Professional Documents
Culture Documents
Christopher Gallagher
A Thesis
MASTER OF ARTS
May 2019
Committee:
Eric Dubow
Scott Highhouse
© 2018
Christopher M Gallagher
employees in the workplace, the effects are not necessarily entirely positive. At times, the
demands our coworkers make of us, either explicitly or implicitly, may culminate in dysfunction.
The construct social burden represents these demands and is predictive of dysfunction when
the likelihood that social burden leads to counterproductive behaviors. Each of these individual
differences has an intensifying effect when taken separately and when taken jointly (i.e., a three-
way interaction). To conclude, I discuss the implications of these findings and directions for
TABLE OF CONTENTS
Page
INTRODUCTION ................................................................................................................. 1
Hypothesis 1................................................................................................... 5
Hypothesis 2................................................................................................... 6
Hypothesis 3....................................................................................... 8
Hypothesis 4....................................................................................... 9
Hypothesis 5....................................................................................... 10
METHOD .............................................................................................................................. 11
Participants................................................................................................................. 11
Measures .................................................................................................................... 11
CWB .............................................................................................................. 12
HAB ............................................................................................................... 12
NR .................................................................................................................. 13
RESULTS .............................................................................................................................. 14
Hypothesis 1................................................................................................... 15
vi
Hypothesis 2................................................................................................... 15
Hypothesis 3................................................................................................... 15
Hypothesis 4................................................................................................... 16
Hypothesis 5................................................................................................... 17
DISCUSSION ........................................................................................................................ 19
REFERENCES ...................................................................................................................... 23
APPENDIX D. MEASURES................................................................................................ 40
INTRODUCTION
Our colleagues and coworkers, superiors and subordinates, all demand from us a certain
amount of supportive action. Whether this comes from the design of the position (e.g., a highly
interdependent team) or outside of the normal process of work (e.g., simple favors), it is
undeniable that this interconnectedness has some effect on all of the actors involved (i.e., the
helper and the helped; Bowler & Brass, 2006). Moreover, the social workplace likely affects all
levels of organizations: individuals, work groups, departments, and so on. In fact, Bandiera,
Barankay, and Rasul (2009) found that increased social connectedness of managers positively
affected productivity at the individual worker level all the way up to the firm level. This result is
not surprising. We intuit that working more closely—even asking favors or disclosing secrets—
brings us closer together. On average, increasing connectedness may be beneficial, but this good
may not be unalloyed. Although these studies support the notion of the benefit of increased
social contact, their design is sure to obscure the complexities of the social environment. In the
present study, I highlight the complexity of social interaction at work by focusing on a construct
Social Burden
Social connectedness is not only related to increased productivity, but group effectiveness
as well (Oh, Chung, & Labianca, 2004). For example, increased social interaction aids in the
development of social skills via task interdependence (Greenberger & Steinberg, 1981) which
likely has positive effects for future interactions within groups at work. However, some scholars
are starting to demonstrate that not all social interactions are beneficial. Social burden is defined
as “the perceived presence of colleagues’ behaviors that elicit the focal employees’ social
support” (Yang, Liu, Nauta, Caughlin, & Spector, 2016, p. 2). In other words, the experience of
SOCIAL BURDEN, HAB, NR INTERACTION 2
social burden occurs when focal employees (burdenees) are exposed to situations where help is
perceived to be needed. Further, it occurs regardless of the intentions of those creating the
burdensome situation (burdeners), whether burdenees provide support or not, and whether the
support provided was effective. An example of a situation in which social burden could occur is
illustrative. A colleague erupts in a highly emotional display (e.g., crying) after giving a
presentation, but this employee does not direct his or her display at anyone remaining in the
room. Typical social norms suggest that someone help the struggling coworker (Stryker &
Statham, 1985), but social burden may be experienced regardless of whether these norms were
Though social burden may have similarities with other organizational or social
constructs, it is conceptually distinct. Unlike other negative social exchanges (e.g., workplace
aggression, incivility), social burden represents behaviors bearing no explicit negativity directed
at the focal employees. For example, although burdeners may be upset, they are not lashing out
at burdenees. Further, whereas other negative social exchanges can only take a negative form,
social burden can take negative, neutral, or positive forms (Basch & Fisher, 2000; Yang et al.,
2016). As a result, the social burden construct may also appear to be a reframing of social
support. While social burden differs from negative social exchange, it also differs from social
support. As illustrated by the emotional outburst example above, helping employees in response
to “social burden scenarios” isn’t required for social burden to have occurred. That is, social
support refers to the provision of the supporting behaviors (e.g., comforting the crying
coworker), whereas social burden refers to only the elicitation of some type of support (e.g., the
because there are potentially differential antecedents and outcomes for these two related
concepts.
Much like the conceptual foundation of the construct, the measurement of social burden
is of great importance. Yang and colleagues (2016) detailed the creation of a short measure of
social burden. This measure is from the perspective of those being exposed (i.e., burdenee),
much like popular measures of incivility (e.g., Cortina, Magley, Williams, & Langhout, 2001).
The authors found social burden to correlate with several undesirable conditions, such as anxiety,
irritation, turnover intentions, and decreased job satisfaction. Further, social burden added
explanatory power in the prediction of these outcomes when controlling for known correlates of
the outcomes of interest, such as age and tenure. Moreover, social burden predicted behaviors
important to the present study: counterproductive work behavior (CWB), a particularly negative
work outcome.
Several theoretical models can be used to aid in the understanding of why social burden
may have primarily negative psychological and behavioral outcomes. According to symbolic
interactionist role theory (Stryker & Statham, 1985) employees hold various formal roles at work
(e.g., coworker, subordinate, or supervisor) and different roles come with different sets of
expectations. These role expectations both guide and constrain behavior during social exchanges.
However, employees can also adopt informal roles such as “being a good colleague” (Stryker &
Vryan, 2006). Although the expectations for informal roles may not be as normative as formal
roles, they provide general guidelines to be kind to others and help, within reason, when needed.
This theoretical perspective highlights the pressures to acquiesce to—and the toll that may arise
from resisting—coworkers eliciting support when people adopt he admirable role of being a
SOCIAL BURDEN, HAB, NR INTERACTION 4
good colleague. It is unlikely that identifying with a “good colleague” role would lead directly to
counterproductive behaviors, but the effect of resisting such a role eventually may lead to such
an outcome.
It is through models of resources (e.g., Bakker & Demerouti, 2007; Hobfoll, 1989) that
we find a path from resisting the “good colleague” role to potential dysfunction. According to the
job demands-resources model, some aspects of employees’ jobs can be demanding while others
can serve as resources (Bakker & Demerouti, 2007). A key component to this model is that job
demands deplete employees’ personal resources (e.g., time or energy) which can impair their
ability to carry out essential job tasks. Further, job demands can negatively impact employees’
health. Conservation of resources theory suggests that investing time and energy—types of
resources—is likely to result in strain and negative outcomes, particularly if resources are not
Underdahl, & Westman, 2014). Therefore, social burden can be viewed as a job demand in that
repetitious help requests from others in the workplace can become burdensome, where attending
to them gradually depletes burdenees’ resources. Even when help is not provided, burdenees may
experience negative emotions like guilt or regret. As burdenees’ affective, cognitive, and
physical resources become depleted, they may be unable to regulate their work behavior, such as
those consistent with workplace norms (Baumeister & Vohs, 2007). These resource-based
theoretical perspectives lead from the pressures felt in the symbolic interactions perspective
detailed above into a process whereby those pressures are converted into work-related
dysfunction such as poor performance on job tasks or, importantly to the present study, increased
(CWB) is “volitional acts by employees that violate the legitimate interests of, or do harm to an
organization, its members, or its stakeholders” (Marcus, Taylor, Hastings, Sturm, & Weigelt,
2016, p. 204). CWB can be viewed from multiple levels of specificity: a general CWB factor
(Spector, Bauer, & Fox, 2010), person and organization factors (Bennett & Robinson, 2000;
Spector & Fox, 2003), and a multifaceted measure of specific behaviors. Many aspects of CWB
fall within a taxonomy of overall work performance (Campbell & Wiernik, 2015), underscoring
The preceding describes how elicitations of social support can be stressors to those being
elicited, whether or not support is provided. The stressor-emotion model of CWB (Spector &
Fox, 2005) and similar models that view CWB as arising from exposure to stressors can link the
above explanations to the commission of CWB. In the case of the stressor-emotion model (see
Figure A1 in Appendix A), the external stressor is perceived as a source of stress which gives
rise to negative emotion. The negative emotions then lead to CWB. Applying this model to the
social burden situation, we can see that social burden may lead to the depletion of resources
which may lead to increases in negative emotions and counterproductive behaviors. A revised
version of the stressor-emotion model of CWB that highlights where the present study fits into
the model is presented in Figure A2 in Appendix A. Empirical evidence has supported the link
between personal resource depletion and CWB (Christian & Ellis, 2011); therefore, the following
is proposed:
Although Yang and colleagues (2016) did not examine whether social burden was related
to CWB in general, they did test whether social burden was related to CWB targeting people in
SOCIAL BURDEN, HAB, NR INTERACTION 6
the workplace. Following the precedent of the deviant workplace behavior literature, CWB
researchers often divide behaviors under the construct’s umbrella into behaviors directed at
people (CWB-I) and behaviors directed at the organization (CWB-O). Due to the specifically
interpersonal nature of social burden, the authors focus on CWB-I as an outcome. This is an
intuitive link—if a burdenee seeks to remedy or lash out at the source of burden, it would
logically be at the individual burdener and not the broader organization. Therefore, to replicate
Hypothesis 2. Social burden has a stronger relation with CWB-I than CWB-O.
Ambiguity is a characteristic shared by the social exchanges that fall into both the social
burden construct and those of workplace incivility. Incivility entails violating workplace norms
for mutual respect with an ambiguous intent to harm (Cortina et al., 2001) and is often
exemplified by insulting remarks and social exclusion. Incivility could overlap with social
burden if emotional display rules at work were dictated tightly by norms, though measures of
incivility do not typically consider such unique norms (Cortina et al., 2001; Matthews & Ritter,
2016). Wu, Zhang, Chiu, Kwan, and He (2014) identified two variables that would interact with
incivility in the prediction of deviant workplace behaviors: hostile attribution bias (HAB) and
negative reciprocity beliefs (NR). In this study, both HAB and NR increased the positive relation
between experienced incivility and deviant behaviors at work. Although other individual
differences are potentially important to social burden, HAB and NR are individual difference
variables particularly suited to affect the outcomes of the social burden dynamic due to their
Hostile attribution bias. The way individuals attribute intentions and other
attributional style (Douglas, Martinko, Gundlach, & Douglas, 2002). Though the idea that
attributions may be incorrect and be responsible for dysfunction (e.g., fundamental attribution
error; Ross, 1977; Heider, 1958) predates the hostile attribution bias (HAB) construct, HAB is a
less generalized attributional tendency whereby people tend to attribute hostile intentions to
ambiguous situations (Dodge & Frame, 1982). More plainly stated, people high in HAB tend to
Likely due to maturity effects and relatively pronounced importance in childhood, hostile
attributions in the workplace have not received as much attention as many other individual
differences. However, there is a growing base of evidence for the construct’s importance to
organizational scholars. Spector and Zhou (2014) found that, given a battery of individual
differences, HAB was among only conscientiousness and agreeableness as significant predictors
of the three types of CWB measured. This suggests that HAB may be an overlooked variable in
the CWB area, as conscientiousness and agreeableness have decades of evidence supporting their
use (Ones & Viswesvaran, 2003). Wu and colleagues (2014) found that HAB strengthened the
positive relation between incivility and deviance. Similarly, Hoobler and Brass (2006) found that
supervisors higher in HAB were more likely to respond with abuse toward subordinates than
those low in HAB. These studies suggest that attending to individuals’ attributional tendencies is
important for researchers interested in social dynamics at work. Due to the nature of HAB, it is
particularly important when the social phenomenon is potentially ambiguous, as in the case with
Hypothesis 3. HAB will interact with social burden in the prediction of CWB such that
Negative reciprocity. Reciprocity is a widespread norm in society which states that one
should “return the favor” when presented with benefits (Gouldner, 1960). Individuals who
endorse the return of negative acts in kind are said to hold negative reciprocity beliefs (NR)
(Eisenberger, Lynch, Aselage, & Rohdieck, 2004). Eisenberger et al. went beyond the common
notion of reciprocity by demonstrating that individuals differed on the extent to which they
believe that reciprocity is an idea that pertains to the positive, the negative, or both. This is an
important distinction, as previous work did not distinguish between reciprocity for positive acts
and reciprocity for negative acts, sometimes ignoring the possibility for a negative reciprocity at
all. Eisenberger and colleagues found that high NR in situations of mistreatment predicted
increased anger and behaviors such as ridicule of those who mistreated them, suggesting that NR
beliefs should be taken into consideration by researchers in situations where people may feel
motives. In particular, NR has been shown to predict willingness to avenge a wrong (Barclay,
Whiteside, & Aquino, 2014; Garcia, Restubog, & Denson, 2010), though it is often shown as
interacting with other variables. Many organizational studies have used the construct in relation
to negative outcomes such as deviance (e.g., Mitchell & Ambrose, 2007). As long as the social
exchanges viewed by individuals with negative reciprocity beliefs are perceived as negative, they
should be more likely to “seek revenge” or respond in some similarly negative manner
As Yang and colleagues (2016) suggest, the negative effects of social burden may arise
from the accumulation of support-seeking behaviors. That is, any single event of social burden
may not have an adverse effect, but the accumulation of these events is associated with the
undesirable constructs found in the small body of social burden research. When this
accumulation is seen as negative by the burdenee, belief in negative reciprocity becomes relevant
and likely to lead to CWB. This dynamic leads to the fourth hypothesis for the present study.
Hypothesis 4. NR will interact with social burden in the prediction of CWB such that NR
The stressor-emotion model of CWB (Spector & Fox, 2005) can account for the
hypothesized effect of both of the moderators proposed here. First, the event must be perceived
by the individual to be a source of stress, which the model considers to be affected by appraisal
constructs (e.g., HAB). In the case of the proposed study, Hypothesis 3 states that those high in
HAB are more likely to appraise these events as stressors. Later in the model, the experience of
negative emotion following the stressor leads to CWB. It is here that Hypothesis 4 fits, as
negative reciprocity beliefs cause people to take action following negative events. As an
extension of finding CWB linked directly to social burden, the inclusion of these two individual
Wu et al. (2004) uncovered a significant three-way interaction with HAB, NR, and
incivility in the prediction of deviance. Similarly, there is potential for interplay between these
variables (i.e., HAB and NR) in predicting CWB from social burden in the present study. Indeed,
HAB becomes essential to a hypothetical series of events culminating in lashing out with
outburst after a presentation, we can expect that, of the coworkers remaining in the room to
witness, employees higher in HAB will be likelier to view the outburst as a negative act directed
at others. Without this attribution, the belief in NR has less influence on subsequent behaviors.
By definition, people who are high in HAB are involved in more of these self-perceived negative
events, and those high in HAB who are also high in NR could then be expected to be motivated
to respond with counterproductive behaviors beyond those high on only one of these two
variables.
Hypothesis 5. There will be a three-way interaction between HAB, NR, and social burden
in prediction of CWB, such that those highest in HAB and NR will demonstrate the highest
METHOD
Participants
Participants were recruited via Amazon’s Mechanical Turk (MTurk) service and were
required to work at least 20 hours per week in the United States, not including their MTurk work.
Participants were directed to a Qualtrics survey including an item to screen out participants who
use MTurk for a major portion of their income and three attention-check items to protect against
regression is concerning. Although many factors that can be difficult to plan for can affect the
power to detect a true effect in moderated regression (e.g., violation of assumptions or reliability
of predictors; Aguinis, Beaty, Boik, & Pierce, 2005), it is possible to broadly estimate the
necessary sample size given an estimated effect size and characteristics of the regression model.
I estimated the sample size necessary to find an effect given effect size of f2 = .01 and f2 = .02,
where f2 = .02 reflects a small effect size as defined by Cohen (1988) and 80% power. This
Given these considerations, an initial pool of 525 participants completed the survey, of
which 491 remained after screening out participants who incorrectly responded to 2 out of 3
attention-check items. The participants used in all analyses (N = 491) were 81% White, 7%
Black, 7% Asian, and 5% other or multiple ethnicities. The average age of the sample was 37
years, ranging from 18 to 69 with a standard deviation of 11 years, and 60% female.
Measures
The items and instructions for all measures can be found in Appendix B.
SOCIAL BURDEN, HAB, NR INTERACTION 12
and asks participants to rate the frequency with which they experienced these over the past
month. The measure contains 4 items of response range 1 to 5, including “Acted emotionally
behaviors, often combining items from multiple dimensions to yield the overall CWB scores. I
used the Spector and colleagues (2006) multidimensional, 32-item measure of CWB, ultimately
split between the organizationally focused and individually focused items. An example of a
CWB-I scale item is “Blamed someone at work for an error you made,” and a CWB-O item is
“Purposely dirtied or littered your place of work.” This measure asks for a frequency with which
vignettes, the burden on organizational researchers’ participants creates the need for a more
abbreviated way of measuring HAB. In order to test the relative viability of measuring HAB in
the traditional manner by organizational researchers, I used the 7-item Workplace Hostile
Attributions Scale (WHAS; Bal & O’Brien, 2010). An example item from this measure is “If
coworkers ignore me, it is because they are being rude.” The response format for this scale is
instrument such as those used by researchers in developmental and criminal psychology areas,
yet developed for the average person. To this end, I used the ambiguous-intentionality items
created by Tremblay and Belchevski (2004), which contain seven scenarios such as “You go to
your work and you are in a bad mood. As you walk in the office, a coworker teases you about
SOCIAL BURDEN, HAB, NR INTERACTION 13
something that you are wearing.” (p. 423). After reading each scenario, participants rate how
intentional and hostile each scenario is on a scale from 1 to 10. The items relating to hostility are
NR. The shortened NR scale contains seven items, such as “If a person despises you, you
should despise them” (Eisenberger et al., 2004). This scale has response options ranging from 1
to 7.
SOCIAL BURDEN, HAB, NR INTERACTION 14
RESULTS
The descriptive statistics for the seven variables included in this study can be found in
Table 1. Pearson correlations between all variables can be found in Table 2. Along the diagonal
Data were collected using both the vignette-style and Likert-style HAB instruments to
compare the relative strengths and weaknesses of these approaches in an organizational context.
The question of which measure is most appropriate or superior is often subjective and context
dependent. One way to answer that question is to examine whether the measures create different
results in statistical analyses with other constructs. Another is to compare the measurement
properties of both scales. Taken together, this evidence can offer some guidance as to the most
To begin comparing the WHAS with the vignette-based measure of HAB, I examined the
internal consistency of each measure. The WHAS is considerably more internally consistent than
the vignettes (i.e., Cronbach’s alphas .71 versus .85, respectively). Relatedly, both measures
present a similar pattern of item-total correlations, i.e., no items problematically low in their
discrimination, but the WHAS items have generally higher item-total correlations (see Table 3).
the same item quantity, is to examine the variance extracted via factor analysis (Henson &
Roberts, 2006). The two measures load as expected, likely due to contextual (i.e., work-specific
vs. life-general) and format (i.e., vignette vs. agreement) factors, on two separate factors which
are correlated at .41. The WHAS accounts for, after rotation, more variance explained in the
indicators than the vignettes (3.70 versus 2.65, respective sum of squared loadings, see Table 3).
The convergence of all this evidence, and the absence of evidence to the contrary, strongly
SOCIAL BURDEN, HAB, NR INTERACTION 15
suggests that the WHAS is the most appropriate measure for the current study. All analyses
Tests of Hypotheses
Hypothesis 1. With the intention of replicating the findings of the Yang et al. (2016)
article, I hypothesized a positive relationship between social burden and CWB. Indeed, the
correlation between the two variables was significant and positive: r(489) = .41, p < .001. Thus,
Hypothesis 2. Next, I posited that, due to the interpersonal nature of both social burden
and individually focused CWB, social burden and CWB-I would be more strongly related than
social burden with CWB-O. Referencing the correlation table, at two digits the correlations are
equivalent (i.e., r = .40) suggesting that there is no support for this hypothesis. A more robust
test of this hypothesis is Williams’s Test of dependent correlations that share one variable
(Steiger, 1980). As suggested by a glance, this test also fails to support the hypothesis: t(487) =
0.06, p = .52. Thus, Hypothesis 2 was not supported. Given the apparent unity between CWB-I
and CWB-O in the current sample, and in order to retain information from both dimensions,
CWB will be measured as an overall construct for the remainder of the study unless otherwise
noted.
Hypothesis 3. The third hypothesis states that HAB will interact with social burden in the
prediction of CWB such that the social burden-CWB relation will be stronger when HAB is high.
In order to test this, moderated multiple regression was used with consideration to suggestions
made by Cortina (1993). This consists of including a squared term for each interacting variable.
For example, a regression with predictors X, Z, and their interaction term XZ, should
additionally include X2 and Z2 to ensure that the results of the test of interaction terms is not due
SOCIAL BURDEN, HAB, NR INTERACTION 16
to variance shared with the non-linear (i.e., quadratic) term, which is particularly important as
predictors becoming increasingly correlated (Cortina, 1993). The interaction term’s inclusion in
the final step of the regression offers a significance test for the interaction, and the interaction
Hypothesis 3 was supported using the above analytic strategy. Specifically, although the
quadratic terms accounted for significant variance explained beyond the lower-order variables
(ΔR2 = 0.015, F[2, 486] = 5.35, p = .005), the HAB by social burden interaction term also
accounted for significant additional variance (ΔR2 = 0.061, F[1, 485] = 47.84, p ≤ .001) beyond
both the nonlinear effects and main effects. The entire regression results can be found in Table 4.
To better understand how the two variables interact in the prediction of CWB, I plotted the
interaction in the original metric of the scales (see Figure 1) at low, average, and high levels of
HAB. Average and high HAB, represented by the mean and one standard deviation above the
mean, demonstrate a progressively stronger impact on predicted CWB via social burden, further
supporting the hypothesis. As a final note, the y-axis—representing CWB—of all interaction
plots should be taken into consideration. The plots are scaled to focus on simple slopes and not
prediction across the possible range of CWB. Indeed, the range of CWB is very low, as self-
Hypothesis 4. Using the same analytic process described for Hypothesis 3, I tested the
hypothesis that negative reciprocity interacts with social burden in the prediction of CWB such
that the social burden-CWB relation is stronger when negative reciprocity is higher. Although
the nonlinear terms accounted for significant variance in the preceding hypothesis, they did not
do so in the test of this hypothesis (ΔR2 = 0.01, F(2, 486) = 2.46, p = .086). As in Hypothesis 3,
the interaction term explained significant additional variance beyond the lower-order and
SOCIAL BURDEN, HAB, NR INTERACTION 17
nonlinear effects (ΔR2 = 0.056, F(1, 485) = 41.05, p ≤ .001). The entire regression result can be
found in Table 5. Like for the previous hypothesis, I plotted the interaction to better understand
its effects (see Figure 2) at the low, average, and high NR. From the plot, we can see that the
highest NR scores have the greatest relation between social burden and CWB. This lends further
support to Hypothesis 4.
be troublesome. Between the compounding of measurement error, and the rarity of extreme
scorers on three combined variables, the statistical power to find a three-way interaction is
severely diminished (McClelland & Judd, 1993). Explaining a three-way interaction once
relation of a predictor and outcome depending jointly on the level of two other predictors. For
the present study, I predicted that the social burden-CWB relation would depend on HAB and
NR jointly. Further, I predicted that those high in HAB and NR would demonstrate the largest
To test whether the three-way interaction was significant, the suggestion to include
quadratic terms to control for potential nonlinear effects sharing variance with the interaction
terms (Cortina, 1993) was discarded as the inclusion made no appreciable difference at the two-
way interaction level. This has the benefit of simplifying interpretations of the results of the
regression. Otherwise, the hierarchical procedure was retained, and the results of the
regression(s) can be found in Table 6. The final step in the hierarchical regression supported the
hypothesis that the three-way interaction was significant (ΔR2 = 0.016, F[1, 483] = 13.73, p <
.001). Because three-way interactions can be difficult to understand, I present two graphs that
offer complementary perspectives. The first, Figure 3, captures the progressive slopes involved
SOCIAL BURDEN, HAB, NR INTERACTION 18
in the social burden-by-HAB across a range of NR values. In this graph, wee see that each
Figure 4 plots these slopes in combinations and aids in understanding the test of
Hypothesis 5. To test the hypothesis that the greatest social burden-CWB relationship would be
found in those with both high HAB and high NR scores, a test of the difference between slopes is
necessary (Dawson & Richter, 2006). Specifically, this requires three tests: test 1, a test of
difference between slopes of high HAB-high NR (i.e., slope 1 in Figure 4) and high HAB-low
NR (i.e., slope 2 in Figure 4; t[483] = 3.73, p < .001), test 2 between high HAB-high NR and
low HAB-high NR (i.e., slopes 1 and 3; t[483] = 6.40, p < .001) and test 3 between high HAB-
high NR and low HAB-low NR (i.e., slopes 1 and 4; t[483] = 6.40, p < .001). The results of these
DISCUSSION
are likely to arise. Potentially ambiguous social situations such as those captured by the social
burden construct may have a negative impact per se, but coupled with a dysfunctional attribution
tendency (e.g., HAB), the outcomes are more detrimental. The level of evidence presented here
may not be strong enough to warrant organizational interventions, but it does shed light on
complexities of the organizational social environment that could lead managers or human
resources professionals to pay attention to what employees are asking of one another.
Four out of the present study’s five hypotheses for the current study were supported.
Hypothesis 1 added to existing evidence that social burden is related to CWB by replicating the
results of the original (Yang et al., 2016) study in a broader sample. I was not, however, able to
demonstrate that the statistical effect of social burden on CWB is greater for CWB toward people
than CWB toward the organization. This may be due to the measure of CWB having nearly
unified dimensions, or it could be for a more behaviorally oriented explanation. This question is
Hypotheses 3-5 tested the significance of two individual difference variables in the
prediction of CWB via social burden, considered both apart and jointly. Testing these hypotheses
apart, which would be standard if only NR or HAB were proposed as significant constructs, as
well as together in a full factorial three-way interaction, allows for two important
demonstrations. First, considering the constructs separately allowed for the inclusion of quadratic
terms without further straining the power of a full regression including a three-way interaction
terms and three quadratic terms. These quadratic terms demonstrate that the significant
explanatory power of the interactions is not due to variance shared with quadratic terms, a worry
SOCIAL BURDEN, HAB, NR INTERACTION 20
for predictors that are relatively highly correlated (Cortina, 1993). Moreover, it illustrates the
importance of testing three-way interactions despite the difficulty of uncovering them. Had
researchers only tested the HAB and NR as interactions separately their effect on the social
burden-CWB relation would have looked highly similar. This similarity is diminished when
The full factorial regression including the three-way interaction demonstrates that the
relation between social burden and CWB was impacted by the joint relation between HAB and
NR (Hypothesis 5). Though the three-way interaction term was significant, a more detailed
examination of the resulting slopes was more informative. First, a visual inspection of the slopes
can be found in Figure 3. The slopes that stand out from the others are those for high HAB-high
NR (slope 1) and high HAB-low NR (slope 2). These slopes appear associated with considerably
more CWB in conditions of high social burden, though not in conditions of low social burden.
This is consistent with the idea that attributions of hostility to ambiguous social situations, or
people therein, are of primary importance to the detrimental outcomes of social burden. A more
rigorous test of these slopes suggests that this conclusion is hastily drawn. That is, the high
HAB-low NR slope, while visually striking, does not differ from the others statistically. This
suggests that while social burden and HAB may be the most important (i.e., impart the greatest
statistical effect) constructs at play here, NR beliefs should also be considered when CWB is an
outcome of interest, particularly when it is likely that social interactions will be interpreted
negatively. It should be noted that the slope difference test used (Dawson & Richter, 2006) may
exhibit reduced power when certain factors (e.g., data distribution or effect size of slopes) are
sub-optimal, as may be the case with the present data. As such, conclusions offered by the more-
SOCIAL BURDEN, HAB, NR INTERACTION 21
rigorous hypothesis testing may not be viewed as absolutely superior to visual inspection of the
A further contribution of this study lies in the analysis of simultaneously measured HAB
instruments. The measurement of HAB in the literature from which it arises (i.e., childhood
development) entails several procedures, most dominantly the use of vignettes (De Castro,
Veerman, Koops, & Monshouwer, 2002). In organizational research, however, there is no such
psychometric properties of the vignette measure were inferior to the WHAS (Bal & O’Brien,
2010). This may be due to the stability of Likert-based measures versus vignettes as a result of an
increased reading demand, or it may be due to a lack of well-established vignettes created for
adult populations. Whatever the reasons, the results suggest that at least the WHAS is viable in
Despite the contributions of the present study, some limitations must be recognized. First,
the study used a cross-sectional design which means that no causal inference nor an establishing
of temporal order is appropriate. Next, the correlations between constructs, as seen in Table 2,
are all positive and nearly all significant. It is possible that an acquiescence response style has
inflated the relations between observed variables (Weijters, Geuens, & Schillewaert, 2010).
In the face of such limitations, there are many interesting questions raised by the results
of this study, and these questions should be met with future research. First, the construct space
and measurement of social burden should be clarified and refined. The items that form the social
burden measure are winnowed from a larger measure of negative social interactions—hence the
SOCIAL BURDEN, HAB, NR INTERACTION 22
name social burden. Future research should expand the items to reflect the construct’s definition,
which includes no valence. That is, items should not come from a pool of negative, or for that
matter positive, interactions. Items should be added to the pool based on their consonance with
the social burden definition. Next, as suggested by Figure A2, emotional responses to ambiguous
social burden events may play an influential role in the negative outcomes of social burden. A
process model of emotional response before and after social burden events using event sampling
could be tested to determine the temporal precedence, increasing the quality of evidence beyond
purely cross-sectional data. Additionally, due to the ambiguity inherent to some social burden
scenarios, future studies should include theoretically supported positive outcomes to social
burden. Finally, due to the conceptual proximity to social support (i.e., social support is likely to
occur following elicitation of support), one might question whether the negative effects of social
burden are mitigated or exacerbated when coworkers acquiesce to elicitations of the support.
That is, does actually providing support after episodes of burden make the outcomes worse,
The findings related to the concept of social burden throw into question whether
increased social interconnectedness is absolutely positive for the workplace and members
thereof. These social environments are complex and relatively poorly understood. This study
found that increases in a certain type of social interaction, a type that is not necessarily
pernicious on its face, was related to increases in deviant behavior which was exacerbated by
employees’ tendencies to appraise social interactions in a negative light. This is but one
demonstration of the complexity of the social environment of the workplace, and the unexpected
REFERENCES
Aguinis, H., Beaty, J. C., Boik, R. J., & Pierce, C. A. (2005). Effect Size and Power in Assessing
Bandiera, O., Barankay, I., & Rasul, I. (2009). Social connections and incentives in the
Barclay, L. J., Whiteside, D. B., & Aquino, K. (2014). To avenge or not to avenge? Exploring
the interactive effects of moral identity and the negative reciprocity norm. Journal of
Bakker, A. B., & Demerouti, E. (2007). The job demands-resources model: State of the
https://doi.org/10.1108/02683940710733115
Bal, A., & O’Brien, K. E. (2010). Validation of the hostile attributional style short form. Paper
presented at the 25th conference for the Society for Industrial and Organizational
Basch, J., & Fisher, C. D. (2000). Affective events–emotions matrix: A classification of work
Emotions in the workplace: Research, theory, and practice (pp. 36–48). Westport, CT:
Quorum.
Baumeister, R. F., & Vohs, K. D. (2007). Self‐Regulation, ego depletion, and motivation. Social
Campbell, J. P., & Wiernik, B. M. (2015). The modeling and assessment of work performance.
http://doi.org/10.1146/annurev-orgpsych-032414-111427
Christian, M. S., & Ellis, A. P. (2011). Examining the effects of sleep deprivation on workplace
934. https://doi.org/10.5465/amj.2010.0179
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ:
Erlbaum.
Cortina, L. M., Magley, V. J., Williams, J. H., & Langhout, R. D. (2001). Incivility in the
Dawson, J. F., & Richter, A. W. (2006). Probing three-way interactions in moderated multiple
De Castro, B. O., Veerman, J. W., Koops, W., Bosch, J. D., & Monshouwer, H. J. (2002).
Dodge, K. A., & Frame, C. L. (1982). Social cognitive biases and deficits in aggressive
Douglas, S. C., Martinko, M. J., Gundlach, M. J., & Douglas, S. C. (2002). Toward an
http://doi.org/10.1111/1468-2389.00192
Eisenberger, R., Lynch, P., Aselage, J., & Rohdieck, S. (2004). Who takes the most revenge?
Garcia, P. R. J. M., Restubog, S. L. D., & Denson, T. F. (2010). The moderating role of prior
Greenberger, E., & Steinberg, L. D. (1981). The workplace as a context for the socialization of
Halbesleben, J. R., Neveu, J. P., Paustian-Underdahl, S. C., & Westman, M. (2014). Getting to
Heider, F. The psychology of interpersonal relations. New York: Wiley, 1958. Holzinger, K. J.
(1941). Review of talents and tasks: Their conjunction in a democracy for wholesome
doi:10.1037/h0050460
Henson, R. K., & Roberts, J. K. (2006). Use of exploratory factor analysis in published research:
Hoobler, J. M., & Brass, D. J. (2006). Abusive supervision and family undermining as displaced
9010.91.5.1125
Marcus, B., Taylor, O. a., Hastings, S. E., Sturm, A., & Weigelt, O. (2016). The structure of
https://doi.org/10.1177%2F0149206313503019
Matthews, R. A., & Ritter, K. J. (2016). A concise, content valid, gender invariant measure of
McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and
http://psycnet.apa.org/doi/10.1037/0033-2909.114.2.376
Mitchell, M. S., & Ambrose, M. L. (2007). Abusive supervision and workplace deviance and the
1159. http://psycnet.apa.org/doi/10.1037/0021-9010.92.4.1159
Oh, H., Chung, M., & Labianca, G. (2004). Group Social Capital and Group Effectiveness: The
Role of Informal Socializing Ties. The Academy of Management Journal, 47(6), 860-875.
Ones, D. S., & Viswesvaran, C. (2003). The big-5 personality and counterproductive behaviours.
Macmillan UK.
Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution
Spector, P. E., Bauer, J. A., & Fox, S. (2010). Measurement artifacts in the assessment of
http://doi.org/10.1037/a0019477
Spector, P. E., & Fox, S. (2003, November). Emotional experience at work: Assessing emotions
with the Job-related Adjective Well-being Scale (JAWS). Paper presented at the meeting
Spector, P. E., & Fox, S. (2005). A model of counterproductive work behavior. In S. Fox, &
Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh, A., & Kessler, S. (2006). The
http://doi.org/10.1016/j.jvb.2005.10.005
Spector, P. E., & Zhou, Z. E. (2014). The moderating role of gender in relationships of stressors
Steiger, J.H. (1980), Tests for comparing elements of a correlation matrix. Psychological
Stryker, S., & Statham, A. (1985). Symbolic interactionism and role theory. In G. Lindzey and E.
Aronson (Eds.), The handbook of social psychology (pp. 311–378). New York: Random
House.
Stryker, S., & Vryan, K. D. (2006). The symbolic interactionist frame. In Handbook of social
SOCIAL BURDEN, HAB, NR INTERACTION 28
Thomas, K. A., & Clifford, S. (2017). Validity and mechanical turk: An assessment of exclusion
Tremblay, P. F., & Belchevski, M. (2004). Did the instigator intend to provoke? A key
moderator in the relation between trait aggression and aggressive behavior. Aggressive
Yang, L. Q., Liu, C., Nauta, M. M., Caughlin, D. E., & Spector, P. E. (2016). Be mindful of what
you impose on your colleagues: Implications of social burden for burdenees' well‐being,
attitudes and counterproductive work behaviour. Stress and Health, 32(1), 70-83.
http://doi.org/10.1002/smi.2581
Weijters, B., Geuens, M., & Schillewaert, N. (2010). The individual consistency of acquiescence
Wu, L. Z., Zhang, H., Chiu, R. K., Kwan, H. K., & He, X. (2014). Hostile attribution bias and
APPENDIX A: TABLES
Table 1
Table 2
Note. N = 490. WHAS = Workplace Hostile Attributions Scale; Neg Rec = Negative Reciprocity; CWB =
counterproductive work behavior; CWB-I = CWB toward individuals; CWB-O = CWB toward the
organization; V-HAB = HAB vignettes measure. Cronbach’s alpha on diagonal.
* p < .05. ** p < .01. *** p < .001.
SOCIAL BURDEN, HAB, NR INTERACTION 31
Table 3
Item-total Correlations, Pattern Matrix via combined Factor Analysis for HAB
measures
Items ITC F1 F2
Table 4
Hierarchical Moderated Regression Predicting CWB from Social Burden, HAB, and the
Interaction of Social Burden with HAB
Predicting CWB
Model Predictors B SE t F
Step 1
(intercept) 1.35 0.02 64.1** 102.8** R2
SB 0.14 0.02 6.49** 0.30
HAB 0.21 0.02 9.38**
Step 2
(intercept) 1.27 0.03 40.84** 55.0** R2
SB 0.13 0.02 5.33** 0.31
HAB 0.19 0.02 7.76**
HAB2 0.04 0.02 2.50* (ΔR2)
SB2 0.03 0.02 1.84 0.01**
Step 3
(intercept) 1.31 0.03 43.3** 57.8** R2
SB 0.14 0.02 5.87** 0.37
HAB 0.20 0.02 8.24**
SB2 0.00 0.02 -1.01 (ΔR2)
HAB2 -0.02 0.02 0.25 0.06**
SB*HAB 0.16 0.02 6.92**
Note. N = 491., steps df 1: (2, 488) 2: (2, 466), 3: (2, 485). SB = Social burden; NR = Hostile
attribution bias.
* = p < .05; ** = p < .01
SOCIAL BURDEN, HAB, NR INTERACTION 33
Table 5
Hierarchical Moderated Regression Predicting CWB from Social Burden, NR, and the
Interaction of Social Burden with NR
Predicting CWB
Model Predictors B SE t F
Step 1
(intercept) 1.35 0.02 62.3** 82.65** R2
SB 0.20 0.02 6.49** 0.25
NR 0.16 0.02 9.38**
Step 2
(intercept) 1.31 0.02 39.54** 42.81** R2
SB 0.18 0.02 7.29** 0.26
NR 0.17 0.02 6.89**
NR2 0.00 0.02 0.20 (ΔR2)
SB2 0.04 0.02 2.17* 0.01
Step 3
(intercept) 1.31 0.03 45.28** 45.28** R2
SB 0.19 0.02 7.95** 0.32
NR 0.16 0.02 7.10**
NR2 -0.00 0.02 -0.07 (ΔR2)
SB2 0.03 0.02 1.46 0.06**
SB*NR 0.13 0.02 6.41**
Note. N = 491. Steps df 1: (2, 488) 2: (2, 486), 3: (2, 485). SB = Social burden; NR =
negative reciprocity.
* = p < .05; ** = p < .01
SOCIAL BURDEN, HAB, NR INTERACTION 34
Table 6
Hierarchical Moderated Regression Predicting CWB from Social Burden, HAB, NR, all Two-
way Interactions, and a Three-way Interaction of all Predictors
Predicting CWB
Model Predictors B SE t F
Step 1
(intercept) 1.35 0.02 65.4** 77.93** R2
SB 0.15 0.02 6.52** 0.32
HAB 0.17 0.02 7.17**
NR 0.10 0.02 4.49
Step 2
(intercept) 1.28 0.02 58.83** 55.68** R2
SB 0.13 0.02 6.24** 0.41
HAB 0.15 0.02 6.37**
NR 0.09 0.02 4.36** (ΔR2)
HAB*SB 0.11 0.02 5.16** 0.08**
NR*SB 0.04 0.02 1.87
HAB*NR 0.05 0.02 2.16*
Step 3
(intercept) 1.28 0.02 59.41** 45.28** R2
SB 0.10 0.02 4.53** 0.42
HAB 0.15 0.02 6.35**
NR -0.00 0.02 2.71** (ΔR2)
HAB*SB 0.10 0.02 4.77** 0.02**
NR*SB 0.04 0.02 1.56
HAB*NR 0.04 0.02 1.96*
HAB*NR*SB 0.07 0.02 3.71**
Note. N = 491, steps df 1: (3, 487) 2: (3, 484), 3: (1, 483). SB = Social burden; HAB = hostile
attribution bias; NR = negative reciprocity.
* = p < .05; ** = p < .01
SOCIAL BURDEN, HAB, NR INTERACTION 35
APPENDIX B. FIGURES
2.4
2.2
(1) H HAB, H NR
2
Predicted CWB
(2) H HAB, L NR
1.8
(3) L HAB, H NR
1.6
(4) L HAB, L NR
1.4
1.2
1
Low Social Burden High Social Burden
Figure 4. Interaction of Social Burden, HAB, and NR in the prediction CWB, as Visualized for
Slope Difference Test.
L = - 1 SD, H = +1 SD. CWB = counterproductive work behavior; SB = social burden; HAB =
hostile attribution bias; NR = negative reciprocity.
SOCIAL BURDEN, HAB, NR INTERACTION 39
Perceived
Control
Individual
Appraisal
Differences
Figure A1. General Stressor-emotion model of CWB, adapted from Spector & Fox (2005).
Perceived as Negative
Social Burden CWB
stressor Emotion
Hostile Negative
NR*HAB
Attribution Bias Reciprocity
Interaction
beliefs
Figure A2. Stressor-emotion model of CWB Spector & Fox (2005) adapted to incorporate and
highlight (using bold font) constructs specifically included in hypotheses tested in the present
study.
SOCIAL BURDEN, HAB, NR INTERACTION 40
APPENDIX D. MEASURES
Social Burden
Our colleagues (co-workers or supervisors) engage in many behaviors in the workplace. Please
indicate how often your colleagues have engaged in each behavior in the previous 30 days or so.
How often have you done each of the following things on your present job over the past month?
4. Stayed home from work and said you were sick when you weren’t
27. Said something obscene to someone at work to make them feel bad
Negative Reciprocity
Please indicate to what extent you agree or disagree with the following statements.
3. If someone says something nasty to you, you should say something nasty back.
4. If a person wants to be your enemy, you should treat them like an enemy.
6. If someone treats you badly, you should treat that person badly in return.
7. You should not give help to those who treat you badly.
Indicate how much you agree or disagree with these workplace statements.
3. If coworkers work slowly on a task I assigned them, it is because they don’t like me.
First you will be asked to rate how sure you are that the actions of the other person in each
scenario were intentional acts (as opposed to accidental). You will rate this on a scale of 1 to 6
where 1 = I am not at all sure the action was intentional, and 6 = I am very sure the action
was intentional.
Then you will rate how sure you are that these actions were hostile (intended to bring some harm
or discomfort to you, as opposed to harmless). Again, you will rate this on a scale from 1 to 6
where 1 = I am not at all sure the action was hostile, and 6 = I am very sure the action
was hostile.
SOCIAL BURDEN, HAB, NR INTERACTION 43
1. You are carrying a heavy load of groceries up to a check-out line at the grocery store
and just as you are about to enter in line, someone cuts in front of you. You end up
2. You and your friends have been waiting in line for over half an hour to get into a bar.
You are to be the next ones to get in but two guys who appear to be very intoxicated
3. You are having a discussion with a couple of co-workers. You disagree with them
and express your opinion. One of the male co-workers tells you that you are not
4. You are at a large party and you are introduced to several people you don’t know.
One guy starts talking to you and tells you something that you find insulting.
5. You are at a party with your boyfriend/girlfriend/partner and you go get some drinks
for both of you. When you return you see a woman who appears to be somewhat
intoxicated place her arm around your date. Your date looks quite upset and turns
away from the woman but the woman does not leave.
6. You go to your work and you are in a bad mood. As you walk in the office, a
7. You are at a local dance club. While you are dancing, a male stranger bumps into you
very roughly.
SOCIAL BURDEN, HAB, NR INTERACTION 44
Thank you for your submission of Revision materials for this project. This project is still exempt from IRB
review according to federal regulations AND that the proposed research has met the principles outlined
in the Belmont Report. You may now begin the research activities.
If you have any questions, please contact the Office of Research Compliance at 419-372-7716 or
orc@bgsu.edu. Please include your project title and reference number in all correspondence with this
committee.
This letter has been electronically signed in accordance with all applicable regulations, and a copy is retained within Bowling Green State
University Institutional Review Board's records.