Journal of Applied Psychology Copyright 2000 by the American Psychological Association, Inc.

2000, Vol. 85, No. 3, 349-360 0021-9010/00/S5.00 DOI: 10.I037//0021-9010.85.3.349

Development of a Measure of Workplace Deviance

Rebecca J. Bennett Sandra L. Robinson

University of Toledo University of British Columbia

The purpose of this research was to develop broad, theoretically derived measure(s) of deviant behavior
in the workplace. Two scales were developed: a 12-item scale of organizational deviance (deviant
behaviors directly harmful to the organization) and a 7-item scale of interpersonal deviance (deviant
behaviors directly harmful to other individuals within the organization). These scales were found to have
internal reliabilities of .81 and .78, respectively. Confirmatory factor analysis verified that a 2-factor
structure had acceptable fit. Preliminary evidence of construct validity is also provided. The implications
of this instrument for future empirical research on workplace deviance are discussed.

Workplace deviance is a pervasive and expensive problem for Understanding Workplace Deviance
organizations. For example, 75% of employees have reportedly
stolen from their employer at least once (McGurn, 1988), and it Workplace deviance has been defined as voluntary behavior that
has been estimated that 33% to 75% of all employees have en- violates significant organizational norms and, in so doing, threat-
gaged in behaviors such as theft, fraud, vandalism, sabotage, and ens the well-being of the organization or its members, or both
voluntary absenteeism (Harper, 1990). In recent studies, almost (Robinson & Bennett, 1995). Workplace deviance refers to volun-
25% of an employee sample indicated knowledge of illicit drug tary behavior in that employees either lack motivation to conform
use among coworkers during the past year (Lehman, Wolcom, & to, and/or become motivated to violate, normative expectations of
Simpson, 1990), 42% of a surveyed sample of women reported the social context (Kaplan, 1975). Organizational norms consist of
basic moral standards as well as other traditional community
experiencing sexual harassment at work (Webb, 1991), and 7% of
standards, including those prescribed by formal and informal or-
a sample of employees reported being victims of physical threats
ganizational policies, rules, and procedures (Feldman, 1984).
(Northwestern Life Insurance Company, 1993).
For scales to be valid, it is essential that there be at least a
It is not surprising that the prevalence of workplace deviance
tentative theoretical model to guide scale development (Churchill,
poses a serious economic threat to organizations. The annual costs
1988; DeVellis, 1991). It is argued here that deviant behaviors fall
of workplace deviance have been estimated to be as high as $4.2
into clusters or families (Robinson & Bennett, 1997; Roznowski &
billion for workplace violence alone (Bensimon, 1994), $40 to Hulin, 1992). Any specific deviant behavior can be placed into one
$120 billion for theft (Buss, 1993; Camara & Schneider, 1994), of these behavioral families. We make this assumption because we
and $6 to $200 billion for a wide range of delinquent organiza- believe that although there are a myriad of different manifestations
tional behavior (Murphy, 1993). of deviant behaviors, research suggests that some of these mani-
Despite the prevalence and costs of workplace deviance, our festations are similar in nature to one another, share similar ante-
current understanding of workplace deviance remains limited, and cedents, and may thus be functional substitutes for one another
much empirical research has yet to be done. This empirical re- (i.e., they serve the same goals; Robinson & Bennett, 1997).
search may be enhanced by the availability of a validated measure Research suggests a wide range of reasons why employees
of workplace Deviance. The purpose of this study is to produce engage in deviant behavior (Bennett, 1998a, 1998b; Robinson &
such a measure. Bennett, 1997; Robinson & Greenberg, 1999), ranging from reac-
tions to perceived injustice, dissatisfaction, role modeling, and
thrill-seeking. Yet, deviant organizational behavior is distinct in
that it is usually behavior that is very constrained in the workplace.
Employees in a given time period or context are very limited in
Rebecca J. Bennett, Department of Management, University of Toledo; terms of the type of deviant behavior in which they can engage.
Sandra L. Robinson, Faculty of Commerce and Business Administration, Thus, they may be motivated to engage in deviance, but that
University of British Columbia, Vancouver, British Columbia, Canada. deviance will take different manifestations depending on the con-
Support for this research was provided by an Academic Challenge Grant straints of the situation. We would argue then that an employee
Award from the Department of Management at the University of Toledo.
may choose from among deviant behaviors within a family that are
We thank Lynn Shore, Paul Spector, and Linn Van Dyne for their helpful
advice.
functionally equivalent, choosing the one that is least constrained,
Correspondence concerning this article should be addressed to Rebecca most feasible, or least costly, given the context (Robinson &
J. Bennett, Department of Management, University of Toledo, Toledo, Bennett, 1997).
Ohio 43606-3390. Electronic mail may be sent to rebecca.bennett@ If an individual engages in one behavior from a family, he or she
utoledo.edu. is more likely to engage in another behavior from that family than
349
350 BENNETT AND ROBINSON

to engage in a behavior within another family. We assume em- within each content area, just as the items on our deviance scale
ployees may engage in behavioral switching within families be- vary in terms of the type of deviance represented as well as their
cause the behaviors within each are substitutable and functionally seriousness. A summation of items reflecting each type of devi-
equivalent (Robinson & Bennett, 1997). Employees then may ance should indicate the participation levels of each form of
engage in one or several behaviors from a wide set. deviance, much like summing the numbers of math items versus
If we apply the family of behavior metaphor to deviant behav- spelling items would indicate the knowledge level in those differ-
iors, what might those families of deviance look like? Robinson ent areas.
and Bennett (1995, 1997) identified a typology of workplace Given the above, we propose that workplace deviance can be
deviance that may provide insight into this question. They argued captured with two general factors: interpersonal deviance and
that an important distinction between types of deviance was organizational deviance (both serious and minor forms of each
whether the deviance was directed or targeted at either the orga- type are represented within each family).
nization (organizational deviance) or at members of the organiza-
tion (interpersonal deviance). The target of deviance is an impor-
Present Study
tant element of deviance for several reasons. First, it is posited that
this dimension of deviance identifies an important qualitative To develop a measure of workplace deviance, three multiphase
difference between deviant acts; individuals prone toward devi- studies were conducted. In the first study, a pool of 314 deviant
ance directed at the organization are likely to be different than workplace behaviors was generated and these behaviors were
those individuals prone toward deviance directed at other individ- reviewed and assessed by a panel of experts. In the second study,
uals. Numerous behavioral constructs, from conflict to dissatisfac- a subset of 58 of the deviant behavior items was further refined
tion behavior to citizenship behavior, have been classified in terms to 23 items by analyzing the interitem correlations, variances, and
of their targets (C. D. Fisher & Locke, 1992; Organ, 1988, 1990; factor loadings of each item. Finally, a third study was conducted
Roznowski & Hulin, 1992; Williams & Anderson, 1991). The using confirmatory factor analysis (CFA) to verify the proposed
domain of workplace deviance is no exception. Most conceptual dimensionality of the remaining 23 items and to begin the process
approaches to workplace deviance have explicitly acknowledged of construct validation.
that deviance may be directed at either the organization itself or its
members, or both (Baron & Neuman, 1996; Giacalone & Green-
berg, 1997; O'Leary-Kelly, Griffin, & Glew, 1996; Robinson &
Study 1: Instrument Development
Greenberg, 1999; Skarlicki & Folger, 1997). Both Green (1997) Phase 1: Item Generation
and Turner and Stephenson (1993) have conceptualized organiza-
tional crimes in terms of targets. A similar distinction has been The purpose of this phase of the study was to create a large,
drawn regarding conceptualizations of more specific types of de- inclusive pool of exemplars of deviant behavior, so that together
viant acts as well. For example, Greenberg and Scott (1996) have they reflect the domain of behaviors meeting our definition of
distinguished between employee theft directed at other employees workplace deviance. Consistent with this purpose, types of deviant
(e.g., taking money from a coworker's wallet) and that directed at behaviors were generated in a variety of ways. Two separate
the organization (e.g., taking money from the cash register). Using samples of employees were asked for examples of various forms of
the above example, it makes sense to avoid referring to both workplace deviance. Previous research and theory were also ex-
behaviors as forms of theft (e.g., Snyder, Blair, & Arndt, 1990) and amined for further examples of behaviors that fit our definition of
attempting to interpret them in a similar fashion. Indeed, despite workplace deviance. From these sources, an initial pool of 314
similarities between them, there is good reason to believe that examples of deviant behaviors was obtained, taking into account
these two forms of deviance are motivated by different factors considerable redundancy.
(Bies, Tripp, & Kramer, 1997; Giacalone, Riordan, & Rosenfeld, Procedure 1. We recruited 70 respondents from four sources
1997). in Toledo, Ohio: a university office (n = 7), a technical staff office
Robinson and Bennett (1997) also noted that deviance may vary within an industrial company (n = 10), a neighborhood (n = 38),
along a continuum of severity, from minor forms of deviance to and an evening masters of business administration (MBA) class
more serious forms. Unlike the interpersonal versus organizational (n = 15). All of the respondents worked full time. Of the partic-
distinction, however, this is more a quantitative distinction rather ipants, 61% were women. These employees had an average age
than a qualitative one. Thus, although one would expect that of 37 years (SD = 14.69) and had worked an average of 15.69
interpersonal and organizational deviance would fall into distinct years (SD = 12.10).
clusters or families representing two qualitatively different forms Respondents were asked to describe two incidents of "someone
of deviance, both families of deviance contain both serious and at work engaging in something considered to be deviant at their
minor forms of deviance. Serious and minor deviant behaviors workplace (i.e., something that is considered to be wrong)." A total
would not, by themselves, reflect two different types of deviance. of 45 unique behaviors were generated in this manner.
Thus, for example, both spreading rumors and physical violence Procedure 2. In all, 62 upper-level undergraduate students
would fall into the interpersonal deviance family, just as both also provided examples of workplace deviance that they had
sabotaging equipment and littering one's work environment would observed or experienced. Their average age was 24 years
fall into the organizational deviance family. This distinction is (SD = 5.47), and 42% were women. All worked at least 20 hr per
analogous to individual items on an achievement test. Items on the week in a variety of jobs, such as retail clerks, bank tellers,
test vary qualitatively in terms of knowledge content area (e.g., waitstaff, sorters and loaders for a major shipping company, man-
spelling vs. math items) and quantitatively in terms of difficulty agers for small and large companies, clerical positions, and child
 MEASURING WORKPLACE DEVIANCE 351

care. Respondents reported an average of 6.46 years of work Procedure

experience (SD = 5.48).
Respondents were asked to describe an example of each type of Professors who were not involved in the research asked their
deviant workplace behavior that Robinson and Bennett (1995, students to participate. A graduate student briefly explained that
1997) identified. From this procedure, 45 unique behaviors were the surveys were both voluntary and completely anonymous. Stu-
generated. dents were given 20 min of class time to complete the survey.
Procedure 3. The researchers also independently generated 68 The nonstudent respondents were approached by a graduate
items on the basis of 23 previously published theoretical and student, who briefly explained that a professor at the university
empirical investigations of deviant behaviors in the workplace. was conducting research on attitudes and behaviors at work and
After eliminating redundant behaviors, 113 examples of deviant would appreciate their completion of a brief survey. Because no
behaviors in organizations remained. names or code numbers were associated with the surveys, respon-
dents were assured that their responses were anonymous.
The survey was composed of a list of the 58 deviant workplace
Phase 2: Item Review behavior items. Respondents were asked to indicate on a 7-point
Likert scale the extent to which they had engaged in each of the
Sample. The 113 items were reviewed by nine judges with behaviors in the last year. The scale anchors were as follows: 1
different but related areas of expertise: industrial psychology, labor (never), 1 (once a year), 3 (twice a year), 4 (several times a
relations, marketing, organizational behavior, strategic manage- year), 5 (monthly), 6 (weekly), and 7 (daily). Respondents also
ment, and organizational communication. Seven of the judges answered several demographic questions.
were academics with a doctoral degree and management consult-
ing experience, and the remaining two judges were practicing
managers. Phase I: Item Selection Process
Procedure. The judges reviewed the items on the basis of
several criteria. First, the judges rated each behavior in terms of Evaluation of items was made on the basis of two criteria:
whether it was consistent with the definition of deviance used here; item-total correlations and item variances. Because a scale should
that is, whether the item reflected behavior that is voluntary; be composed of highly interrelated items (DeVellis, 1991), items
with high interitem correlations with items theorized to be in the
behavior that is potentially harmful to organizations or its mem-
same behavioral family were selected to be included in the sub-
bers, or both; and behavior that would violate significant norms in
scales. The variance of the workplace deviance items was also
most organizations. Second, the judges rated each behavior in
considered in determining which items to select for the scales.
terms of its clarity and consciousness. Third, the judges rated the
Because items with extremely low variances do not allow discrim-
degree to which each item reflected a behavior that would be
ination between individuals on the construct of interest (DeVellis,
relevant to a wide variety of occupations and organizations. Judges
1991), items with variances below 1.5 were eliminated. This
used a 7-point Likert scale ranging from 1 (strongly disagree) to 7
process resulted in the removal of 30 items, leaving us with 28.
(strongly agree) to rate the behaviors on each of these criteria.
The means, standard deviations, and participation rates for
Items that received a mean score of 3.0 or less on any of the rating these 28 items are presented in Table 1. It is noteworthy that 19 of
dimensions were either rewritten or eliminated. As a result of this the 28 behaviors had participation rates of 50% or more, indicating
process, 58 items survived. that 50% or more of the respondents had engaged in those behav-
iors in the last year.
Study 2: Instrument Refinement
Phase 2: Preliminary Factor Analysis
Sample
We conducted a principal factor analysis to analyze the interre-
A total of 226 respondents participated in Study 2. Of these lationships of the items and to suggest additional items for deletion
respondents, 126 were full-time employees and 100 were MBA (Ford, MacCallum, & Tail, 1986; Schwab, 1980). Our guiding
students. Inasmuch as the MBA students were entering their first theory suggested two related forms of workplace deviance, and
term in the program, all had worked full time a few months prior hence, we used a principal axis factoring procedure with oblique
to their participation. The average age (for the combined sample) rotation to impose a two-factor solution (Ford et al., 1986; Kim &
was 28.3 years (SD = 7.09), the average years of work experience Mueller, 1978). In order to ensure that each item represented the
was 7.14 (SD = 6.40), and they had worked an average of 44.73 construct underlying each factor, we used a factor weight of .40 as
(SD = 12.51) hr per week. Of the respondents, 44% were women. the minimum cutoff. Second, we required each item to be clearly
The respondents worked in the following industries: retail defined by only one factor and, thus, maintained that the difference
(19.5%), manufacturing (11.5%), public or government service between weights for any given item was more than .10 across
(10.2%), hotel and restaurant (5.3%), education (4.4%), and other factors. Four items were dropped because they did not meet these
service industries (45.6%); 3.5% did not report the industry in criteria, thus 24 items remained. As shown in Table 2, these two
which they worked. Respondents worked in the following occu- factors appear to represent interpersonal and organizational forms
pations: professional (19%), service provider (18%), manager of deviance. From these results, we developed two scales: One
(16%), technician (15%), clerical (12%), sales (11%), crafts/repair labeled Organizational Deviance was composed of 16 items, and
(3%), and production (1%). one labeled Interpersonal Deviance was composed of 8 items.
352 BENNETT AND ROBINSON

Table 1
Means, Standard Deviations, and Participation Rates of Deviant Workplace
Behaviors in Study 2

Item M SD Participation rate"

1. Worked on a personal matter instead of work for your 4.40 1.96 84.3
employer
2. Taken property from work without permission 2.39 1.69 51.8
3. Spent too much time fantasizing or daydreaming instead of 4,00 2.11 77.4
working
4. Made fun of someone at work 4.29 2.29 77.8
5. Falsified a receipt to get reimbursed for more money than 1.69 1.34 24.6
you spent on business expenses
6. Said something hurtful to someone at work 2.40 1.60 55.2
7. Taken an additional or a longer break than is acceptable at 3.95 1.98 78.5
your workplace
8. Repeated a rumor or gossip about your company 3.66 1.91 72.5
9. Made an ethnic, religious, or racial remark or joke at work 2.69 1.96 52.5
10. Come in late to work without permission 3.39 1.91 70.0
11. Littered your work environment 1.97 1.80 28.5
12. Cursed at someone at work 2.73 2.09 50.5
13. Called in sick when you were not 2.30 1.36 57.8
14. Told someone about the lousy place where you work 2.93 1.99 58.9
15. Lost your temper while at work 3.44 1.82 78.8
16. Neglected to follow your boss's instructions 2.78 1.78 60.6
17. Intentionally worked slower than you could have worked 2.71 1.90 54.1
18. Discussed confidential company information with an 1.90 1.51 33.3
unauthorized person
19. Left work early without permission 2.65 1.84 51.9
20. Played a mean prank on someone at work 1.94 1.56 35.7
21. Left your work for someone else to finish 2.21 1.52 48.6
22. Acted rudely toward someone at work 2.70 1.85 53.0
23. Repeated a rumor or gossip about your boss or coworkers 3.60 1.87 69.1
24. Made an obscene comment at work 2.81 2.23 48.4
25. Used an illegal drug or consumed alcohol on the job 1.70 1.44 25.9
26. Put little effort into your work 2.94 1.84 64.0
27. Publicly embarrassed someone at work 1.84 1.43 33.9
28. Dragged out work in order to get overtime 1.77 1.48 26.0

Note. Responses ranged from 1 (never) to 7 (daily). N = 226.

a
Percentage of respondents who indicated that they had participated in the behavior at least once in the last year.

Study 3: Instrument Validation income ranged from under $10,000 to above $100,000, with the
mode being a salary between $15,000 and $30,000. Occupational
Sample titles of these respondents were as follows: 26% (26%) profession-
Two samples were used. One larger sample (n = 352) was used als; 15% (16%) managers, 9% (9.2%) clerical workers, 8%
for the purposes of CFA, and a subset of that sample (n = 133) was (10.7%) production workers, 8% (6.1%) service providers, 7%
used for the purposes of preliminary construct validity testing. (5.3%) technicians, 6% (6.1%) salespersons, and 4% (4.6%) crafts-
Thus, the larger sample provided self-reports of their deviant men or repairmen; 16% (16%) did not specify their occupational
behavior, whereas the subsample provided both self-reports of title.
their deviant behavior and self-reports of other constructs. The
reason for this split was to balance our need to obtain a sufficiently Procedure
large sample to conduct CFA with our need to keep the survey as A total of 1,000 individuals were chosen from the Toledo, Ohio,
short as possible so as to increase our response rate. Descriptive area phone directory using a systematic random sampling proce-
statistics of these samples are provided below (the statistics in dure (Fowler, 1993). In order to maximize our response rate, we
parentheses apply to the subsample). followed most of the recommendations put forth by Dillman's
The sample was composed of 352 (133) full-time employees (1972, 1978) total design method. Each potential respondent was
from the Toledo, Ohio, area. Of the respondents, 44% (49%) were mailed a cover letter, a survey, and a self-addressed stamped
women. Average age was 43.51 years (SD = 10.5; 43.58 years, envelope to their home address. A follow-up postcard was sent 2
SD = 10.94). The education level of the respondents varied: 29% weeks after the survey to encourage participation. Another letter,
(27%) had only high school level education, 39% (37%) had some survey, and self-addressed stamped envelope were mailed 4 weeks
college training, 23% (28%) had a bachelor's degree, and 9% (8%) after the initial survey was sent.
had a graduate degree. Average years of work experience for this All of the respondents were given a survey that included a list of
sample was 23.41, SD = 10.17 (23.52, SD = 10.57). Annual the 24 deviant workplace behaviors. Respondents were asked to
 MEASURING WORKPLACE DEVIANCE 353

Table 2
Principal Axis Factor Analysis (Oblimin Rotation)

Factor loadings

Organizational Interpersonal
Item deviance deviance

1. Worked on a personal matter instead of work for your employer .50 .23
2. Taken property from work without permission .56 .33
3. Spent too much time fantasizing or daydreaming instead of .61 .38
working
4. Made fun of someone at work .32 .71
5. Falsified a receipt to get reimbursed for more money than you .43 .16
spent on business expenses
6. Said something hurtful to someone at work .26 .57
7. Taken an additional or a longer break than is acceptable at your .68 .37
workplace
8. Repeated a rumor or gossip about your company .65 .54
9. Made an ethnic, religious, or racial remark or joke at work .25 .55
10. Come in late to work without permission .66 .41
11 . Littered your work environment .45 .19
12 . Cursed at someone at work .38 .63
13 . Called in sick when you were not .49 .44
14 . Told someone about the lousy place where you work .48 .36
15 . Lost your temper while at work .33 .44
16. Neglected to follow your boss's instructions .65 .46
17 . Intentionally worked slower than you could have worked .65 .40
18 . Discussed confidential company information with an .53 .24
unauthorized person
19 . Left work early without permission .68 .31
20 . Played a mean prank on someone at work .42 .58
21 . Left your work for someone else to finish .56 .42
22 . Acted rudely toward someone at work .49 .71
23 . Repeated a rumor or gossip about your boss or coworkers .32 .30
24 . Made an obscene comment at work .52 .61
25 . Used an illegal drug or consumed alcohol on the job .54 .26
26 . Put little effort into your work .68 .44
27 . Publicly embarrassed someone at work .30 .56
28 . Dragged out work in order to get overtime .50 .30
Eigenvalue 8.76 2.18
% variance explained (unrotated factors) 31.28 7.79
% variance explained (rotated factors) 29.05 5.54

Note. Numbers in boldface indicate dominant factor loadings.

indicate on a 7-point Likert scale ranging from 1 (never) to 7 reported in Table 3. It is worth noting that the participation rates
(daily) the extent to which they had engaged in each of the for these behaviors were relatively high, but were significantly
behaviors in the last year. Respondents also answered several lower than the participation rates reported in Study 2.
demographic questions. The subsample of 133 respondents were The construct-validation approach used consisted of several
also asked to respond to a variety of self-report attitudinal and stages: (a) demonstrating dimensionality and internal consistency,
behavioral scales. No names or code numbers were associated with (b) demonstrating convergent validity by showing high correla-
the surveys, so respondents were assured that their responses were tions with alternative measures of similar constructs, and (c) dem-
completely anonymous. onstrating discriminant validity by showing not-too-high correla-
Respondents were asked to mail the completed survey directly tions with unrelated constructs (Campbell & Fiske, 1959; Schwab,
back to the researchers. Only those respondents who were cur- 1980). Each of these stages is discussed below.
rently employed full time were eligible to participate. As such,
potential respondents who were unemployed or retired were asked Phase 1: Dimensionality
to return the survey uncompleted with an indication of their
employment status. In all, 542 surveys were returned; 190 of these We performed a CFA to cross-validate the two-factor solution
were returned uncompleted by potential respondents who were not obtained in the exploratory factor analysis. LISREL 8 (Joreskog &
currently employed full time. In summary, 352 completed, usable S6rbom, 1993) was used to evaluate the fit of the measurement
surveys were obtained (a response rate of 43%). The means, model. The covariance matrix from the random sample of Toledo
standard deviations, and participation rates for these items are residents was used as input for the CFA. We followed Bollen's
354 BENNETT AND ROBINSON

Table 3
Means, Standard Deviations, and Participation Rates of Deviant Workplace
Behaviors in Study 3

Item M SD Participation rates'

1. Worked on a personal matter instead of work for your 2.66 1.70 60.0
employer
2. Taken property from work without permission 1.43 0.93 20.5
3. Spent too much time fantasizing or daydreaming instead of 2.39 1.75 47.3
working
4. Made fun of someone at work 2.93 1.95 57.3
5. Falsified a receipt to get reimbursed for more money than 1.07 0.46 3.8
you spent on business expenses
6. Said something hurtful to someone at work 1.79 1.15 38.0
7. Taken an additional or a longer break than is acceptable at 2.50 1.77 51.8
your workplace
8. Made an ethnic, religious, or racial remark or joke at work 2.17 1.65 43.0
9. Come in late to work without permission 1.69 1.22 32.7
10. Littered your work environment 1.42 1.22 14.4
11. Cursed at someone at work 1.87 1.53 29.8
12. Told someone about the lousy place where you work 2.12 1.59 35.8
13. Lost your temper while at work 2.64 1.44 69.2
14. Neglected to follow your boss's instructions 1.83 1.34 37.3
15. Intentionally worked slower than you could have worked 1.81 1.46 30.6
16. Discussed confidential company information with an 1.23 0.65 13.5
unauthorized person
17. Left work early without permission 1.44 1.03 21.3
18. Played a mean prank on someone at work 1.17 0.72 7.3
19. Left your work for someone else to finish 1.30 0.77 14.9
20. Acted rudely toward someone at work 1.80 1.20 39.6
21. Used an illegal drug or consumed alcohol on the job 1.10 0.54 3.3
22. Put little effort into your work 1.75 1.34 3.1
23. Publicly embarrassed someone at work 1.20 0.61 10.8
24. Dragged out work in order to get overtime 1.12 0.61 4.7

Note. Responses ranged from 1 (never) to 7 (daily). N — 352.

a
Percentage of respondents who indicated that they had participated in the behavior at least once in the last year.

(1989) recommendation to interpret multiple indexes of fit. We The 7 interpersonal deviance items were aggregated to form a
examined LISREL fit statistics, including the chi-square test and scale, as were the 12 organizational deviance items. The scales
the root-mean-square residual (RMSR). Supplementing these in- showed acceptable internal consistency, with Cronbach's alpha
dexes, we examined the normed fit index (NFI; Bentler & Bonnet, reliabilities of .81 for the Organizational Deviance scale and .78
1980), the goodness of fit index (GFI; Joreskog & Sorbom, 1986) for the Interpersonal Deviance scale. The average total score for
and the comparative fit index (CFI, Bentler, 1990). The latter two the Organizational Deviance scale was 19.86, (SD = 7.31). The
indexes have been shown to be relatively stable in sample sizes average total score for the Interpersonal Deviance scale was 12.98
smaller than 250 (Hu & Bentler, 1995). (SD = 5.97). The correlation between these two scales was mod-
The CFA showed only a moderate fit for the two-factor model, erate (r2 — .46, p < .01). This suggests that the two types of
^(251, N = 143) = 422.20, p < .05, RMSR = .07, GFI = .80, workplace deviance are distinct but related. The final scale items
CFI = .75, NFI = .56. However, inspection of the modification are reported in the Appendix.
indexes, standardized residuals, and factor loadings indicated that
a better fit could be obtained by removing five problematic indi- Phase 2: Convergent and Discriminant Validity
cators. The items "worked on a personal matter instead of work for Assessment
your employer," "left work early without permission," "told some-
one about the lousy place you work," and "left your work for A measure has convergent validity to the extent that it covaries
someone else to finish" were dropped from the Organizational with other measures purported to measure the same or similar
Deviance scale. "Lost your temper at work" was dropped from the constructs (Campbell & Fiske, 1959). To assess the convergent
Interpersonal Deviance scale. Deleting these items improved validity of our deviant workplace behavior scales, we first com-
model fit, x*(l47, N = 143) = 198.37, p < .00, RMSR = .05, pared the scores on our scales with scores on other scales measur-
GFI = .87, CFI = .90, and NFI = .88, and resulted in nearly all ing deviant workplace behaviors or behaviors that are conceptually
of the indexes falling within acceptable ranges (Anderson & Gerb- similar to workplace deviance: modified scales of production
ing, 1988). deviance and property deviance developed by Hollinger and Clark
 MEASURING WORKPLACE DEVIANCE 355

Table 4
Correlations Between Interpersonal and Organizational Deviance Scales and Measures of
Similar Behaviors, Theoretically Related Behaviors, and Dissimilar Behaviors

Observed correlations

Interpersonal Organizational
Comparison measure deviance deviance

Similar behaviors
Hollinger & Clark (1982, 1983a, 1983b)
Property deviance .29** .59'
Production deviance .39* .70*
Lehman & Simpson (1992)
Physical withdrawal .23* .79'
Psychological withdrawal .40** .65'
Antagonistic work behavior .62** .42'
Neglect (Farrell & Rusbult, 1986) .39** .48'
Theoretically related behaviors
Frustration (Peters, O'Conner, & Rudolf, 1980) .21* .01
Procedural justice (Niehoff& Moorman, 1993) -.33** -.32**
Distributive justice (Niehoff & Moorman, 1993) -.12 -.08
Interactional justice (Niehoff & Moorman, 1993) -.35** -.33
Normlessness (Dean, 1961) .21* .13
Machiavellianism (Christie & Geis, 1970) .39** .26*
Citizenship behavior (Podsakoff, MacKenzie, Moorman, & Fetter, 1990)
Courtesy .41** -.22*
Conscientiousness -.28** -.35**
Dissimilar behaviors
Voice (Parrel & Rusbult, 1986) -.09 -.14
Exit (Farrell & Rusbult, 1986) .11 .17
Loyalty (Farrell & Rusbult, 1986) -.21* -.13

Note. N = 133.
*p < .05. **p < .01.

(1982, 1983a, 19831)),1 and scales developed by Lehman and were more closely related to scores on the Organizational Devi-
Simpson (1992) to measure physical withdrawal, psychological ance Scale (r = .65, p < .01) than to scores on the Interpersonal
withdrawal, and antagonistic work behaviors. Deviance Scale (r = .40, p < .01). Scores on the Physical With-
Evidence for convergent validity would be demonstrated if drawal Scale were also more closely related to the Organizational
scores on these scales were relatively highly correlated with scores Deviance Scale (r = .79, p < .01) than to the scores on the
on our scales of workplace deviance. Furthermore, we would Interpersonal Deviance Scale (r = .23, p < .05). Finally, the
expect scores on our Organizational Deviance Scale and scores on Property and Production Deviance Scales were also, as expected,
our Interpersonal Deviance Scale to have differential relationships more closely related to the Organizational Deviance Scale (r =
with scores on these related scales. Specifically, we would expect .59, p < .01; r = .70, p < .01) than to the Interpersonal Deviance
that scores on the Interpersonal Deviance Scale, compared with Scale (r = .29, p < .01; r = .39, p < .01). In summary, scores on
scores on the Organizational Deviance Scale, to be more strongly our two workplace deviance scales are strongly to moderately
related to scores on the Antagonistic Work Behaviors Scale, as correlated with scores on these similar scales in a predictable
both of these scales include interpersonal behaviors that are po- fashion, exhibiting a mean true score correlation of .50.
tentially harmful to other individuals. Conversely, we would ex-
We also examined the relationship between our instruments and
pect scores on the Organizational Deviance Scale, compared with
theoretically relevant constructs. We would expect these relation-
scores on the Interpersonal Deviance Scale, to be more strongly
ships to be moderate, but not as strong as the relationships found
related to scores on the Production Deviance Scale, the Property
between the instrument being validated and other measures pur-
Deviance Scale, the Psychological Withdrawal Scale, and the
porting to assess the same or similar constructs. Specifically, we
Physical Withdrawal Scale, because all of these scales reflect
looked at the degree of association between the scores on our
organizationally directed forms of workplace deviance. Such a
pattern of findings would also provide some evidence of discrimi- workplace deviance scales and scores on the following scales:
nant validity. Frustration (Peters, O'Connor, & Rudolf, 1980); Procedural, Dis-
As Table 4 reveals, the relationships between our measures and
these similar measures follow our predictions. Scores on the An- 1
Hollinger & Clark's (1982) scales of production and property deviance
tagonistic Work Behaviors Scale were more closely related to were designed for specific occupations in retail, hospitals, and manufac-
scores on the Interpersonal Deviance Scale (r = .62, p < .01) than turing environments. We modified these scales to make them applicable to
scores on the Organizational Deviance Scale (r = .42, p < .01). the wide range of occupations in which the respondents of this study
Scores on the Psychological Withdrawal Scale, on the other hand, worked.
356 BENNETT AND ROBINSON

tributive and Interactional Injustice (Niehoff & Moorman, 1993); scores on the Machiavellianism Scale would be more highly re-
Normlessness (Dean, 1961); Machiavellianism (Christie & Geis, lated to scores on the Interpersonal Deviance Scale and less related
1970); and two subscales of Organizational Citizenship Behavior to scores on the Organizational Deviance Scale. We found that
(Podsakoff, MacKenzie, Moorman, & Fetter, 1990). See Table 4 scores on the Machiavellianism Scale were related to scores on our
for the results. Interpersonal Deviance Scale (r = .39, p < .01) and to scores on
Frustration. Strong relationships have been found between our Organizational Deviance Scale (r = .26, p < .05).
frustration and aggression in general (Berkowitz, 1978; Dollard, Organizational citizenship behavior. Organizational citizen-
Doob, Miller, Mowrer, & Sears, 1939). In the workplace, frustra- ship behavior (OCB) refers to extra-role behavior that promotes
tion has been found to be associated with spreading rumors, organizational effectiveness but is not explicitly recognized by an
vandalism, theft, aggression, and sabotage (Spector, 1975; Storms organization's reward system (Organ, 1988, 1990). Organizational
& Spector, 1987). As such, we would expect scores on a frustration deviance is volitional behavior that is potentially harmful to the
scale to be associated both with acts directly harmful to the organization, whereas OCB reflects prosocial, voluntary behavior
organization (as measured by our Organizational Deviance Scale)
that is beneficial to the organization. As such, we would expect
as well as with acts directly harmful to other individuals within the
workplace deviance and OCB to be negatively and moderately
organization (as measured by our Interpersonal Deviance Scale).
related.
As predicted, scores on the Frustration Scale were found to be
We would also expect our two scales of workplace deviance to
positively associated with scores on the Interpersonal Deviance
be more related to some forms of OCB than to others. Specifically,
Scale (r = .21, p < .05) but, contrary to prediction, Frustration
Scale scores were not significantly related to scores on the Orga- we would expect conscientiousness, a type of OCB defined as
nizational Deviance Scale (r = ,01, us), discretionary behavior that goes beyond the minimum role require-
Perceived injustice. Considerations of fairness and justice are ments of the organization (e.g., attendance, obeying rules, and
extremely important in determining how people will respond in a taking breaks), to be more negatively related to organizational
potentially aggressive setting (Donerstein & Hatfield, 1982). Fur- deviance than interpersonal deviance, because organizational de-
thermore, perceptions of inequity and procedural injustice have viance includes behaviors such as slowed or reduced work effort.
been linked to employee theft (Greenberg, 1990, 1993; Siehl, Conversely, we would expect courtesy, OCB aimed at preventing
1987) and vandalism (DeMore, Fisher, & Baron, 1988; J. D. Fisher the occurrence of work-related problems for others, to be more
& Baron, 1982). These acts may be a means by which to "get negatively related to interpersonal deviance than to organizational
even" with the organization or to retaliate against those individuals deviance because interpersonal deviance reflects behavior that is
who have treated one unfairly (Bies et al., 1997; J. D. Fisher & potentially harmful to one's coworkers.
Baron, 1982; Greenberg, 1990, 1993). As such, we would expect As predicted, scores on the Conscientiousness Scale were neg-
perceived workplace injustice to be positively related to both of atively related to scores on our Organizational Deviance Scale
our measures of workplace deviance. (r = -.35, p < .01) as well as to scores on our Interpersonal
Consistent with our expectations, scores on our interpersonal Deviance Scale (r = —.28, p < .01). Also as predicted, scores on
deviance measure were found to be negatively associated with the Courtesy Scale were more strongly negatively related to scores
scores on the scales of Procedural Justice (r = - .33, p < .01) and on our Interpersonal Deviance Scale (r = -.41, p < .01) than to
Interactional Justice (r = —.35, p < .01). Similarly, scores on our scores on our Organizational Deviance Scale (r = —.22, p < .05).
measure of organizational deviance were also found to be nega- This pattern of relationships and nonrelationships with related
tively related to scores on the Procedural Justice Scale (r = — .32, constructs also indicates the discriminant validity of our measure.
p < .01) and to scores on the Interactional Justice Scale (r = —.33, We expect our measures of interpersonal and organizational devi-
p < .01). Contrary to our expectations, however, scores on the ance to have lower correlations with measures of behaviors that are
Distributive Injustice Scale were not related to scores on either of
presumed to be distinct and unrelated. One diverse group of
the deviance scales.
behaviors that should be unrelated to employee deviance behaviors
Normlessness. Normlessness refers to the lack of acceptance
are the diverse responses to dissatisfaction (exit, voice, and loy-
of social expectations about behavior (Shepard, 1972). According
alty) proposed by Hirschman (1970). This categorization was
to bonding theory (e.g., Akers, 1973) and social control models of
further revised and expanded by Farrell (1983) and Rusbult,
deviance (e.g., Hirschi, 1969), individuals who are "bonded" to a
Farrell, Rogers, and Mainous (1988) to include neglect as a fourth
social environment (i.e., individuals who feel attached to social
category of response to dissatisfaction. Neglect is defined as "pas-
conventions and institutions and who have a commitment to con-
formity) will be less inclined to engage in deviant acts. As such, sively allowing conditions to deteriorate through reduced interest
we would expect normlessness to be related to workplace devi- or effort, chronic lateness or absences, using company time for
ance. Consistent with our predictions, we found that scores on the personal business, or increased error rate" (Rusbult et al., 1988, p.
Normlessness Scale were related to scores on our Interpersonal 601). Hence, we would expect interpersonal and organizational
Deviance Scale (r = .21, p < .05); however, they were not deviance to be related to neglect, but not related to exit, voice, and
strongly related to scores on the Organizational Deviance Scale loyalty. The pattern of results gives evidence of discriminant
(r — .13, p = ns), although they were in the expected direction. validity. Both the Interpersonal and Organizational Deviance
Machiavellianism. Machiavellianism refers to a person's gen- Scales have a relatively strong relationship with the Neglect Scale
eral strategy for dealing with people, especially the degree to (r = .39, p < .01 and r = .48, p < .01), but have much lower
which the individual feels other people are manipulable in inter- correlations with the Exit, Voice, and Loyalty Scales (mean cor-
personal situations (Christie & Geis, 1970). We would expect that relation of .14).
 MEASURING WORKPLACE DEVIANCE 357

Discussion may be useful for assessing deviant behavior in organizations, at

least if the respondents are guaranteed anonymity.
The purpose of this study was to develop a survey instrument The measures developed here may be useful for examining
that could assess a wide range of deviant workplace behaviors. We deviance as a more general phenomenon. Most of the limited
believe this goal was accomplished in that considerable support empirical research on workplace deviance has tended to address
was found for the construct validity of our scales. The first stage only one or two deviant behaviors in isolation, such as theft
of the construct validation process was the assessment of the (Greenberg, 1990, 1993), sabotage (Giacalone, 1990; Taylor &
dimensionality of the deviance instrument. A CFA using maxi- Walton, 1971), or sexual harassment (Gutek, 1985). These mea-
mum likelihood estimation in LISREL 8 provided evidence that sures will help future researchers take a broader approach to
the hypothesized two-factor structure fit the data. Two scales of deviance, because these measures cover a range of deviant behav-
deviance emerged, one reflecting interpersonal deviance and the iors. Moreover, measures that take into account families or clusters
other reflecting organizationally directed deviance. These results of behaviors tend to provide more reliable and valid measures of
are consistent with prior conceptual approaches that have sug- the underlying theoretical construct (C. D. Fisher & Locke, 1992;
gested two distinct forms of workplace deviance, that directed at Rosse & Hulin, 1985). Broad measures also aid researchers in
the organization itself and that directed at its members (Baron & better understanding the underlying construct and its many more
Neuman, 1996; Giacalone & Greenberg, 1997; O'Leary-Kelly et specific behavioral manifestations (C. D. Fisher & Locke, 1992;
al., 1996; Robinson & Bennett, 1995, 1997; Robinson & Green- Rosse & Hulin, 1985), even allowing researchers to generalize
berg, 1999; Skarlicki & Folger, 1997). their findings to other "surface variables" not included in the scale
Evidence for the construct validity of our instrument was also that reflect the same underlying construct (Rosse & Hulin, 1985;
found by assessing the relationships between our deviance scales Roznowski & Hulin, 1992). Investigating the complete constructs
and other measures purported to assess similar constructs. Further- rather than dissecting the constructs into individual behaviors will
more, our instrument was found to have moderate relationships give researchers a clearer picture of what causes different types of
with measures of theoretically relevant constructs, such as frustra- deviant behavior in the workplace, as well as when and under what
tion, perceptions of injustice, and citizenship behavior. Finally, circumstances these behaviors might be considered substitutable.
these scales also showed discriminant validity, as they were not Although this study makes numerous contributions, several lim-
highly correlated with measures of unrelated constructs such as itations of this study must also be noted. One possible limitation of
exit, voice, and loyalty. this study is that our response rate was only 43% in Study 3.
It should be noted that the process of validating a construct is However, this is a good response rate when compared with re-
never complete; no measure can ever be said to be validated in any sponse rates obtained in other studies surveying the general pop-
final sense. Only over time and numerous studies can it be argued ulation. Moreover, it is likely more representative of the general
that the evidence leans toward supporting or not supporting the population of employees than is the more typical study that obtains
validity of a particular measure (Nunnally, 1978; Schwab, 1980). a high response rate, but from a sample of employees within a
The results of this study are a first step, suggesting evidence of the single organization.
construct validity of these scales. However, future research is A second noteworthy limitation is that our study was based
necessary to lend additional support to this conclusion. entirely on self-report. Although considerable evidence supports
It is interesting to note the prevalence of the deviant behaviors the validity of self-reports in general (Spector, 1992), and self-
assessed in this study. The relatively high participation rates con- report has also been found to provide accurate assessments of
tradict common assumptions regarding the frequency of these deviant behaviors in particular (Akers, Massey, Clarke, & Lauer,
behaviors as well as the willingness of employees to admit to 1983; Clark & Tift, 1966; Lee, 1993; Ones, Viswesvaran, &
engaging in these behaviors in self-report measures. It is also Schmidt, '1993), some noteworthy criticisms of this methodology
interesting to note that whereas the vast majority of the behaviors have been raised (Sackett, Bums, & Callahan, 1989; Sackett &
that we assessed had participation rates above 50% in Study 2, Harris, 1984). These criticisms center on social desirability biases;
considerably lower participation rates were reported in Study 3. critics fear that respondents may attempt to "fake good," thus
This difference was unexpected because those in Study 2, with biasing the results. This fear may, however, be unwarranted, as
higher participation rates, actually had less anonymity than the Ones et al.'s (1993) meta-analysis of integrity measures suggests
sample in Study 3, because those in Study 3 were random that self-report criteria tended to result in higher estimates of
mailed-in surveys from the general population. One possible ex- validity than external measures of deviance. Their explanation is
planation for this finding is demographic differences between the that (a) many deviant behaviors go undetected, hence limiting the
two samples; Study 2 respondents were younger and had less work validity of external measures, and (b) there is substantial evidence
experience than Study 3 respondents, and thus may be more prone that the correlation between admissions and actual behavior is
to deviant behavior in the workplace. substantial (Ones, Viswesvaran, & Schmidt, 1993). Consequently,
The development of these scales of workplace deviance has it is our belief that self-report can be a valid way of assessing the
potentially important implications for future research on deviance broad variety of deviant behaviors in the workplace, particularly
in the workplace. To date, empirical research on workplace devi- when respondents are assured anonymity.
ance has been limited. One possible explanation of this shortage of Researchers must attempt to assure that respondents remain
research might be the inherent difficulty in measuring this behav- anonymous, or at least assure confidentiality of responses when
ior. The results of this study demonstrate that those employees who using these scales, in order to minimize response biases. One must
completed our unsolicited questionnaire were willing to admit to keep in mind that our scales were developed with anonymous
engaging in socially unacceptable behaviors. Hence, self-report respondents outside the workplace. It is difficult to know to what
358 BENNETT AND ROBINSON

extent the results and validity of our scales would be different if aggression: Evidence on their relative frequency and potential causes.
the instruments were given to respondents who were not assured Aggressive Behavior, 22, 161-173.
anonymity. Even when anonymity is guaranteed, respondents may Bennett, R. J. (1998a). Perceived powerlessness as a cause of employee
provide different reports if the self-reports are collected within an deviance. In R. Griffin, A. O'Leary Kelley, & J. Collins, (Eds.), Dys-
organizational setting. functional workplace behavior (pp. 221-239). Greenwich, CT: JAI
Press.
A further limitation of the resulting scales is that they do not
Bennett, R. J. (1998b). Taking the sting out of the whip: Reactions to
include all possible types of deviant behavior. We decided early on consistent punishments for unethical behaviors. Journal of Applied So-
to follow the traditional path of scale development, which required cial Psychology: Experimental, 4, 1-15.
us to eliminate items with low variances and low interitein corre- Bensimon, H. F. (1994). Crisis and disaster management: Violations in the
lations in order to ensure reliability. Upon reflection, we realize workplace. Training and Development, 28, 27-32.
that we also made an implicit choice at that stage to create a scale Bentler, P. M. (1990). Comparative fit indices in structural models. Psy-
of more common forms of employee deviance. We had to make chological Bulletin, 107, 238-246.
tradeoffs between ensuring reliability and ensuring that the full Bentler, P. M., & Bonnet, D. G. (1980). Significance tests and goodness of
range of deviant behaviors were included in our scale. However, fit in the analysis of covariance structures. Psychological Bulletin, 88,
we believe we made the right choice for several reasons. First, the 588-606.
approach we took was suitable given our theory that deviant Berkowitz, L. (1978). Whatever happened to the frustration-aggression
hypothesis? American Behavioral Scientist, 21, 691-708.
behaviors within a given family are functional substitutes for one
Bies, R. J., Tripp, T. M., & Kramer, R. M. (1997). At the breaking point:
another; thus, an employee may engage in one of several types of Cognitive and social dynamics of revenge in organizations. In R. A.
deviance within a family. This logic suggests that a scale of Giacalone & J. Greenberg (Eds.), Antisocial behavior in organizations
deviant behavior need not include the complete set of possible (pp. 18-36). Thousand Oaks, CA: Sage.
manifestations of deviance. Second, the behaviors we eliminated Bollen, K. A. (1989). Structural equations with latent variables. New
were very rare. Third, had we not eliminated those behaviors, our York: Wiley.
scale would have lacked reliability and, thus, would be unlikely to Buss, D. (1993). Ways to curtail employee theft. Nation's Business, pp. 36,
be used by others. Thus, although a more inclusive scale may 38.
appear more interesting, it is unlikely to have had as much prac- Camara, W. J., & Schneider, D. L. (1994). Integrity tests: Facts and
tical value to field researchers. unresolved issues. American Psychologist, 49, 112-119.
Campbell, D., & Fiske, D. (1959). Convergent and discriminant validation
A final limitation of this study is that, because we have at-
by the multitrait—multimethod matrix. Psychological Bulletin, 56, 81-
tempted to create a scale that is widely applicable to a range of
105.
organizational contexts and occupations, the scale produced here Christie, R., & Geis, F. L. (1970). Studies in Machiavellianism. San Diego,
includes only those behaviors commonly found across such con- CA: Academic Press.
texts. Indeed, in our second study, we assessed and eliminated Churchill, G. (1988). Basic marketing research. Chicago: Dryden Press.
behaviors that our judges deemed not to be deviant across most Clark, J. P., & Tift, L. L. (1966). Polygraph and interview validation of
organizational contexts. As such, this scale is unable to account for self-reported deviant behavior. American Sociological Review, 31,516-
deviant behaviors that are idiosyncratic to specific organizational 522.
contexts and occupations. Thus, even though we conceptually Dean, D. (1961). Alienation: Its meaning and measurement. American
define deviance in terms of the particular normative context in Sociological Review, 25, 753-758.
which it occurs, our methodology was designed to capture only a DeMore, S. W., Fisher, J. D., & Baron, R. M. (1988). The equity-control
model as a predictor of vandalism among college students. Journal of
subset of the behaviors found in any particular organizational
Applied Social Psychology, 18, 80-91.
context. Nonetheless, we believe this approach strikes a balance DeVellis, R. (1991). Scale development: Theory and applications. London:
between our belief that deviance is dependent on the social context Sage.
in which it is defined and our need to produce a generalizable Dillman, D. A. (1972). Increasing mail questionnaire response in large
survey instrument appropriate for the rigors of normal science. samples of the general public. Public Opinion Quarterly, 36, 254-257.
In summary, we hope that this instrument proves useful to the Dillman, D. A. (1978). Mail and telephone surveys: The total design
future study of workplace deviance, helping to facilitate and en- method. New York: Wiley-Interscience.
courage the much-needed empirical research into this significant Dollard, J., Doob, L., Miller, N., Mowrer, O., & Sears, R. (1939). Frus-
form of organizational behavior. tration and aggression. New Haven, CT: Yale University Press.
Donerstein, E., & Hatfield, E. (1982). Aggression and inequity. In J.
Greenberg & R. Cohen (Eds.), Equity and justice in social behavior (pp.
References 309-336). New York: Academic Press.
Farrell, D. (1983). Exit, voice, loyalty, and neglect as responses to job
Akers, R. L. (1973). Deviant behavior: A social learning approach. Bel- dissatisfaction: A multi-dimensional scaling study. Academy of Manage-
mont, MA: Wadsworth. ment Journal, 26, 596-607.
Akers, R. L., Massey, J., Clarke, W., & Lauer, R. M. (1983). Are self- Farrell, D., & Rusbult, C. (1986). Measurement of responses to job dis-
reports of adolescent deviance valid? Biochemical measures, random- satisfaction: Exit, voice, loyalty and neglect. Unpublished manuscript,
ized response, and the bogus pipeline in smoking behavior. Social Western Michigan University.
Forces, 62, 234-251. Feldman, D. C. (1984). The development and enforcement of group norms.
Anderson, J. O., & Gerbing, D. W. (1988). Structural equation modeling in Academy of Management Review, 9, 47-53.
practice: A review and recommended two-step approach. Psychological Fisher, C. D., & Locke, E. A. (1992). The new look in job satisfaction
Bulletin, 103, 411-423. research and theory. In C. J. Cranny, P. C. Smith, & E. F. Stone (Eds.),
Baron, R. A., & Neuman, J. H. (1996). Workplace violence and workplace Job satisfaction (pp. 165-194). New York: Lexington Books.
 MEASURING WORKPLACE DEVIANCE 359

Fisher, J. D., & Baron, R. M. (1982). An equity-based model of vandalism. relationship between methods of monitoring and organizational citizen-
Population and Environment, 5, 182-200. ship behavior. Academy of Management Journal, 36, 528-556.
Ford, J. K., MacCallum, R. C., & Tail, M. (1986). The application of Nunnally, J. C. (1978). Psychometric theory. New York: McGraw-Hill.
explanatory FA in applied psychology: A critical review and analysis, O'Leary-Kelly, A. M., Griffin, R. W., & Glew, D. J. (1996). Organization-
Personnel Psychology, 39, 291-314. motivated aggression: A research framework. Academy of Management
Fowler, F. J. (1993). Survey research methods. Newbury Park, CA: Sage. Review, 21, 225-253.
Giacalone, R. (1990). Employee sabotage: The enemy within. Supervisory Ones, D. S., Viswesvaran, C., & Schmidt, F. L. (1993). Meta-analysis of
Management, pp. 6-7. integrity test validities: Findings and implications for personnel selection
Giacalone, R. A., & Greenberg, J. (1997). Antisocial behavior in organi- and theories of job performance. Journal of Applied Psychology, 78,
zations. Thousand Oaks, CA: Sage. 679-703.
Giacalone, R. A., Riordan, C. A., & Rosenfeld, P. (1997). In R. Giacalone Organ, D. W. (1988). Organizational citizenship behavior. Lexington,
& J. Greenberg (Eds.), Antisocial behavior in organizations (pp. 109- MA: Lexington Books.
129). Thousand Oaks, CA: Sage. Organ, D. W. (1990). The motivational basis of organizational citizenship
Green, G. S. (1997). Occupational crime (2nd ed.). Chicago: Nelson-Hall. behavior. In L. L. Cummings & B. M. Staw (Eds.), Research in orga-
Greenberg, J. (1990). Employee theft as a reaction to underpayment ineq- nizational behavior (pp. 43-72). Greenwich, CT: JAI Press.
uity: The hidden cost of paycuts. Journal of Applied Psychology, 75, Peters, L. J., O'Connor, E. J., & Rudolf, C. J. (1980). The behavioral and
561-568. affective consequences of situational variables relevant to performance
Greenberg, J. (1993). Stealing in the name of justice: Informational and settings. Organizational Behavior and Human Performance, 25, 79-96.
interpersonal moderators of theft reactions to underpayment inequity. Podsakoff, P. M., MacKenzie, S., Moorman, R. H., & Fetter, R. (1990).
Organizational Behavior and Human Decision Processes, 54, 81-103. Transformational leader behaviors and their effects on followers' trust in
Greenberg, J., & Scott, K. S. (1996). Why do workers bite the hands that leader, satisfaction, and organizational citizenship behaviors. Leadership
feed them? Employee theft as a social exchange process. In B. M. Staw Quarterly, I, 107-142.
& L. L. Cummings (Eds.), Research in organizational behavior Robinson, S., & Bennett, R. (1995). A typology of deviant workplace
(Vol. 18, pp. 111-156). Greenwich, CT: JAI Press. behaviors: A multi-dimensional scaling study. Academy of Management
Gutek, B. A. (1985). Sex in the workplace. San Francisco: Jossey-Bass. Journal, 38, 555-572.
Harper, D. (1990). Spotlight abuse-save profits. Industrial Distribu- Robinson, S., & Bennett, R. (1997). Workplace deviance: Its definition, its
tion, 79, 47-51. manifestations, and its causes. Research on Negotiations in Organiza-
Hirschi, T. (1969). Causes of delinquency. Berkeley: University of Cali- tions, 6, 3-27.
fornia Press. Robinson, S., & Greenberg, J. (1999). Employees behaving badly: Dimen-
Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in sions, determinants and dilemmas in the study of workplace deviance. In
firms, organizations, and states. Cambridge, MA: Harvard University D. M. Rousseau & C. Cooper (Eds.), Trends in organizational behavior
Press. (Vol. 5, pp. 1-23). New York: Wiley.
Hollinger, R. C., & Clark, J. P. (1982). Formal and informal social controls Rosse, J., & Hulin, C. (1985). Adaptation to work: An analysis of em-
of employee deviance. The Sociological Quarterly, 23, 333—343. ployee health, withdrawal, and change. Organizational Behavior and
Hollinger, R., & Clark, J. (1983a). Deterrence in the workplace: Perceived Human Decision Processes, 36, 324-247.
certainty, perceived severity, and employee theft. Social Forces, 62, Roznowski, M., & Hulin, C. (1992). The scientific merit of valid measures
398-418. of general constructs with special reference to job satisfaction and job
Hollinger, R., & Clark, J. (1983b). Theft by employees. Lexington, MA: withdrawal. In C. J. Cranny, P. C. Smith, & E. F. Stone (Eds.), Job
Heath. satisfaction (pp. 123-163). New York: Lexington Books.
Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Rusbult, C., Farrell, D., Rogers, G., & Mainous, A. (1988). Impact of
Structural equation modeling: Concepts, issues and applications (pp. exchange variables on exit, voice, loyalty, and neglect: An integrative
76-99). Thousand Oaks, CA: Sage. model of responses to declining job satisfaction. Academy of Manage-
Joreskog, K. G., & Sorbom, D. (1986). LISREL IV: Analysis of linear ment, 31, 599-627.
structural relationships by maximum likelihood, instrumental variables Sackett, P. R., Bums, L. R., & Callahan, C. (1989). Integrity testing for
and least squares methods (4th ed.). Moorseville, IN: Scientific Soft- personnel selection: An update. Personnel Psychology, 42, 491-528.
ware. Sackett, P. R., & Harris, M. M. (1984). Honesty testing for personnel
Kaplan, H. B. (1975). Self-attitudes and deviant behavior. Pacific Pali- selection; A review and critique. Personnel Psychology, 32, 221-245.
sades, CA: Goodyear. Schwab, D. P. (1980). Construct validity in organizational behavior. In
Kim, J., & Mueller, C. (1978). Factor analysis: Statistical methods and B. M. Staw & L. L. Cummings (Eds.), Research in organizational
practical issues. Newbury Park, CA: Sage. behavior (pp. 3-43). Greenwich, CT: JAI Press.
Lee, R. (1993). Doing research on sensitive topics. London: Sage. Shepard, J. M. (1972). Alienation as a process: Work as a case in point. The
Lehmann, W. E. K., Holcom, M. L., & Simpson, D. D. (1990). Employee Sociological Quarterly, 13, 161-173.
health and performance in the workplace: A survey of municipal em- Siehl, E. W. (1987). Garment workers: Perceptions of inequity and em-
ployees of a large southwest city (IBR tech. rep.). Fort Worth: Texas ployee theft. British Journal of Criminology, 27, 174-190.
Christian University, Institute of Behavioral Research. Skarlicki, D. P., & Folger, R. (1997). Retaliation in the workplace: The
Lehman, W., & Simpson, D. (1992). Employee substance abuse and roles of distributive, procedural, and interactional justice. Journal of
on-the-job behaviors. Journal of Applied Psychology, 77, 309-321. Applied Psychology, 82, 416-425.
McGurn, X. (1988). Spotting the thieves who work among us. Wall Street Snyder, N. H., Blair, K. E., & Arndt, T. (1990). Breaking the bad habits
Journal, p. 16a. behind time theft. Business, 40(4), 31-33.
Murphy, K. R. (1993). Honesty in the workplace. Belmont, CA: Brooks/ Spector, P. (1975). Relationships of organizational frustration with re-
Cole. ported behavioral reactions of employees. Journal of Applied Psychol-
Northwestern Life Insurance Company. (1993). Fear and violence in the ogy, 60, 635-637.
workplace. Minneapolis, MN: Author. Spector, P. (1992). A consideration of the validity and meaning of self-
Niehoff, B. P., & Moorman, R. H. (1993). Justice as a mediator of the report measures of job conditions. In C. L. Cooper & I. T. Robertson
360 BENNETT AND ROBINSON

(Eds.), International review of industrial and organizational psychology Turner, D. L., & Stephenson. R. G. (1993, February). The lure of white-
(pp. 123-151). Chichester, England: Wiley. collar crime. Security Management, pp. 57-58.
Storms, P., & Spector, P. (1987). Relationships of organizational frustra- Webb, S. L. (1991). Step forward: Sexual harassment in the workplace.
tion with reported behavioral reactions: The moderating effect of locus New York: MasterMedia.
of control. Journal of Occupational Psychology, 60, 227-234. Williams, L. J., & Andersen, S. E. (1991). Job satisfaction and orga-
Taylor, L., & Walton, P. (1971). Industrial sabotage; Motives and mean- nizational commitment as predictors of organizational citizenship
ings. In S. Cohen (Ed.), Images of deviance (pp. 219-245). Middlesex, behavior and in-role behaviors. Journal of Management, 17, 601-
England: Penguin Books. 617.

Appendix

Final Interpersonal and Organizational

Deviance Scale Items
Measure
Interpersonal Deviance
Made fun of someone at work
Said something hurtful to someone at work
Made an ethnic, religious, or racial remark at work
Cursed at someone at work
Played a mean prank on someone at work
Acted rudely toward someone at work
Publicly embarrassed someone at work

Organizational Deviance
Taken property from work without permission
Spent too much time fantasizing or daydreaming instead of working
Falsified a receipt to get reimbursed for more money than you spent on business expenses
Taken an additional or longer break than is acceptable at your workplace
Come in late to work without permission
Littered your work environment
Neglected to follow your boss's instructions
Intentionally worked slower than you could have worked
Discussed confidential company information with an unauthorized person
Used an illegal drug or consumed alcohol on the job
Put little effort into your work
Dragged out work in order to get overtime

Received November 22, 1996

Revision received June 29, 1999
Accepted July 6, 1999