casper, Tyler, and Fisher
(1988) and Landis and Goodstein (1986) ?nd that felons’
perceptions of distributive justice (i.e., sentencing) in?u-
enced their perceptions of procedural justice. Based on this
discussion, we offer the following hypothesis:
H3: Distributive justice is positively associated with
procedural justice.
Theory suggests that once employees assess procedural
justice, they thenmake evaluations of their ownsatisfaction
with the organization. This evaluation results in behavioral
effects (Lind & Tyler, 1988). Libby (1999) ?nds that when
employees are allowed both voice (e.g., procedural justice)
and explanationinthe budgeting process their performance
increases. Parker and Kohlmeyer (2005) ?nd that employ-
ees’ perceptions of the consistent application of standards
across the decision-making process in audit ?rms in?uence
employees’ turnover intentions. Lau and Sholihin (2005)
?nd that importance placed on performance measures is
associated with procedural justice, which in turn, affects
the employee’s trust inhis/her supervisor andhis/her inten-
tion to leave the organization.
We posit that procedural justice will be associated with
organizational citizenship behaviors (OCBs), which are
behaviors above and beyond the requirements of the job.
Social exchange theory suggests that when employees per-
ceive that either an individual or an entity acts in good
faith towards them, then they reciprocate and return the
favor (Lind & Tyler, 1988). This relationship has strong sup-
port in the literature including Organ (1990), Cohen-Cha-
rash and Spector (2001) who ?nd that procedural justice
is signi?cantly related to OCBs directed towards the orga-
nization (OCBOs), and Colquitt et al. (2001) who ?nd that
procedural justice has a stronger relationship with OCBOs
than with citizenship behaviors directed towards individu-
als (OCBIs).
We argue that OCBOs in turn will result in enhanced
employee job performance. Podsakoff and MacKenzie
(1997) propose several reasons why OCBs enhance em-
ployee performance and effectiveness. OCBs increase pro-
ductivity, direct resources toward more productive tasks,
decrease the need to dedicate resources solely to mainte-
nance functions, enhance coordination between members
of work groups, increase organizations’ ability to attract
and retain good performers, increase the consistency of
performance, and help employees adapt to changes in the
environment. Thus, we hypothesize that OCBs targeting
organizations should lead to higher supervisory ratings of
in-role employee performance. Indeed, empirical research
supports a link between OCBOs and job performance. For
instance, Podsakoff, Ahearne, and MacKenzie (1997) dem-
onstrate a positive link between one type of OCBOs, sports-
manship (tolerance of minor annoyances and
inconveniences that are an inevitable part of work life)
and performance quantity. Additionally, Podsakoff and
MacKenzie (1994) ?nd that sportsmanship and civic virtue
(responsible involvement in governance or other activities
that affect performance, such as the employee attending
meetings) accounts for a signi?cant amount of variance
7
We do not make any claims that causal models are suf?cient conditions
for distributive justice, we argue only that they are necessary. That is, we do
not claim to explain the entire variance in the perceptions of distributive
justice.
310 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
in sales performance. To summarize, the literature sup-
ports the hypothesis that procedural justice affects OCBOs,
which in turn, affect performance. Formally, we hypothe-
size the following:
H4: Procedural justice is positively associated with em-
ployee performance through organizational citizenship
behaviors.
Sample and research method
Description of the setting/performance measurement system
To test our hypotheses, we obtained our data from a
large, southeastern ?nancial services organization that
had implemented an SPMS incentive plan.
8
Speci?cally,
we conducted a survey within one region of the Branch
banking division. This region consisted of 53 branches
ranging in size from one to 25 employees. The unit of anal-
ysis for this study is the employee, appropriate for a test of
theory that relates organizational justice to perceptions of
SPMS characteristics. We surveyed employees and mea-
sured their perceptions of two characteristics of the
SPMS—degree of technical validity and extent to which
the SPMS re?ects a strategic causal model
9
—and organiza-
tional justice. We addressed and overcame one of the lim-
itations of survey research by matching the employee’s
responses to two assessments of performance by their
supervisors (i.e., we constructed an employee-supervisor
dyad). That is, we do not rely on a self-reported measure
of performance.
10
Each employee in the division we studied is evaluated
at two levels. First, on each employee’s hiring anniversary,
supervisors complete a performance evaluation, called a
personal development plan (PDP), which culminates in a
rating of employee performance as: needs immediate
improvement, growth opportunity, performer, very good
performer, and exceptional performer. These PDPs cover
aspects of traditional performance reviews, such as organi-
zational values and core competencies, as well as attain-
ment of objectives. The PDPs serve as the foundation for
annual salary raises. Second, each employee’s performance
is reviewed quantitatively through an incentive matrix
system, which serves as the basis for bonus compensation.
The matrix results are updated monthly and accessible on-
line. Moreover, the Vice-President reviews results with
each branch monthly. Individual bonuses are determined
on a calendar-year basis. Our primary focus is on this latter
bonus process.
During our initial inquiries with the division, the vice-
president reported that the company did not use an SPMS.
However, as our discussions progressed, we discovered
that the incentive matrix system appeared to be a sophis-
ticated SPMS. The incentive matrix system contains multi-
ple measures and is linked to division strategy. We
reviewed the stated objectives of the division and noted
that the strategic objectives are clearly tied to the mea-
sures included in the matrix system.
In 2002 the most recent version of the incentive matrix
system was implemented, resulting in the reliance on the
SPMS. The incentive matrix is structured according to
three fundamental objectives. First, the program is tar-
geted to reward top performers. As such, it is intentionally
designed as a ‘‘tough goal” system. Under this philosophy,
earning a bonus is not inevitable. Second, management is
focused on establishing a balanced approach to perfor-
mance evaluation. The design forces individuals to focus
on the trade-offs inherent among the performance
measures by assigning to each measure a relative weight-
ing, which sum to 100%. The organization selected these
weightings so as to reward balanced performance across
growth, pro?tability, credit quality, and client service
quality. Finally, management encourages teamwork
through inclusion of team-based incentives. Bank stan-
dards have been developed and are used as default values
for the incentive matrixes. However, the default values can
be overridden and tailored to the individual. Thus, a de-
tailed incentive matrix is constructed for each individual
based on his/her job title, branch location, speci?c objec-
tives, and other individual-level characteristics so that
each individual’s incentive matrix includes a mix of default
and unique goals. The details of the incentive matrix sys-
tem were communicated to employees through a 39 page
document.
The incentive matrix uses 8–12 weighted, objective per-
formance indicators across multiple categories and per-
spectives. The weighting serves as the midpoint for the
employee’s ‘‘opportunity”. The employee has an opportu-
nity to earn at three levels: a threshold, a target, and a max-
imum level of performance, which are denoted as goals on
the incentive matrix. These goals are intended to motivate
the employee’s behavior. The year-end performance report
multiplies the performance level for each measure by the
appropriate weight. If the employee achieves the target
on all measures, he or she will earn 100 points. Fewer
points will be awarded for achieving minimumgoals, while
a point total in excess of 100 will be awarded for achieving
goals above the target. Overall performance on the matrix is
calculated as the sumof the individual scores. The incentive
system allows employees to earn up to 50% of their base
salary as a bonus. To illustrate the incentive matrix plan,
consider the following simpli?ed example. Lucy is an area
branch manager. Her incentive matrix plan consists of
two measures: average transactions per teller (ATT) with
a weight of 40 and deposit production (DP) with a weight
8
Due to a con?dentiality agreement, we are unable to disclose the name
of the organization.
9
Anderson and Widener (2007) note that one-?rm ?eld studies may be
an appropriate research method for testing theory if the unit of analysis is
the employee. Furthermore, the variance and variable of interest is in the
perception of the system design. Robbins (2003, p. 145) states, ‘‘individuals
behave in a given manner based not on the way their external environment
actually is but, rather, on what they see or believe it to be. It is the
employee’s perception of a situation that becomes the basis for his or her
behavior. . . Similarly, issues such as fair pay for work performed, the
validity of performance appraisals, and the adequacy of working conditions
are not judged by employees in a way that ensures common perceptions;
nor can we be assured that individuals will interpret conditions about their
jobs in a favorable light. Therefore, to be able to in?uence productivity it is
necessary to assess how workers perceive their jobs”.
10
Nor are we subject to common measure bias on the outcome variable
due to the construction of the employee–supervisor dyad.
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 311
of 60. The opportunity for ATT is 35, 40, and 45 (because the
weight is the midpoint of the employee’s ‘‘opportunity”)
and the opportunity for DP is 50, 60, and 70. Based on Lucy’s
performance during the period, Lucy must score at least an
85 (35 + 50) to earn a bonus. Lucy’s bonus reaches a maxi-
mumof 50% of her salary at 115 points.
11
Thus, the available
bonus will be distributed over the range between 85 and 115
points.
Survey procedures and sample selection
We created two versions of the survey to manipulate
the ordering of questions within the longer sections of
the survey. While the survey only took about 15 min, this
change was undertaken to avoid any potential fatigue is-
sues impacting a particular survey item. We distributed
the survey to 540 employees at 47 branches. Survey pack-
ets and instructions were given to branch managers at a
monthly meeting and managers in turn distributed the
surveys to their subordinates. Employees completed a sur-
vey regarding their perceptions of the characteristics of the
SPMS and organizational justice within their workplace.
Employees also provided information regarding their
demographics and work history. In addition, supervisors
were asked to complete a 5 min survey for each of their
subordinates that evaluated in-role job performance and
organizational citizenship behavior. Surveys were returned
(separately for employees and supervisors) to the
researchers’ university address to ensure con?dentiality.
Employee numbers were used to match the supervisor
and employee surveys. Participants were assured that this
coding system was for matching purposes only and that
they could not be identi?ed individually.
Three hundred and twenty-three employees returned
surveys for a response rate of about 60%. Seventeen sur-
veys were omitted due to missing data. Supervisors did
not complete performance ratings for 64 of the respon-
dents, which reduced the ?nal sample size to 242. Most
participants were female (88%), Caucasian (87% and 10%
African American), and worked full-time (87%). Partici-
pants held a variety of jobs within the bank, but most were
relationship bankers (26%), tellers (23%), senior tellers
(19%), and ?nancial center managers (10%). Sample job
descriptions are provided in Appendix A. On average,
employees were 43 years old and had worked for the bank
for over eight years. Twenty-nine percent held high school
degrees, 34% had some college education, 11% had an asso-
ciate’s degree, 21% had a bachelor’s degree, and 2% held
graduate degrees.
Variable measures
To establish content validity, we thoroughly reviewed
the prior literature. We used established scales when they
existed; otherwise, we relied on the underlying literature
to provide a basis for formulating survey questions.
Once an initial draft of our survey instrument was pre-
pared, colleagues reviewed it for readability and clarity.
After incorporating their suggestions into the survey, we
met with the Vice-President of the division. We discussed
the project to ensure that the questions were appropriate
and worded in a manner that facilitated understanding.
We also reviewed archival documents including the incen-
tive matrix and a binder of documentation about the plan
to ensure that our understanding of the plan contents was
accurate and our language was clear and not misleading.
After reviewing the survey items with him, adjustments
were made to the wording and content to clarify meanings
and to make the items relevant for employees in that par-
ticular division.
We used data from multiple respondents to construct
our variables; thus, reducing the concern for common
measure bias. Speci?cally, we used responses from super-
visors to measure organizational citizenship behavior and
performance, and from employees to measure the four
other variables of interest.
12
As a precaution, we ran a Hart-
man’s one-factor test on the thirty survey questions asked of
the employees. The factor solution yielded ?ve factors with
eigenvalues >1.0. We also ran a Hartman’s one-factor test on
the 14 questions asked of the supervisors. The factor solu-
tion yielded three factors with eigenvalues >1.0. Overall,
the results of the Hartman one-factor tests, along with using
two different respondents, support the absence of signi?-
cant single-source bias (Podsakoff & Organ, 1986). An abbre-
viated version of the survey is found in Appendix B (note
that several questions are reverse coded).
13
Descriptive sta-
tistics for the multi-item variables are reported in Table 1. In
the sub-sections that follow, we brie?y discuss each
variable.
Degree of technical validity (TC)
We measure the degree of technical validity by modify-
ing McGowan’s (1998) technical characteristics of costing
system information scale so it applies to an SPMS. We
asked the respondents to rate the following ?ve questions
using a seven-point scale where 1 = strongly disagree and
7 = strongly agree: the information provided by the incen-
tive matrix is (1) accurate, (2) accessible, (3) reliable, (4)
timely, and (5) understandable. The Cronbach’s a is
0.872, indicating a high degree of internal consistency. To
review the variability of this construct, we created a mea-
sure using the average of the sum of the ?ve questions. The
construct has a mean of 5.27, variance of 1.05, and ranges
from 1.57 to 7.00.
Extent re?ective of strategic causal model (SCM)
To measure SCM, we developed survey items based on
the underlying literature to capture the extent the employ-
ee’s incentive matrix is linked to the ?rm’s strategic objec-
11
The lower and upper limits can be set at different levels depending on
the desired amount of variability around the target for each measure. The
key is that the deviation must be equal on both the upper and lower sides of
the midpoint.
12
Unfortunately, we do not have both supervisor and subordinate ratings
of the items.
13
To ensure con?dentiality, we disguise the identify of the ?rm that
provided the data for this study and replace any reference to the entity with
a generic ‘‘BANK” in the Appendix.
312 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
Table 1
Descriptive statistics.
Latent construct Item
N
Item
mean
Standard
deviation
Minimum Maximum Cronbach’s
a (std.)
Extent re?ective of strategic causal model (SCM) 0.950
My incentive matrix includes measures that are. . .
SL1: . . .chosen to ?t BANK’s strategy (versus using a set of generic measures) 242 5.40 1.122 1 7
SL2: . . .customized for each individual (versus being the same for all individuals
in the bank)
242 4.60 1.645 1 7
SL3: . . .chosen because they relate to BANK’s strategy 242 5.44 1.175 1 7
My incentive matrix helps me understand. . .
SL4: . . .how BANK’s strategy is to be achieved 242 5.23 1.269 1 7
SL5: . . .the ?nancial consequences of my decisions and actions 242 5.29 1.268 1 7
SL6: . . .trade-offs that I encounter among performance measures when I am
making decisions
242 5.15 1.204 1 7
SL9: . . .the cause-and-effect relationship between my actions and BANK’s
performance measurement outcomes
242 5.38 1.196 1 7
The incentive matrix system. . .
TMS1: . . .receives strong support from upper management 242 5.67 1.238 1 7
TMS2: . . .has been closely tied to the competitive strategies of BANK 242 5.51 1.188 2 7
MISC1: . . .impacts the way I perform my job responsibilities 242 5.32 1.394 1 7
MISC2: . . .is an appropriate system to evaluate my individual performance 242 4.80 1.622 1 7
USE2: . . .provides me with information that re?ects BANK’s strategy 242 5.33 1.300 1 7
USE3: . . .helps me make decisions that create long-term value for BANK, as
opposed to focusing solely on short-term results
242 5.29 1.367 1 7
USE5: . . .is useful because it uses multiple measures to evaluate performance 242 5.39 1.245 1 7
Degree of technical validity (TC) 0.872
The information provided by the incentive matrix is. . .
TC1:. . .accurate 242 5.34 1.249 1 7
TC2:. . .accessible 242 5.49 1.338 1 7
TC3:. . .reliable 242 5.26 1.256 1 7
TC4:. . .timely 242 4.99 1.469 1 7
TC5:. . .understandable 242 5.09 1.554 1 7
Distributive justice (DJ) 0.662
DJ1: My outcomes (e.g., pay, recognition, promotions) re?ect the effort I have
put into my work
242 4.72 1.988 1 7
DJ2: My outcomes are appropriate for the work I have completed 242 4.60 1.544 1 7
DJ3: My outcomes re?ect what I have contributed to BANK 242 3.81 1.963 1 7
DJ4: My outcomes are justi?ed, given my performance 242 4.54 1.575 1 7
Procedural justice (PJ) 0.793
PJ1: I am able to express my views and feelings during procedures used to make
decisions at BANK
242 5.01 1.613 1 7
PJ2: I have in?uence over outcomes arrived at by procedures used to make
decisions at BANK
242 4.12 1.599 1 7
PJ3: Procedures used to make decisions at BANK are applied consistently 242 4.81 1.613 1 7
PJ4: Procedures used to make decisions at BANK are free of bias 242 4.16 1.578 1 7
PJ5: Procedures used to make decisions at BANK are based on accurate
information
242 5.37 1.317 1 7
PJ6: I am able to appeal outcomes arrived at by procedures used to make
decisions at BANK
242 4.40 1.360 1 7
PJ7: Procedures used to make decisions at BANK uphold ethical and moral
standards
242 6.07 1.109 1 7
Organizational citizenship behaviors – organization (OCBO) 0.722
OCBO1: Attendance at work is above the norm 242 5.63 1.581 1 7
OCBO2: Gives advance notice when unable to come to work 242 5.95 1.270 1 7
OCBO3: Takes undeserved work breaks 242 5.85 1.478 1 7
OCBO4: Spends a great deal of time with personal phone calls 242 5.46 1.510 1 7
OCBO5: Complains about insigni?cant things at work 242 5.24 1.752 1 7
OCBO6: Conserves and protects organizational property 242 6.08 0.950 2 7
OCBO7: Adheres to informal rules devised to maintain order 242 5.77 1.052 2 7
In-role performance (PERF) 0.899
PERF1: Adequately completes assigned duties 242 5.92 1.151 1 7
PERF2: Ful?lls responsibilities speci?ed in job description 242 5.85 1.220 1 7
PERF3: Performs tasks that are expected of him/her 242 5.90 1.211 1 7
PERF4: Meets formal performance requirements of the job 242 5.85 1.274 2 7
PERF5: Engages in activities that will directly affect his/her performance
evaluation
242 5.01 1.569 1 7
PERF6: Neglects aspects of the job he/she is obligated to perform 242 5.69 1.573 1 7
PERF7: Fails to perform essential duties 242 6.02 1.387 2 7
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 313
tives (e.g., Kaplan & Norton, 1996). We used a 14-item
scale
14
that captures various details of the performance
measurement system, such as the extent to which the per-
formance measures are selected to ?t organizational strat-
egy and whether the measures increase employee
understanding of organizational functioning. The respon-
dents rated each question on a seven-point scale where
1 = strongly disagree and 7 = strongly agree. The Cronbach’s
a is 0.95, which indicates high internal consistency. To re-
view the variability of this construct, we created a measure
using the average of the sum of the 14 questions. The con-
struct has a mean of 5.23, variance of 1.29, and ranges from
1.00 to 7.00.
Organizational justice (PJ and DJ)
We measured justice using a multidimensional scale
developed by Colquitt (2001). Distributive justice (DJ) con-
sists of a four-item scale assessing the degree to which
employees perceive they are fairly rewarded when consid-
ering their effort and level of performance. The procedural
justice (PJ) scale uses Leventhal’s (1980) six procedural
rules to evaluate employees’ fairness perceptions of proce-
dures used by organizations to make decisions. The seven-
item scale assesses the presence of voice, consistency,
accuracy, appeal processes, bias, and ethical treatment to
determine if procedures are fair. The Cronbach’s as of
0.662 and 0.793, respectively, for DJ and PJ, indicate high
internal consistency.
Organizational citizenship behaviors (OCBOs)
Organizational citizenship behaviors are often charac-
terized as ‘‘extra-role” behaviors (Williams & Anderson,
1991). We use a seven-item scale developed by Williams
and Anderson (1991) to measure organizational citizen-
ship behaviors directed at the organization. Questions in-
clude such items as whether the employee complains
about insigni?cant things at work (a reverse-coded item),
conserves and protects organizational property, and ad-
heres to informal rules devised to maintain order. The
OCBO scale has a Cronbach’s a of 0.722, which indicates
high internal consistency.
Performance (PERF)
We use a seven-item scale developed by Williams and
Anderson (1991) to measure performance. This measure
captures ‘‘in-role” behavior, de?ned as ‘‘behaviors that
are recognized by formal reward systems and are part of
the requirements as described in job descriptions” (Wil-
liams & Anderson, 1991, p. 606). We asked supervisors to
rate their employees’ completion of required job tasks.
Speci?cally, we asked questions such as whether the em-
ployee ful?lls responsibilities speci?ed in the job descrip-
tion (see Appendix A for a complete listing). The
Cronbach’s a of 0.899 indicates high internal consistency.
To establish that the measure is acting plausibly, we corre-
late PERF to an objective measure, which is the employee’s
rating received on his/her most recent annual performance
PDP evaluation. The ?rm’s evaluation process assesses per-
formance using a scale as follows: growth opportunity,
performer, very good performer, or exceptional per-
former,
15
which we converted to a four-point scale. The
bi-variate correlation between PERF and the assessment of
performance based on the employee’s most recent evalua-
tion is 0.549 (p < 0.01). We conclude that the survey mea-
sure of PERF is behaving plausibly.
16
Results
The correlation matrix is presented in Table 2 and pro-
vides univariate results. As expected, we see that TC and
SCM are correlated with each of PJ and DJ. We also see that
DJ and PJ are correlated with OCBOs, and OCBOs are corre-
lated with PERF.
17
We use AMOS 4.0 to estimate the structural equation
model shown in Fig. 1. Moreover, we model the default
assumption in SEM, allowing the exogenous variables to
covary. The measurement and structural models are jointly
estimated using the full information maximum likelihood
estimation method.
18
In Table 3, we present the measure-
ment model results. Since latent variables are not directly
measured, we assign each of thema scale by setting the load-
Table 2
Correlation matrix.
SCM TC DJ PJ OCBO
SCM
TC 0.612
**
DJ 0.331
**
0.367
**
PJ 0.499
**
0.585
**
0.511
**
OCBO 0.122 0.146
*
0.138
*
0.146
*
PERF 0.098 0.039 0.102 0.056 0.665
**
Where SCM – extent re?ective of a strategic causal model; TC – degree of
technical validity; DJ – distributive justice; PJ – procedural justice; OCBO
– organizational citizenship behaviors – organization; PERF – in-role
performance.
*
Signi?cant at 0.05.
**
Signi?cant at 0.00.
14
The survey contains 20 questions regarding the incentive matrix
system. We discarded three of the questions that speci?cally ask about
bene?ts to the organization as opposed to the employee. Since we
developed this construct speci?cally for this study, we ran an exploratory
factor analysis on the remaining 17 questions. The factor analysis revealed
two factors with eigenvalues >1, with three questions crossloading on both
factors. We removed those three questions, reran the factor analysis and
noted that the factor solution only contained one factor, which we used in
the ?nal analysis.
15
In describing the PDP earlier, we noted ?ve potential evaluation ratings
while here we list four ratings. The ‘‘needs immediate improvement” option
in the PDP was not given for any of the respondents in our sample.
16
If supervisors engage in leniency and in?ate subordinates’ evaluations,
the overall mean performance level will increase; however, we are not
interested in exploring the mean effect, rather we are interested in
investigating the association between perceptions of justice and employee
performance. The more signi?cant concern in this type of study is common
measurement bias, which we overcame by creating an employee–super-
visor dyad and removing the concern of common measure bias (i.e., the
employee provides the perceptual measure of justice while the employee’s
supervisor rates his/her performance).
17
We reviewed the variance in?ation factors and tolerance levels and
found no evidence of multicollinearity.
18
A review of kurtosis and skewness reveals that all variables are well
below 10.0 and 3.0, respectively, indicating that the data is within tolerable
levels of univariate normality (Kline, 1998).
314 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
ing of one itemfor each latent construct to 1.0 (i.e., otherwise
known as identi?cation) (Kline, 1998). This con?rmatory
analysis yields standardized loadings greater than 0.30
(unstandardized >0.40), all of which are signi?cant
(p-value < 0.01), providing evidence of convergent validity.
We also note no excessive correlations between the
Table 3
Measurement model. This table reports the descriptive statistics along with the results of maximum likelihood estimation of the measurement model for the
latent variables used in Fig. 1. The referent indicators that de?ne the scale of the latent variables are denoted by an asterisk (Ã).
Latent variables N Mean Standard
deviation
Standardized factor
loading
Extent re?ective of strategic causal model (SCM)
My incentive matrix includes measures that are. . .
SL1: . . .chosen to ?t BANK’s strategy (versus using a set of generic measures) 242 5.40 1.122 0.782
SL2: . . .customized for each individual (versus being the same for all individuals in the bank) 242 4.60 1.645 0.659
SL3: . . .chosen because they relate to BANK’s strategy 242 5.44 1.175 0.784
My incentive matrix helps me understand. . .
SL4: . . .how BANK’s strategy is to be achieved 242 5.23 1.269 0.806
SL5: . . .the ?nancial consequences of my decisions and actions 242 5.29 1.268 0.872
SL6: . . .trade-offs that I encounter among performance measures when I am making decisions 242 5.15 1.204 0.810
SL9: . . .the cause-and-effect relationship between my actions and BANK’s performance
measurement outcomes
242 5.38 1.196 0.864
The incentive matrix system. . .
TMS1: . . .receives strong support from upper management 242 5.67 1.238 0.651
TMS2: . . .has been closely tied to the competitive strategies of BANK 242 5.51 1.188 0.719
MISC1: . . .impacts the way I perform my job responsibilities
*
242 5.32 1.394 0.760
MISC2: . . .is an appropriate system to evaluate my individual performance 242 4.80 1.622 0.746
USE2: . . .provides me with information that re?ects BANK’s strategy 242 5.33 1.300 0.732
USE3: . . .helps me make decisions that create long-term value for BANK, as opposed to focusing
solely on short-term results
242 5.29 1.367 0.760
USE5: . . .is useful because it uses multiple measures to evaluate performance 242 5.39 1.245 0.827
Degree of technical validity (TC)
The information provided by the incentive matrix is. . .
TC1:. . .accurate 242 5.34 1.249 0.849
TC2:. . .accessible 242 5.49 1.338 0.733
TC3:. . .reliable 242 5.26 1.256 0.907
TC4:. . .timely 242 4.99 1.469 0.610
TC5:. . .understandable
*
242 5.09 1.554 0.723
Distributive justice (DJ)
DJ1: My outcomes (e.g., pay, recognition, promotions) re?ect the effort I have put into my work 242 4.72 1.988 0.598
DJ2: My outcomes are appropriate for the work I have completed 242 4.60 1.544 0.741
DJ3: My outcomes re?ect what I have contributed to BANK 242 3.81 1.963 0.338
DJ4: My outcomes are justi?ed, given my performance
*
242 4.54 1.575 0.700
Procedural justice (PJ)
PJ1: I am able to express my views and feelings during procedures used to make decisions at
BANK
242 5.01 1.613 0.631
PJ2: I have in?uence over outcomes arrived at by procedures used to make decisions at BANK 242 4.12 1.599 0.566
PJ3: Procedures used to make decisions at BANK are applied consistently 242 4.81 1.613 0.706
PJ4: Procedures used to make decisions at BANK are free of bias 242 4.16 1.578 0.448
PJ5: Procedures used to make decisions at BANK are based on accurate information 242 5.37 1.317 0.759
PJ6: I am able to appeal outcomes arrived at by procedures used to make decisions at BANK 242 4.40 1.360 0.428
PJ7: Procedures used to make decisions at BANK uphold ethical and moral standards
*
242 6.07 1.109 0.663
Organizational citizenship behaviors – organization (OCBO)
OCBO1: Attendance at work is above the norm 242 5.63 1.581 0.521
OCBO2: Gives advance notice when unable to come to work 242 5.95 1.270 0.590
OCBO3: Takes undeserved work breaks 242 5.85 1.478 0.386
OCBO4: Spends a great deal of time with personal phone calls 242 5.46 1.510 0.465
OCBO5: Complains about insigni?cant things at work 242 5.24 1.752 0.544
OCBO6: Conserves and protects organizational property 242 6.08 0.950 0.615
OCBO7: Adheres to informal rules devised to maintain order
*
242 5.77 1.052 0.604
In-role performance (PERF)
PERF1: Adequately completes assigned duties 242 5.92 1.151 0.880
PERF2: Ful?lls responsibilities speci?ed in job description 242 5.85 1.220 0.939
PERF3: Performs tasks that are expected of him/her 242 5.90 1.211 0.933
PERF4: Meets formal performance requirements of the job 242 5.85 1.274 0.858
PERF5: Engages in activities that will directly affect his/her performance evaluation 242 5.01 1.569 0.311
PERF6: Neglects aspects of the job he/she is obligated to perform 242 5.69 1.573 0.705
PERF7: Fails to perform essential duties
*
242 6.02 1.387 0.809
L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321 315
constructs and a review of the modi?cation indices reveals
no signi?cant crossloadings, evidence of discriminant
validity.
19
In Table 4, we report the results of the structural
model. We evaluate the model using the Chi-square test
statistic (1753.403), the Chi-square divided by the model
degrees of freedom (1.961), the comparative ?t index
(0.976), the root mean square error of approximation
(0.063) and Hoelter’s critical N (137). The model appears
to ?t the data reasonably well.
20
The path coef?cients reported in Table 4 provide evi-
dence on the hypotheses. We ?nd support for our hypoth-
esized model as depicted in Fig. 2. H1a and H2a predict
that the extent to which an SPMS re?ects a strategic causal
model is associated with higher levels of both perceived
procedural justice and distributive justice. We ?nd that
both of these hypotheses are supported (p < 0.05 and
p < 0.01, respectively). H1b and H2b predict that the de-
gree to which an SPMS is technically valid is associated
with higher levels of perceived procedural justice and dis-
tributive justice. We ?nd that both of these hypotheses are
supported (p < 0.01 and p < 0.05, respectively). These re-
sults show that employees’ perceptions of the design char-
acteristics of an SPMS incentive plan enhance perceptions
of organizational justice.
H3 predicts that distributive justice is associated with
procedural justice and the results show that H3 is sup-
ported (p < 0.01). We also ?nd support for H4, which pre-
dicts that procedural justice affects performance through
organizational citizenship behaviors (PJ to OCBO, p < 0.05;
OCBO to PERF, p < 0.01). To provide further evidence on
this mediation hypothesis, we also perform the steps sug-
gested by Baron and Kenny (1986). We ?rst estimate the
direct path from PJ to PERF and ?nd that it is not signi?-
cant. Second, we estimate the direct path from PJ to OCBO
and ?nd a signi?cant relation (standardized b = 0.192,
p < 0.05, two-tailed). Finally, we estimate the direct effects
of both PJ and OCBO on PERF. We ?nd that PJ is insigni?-
cant, while OCBO is signi?cant even after controlling for
PJ (standardized b = 0.802, p < 0.01, two-tailed). According
to Kenny (2006), the necessary steps to show mediation
are steps 2 and 3.
21
Overall, we ?nd evidence that supports our hypothe-
sized model shown in Fig. 1. However, common to cross-
sectional survey research, we rely on theory to establish
the expected directions of the paths between constructs.
Therefore, in the next section we test and provide evidence
on possible reverse causation. We also discuss this issue
further in the conclusion.
Alternative models
Although the model presented in Table 4 is well ?tting,
there is no assurance that it is the only model. The purpose
of this section is to compare the base model to alternative
models to rule out alternative model speci?cations (Kline,
1998). The alternative model speci?cations in Table 5 build
the model by adding one or more paths. When building a
model, a signi?cant X
2
difference test indicates that the
alternative model is better ?tting than the base model
(Kline, 1998).
Alternative model 1 adds the direct paths from each of
the two SPMS characteristics—TC and SCM—to PERF. The
results show that the path coef?cients on the additional
two paths and the X
2
difference test are insigni?cant,
which indicates that the alternative model is not better ?t-
ting than the original model, and, thus we reject alterna-
tive model 1. In alternative model 2, we investigate an
additional direct path from DJ to OCBO. This path does have
some support in the literature, although the conclusion is
that the effect of PJ on OCBOs is much stronger than that
of DJ on OCBOs (Colquitt et al., 2001). The results show that
the path coef?cient from DJ to OCBO is insigni?cant and
the model is not better ?tting. Thus, we reject alternative
model 2.
Table 4
Structural model. This table reports the results of maximum likelihood
estimation of the structural equation model depicted in Fig. 1. Each cell
contains the standardized coef?cient.
Dependent
variable (R
2
)
Independent
variable
Hypothesis Hypothesized
direction
Coef?cient
Base model
DJ (0.274) SCM H2a + 0.312
***
PJ (0.652) SCM H1a + 0.179
**
DJ (++) TC H2b + 0.257
**
PJ (++) TC H1b + 0.365
***
PJ (++) DJ H3 + 0.422
***
OCBO (0.028) PJ H4 + 0.169
**
PERF (++) OCBO H4 + 0.793
***
Model ?t statistics
X
2
1753.403
***
DF 894
CMINDF 1.961
CFI 0.976
RMSEA 0.063
Hoelter’s
critical N
137 (0.01)
++ R
2
is reported on another row since this dependent variable is repeated
in the table.
ÃÃÃ, ÃÃ, and à indicate the signi?cance of the p-value at 0.85. Moreover, we
did perform additional robustness checks by running factor analyses with
both TC and SCM indicators; and one with both PJ and DJ indicators. In both
analyses, the factor solution revealed two factors, all indicators loaded
appropriately, and there were no crossloadings.
20
An insigni?cant Chi-square (Joreskog, 1969), a CMINDF ratio less than 5
(Wheaton, Muthen, Alwin, & Summers, 1977), a CFI close to 1 (Arbuckle &
Wothke, 1999; Bentler, 1990), and an RMSEA of less than 0.08 (Browne and
Cudeck, 1993) indicate good ?t. Although we report the Chi-square test
statistic, recent evidence shows that the Chi-square test statistic is sensitive
to sample size and may be misleading (Byrne, 2001).
21
In our study, the direct link between PJ and PERF is weak (step 1) and
thus the effect of PJ on PERF can only be seen through the effect that PJ has
on OCBO. This ?nding is also consistent with Luft and Shields (2003) who
state that including appropriate paths in causal links can help reveal effects.
We also perform the Sobell’s test for mediation and ?nd that OCBO is a
signi?cant mediator in the relation between PJ and PERF (p < 0.01)
(Preacher & Hayes, 2004).
316 L.L. Burney et al. / Accounting, Organizations and Society 34 (2009) 305–321
Alternative models 3–6 consider the possibility that
PERF drives the employees’ perceptions of justice and the
two characteristics of the SPMS. For example, some argue
that higher ratings are associated with higher perceived
PJ and DJ (e.g., Erdogan, 2002; Stoffey & Reilly, 1997); how-
ever, it is unclear whether it is a measurement issue in that
the scale did not capture both favorable and unfavorable
inequity or whether the relation holds (Erdogan, 2002).
Nevertheless, to rule out the plausible alternative of re-
verse causality, we allow the employee’s performance to
‘‘in?uence” all four of the employee’s perceptions. That is,
in each of the four alternative models, we allow for the
possibility of ‘‘feedback” loops or bidirectional causality.
The results show that the path coef?cients on the addi-
tional paths are not signi?cant and the alternative models
are not better ?tting than the base model, thus we reject
the alternative models.
In sum, after testing six alternative models we conclude
that our base model is the best-?tting model.
22
Validity tests for statistical robustness
To ensure that the SEM model results are robust, we
perform two validity tests. First, an untabulated con?rma-
tory factor analysis on OCBO revealed two dimensions. We
run the SEM model using only the ?rst dimension of OCBO
(items 1, 2, 6, and 7) and ?nd that our statistical inferences
are unchanged. Second, although earlier statistical checks
Table 5
Comparison of base model to alternative models. This table presents the maximum likelihood estimation of the structural equation model in Table 4 to three
alternative model speci?cations. The ?rst alternative builds the model by adding two additional paths. The next 5 alternatives build the model by adding one
additional path. A signi?cant X
2
difference test indicates that the alternative model is better ?tting than the base model. Each cell contains the standardized
coef?cient.
Dependent variable Independent variable Base model Alt. 1 Alt. 2 Alt. 3 Alt. 4 Alt. 5 Alt. 6
DJ SCM 0.312
***
0.314
***
0.309
***
0.311
***
0.311
***
0.352
***
0.353
***
PJ SCM 0.179
**
0.178
**
0.183
**
0.182
**
0.176
**
0.245
***
0.247
***
DJ TC 0.257
**
0.256
**
0.260
***
0.255
**
0.258
**
0.264
***
0.263
***
PJ TC 0.365
***
0.367
***
0.363
***
0.366
***
0.368
***
0.373
***
0.370
***
PJ DJ 0.422
***
0.423
***
0.419
***
0.417
***
0.431
***
0.449
***
0.449
***
OCBO PJ 0.169
**
0.197
**
0.091 0.151
*
0.218
**
0.157
*
0.153
***
PERF OCBO 0.793
***
0.811
***
0.794
***
0.792
***
0.798
***
0.792
***
0.792
***
Alternative model speci?cation
PERF SCM – 0.026 – – – – –
PERF TC – À0.112 – – – – –
OCBO DJ – – 0.103 – – – –
DJ Perf – – – 0.045 – – –
PJ Perf – – – – À0.072 – –
TEC Perf – – – – – À0.014 –
SPMS Perf – – – – – – À0.003
Model ?t
X
2
1753.403 1750.227 1752.789 1753.037 1752.010 1875.666 1875.698
p-Value 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Df 894 892 893 893 893 893 893
CMINDF 1.961 1.962 1.963 1.963 1.962 2.098 2.098
CFI 0.976 0.976 0.976 0.976 0.976 0.972 0.972
RMSEA 0.063 0.063 0.063 0.063 0.063 0.068 0.068
++ R
2
is reported on another row since this dependent variable is repeated in the table.
ÃÃÃ, ÃÃ, and à indicate the signi?cance of the p-value at