Journal of Vocational Behavior 68 (2006) 446–460
www.elsevier.com/locate/jvb
The dimensionality of counterproductivity: Are all
counterproductive behaviors created equal?
Paul E. Spector a,¤, Suzy Fox b, Lisa M. Penney c, Kari Bruursema a,
Angeline Goh a, Stacey Kessler a
a
b
University of South Florida, USA
Loyola University Chicago, USA
c
University of Houston, USA
Received 21 July 2005
Available online 6 December 2005
Abstract
Most studies of counterproductive work behavior (CWB) assess it as one or two overall dimensions that might obscure relationships of potential antecedents with more speciWc forms of behavior.
A Wner-grained analysis of the relationship between counterproductive work behavior and antecedents was conducted with the Wve-subscales (abuse toward others, production deviance, sabotage,
theft, and withdrawal) taken from the 45-item Counterproductive Work Behavior Checklist, a measure that has been used in a number of prior studies. Described is the rationale for each of the Wve
dimensions, which have been discussed individually in the literature. Data from three combined studies provide evidence for diVerential relationships with potential antecedents that suggest the use of
more speciWc subscales to assess CWB. Most notably, abuse and sabotage were most strongly related
to anger and stress, theft was unrelated to emotion, and withdrawal was associated with boredom
and being upset. Finally, the distinct forms of CWB may suggest distinct underlying dynamics, that
vary in their balance of hostile and instrumental motivational systems.
© 2005 Elsevier Inc. All rights reserved.
Keywords: Counterproductive work behavior; Aggression; Sabotage; Theft; Withdrawal; Job satisfaction;
Emotion
*
Corresponding author. Fax: +1 813 974 4617.
E-mail address: spector@shell.cas.usf.edu (P.E. Spector).
0001-8791/$ - see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jvb.2005.10.005
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
447
1. Introduction
Counterproductive work behavior (CWB) has emerged as a major area of concern
among researchers, managers, and the general public. These behaviors are a set of distinct
acts that share the characteristics that they are volitional (as opposed to accidental or mandated) and harm or intend to harm organizations and/or organization stakeholders, such
as clients, coworkers, customers, and supervisors (Spector & Fox, 2005). CWB has been
studied under diVerent labels from a variety of theoretical perspectives, such as aggression
(Douglas & Martinko, 2001; Fox & Spector, 1999; Neuman & Baron, 1997; O’Leary-Kelly,
GriYn, & Glew, 1996), deviance (Hollinger, 1986; Robinson & Bennett, 1995), retaliation
(Skarlicki & Folger, 1997), and revenge (Bies, Tripp, & Kramer, 1997). An inspection of
various scales used by diVerent groups of researchers across studies shows that they each
contain an overlapping set of behaviors (Spector & Fox, 2005) that include disparate acts
that have diVerent targets. In most cases researchers combine a checklist of behaviors into
a single index or at most two indices, distinguishing only between behaviors targeting the
organization and those targeting persons in the organization. EVorts to empirically test a
further subdivision of these two categories into minor versus serious (based upon the Robinson–Bennett typology) have been stymied by extremely low baserates of reports of serious behaviors, such as physical violence. As a result, behaviors as diVerent as spreading
rumors and stealing from coworkers, or coming late to work and destroying organization
property, are combined (and implicitly equated) within a single index. Few researchers
have divided CWBs into more speciWc categories, although some have focused attention on
a single form of behavior.
1.1. How CWB has been categorized
Considering CWB broadly as intentional behavior that harms or intends to harm organizations and its members, there are a number of related research streams that would be
included. Some of this research has focused on single, narrowly deWned behaviors such as
absence (Dalton & Mesch, 1991), client abuse (Perlow & Latham, 1993), or theft (Greenberg, 1990), whereas others have combined disparate behaviors into broader categories
(e.g., Chen & Spector, 1992; Hanisch, Hulin, & Roznowski, 1998). More recently some
researchers have adopted the Robinson and Bennett (1995) distinction of behaviors targeting the organization versus people (e.g., Bennett & Robinson, 2000; Fox & Spector, 1999;
Fox, Spector, & Miles, 2001), with disparate acts falling particularly into the organizational category. Many researchers, however, combine many diVerent forms of CWB into a
single index (e.g., Martinko, Gundlach, & Douglas, 2002; Miles, Borman, Spector, & Fox,
2002; Penney & Spector, 2002), although perhaps calling it by a diVerent term (e.g., retaliation, Skarlicki & Folger, 1997). In some cases a speciWc argument was made that the behaviors all represented a single underlying construct that was likely driven by the same
antecedents. For example, retaliation is theorized to be a response to injustice that can take
on many diVerent forms (Folger & Skarlicki, 2005). This suggests that diVerent behavioral
forms of retaliation (CWB) can be interchangeable, and are chosen perhaps based on
opportunity.
However, there is reason to question whether all forms of CWB have the same antecedents, and thus it may be fruitful to create multiple indices rather than combine all items
into one. For example, Fox et al. (2001) tied justice more to CWB directed toward
448
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
organizations and interpersonal conXict more to CWB directed toward other people. Lee
and Spector (in press) showed that conXict with coworkers was more likely to result in
CWB directed toward people, whereas conXict with supervisors was more likely to result in
CWB directed toward the organization. Taking the perspective of worker resistance, Roscigno and Hodson (2004) conducted a content analysis of 82 workplace ethnographies and
showed that diVerent combinations of work conditions were associated with diVerent
resistance behaviors that correspond to CWBs of sabotage, theft, and withdrawal. More
Wne-grained research on the single behavior of absence found that diVerent types could be
distinguished according to reasons (Dalton & Mesch, 1991).
1.2. Five dimensions of CWB
Much of the work on CWB has roots in the study of human aggression. Most such theories have linked it to negative emotions, such as anger and/or frustration in response to
environmental conditions in both the social psychological (e.g., Anderson, Deuser, & DeNeve, 1995; Berkowitz, 1998) and workplace (Fox & Spector, 1999; Neuman & Baron, 1997)
literatures. Over time the range of precipitating conditions has expanded to include a
broad range of job stressors that induce a variety of negative emotions (Chen & Spector,
1992; Fox et al., 2001). There are similarities between aggression-based models and models
that focus primarily on injustice (e.g., Folger & Skarlicki, 2005) in that the role of negative
emotions in response to perceived situations is important. Fox et al. (2001) made this connection explicit in considering perceived injustice as a workplace stressor.
The literature on aggression distinguishes two forms based on underlying motive—hostile versus instrumental (Berkowitz, 1998). Hostile aggression is associated with negative
emotions, most typically anger, is often impulsive, and has harm as its primary motive.
Instrumental aggression is not necessarily associated with emotion and has some additional goal beyond harm. Anderson and Bushman (2002) note that the proximal motive of
all aggression is harm, but instrumental has a more distal motive, such as robbery. In fact,
some scholars add that some forms of aggression such as organizational retaliatory behavior (Folger & Skarlicki, 2005) and revenge (Bies & Tripp, 2005) may have functional or
prosocial instrumental motivation and consequences.
1.2.1. Abuse against others
Abuse consists of harmful behaviors directed toward coworkers and others that harm
either physically or psychologically through making threats, nasty comments, ignoring the
person, or undermining the person’s ability to work eVectively. Such behaviors are direct
forms of aggression, although in the workplace physical aggression tends to be infrequent
so most research studies have mainly assessed nonphysical forms. Berkowitz (1998) discusses how stressors and other unpleasant situations are linked to hostile aggression
through negative emotion. Included are things that cause physical pain as well as stressors
that might just induce psychological discomfort. Links between stressors at work and
CWB directed toward others have been shown (e.g., Fox & Spector, 1999; Fox et al., 2001).
Geen (1998) notes that aggression is a byproduct of both dispositional and environmental
variables, with some research directed toward identifying an aggressive personality type.
Keashly and Harvey’s (2005) work on emotional abuse in the workplace builds on the
aggression literature, but notes additional factors that contribute to such behavior. They
note how social norms and organizational culture can either support or inhibit such
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
449
behaviors. Thus stressful work conditions might lead to abuse in settings where such
behavior is considered acceptable, or at least is not unacceptable.
We expect abuse to be mainly due to hostile motives, and therefore we hypothesize it
will relate to both stressors and negative emotions. On the stressor side, interpersonal
conXict should most strongly relate to abuse because it reXects the extent to which an
individual directly experiences nastiness from other employees, thus inducing a response in
kind or spiral to even more intense forms (Pearson, Andersson, & Porath, 2005).
1.2.2. Production deviance and sabotage
Production deviance is the purposeful failure to perform job tasks eVectively the way
they are supposed to be performed. It is based in part on Hollinger’s (1986) concept of production deviance, except that we have separated withdrawal into a separate category. Sabotage is defacing or destroying physical property belonging to the employer (Chen &
Spector, 1992). Although these two forms of behaviors are diVerent, with the former being
in many cases more passive (failure to do a task or do it correctly) and the latter being
more active (intentionally destroying something), they are linked theoretically and may
have some of the same underlying causes.
Some workplace aggression researchers have considered production deviance and sabotage to be displaced forms of aggression, directed toward safer inanimate organizational
targets rather than people (Neuman & Baron, 1997). Of course, production deviance would
be safer than sabotage, since the destruction of property is more likely to be sanctioned by
the organization and could result in arrest depending upon the severity of the act. Production deviance being more passive is less visible and can be diYcult to prove. Whereas
aggression toward inanimate objects is likely to be seen as safer than direct physical
aggression, it seems less likely that it would be preferred to verbal aggression for at least
two reasons. First, verbal aggression is unlikely to be punished, and second, it enables the
individual to aggress against the agent of their provocation.
Some researchers have noted that sabotage can be done for instrumental purposes as
well as in response to anger and hostile feelings (Ambrose, Seabright, & Schminke, 2002;
Crino, 1994). For example, such acts can be done to draw attention to a problem, aVect
organizational change, receive peer acceptance, or gain competitive advantage over peers
(e.g., sabotaging their chances for a promotion you want). However, little research has
investigated such motives for production deviance or sabotage.
Given their link to hostile aggression, but noting their possible instrumentality, it is
hypothesized that production deviance and sabotage will relate to the same variables as
abuse, although less strongly. Furthermore, it is hypothesized that production deviance will
have stronger relationships with other variables due to it being seen as safer than sabotage.
1.2.3. Theft
Theft by employees is recognized as a major problem for organizations. As with sabotage, some researchers have suggested that theft can be a form of aggression against an
organization, undertaken in an attempt to harm (e.g., Neuman & Baron, 1997). This suggests that theft would be caused by many of the same factors as abuse. However, most
treatments of employee theft consider other factors. For example, Payne and Gainey
(2004) listed a number of potential causes, including perception that theft was appropriate,
low self-control, injustice, personality, work environment, demographic characteristics,
and stress. Mustaine and Tewksbury (2002) discussed three major reasons for theft:
450
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
economic need, job dissatisfaction, and injustice. Of the three, injustice has perhaps
received the greatest amount of research attention, with several studies showing a clear link
between perceptions of inequity and theft (e.g., Greenberg, 1990, 2002). Although injustice
might be considered a stressor that aVects theft through anger and other negative emotions
(Fox et al., 2001), it is also possible that injustice acts through more instrumental means. In
other words an individual steals, not to harm the organization, but to achieve a state of
equity, or simply for desired economic gain.
It is assumed that theft has more instrumental than hostile motives. Based on prior literature, it is hypothesized to relate to justice and job satisfaction, but we do not expect it to
relate to our other variables.
1.2.4. Withdrawal
Withdrawal consists of behaviors that restrict the amount of time working to less than is
required by the organization. It includes absence, arriving late or leaving early, and taking
longer breaks than authorized. We omitted turnover because it is a single act of permanent
withdrawal rather than a potentially ongoing series of behaviors by current employees. Both
absence and lateness have been studied separately and have their own literatures, although a
number of studies have linked various forms of withdrawal (e.g., Hanisch et al., 1998).
Whereas it might be possible for withdrawal to be motivated by a desire to hurt the organization, most treatments of this form of CWB have taken a diVerent approach.
Of the various forms of withdrawal behavior, absence has received the most attention.
Early models considered it to be a response to dissatisfaction at work, although such connections have been modest at best (Johns, 1997). More recent research has suggested that
absence can occur for a variety of reasons. Johns (1997) lists health, psychological disorders, stress, social norms, culture, labor–management conXict, and individual diVerences as
potential inXuences. The work–family conXict literature (Koslowsky, 2000) reminds us that
withdrawal may indeed be instrumental, chosen as a means of coping with conXicting work
and nonwork obligations.
Withdrawal contrasts with these other forms of behavior because it is an attempt to
avoid or escape a situation rather than do direct harm. An individual might wish to escape
stressors, injustice, dissatisfaction or situations that induce negative emotions. We hypothesize withdrawal to relate to all of our study variables.
1.3. The current paper
For the current paper, we compiled data from three of our prior studies (Bruursema,
Kessler, Fox, & Spector, 2004; Goh, Bruursema, Fox, & Spector, 2003; Penney & Spector,
2005) that used our 45-item Counterproductive Work Behavior Checklist (CWB-C) as well
as measures of stressors, justice, job satisfaction, and negative emotions at work. These
prior studies combined the items to create scores of CWB directed toward organizations or
people and not our Wve categories which is an approach new to this paper. We had subject
matter experts (industrial/organizational psychology doctoral students) place the speciWc
behaviors into our Wve categories for which we computed subscale scores. Correlations
were compared among the subscales on several potential antecedents, including two job
stressors (interpersonal conXict and organizational constraints), two forms of justice
(distributive and procedural), job satisfaction, and two measures of negative emotion. We
compared results across the Wve categories, expecting diVerences as discussed earlier.
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
451
2. Method
2.1. Determination of items for subscales
For the placement of items into dimensions, we had 12 industrial/organizational psychology doctoral students serve as subject matter experts (SMEs) to sort the CWB-C items.
None of the SMEs were otherwise involved with this project. The SMEs were given deWnitions of the Wve categories and were asked to indicate for each item into which category it
Wt. To place an item into a category, we set a threshold of at least 80% agreement (speciWcally 10 of 12 judges or 83%). We were able to classify 33 of 45 behaviors (see Table 3). We
also asked our SMEs to sort the items into categories according to target, speciWcally
CWB-O directed toward the organization and CWB-P directed toward people. All but two
of the 45 items were placed into one of the two categories.
We chose expert judgment of item content over factor analysis of items for two reasons.
First, item checklists such as our CWB-C are causal indicator scales for which items are
not interchangeable measures of a single underlying construct (Bollen & Lennox, 1991;
Edwards & Bagozzi, 2000). Such scales often are comprised of items that are not highly
related and thus do not form the sorts of factors that the more typical eVect indicator
scales produce. Second, the items ask respondents to report frequency of engaging in each
behavior. As seen in Table 3, there was considerable variability in the percentage of people
who engaged in each behavior, with many of the items rarely endorsed. This produced
diVerential skew in the distribution of many items, and diVerences in underlying distribution shape can cause distortions in factor structures (e.g., Spector, Van Katwyk, Brannick,
& Chen, 1997).
2.2. Participants and procedure
For the Goh et al. (2003) sample, surveys were distributed in Wve organizations in Chicago and Tampa. Participants were mailed survey packets with instructions to return them
by mail. In addition the snowball method was used in Chicago asking Master’s in Human
Resources and Industrial Relations students, all full-time employees and mostly managers,
to distribute questionnaires at work. The Wve samples were support staV of two universities
and employees of a Wnancial consulting Wrm, an accounting Wrm, and a behavioral health
services company. With the Wve organizations the response rate was 23%. It should be
noted that with one university, the response rate was based on the number who returned
completed questionnaires out of the number who agreed to do so after a phone request.
With the snowball method it is diYcult to be precise with the response rate as students who
picked up a questionnaire might not have asked anyone to participate, or might have asked
more than one. However, based on the number of distributed questionnaires, the response
rate was 32% for employees. The relatively low response rate was in part due to the fact
that participants were requested to provide a matching coworker evaluation of their
behavior (for use in a diVerent study).
The total sample size was 169 (21.6% male, 26.4% managerial positions, and 91.3%
white-collar). Due to the sensitive nature of the questionnaire items, no further demographic data were collected.
The Penney and Spector (2005) sample consisted of employed individuals taking
mostly night classes at the University of South Florida in chemistry, computer science,
452
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
engineering, interdisciplinary social science, management, and psychology. Criteria for
inclusion were working at least 25 h per week and tenure at the current job of at least 2
months. Two hundred ninety-nine respondents met these requirements (eight additional
ones were dropped). The sample was 23% male, with an average age of 23.3 and average
tenure on the job of two years. Approximately half (47%) of the sample worked 30 h or
more per week, with the remainder working between 25 and 29 h. Comparisons were made
between these two groups on all the study variables using independent group t tests, and
only one was found to be signiWcant for organizational constraints (means D 21.36 vs. 19.04
for full-time and part-time, respectively, t D 2.42, p < .05). Participants were recruited in
classes where they completed questionnaires and returned them anonymously to a box in
front of the classroom. Due to the way data were collected, it could not be determined how
many eligible individuals chose not to participate, so a response rate could not be calculated. The Bruursema et al. (2004) sample consisted of 279 employed individuals taking
coursework at the University of South Florida. The criterion for participation was working
at least 20 h per week and having worked at least 6 months at the current job. Twenty-four
percent of the sample was male and 17.4% held management or supervisory positions.
Data were collected in two ways. First, researchers attended three introductory psychology
classes where participants completed questionnaires and returned them at the end of class.
Second, participants made appointments to come by an oYce in the psychology building
where they completed the questionnaire and handed it to the researcher. A response rate
could not be determined since there was no way to know how many potential respondents
chose not to participate.
2.3. Measures
In addition to the CWB-C, participants were given measures of seven variables
expected to relate to at least some of the categories of CWB. All three studies included
job stressor scales (organizational constraints and interpersonal conXict) and a measure
of job satisfaction. Organizational constraints were assessed with the Spector and Jex
(1998) organizational constraints scale (OCS), an 11-item measure that assesses the
major areas of constraints identiWed by Peters and O’Connor (1980). Each item asks for
a rating of how frequently a constraint is encountered using Wve-point scales ranging
from 1 D never to 5 D every day. Examples include poor equipment or supplies, inadequate training, and conXicting job demands. Spector and Jex (1998) reported a mean
internal consistency reliability (coeYcient ) of .85 across eight samples. Interpersonal
conXict was measured with the four-item Interpersonal ConXict at Work Scale (ICAWS;
Spector & Jex, 1998). It assesses the frequency with which employees experience arguments and yelling in their interactions with coworkers, using the same Wve-point
response scale as the OCS. Spector and Jex (1998) reported a mean across 13 samples
of .74.
The three-item Cammann, Fichman, M, Jenkins, and Klesh (1979) job satisfaction measure from the Michigan Organizational Assessment Scale was used. Response choices
ranged from 1 (disagree very much) to 6 (agree very much). Van Katwyk, Fox, Spector, and
Kelloway (2000) found a .90 coeYcient for this scale.
The short 10-item negative emotion subscale of the Job-Related AVective Well-Being
Scale, JAWS (Van Katwyk et al., 2000) was used in the Goh et al. (2003) sample to assess
people’s negative emotional experiences at work. Instructions ask respondents to indicate
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
453
how often they have experienced each of 10 distinct negative emotions (e.g., anger or
boredom) at work, using Wve response choices ranging from 1 D never to 5 D every day.
Spector and Fox (2003) factor analyzed the subscale and further divided it into two factors: Upset consisting of angry, anxious, depressed, discouraged, fatigued, frightened, furious, and gloomy, and Bored which consisted of a single item. They further found that the
boredom item had little relationship with the other items (which were all highly interrelated), and had diVerent correlates. The Upset scale had an internal consistency reliability
in the current sample of .88.
Measures of distributive and procedural justice reported in Moorman (1991) were used
in the Bruursema et al. (2004) and Goh et al. (2003) samples. Distributive justice was
assessed with a six-item scale originally developed by Price and Mueller (1986). The scale
has Wve response choices ranging from 1 (very unfairly) to 5 (very fairly), with high scores
indicating high levels of justice. Procedural justice was measured with Moorman’s (1991)
12-item scale. This scale included interactional justice items that Moorman suggests should
be combined because of poor discriminability from the procedural justice items (R.H.
Moorman, personal communication, February 3, 1999). Five response choices ranged from
1 (strongly disagree) to 7 (strongly agree). Moorman (1991) reported coeYcient s of .94 for
both justice scales.
3. Results
The study variables, upon which the following analyses are based, consist of the Wve-category and two-category CWB subscales, and seven additional variables predicted to relate
to CWB. Data from three of our prior studies were combined to increase statistical power
and simplify interpretation. Before combining the samples, however, we compared them on
means and correlations of the CWB scales with the Wve additional variables common
across at least two samples. A series of one-way ANOVAs were conducted to compare
means on all variables among the samples. Nine of the 12 comparisons were statistically
signiWcant (job satisfaction, procedural justice, and withdrawal were not), which is not surprising considering the power to detect mean diVerences with our sample sizes. EVect sizes,
however, were not large in most cases, with a median r2 of .04 without the three nonsigniWcant variables. The Bruursema et al. sample tended to report the most CWB and the Goh
et al. sample the least. We also compared the samples pairwise using z tests for independent
correlations on all four stressors and job satisfaction versus the seven CWB subscales (81
comparisons across the 3 samples). Only seven were statistically signiWcant (8.6%), all but
one involving interpersonal conXict. In three of the seven cases one sample was signiWcant
and the other nonsigniWcant. Given the small number of signiWcant diVerences, it seems
likely that they were due to Type 1 errors.
Table 1 contains descriptive statistics for the variables in the study, including means,
standard deviations, observed and possible ranges, and coeYcient s. CoeYcient s
exceeded the generally accepted minimum of .70 for all but some of the CWB scales. It
should be kept in mind, however, that behavior checklists are best considered causal indicator scales (Bollen & Lennox, 1991; Edwards & Bagozzi, 2000) in which items are not
interchangeable indicators of a single underlying construct. Thus spreading a rumor is not
the same thing as threatening someone with harm. This often results in low internal consistencies for this sort of scale because the items deWne the construct rather than the items
being the reXection of the construct.
454
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
Table 1
Descriptive statistics for constraints, interpersonal conXict, job satisfaction, emotions, and CWB subscales
Variable
N
Mean
SD
Observed range
Possible range
CoeYcient Interpersonal conXict
Organizational constraints
Distributive justice
Procedural justice
Job satisfaction
Upset
Boredom
Abuse
Production deviance
Sabotage
Theft
Withdrawal
CWB-O
CWB-P
739
733
441
435
737
166
166
735
737
738
738
738
735
735
6.1
22.9
19.6
59.7
13.3
18.3
2.7
24.1
3.7
3.6
5.8
6.9
31.1
26.3
2.6
8.3
6.2
16.3
4.1
6.5
1.3
6.7
1.3
1.1
1.5
2.3
7.4
6.3
4–19
11–55
6–30
12–84
3–18
8–35
1–5
18–82
3–12
3–11
5–20
4–16
21–81
22–88
4–20
11–55
6–30
12–84
3–18
8–40
1–5
18–90
3–15
3–15
5–25
4–20
21–105
22–110
.77
.88
.93
.96
.90
.88
na
.81a
.61
.42
.58
.63
.84
.85a
Note. CWB-O, CWB directed against organizations; CWB-P, CWB directed against people.
a
N D 460 for coeYcient due to one missing item in Sample 3; na, not applicable.
Table 2 shows correlations among the variables in the study. Of particular concern to
this discussion is the pattern of relationships between the CWB subscales and other variables, and whether or not they matched our hypotheses. Abuse had the strongest correlations for stressors, justice, and negative emotions, and it correlated most strongly with
interpersonal conXict. This Wt our expected pattern since we assumed that abuse represents
hostile aggression directed toward the most immediate causes of stressors and emotion. It
failed to correlate signiWcantly with boredom, suggesting that it is unlikely employees
engage in these behaviors to cope with being bored.
Production deviance, as hypothesized, showed a similar pattern of relationships to
abuse with all but boredom being signiWcant. Correlations tended to be smaller in magniTable 2
Correlations among all study variables
Variable
1
2
3
4
5
1. ConXict
2. Constraints
3. Distributive justice
4. Procedural justice
5. Job satisfaction
6. Upset
7. Boredom
8. Abuse
9. Production deviance
10. Sabotage
11. Theft
12. Withdrawal
13. CWB-O
14. CWB-P
.48*
¡.34*
¡.46*
¡.25*
.55*
.04
.54*
.28*
.26*
.19*
.14*
.35*
.46*
¡.38*
¡.50*
¡.36*
.61*
.01
.32*
.23*
.19*
.15*
.18*
.31*
.26*
.51*
.45*
¡.43*
¡.09
¡.14*
¡.13*
¡.04
¡.06
¡.12*
¡.18*
¡.10*
.46*
¡.42*
¡.18*
¡.31*
¡.21*
¡.19*
¡.16*
¡.19*
¡.31*
¡.25*
¡.54*
¡.28*
¡.23*
¡.24*
¡.14*
¡.08*
¡.22*
¡.35*
¡.14*
6
.17*
.28*
.22*
.14
.05
.24*
.37*
.16*
7
8
9
10
11
12
13
.11
.06
.09
.06
.27*
.33*
.06
.49
.49
.43*
.35*
.65*
.94*
.53
.40*
.37*
.71*
.48*
.46*
.29*
.65*
.51*
.40*
.65*
.45*
.74*
.33*
.62*
Note. Sample sizes were 164–166 for correlations with emotions; 434–440 for correlations with justice; 732–738 for
the remaining variables. CWB-O, CWB directed against organizations; CWB-P, CWB directed against people.
¤
p< .05.
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
455
tude except for job satisfaction, particularly for interpersonal conXict, constraints, and
procedural justice. For sabotage all correlations were smaller in magnitude than for abuse,
but distributive justice, upset, and bored were all nonsigniWcant. Correlations were somewhat lower than production deviance for some but not all variables, thus only partially
supporting our hypotheses.
Theft showed small correlations with the seven variables, with four being signiWcant. In fact
the largest correlation was .19 with interpersonal conXict, with a ¡.16 for procedural justice and
¡.08 for job satisfaction, both of which supported our hypothesis. Withdrawal showed a unique
pattern of relationships, being signiWcantly related to all seven other variables as hypothesized
and being the only subscale to relate signiWcantly to boredom. It was unique in correlating more
strongly with job satisfaction and emotions than to stressors and justice.
For comparison we included correlations with CWB-O and CWB-P that combine items
across the Wve subscales. CWB-P consists mainly of abuse items, whereas CWB-O combines items from the other four subscales. As can be seen in the table, only 1 of 24 correlations with the other variables was nonsigniWcant (CWB-P with bored). The patterns of
correlations with stressors were quite similar across the two CWB measures, but CWB-O
tended to correlate more strongly with emotion and job satisfaction than did CWB-P.
Table 3 shows the individual items by the Wve subscales and the percent of employees
who reported to have engaged in each CWB. The mean percentage across all 33 items was
21.3% (standard deviation D 17.0%) and the median was 15.4%. As can be seen, there was
considerable variation in the frequency with which these behaviors are reported, ranging
from less than 2.6% of the sample for stealing from someone to 61.6% of the sample who
admitted to taking a longer break than permitted. There was also variation within the subscales, with a median percent reporting of 52% across the withdrawal items and less than
15% for each of the other items within each subscale.
4. Discussion
Most recent organizational research on CWB has taken a global perspective of combining diverse behaviors into one or two (organization vs. person target) indices of CWB,
although often called a diVerent term. This is based on theories that consider these various
CWBs to be manifestations of the same underlying motive such as aggression (Neuman &
Baron, 1997) or revenge (Bies et al., 1997). Indeed there is evidence that motives can be
shared across types of behavior. For example, injustice has been raised as a potential cause
of each of our Wve types. However, there also exists work that suggests diVerences in causes
and motives as well. As noted earlier, Roscigno and Hodson (2004) provided evidence that
speciWc types of CWB can occur under diVerent organizational conditions. Furthermore,
the motives underlying some forms of CWB (e.g., aggression) are more likely to be hostile
than other forms (e.g., theft) which are more likely to be instrumental.
Our results supported the idea that the Wve types of CWB tend to have diVerent potential antecedents. Abuse tends to be associated more with job stressors than psychological
strains, and it was related more strongly to upsetting emotions than boredom. In fact, an
examination of the individual emotion items within the upset subscale showed that abuse
was most strongly related to the items angry (r D .37) and furious (r D .33), with correlations less than .21 for the other six negative emotions. Angry and furious were correlated
substantially lower with the four other CWB subscales with all but two correlations below
.20 (angry D .23 with withdrawal and furious D .21 with sabotage). Abuse correlated most
456
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
Table 3
Counterproductive work behaviors by subscale and percent reported
CWB item number and item
Dimension
%a
Purposely wasted your employer’s materials/supplies
Purposely damaged a piece of equipment or property
Purposely dirtied or littered your place of work
Came to work late without permission
Stayed home from work and said you were sick when you were not
Taken a longer break than you were allowed to take
Left work earlier than you were allowed to
Purposely did your work incorrectly
Purposely worked slowly when things needed to get done
Purposely failed to follow instructions
Stolen something belonging to your employer
Took supplies or tools home without permission
Put in to be paid for more hours than you worked
Took money from your employer without permission
Stole something belonging to someone at work
Told people outside the job what a lousy place you work for
Started or continued a damaging or harmful rumor at work
Been nasty or rude to a client or customer
Insulted someone about their job performance
Made fun of someone’s personal life
Ignored someone at work
Blamed someone at work for error you made
Started an argument with someone at work
Verbally abused someone at work
Made an obscene gesture (the Wnger) to someone at work
Threatened someone at work with violence
Threatened someone at work, but not physically
Said something obscene to someone at work to make them feel bad
Did something to make someone at work look bad
Played a mean prank to embarrass someone at work
Looked at someone at work’s private mail/property without permission
Hit or pushed someone at work
Insulted or made fun of someone at work
Sabotage
Sabotage
Sabotage
Withdrawal
Withdrawal
Withdrawal
Withdrawal
Production deviance
Production deviance
Production deviance
Theft
Theft
Theft
Theft
Theft
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
Abuse
29.8
3.0
7.9
54.1
49.9
61.6
43.0
11.2
29.2
12.7
11.8
26.1
15.4
3.5
2.6
48.7
10.1
32.5
26.0
28.0
50.1
15.7
26.0
10.7
18.5
2.8
6.4
8.5
8.1
7.0
12.2
3.4
26.4
The CWB-C is Copyright Suzy Fox and Paul E. Spector, All rights reserved 2003, items reproduced by permission.
a
Percent of sample reporting engaging in behavior.
strongly with interpersonal conXict, which should not be surprising since many of the
abuse items overlap with conXict. The major diVerence between the two scales is that the
conXict scale asks about what the individual has experienced while the CWB scale asks
what the individual has done himself or herself.
Production deviance had a similar pattern to abuse, although several of the correlations
tended to be smaller. This supports the idea that it may share a similar underlying motivation and may reXect displaced aggression from an individual to the inanimate organization. It correlated signiWcantly with six of the eight individual emotions, with the largest
correlations being for fatigue (r D .21), gloomy (r D .20), and furious (r D .19).
Sabotage was more modestly correlated with the other variables than abuse or production deviance, with only one correlation for interpersonal conXict being larger than .20. It
was not signiWcantly correlated with upset or bored. The smaller correlations may reXect
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
457
the inhibition of sabotage since destroying property is something that is likely to be visible
and might result in punishment.
Theft showed the least correlation with the other variables, with the largest correlation
with interpersonal conXict (r D .19), and only half being signiWcant. The remaining signiWcant correlations ranged from .08 to .16 in absolute value. It failed to correlate signiWcantly
with either of the emotion variables (or individual items), and only correlated a tiny but
signiWcant ¡.08 with job satisfaction. Most of the motives for theft are probably instrumental, and our results support the idea that connections with stressors are weaker than
the other CWB forms and there was no connection with emotion. We conducted one additional analysis with the theft scale, separating the single item that asked about stealing
from someone at work from the other items that asked about stealing from the employer.
The single item correlated .26 with interpersonal conXict versus .17 for the remaining items
(both were signiWcant), suggesting that theft might have a hostile motive in some cases,
although it is not clear whether theft was the cause or eVect of the conXict. Considering the
established link between injustice and theft (e.g., Greenberg, 1990, 2002), it was surprisingly
that the correlation with justice was quite modest, with a signiWcant correlation only with
procedural justice of ¡.16. This might be because few of the participants in our study experienced the level of injustice that would motivate theft, although theft itself was reported
by our participants at about the same frequency level as production deviance, sabotage
and withdrawal, all of which correlated signiWcantly with emotion.
Withdrawal correlated signiWcantly with all variables. Opposite to abuse, it tended to
correlate higher with strains (rs ranged in absolute value from .22 to .27) than stressors
(rs ranged in absolute value from .12 to .19). It was unique in being related signiWcantly to
boredom and at about the same magnitude as with upset. An inspection of the individual
emotion items suggested that withdrawal is primarily associated with boredom and feelings of depression. Thus in contrast to abuse and sabotage, individuals who engage in withdrawal might be escaping a diVerent set of emotions. This goes back to the original Wght or
Xight idea with those experiencing anger being more likely to engage in abuse directed
toward others or sabotage directed toward the organization (Wght) whereas those experiencing boredom and depression respond with withdraw (Xight).
By contrast, there were far fewer distinctions between CWB-O and CWB-P. CWB-O
which was a combination of the items from all but abuse correlated in the .30s with all but
distributive justice. Throwing together production deviance, sabotage, theft, and withdrawal may give the false impression that all four forms of CWB relate to the same stressors and strains. An examination of the Wner-grained individual subscales shows that this is
not the case.
The current CWB literature has tended to group diVerent behaviors into an overall
composite, considering them mainly as responses to injustice and work stressors, and associating these behaviors with anger. More research needs to consider instrumental motives
for these behaviors that go beyond emotions. Thus while a Stressor-Emotion model may
best depict certain forms of CWB, perhaps a clearly distinct theoretical framework may be
need to understand instrumental CWB, such as the Theory of Planned Behavior (Ajzen &
Fishbein, 1977). This may parallel the distinction highlighted in the work stress literature
between emotion-focused and problem-focused coping (Lazarus, 1995). Furthermore, the
interplay of such instrumental motives with more emotion-driven motives should be considered. For example, an individual might use the fact that something elicited anger to
rationalize stealing from the employer or taking unauthorized breaks. In such a case anger
458
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
might have been a releasing condition that allowed the individual to do something he or
she wanted to do, but was inhibited by internal ethics and values until a suYcient justiWcation occurred.
The limitations of this study should be kept in mind. Of particular concern is that the
samples are comprised largely of employed students, and it is not certain that their
responses would match other employed groups, although we found few diVerences with the
nonstudent sample. Of course, few studies have representative samples and so there is often
a concern about generalizability. Clearly additional studies with diVerent employed populations are needed.
The past few years has seen a rapid development in the study of CWB, producing a
number of Wndings and theoretical perspectives that have enhanced our understanding of
this phenomenon. Our results here suggest the time has come to begin to look more microscopically at the general class of behaviors to see where there are similarities and diVerences in underlying causes and consequences by speciWc types of CWB. Some of this
research has already suggested that employees engage in many of these behaviors for a
variety of motives elicited by many diVerent conditions. Models that consider multiple reasons for diVerent forms of CWB should help us develop a deeper understanding of this
phenomenon that will ultimately help in minimizing harm to employees and organizations.
References
Ajzen, I., & Fishbein, M. (1977). Attitude–behavior relations: A theoretical analysis and review of empirical
research. Psychological Bulletin, 84, 888–918.
Ambrose, M. L., Seabright, M. A., & Schminke, M. (2002). Sabotage in the workplace: The role of organizational
injustice. Organizational Behavior and Human Decision Processes, 89, 947–965.
Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27–51.
Anderson, C. A., Deuser, W. E., & DeNeve, K. M. (1995). Hot temperatures, hostile aVect, hostile cognition,
and arousal: Tests of a general model of aVective aggression. Personality and Social Psychology Bulletin, 21,
434–448.
Bennett, R. J., & Robinson, S. L. (2000). Development of a measure of workplace deviance. Journal of Applied
Psychology, 85, 349–360.
Berkowitz, L. (1998). AVective aggression: The role of stress, pain, and negative aVect. In R. G. Geen & E. Donnerstein (Eds.), Human aggression: Theories, research, and implications for social policy (pp. 49–72). San Diego:
Academic Press.
Bies, R. J., & Tripp, T. M. (2005). The study of revenge in the workplace: Conceptual, ideological, and empirical
issues. In S. Fox & P. E. Spector (Eds.), Counterproductive workplace behavior: Investigations of actors and
targets (pp. 83–105). Washington, DC: APA.
Bies, R. J., Tripp, T. M., & Kramer, R. M. (1997). At the breaking point: Cognitive and social dynamics of revenge
in organizations. In R. A. Giacalone & J. Greenberg (Eds.), Antisocial behavior in organizations (pp. 18–36).
Thousand Oaks, CA: Sage.
Bollen, K., & Lennox, R. (1991). Conventional wisdom on measurement: A structural equation perspective.
Psychological Bulletin, 110, 305–314.
Bruursema, K., Kessler, S., Fox, S., & Spector, P. E. (2004). The connection between counterproductive work
behavior and organizational citizenship performance. Unpublished data set.
Cammann, C., Fichman, M, Jenkins, D., & Klesh, J. (1979). The Michigan organizational assessment questionnaire.
Unpublished manuscript, University of Michigan, Ann Arbor.
Chen, P. Y., & Spector, P. E. (1992). Relationships of work stressors with aggression, withdrawal, theft and substance use: An exploratory study. Journal of Occupational and Organizational Psychology, 65, 177–184.
Crino, M. D. (1994). Employee sabotage: A random or preventable phenomenon? Journal of Managerial Issues, 6,
311–337.
Dalton, D. R., & Mesch, D. J. (1991). On the extent and reduction of avoidable absenteeism: An assessment of
absence policy provisions. Journal of Applied Psychology, 76, 810–817.
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
459
Douglas, S. C., & Martinko, M. J. (2001). Exploring the role of individual diVerences in the prediction of workplace aggression. Journal of Applied Psychology, 86, 547–559.
Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direction of relationships between constructs and measures. Psychological Methods, 5, 155–174.
Folger, R., & Skarlicki, D. P. (2005). Beyond counterproductive work behavior: Moral emotions and deontic
retaliation versus reconciliation. In S. Fox & P. E. Spector (Eds.), Counterproductive workplace behavior: Investigations of actors and targets (pp. 83–105). Washington, DC: APA.
Fox, S., & Spector, P. E. (1999). A model of work frustration-aggression. Journal of Organizational Behavior, 20,
915–931.
Fox, S., Spector, P. E., & Miles, D. (2001). Counterproductive work behavior (CWB) in response to job stressors
and organizational justice: Some mediator and moderator tests for autonomy and emotions. Journal of Vocational Behavior, 59, 291–309.
Geen, R. G. (1998). Processes and personal variables in aVective aggression. In R. G. Geen & E. Donnerstein
(Eds.), Human aggression: Theories, research, and implications for social policy (pp. 1–21). San Diego: Academic Press.
Goh, A., Bruursema, K., Fox, S., & Spector, P. E. (2003, April). Comparisons of self and coworker reports of counterproductive work behavior. Poster session at the meeting of the Society for Industrial and Organizational
Psychology, Orlando.
Greenberg, J. (1990). Employee theft as a reaction to underpayment inequity: The hidden cost of pay cuts. Journal
of Applied Psychology, 5, 561–568.
Greenberg, J. (2002). Who stole the money, and when? Individual and situational determinants of employee theft.
Organizational Behavior and Human Decision Processes, 89, 985–1003.
Hanisch, K. A., Hulin, C. L., & Roznowski, M. (1998). The importance of individuals’ repertoires of behaviors:
The scientiWc appropriateness of studying multiple behaviors and general attitudes. Journal of Organizational
Behavior, 19, 463–480.
Hollinger, R. C. (1986). Acts against the workplace: Social bonding and employee deviance. Deviant Behavior, 7,
53–75.
Johns, G. (1997). Contemporary research on absence from work: Correlates, causes and consequences. In C. L.
Cooper & I. T. Robertson (Eds.), International review of industrial and organizational psychology 1997 (pp.
115–173). Chichester, UK: John Wiley.
Keashly, L., & Harvey, S. (2005). Emotional abuse in the workplace. In S. Fox & P. E. Spector (Eds.), Counterproductive workplace behavior: Investigations of actors and targets (pp. 201–235). Washington, DC: APA.
Koslowsky, M. (2000). A new perspective on employee lateness. Applied Psychology: An International Review, 49,
390–407.
Lazarus, R. S. (1995). Psychological stress in the workplace. In R. Crandall & P. L. Perrewe’ (Eds.), Occupational
stress (pp. 3–14). Washington, DC: Taylor & Francis.
Lee, V. B., & Spector, P. E. (in press). The social stressors-counterproductive work behaviors link: Are conXicts
with supervisors and coworkers the same? Journal of Occupational Health Psychology.
Martinko, M. J., Gundlach, M. J., & Douglas, S. C. (2002). Counterproductive workplace behavior: A causal reasoning perspective. International Journal of Selection and Assessment, 10, 36–50.
Miles, D. E., Borman, W. E., Spector, P. E., & Fox, S. (2002). Building an integrative model of extra role work
behaviors: A comparison of counterproductive work behavior with organizational citizenship behavior. International Journal of Selection and Assessment, 10, 51–57.
Moorman, R. H. (1991). Relationship between organizational justice and organizational citizenship behaviors:
Do fairness perceptions inXuence employee citizenship? Journal of Applied Psychology, 76, 845–855.
Mustaine, E. E., & Tewksbury, R. (2002). Workplace theft: An analysis of student-employee oVenders and job
attributes. American Journal of Criminal Justice, 27, 111–127.
Neuman, J. H., & Baron, R. A. (1997). Aggression in the workplace. In R. A. Giacalone & J. Greenberg (Eds.),
Antisocial behavior in organizations (pp. 37–67). Thousand Oaks, CA: Sage.
O’Leary-Kelly, A. M., GriYn, R. W., & Glew, D. J. (1996). Organization-motivated aggression: A research framework. Academy of Management Review, 21, 225–253.
Payne, B. K., & Gainey, R. R. (2004). Ancillary consequences of employee theft. Journal of Criminal Justice, 32,
63–73.
Pearson, C. M., Andersson, L. M., & Porath, C. L. (2005). Workplace incivility. In S. Fox & P. E. Spector (Eds.),
Counterproductive workplace behavior: Investigations of actors and targets (pp. 177–200). Washington, DC:
APA.
460
P.E. Spector et al. / Journal of Vocational Behavior 68 (2006) 446–460
Penney, L. M., & Spector, P. E. (2002). Narcissism and counterproductive work behavior: Do bigger egos mean
bigger problems? International Journal of Selection and Assessment, 10, 126–134.
Penney, L. M., & Spector, P. E. (2005). Job stress, incivility, and counterproductive work behavior (CWB): The
moderating role of negative aVectivity. Journal of Organizational Behavior, 26, 777–796.
Perlow, R., & Latham, L. L. (1993). Relationship of client abuse with locus of control and gender: A longitudinal
study. Journal of Applied Psychology, 78, 831–834.
Peters, L. H., & O’Connor, E. J. (1980). Situational constraints and work outcomes: The inXuences of a frequently
overlooked construct. Academy of Management Review, 5, 391–397.
Price, J. L., & Mueller, C. W. (1986). Handbook of organizational measurement. MarshWeld, MA: Pittman.
Robinson, S. L., & Bennett, R. J. (1995). A typology of deviant workplace behaviors: A multidimensional scaling
study. Academy of Management Journal, 38, 555–572.
Roscigno, V. J., & Hodson, R. (2004). The organizational and social foundations of worker resistance. American
Sociological Review, 69, 14–39.
Skarlicki, D. P., & Folger, R. (1997). Retaliation in the workplace: The roles of distributive, procedural, and interactional justice. Journal of Applied Psychology, 82, 434–443.
Spector, P. E., & Fox, S. (2003, November). Emotional experience at work: Assessing emotions with the Job-related
AVective Well-being Scale (JAWS). Paper presented at the meeting of the Southern Management Association, Clearwater Beach, FL.
Spector, P. E., & Fox, S. (2005). A model of counterproductive work behavior. In S. Fox & P. E. Spector (Eds.),
Counterproductive workplace behavior: Investigations of actors and targets (pp. 151–174). Washington, DC:
APA.
Spector, P. E., & Jex, S. M. (1998). Development of four self-report measures of job stressors and strain: Interpersonal conXict at work scale, organizational constraints scale, quantitative workload inventory, and physical
symptoms inventory. Journal of Occupational Health Psychology, 3, 356–367.
Spector, P. E., Van Katwyk, P. T., Brannick, M. T., & Chen, P. Y. (1997). When two factors don’t reXect two constructs: How item characteristics can produce artifactual factors. Journal of Management, 23, 659–678.
Van Katwyk, P. T., Fox, S., Spector, P. E., & Kelloway, E. K. (2000). Using the job-related aVective well-being
scale (JAWS) to investigate aVective responses to work stressors. Journal of Occupational Health Psychology,
5, 219–230.