It's Only Abusive If I Care
The Effect of Organizational Concern on Abusive Supervision, Stress, and Work Behaviors
Abstract
Abstract. Drawing on conservation of resources (COR) theory, we propose that abusive supervision increases stress responses in targets, which, in turn, diminishes their ability to perform extra- and in-role work behaviors. However, based on COR theory, we argue that followers who are driven by low rather than high organizational concern motives place less value on their work and the social context in which technical activities occur. As such, feeling low organizational concern should make people less susceptible to abusive supervision rather than more so. Thus, organizational concern was proposed to moderate the abuse–stress relationship. Across two multisource studies, we found support for most of our hypotheses. Abusive supervision negatively affected organizational citizenship behaviors via increased stress, and low organizational concern was found to attenuate the detrimental effects of abusive supervision. Implications for leadership literature and future research are discussed.
For nearly two decades, the topic of abusive supervision has received a great deal of research attention. It is defined as the supervisors' sustained display of hostile verbal and nonverbal behaviors, excluding physical contact (Tepper, 2000). Abusive supervision has been associated with a myriad of negative outcomes (Mackey et al., 2017; Martinko et al., 2013; Tepper, 2007). To date, the emphasis within the abusive supervision literature has been to explain how and when these acts of abuse affect victims' work behaviors (for a recent review, Tepper, Simon, & Park, 2017). An emerging line of research has started to look at psychological processes within the targets of mistreatment, in particular how stress responses serve as a mechanism that mediates the effect of an abusive supervisor on victim outcomes (Tepper, 2007; Zhang et al., 2019). Viewed through a conservation of resource (COR) theory lens (Hobfoll, 1989, 2001; Zhang et al., 2019), ongoing abuse by an authority is a salient event that should deplete rather than enhance one's personal resources and thus be experienced by the employee as stress reactions (Halbesleben et al., 2014). In turn, these accumulated stress reactions deplete employees' resources required to cope with workplace demands and potentially jeopardize the employment relationship with an increased risk of detrimental physiological, psychological, and/or behavioral outcomes (Halbesleben et al., 2014).
While a resource perspective on abusive supervision offers fresh insights into victim experience, an unanswered question is whether the resource-depletion process depends on factors that reside within the targets of abuse (Tepper, 2007; Tepper et al., 2017; Zhang et al., 2019). Preliminary evidence suggests that employees who are able to maintain self-control (Yuan et al., 2018) and enact emotion-control strategies (Chi & Liang, 2013) have been found to be cope better with episodes of abuse without experiencing the increased levels of stress. Beyond self-regulation, in this article, we examine whether an individual's orientation toward the organization exposes an individual to the resource-depletion effects of an abusive supervisor. A particularly relevant concept is organizational concern (Rioux & Penner, 2001) that can be defined as a collective-oriented motive to engage in behaviors that enrich and fortify the social-organizational environment in which technical work activities occur. Individuals with high organizational concern exhibit citizenship behaviors because they care what happens to the organization because they feel pride in and commitment to the organization (Rioux & Penner, 2001).
Aim of the Study and Contribution to the Literature
Drawing on COR theory, we propose that targeted abuse might strike hardest at those employees who genuinely care about and enact their concern for the organization – and have the least effect on those who are psychologically detached or distant from the organization and its agents. From the perspective of COR theory, individuals with high organizational concern should strive to retain and protect their valued attachment to the organization. The latter, however, is threatened to be lost when individuals with high organizational concern are exposed to prolonged abuse by a supervisor. In turn, depletion of personal resources should increase stress responses. In contrast, those who are less concerned about the organization should be better equipped to cope with an abusive supervisor as they are psychologically distanced from the abuser. It is this psychological distance that allows targets to shield themselves from the internationalization of the mistreatment. This, in turn, reduces the need to deplete their coping resources than those who are psychologically close to this event and take the abuse to heart. Thus, we expect that when organizational concern is strong rather than weak, we will see a stronger negative relationship between acts of abuse and stress responses. Extending our research model to a model of moderated mediation, we also investigate the downstream consequences of abusive supervision-induced stress responses on employees' tendencies toward resource conservation and investment at work. Specifically, we examine extra- and in-role work behavior as outcomes in our model. This allows us to capture employees' potential resource conservation efforts in terms of pro-organizational acts that fall outside (i.e., decreased extra-role work behavior) and inside (i.e., decreased in-role work behavior) the formal job description and reward system.
We contribute to the literature in different ways. First, we add to the emerging stream of research on boundary conditions of abusive supervision's effects. Several personal resources have been proposed to mitigate abusive supervision's effects (Chi & Liang, 2013; Yuan et al., 2018). We extend this line of research by arguing that organizational concern (Rioux & Penner, 2001) may play an important role as well. Second, we suggest organizational concern can be a personal liability in the sense that these mindsets make employees particularly vulnerable should the organization (or its agents) turn on them. Third, we aim to show that the proposed interaction has implications that go beyond extra-role behaviors (Zhang et al., 2019). In this article, we expand the criterion landscape by considering the effects of resource losses on in-role behavior. Our paper begins with a description of the theories underlying our working model (Figure 1) and our study hypotheses. The results of two empirical studies are then described.
Theoretical Context
The core premise of COR theory is that humans are driven by two fundamental orientations: to acquire, retain, and build resources and to prevent the loss of these valued resources (Hobfoll, 2001). Since its inception, COR theory has viewed stress as a reaction to the environment in which there is (a) the (perceived or actual) threat of a net loss of resources, (b) the net loss of (perceived or actual) resources, or (c) a lack of (perceived or actual) resource gain following the investment of resources (Hobfoll, 1989). According to the theory, stress reactions evoke within individuals an orientation toward activities aimed at minimizing net resource loss. However, when not confronted with threats of resource loss, people will be oriented toward the acquisition or development of resource surpluses to hedge against the possibility of future losses. So, rather than serving a tangential role in COR, stress reactions can be viewed as a way of capturing the distress experienced by real or perceived resource losses. The theory suggests that individuals may use defensive strategies to protect the few resources they have left, such as emotional or physical withdrawal from the resource-depleting (stressful) situation (Hobfoll, 2001). Thus, stressed individuals become more introspective and inwardly focused and less sensitive to the needs of others (Hobfoll, 2001).
Abusive Supervision: Viewed as a Source of Resource Depletion
COR theory has proved very useful to understand and explain how targets of abusive supervision tend to react to such unpleasant experiences. The question at hand concerns the nature of perceived or actual resource losses that are associated with an abusive supervisor. When individuals lose resources at work, they are more likely to experience strain in the form of well-being outcomes (e.g., anxiety, depression, self-esteem, and burnout; Halbesleben et al., 2014). However, as Hobfoll (1989) suggests, resource losses appear in loss spirals. Thus, beyond well-being outcomes, Halbesleben et al. (2014) imply that abusive supervision can erode the material and social environment in which work occurs. Specifically, abusive supervision should be associated with reduced communication, inadequate (or absent) performance feedback, an erosion of one's personal reputation within the team, work unit or company, lower likelihood of advancement, reduced ability to access to organizational rewards (or ability to avoid punishments), and social undermining if the abuse is enacted public in front of coworkers/peers (cf. Mackey et al., 2017; Tepper, 2000; Tepper et al., 2017). We make similar assumptions about the way in which an abusive supervisor can trigger perceived or actual resource losses. Thus, given the broad resource implications for individuals exposed to abuse, we propose stress responses as the most efficient way of capturing (and summarizing) the cumulative effect of these resource losses on the individual.
According to COR theory, stress reactions impair work-related behavior as individuals strive to retain and rebuild lost resources (Hobfoll, 1989). We assert that one way of conserving resources without incurring further losses would be for an individual to reduce the level of attention and effort resources allocated to discretionary (citizenship) behaviors – prosocial/pro-organizational acts that typically fall outside the formal job description and reward system (Organ, 1988; Podsakoff et al., 2009). The fact that individual or organization-directed organizational citizenship behaviors (OCBs) are not mandatory makes this class of work behaviors particularly sensitive as a criterion for COR predictions. It is to be expected that the stress response caused by an abusive supervisor leads to a redirection of remaining resources. Due to low levels of resources, individuals should be reluctant to invest resources into others. COR suggests that stressed individuals become more introspective and shift their perspective from the needs of others toward a more individualistic self-focus (Hobfoll, 1989, 2001). Accordingly, in their efforts to protect their already depleted resources, stressed individuals are likely to care primarily about themselves and disengage from behaviors that are targeted at helping and cooperating with others (Driskell et al., 1999). Zhang et al. (2019) termed this indirect association between abusive supervision and OCB via increased stress “the resource perspective” (p. 2478) of abusive supervision and provided meta-analytical support for the significance of this relation.
Notwithstanding the negative implications of abusive supervision for discretionary behaviors, COR theory suggests that followers' loss-control strategies might also extend to formally required work activities. Accordingly, a narrower and more individualistic self-focus is likely to also affect how followers perform their formally assigned (in-role) work duties and the extent to which they achieve performance goals. When individuals perceive high stress, their psychological resources available for work-related tasks are likely to be reduced. While past research indicates that stress may increase performance in some situations, the vast majority of research suggests that chronic, high stress impedes performance as cognitive functioning is impaired (Byron et al., 2010; Cropanzano et al., 2003; Driskell et al., 1999).
Drawing on the reasoning developed above, we not only aim at replicating the resource perspective regarding the prediction of OCB but also expand this lens toward a broader and more comprehensive perspective on work behavior by including in-role behaviors of followers. Thus, we posit the following hypothesis:
Stress mediates the negative relationship between abusive supervision and extra-role behavior (Hypothesis 1a) and in-role behavior (Hypothesis 1b).
The Moderating Role of Organizational Concern
Human behavior serves specific functions (Snyder, 1993). Put differently, individuals engage in behaviors to satisfy certain needs and motives (Finkelstein, 2006). Rioux and Penner (2001) adapted this functional perspective of behavior and introduced organizational concern as a specific motive for engaging in extra-role behaviors. As alluded to earlier, Rioux and Penner (2001) described organizational concern as a collective-orientated motive. The concept encompasses an individual's emotional identification with an organization's values and goals, the emphasis on maximizing organizational interests, and the willingness to subordinate individual self-interest to such collective interests (see also Halbesleben et al., 2010; Lu et al., 2019). Rather than merely reflecting the strength of an emotional bond with the organization (e.g., organizational commitment) or the perceived fit between the requirements of a work role and beliefs, values, and behaviors (e.g., meaning of work; Spreitzer, 1995), organizational concern captures the reasons for employees wanting to engage in helpful and supportive behavior (because an employee cares what happens to the company, the employee feels pride in the organization, or the employee has genuine interest in their work; Rioux & Penner, 2001).
Extending Rioux and Penner's (2001) work, we argue that especially low organizational concern functions as a buffer to the detrimental effect of abusive supervision and, thus, may help organizations. The logic behind this counterintuitive argumentation is again very consistent with COR theory. The key tenet of COR is that at work individuals try to protect resources that they value (Hobfoll, 1989, 2001). Consistent with this perspective, we argue that the strength of the relationship between abusive supervision and stress is dependent on the exact value work and the organization has for the individual. When employees are very concerned about the organization, they are likely and willing to dedicate large amounts of personal resources toward work. This is manifested not only in extra-role behaviors (Rioux & Penner, 2001) but also in increased knowledge sharing (Lu et al., 2019), organizational commitment, and performance (Halbesleben et al., 2010). Said differently, when employees are driven by a high organizational concern motive, they put high personal value to their work and the organization as a whole. This qualifies high organizational concern as a distinct resource in terms of COR (Hobfoll, 2001). Furthermore, high organizational concern also implies that an individual views the organization as being committed to their own welfare (Halbesleben et al., 2014; Organ, 1990; Rioux & Penner, 2001). We expect that such individuals are particularly vulnerable to abusive supervision because the latter presents an acute threat to the personal resource of high organizational concern. Being exposed to supervisor abuse is a very strong contrast to their own strong dedication. As supervisors often function as the face of the organization to followers (Harms et al., 2017), the latter are confronted with a strong imbalance between what they give to the organization and what they receive in return. In such a setting, feelings of uncertainty and loss of control are likely to occur. As research suggests that uncertainty and uncontrollability lead to follower stress (Ito & Brotheridge, 2003; Matta et al., 2017), we expect the relationship between abusive supervision and stress to be strengthened when followers are highly concerned for the organization.
In contrast, when individuals are less motivated by organizational concern, they are unlikely to have a collective self-concept that is overly focused on maximizing organizational welfare (Takeuchi et al., 2015). We argue that the less the employees are focused on the organization, the less value they should place on what happens in the organizational context. Consequently, it should be unlikely that they experience resource loss when faced with an abusive leader. When the abusive behavior represents less of a threat to followers' personal self-concept, it will be easier for followers to retain and replenish their emotional resources. In this way, low organizational concern suggests distance as a form of protection (Krischer et al., 2010). Previous work has already shown that disidentification as a related form of low organizational concern may break the detrimental cycle starting with abusive supervision (Taylor et al., 2019). Furthermore, Harris et al. (2007) showed that, compared to their counterparts, individuals who derive little meaning from work had less performance decreases in response to abusive supervision. While the authors positioned their study in a COR framework, they did not posit or capture stress as the key process driving abusive supervision's downstream consequences. This, however, would be important as stress responses to resource loss are the key tenet of COR theory. Therefore, we extend this line of reasoning by suggesting that low organizational concern helps to protect an individual and, in doing so, mitigates the effect of an abusive supervisor on follower stress (Krischer et al., 2010). This may go as far as absorbing all of the damage caused by destructive treatment (Krischer et al., 2010) leaving followers' stress level steady. In sum, we expect the positive relation between abusive supervision and stress to increase as a function of organizational concern and posit the following hypotheses:
Organizational concern moderates the relationship between abusive supervision and stress. The positive relation is strengthened as a function of organizational concern.
The negative indirect effect of abusive supervision on extra-role behavior (Hypothesis 3a) and in-role behavior (Hypothesis 3b) via increasing stress is moderated by organizational concern, such that the relationship between abusive supervision and stress is stronger when organizational concern is high, strengthening the indirect effects.
Empirical Validation
We tested our predictions in two empirical studies. In Study 1, we focused exclusively on extra-role behavior as an outcome and analyzed data from 95 employee–coworker dyads. We conducted Study 2 to replicate findings. To also test whether the interaction pattern regarding abusive supervision and organizational concern is generalizable across measures, we used a different operationalization of stress. Testing the full model, we considered extra-role and in-role behaviors as outcomes with the former being measured broader than in Study 1. We analyzed three-source, multilevel data from 78 teams and 232 employees.
Study 1
Method
Sample and Procedure
Data for Study 1 were collected in Germany. In partial fulfillment of course requirements, students working on a research project recruited employees from different organizations to participate in our study. We collected data from employee–coworker dyads to alleviate bias from same source data (Podsakoff et al., 2012). Focal employees rated abusive supervision, organizational concern, and stress, whereas coworkers rated focal individuals' extra-role behavior. After respondents had agreed to participate, we distributed survey packages that contained coded questionnaires for both employees and their coworkers. In total, we received 96 completed employee–coworker matched surveys. All employees (52% female) worked full-time, were on average 34.28 years of age, and had an average organizational tenure of 5.69 years.
Measures
Abusive supervision was measured with 10 items from Tepper's (2000) scale. Sample items included “My leader gives me the silent treatment” and “My leader puts me down in front of others.” We applied a 5-point Likert-type scale from 1 (strongly disagree) to 5 (strongly agree). The coefficient α for this scale was .77.
Organizational concern was assessed using 10 items from Rioux and Penner (2001). Following their recommendations, we presented several examples of OCBs in the stem of the scale and informed respondents that our aim was to learn more about the reasons employees engage in such behaviors. The introduction to the items then read: “I show such behaviors because….” Examples of the items include “I care what happens to the company” and “The organization values my work.” Responses were provided on 7-point Likert-type scales from 1 (strongly disagree) to 7 (strongly agree). The coefficient α for this scale was .90.
Stress was operationalized as burnout using the Copenhagen Burnout Inventory (Kristensen et al., 2005). Consistent with COR theory's notion of stress, burnout is conceptualized as an aversive reaction to environmental stressors (Maslach, 1982). It has been shown to be an outcome of threats of loss of resources (e.g., uncertainty and social identity threats; Garrett & McDaniel, 2001; Hall et al., 2015) as well as of actual resource loss (e.g., lack of social support; Halbesleben, 2006). It should be noted that depending on the definition of stress, stress and burnout can be seen as different constructs. For example, studies have looked at the different antecedents and outcomes of burnout and stress (Pines & Keinan, 2005). Yet, in our study, we define stress in terms of COR theory that allows for a rather broad conceptual scope and does not preclude burnout. Indeed, with its conceptualization as an aversive reaction to environmental threats of resource losses, we view burnout to be within COR theory's understanding of stress. Consistent with this, Halbesleben (2006) noted that COR is the leading theory for understanding the processes leading to burnout. The scale encompasses seven items such as “How often do you feel tired?” or “How often are you emotionally exhausted?”. The response scale ranged from 1 (never) to 5 (very often). The coefficient α for this scale was .87.
Extra-role behavior was operationalized as the employee's helping behavior rated by the coworker. Helping behavior was measured using five items from Staufenbiel and Hartz's (2000) OCB questionnaire, which is consistent with Organ's (1997) conceptualization. The subscale for helping includes items such as “This employee helps coworkers when they are overburdened.” We applied a 5-point Likert-type scale from 1 (strongly disagree) to 5 (strongly agree). The coefficient α for this scale was .81.
Construct Validity of the Proposed Measurement Model
We conducted a set of confirmatory factor analyses (CFA) to establish discriminant validity for the study variables. Before doing so, we created three parcels each for every study variable to maintain a favorable indicator-to-sample-size ratio (Bandalos, 2002). This resulted in a combined 12 parcels for four variables. We built the parcels following the item-to-construct balance approach suggested by Little et al. (2002). Our target model (χ2 = 77.54, p < .01, df = 48) included the four factors of abusive supervision, organizational concern, stress, and extra-role behavior. RMSEA = .08, standardized root mean square residual (SRMR) = .06, and the comparative fit index (CFI) = .94 all indicated good to acceptable model fit (Hu & Bentler, 1999). Our target model outperformed competing models. A model in which all indicators loaded on a single factor showed inferior fit (Δχ2 = 327.89, p < .01, Δdf = 6; RMSEA = .26; SRMR = .19; CFI = .33), as did another model in which indicators of organizational concern and extra-role behavior were collapsed into a single OCB factor (Δχ2 = 107.73, p < .01, Δdf = 3; RMSEA = .17; SRMR = .13; CFI = .74).
Results
Table 1 lists the descriptive statistics and correlations for all variables of Study 1.
We estimated our model of moderated mediation using path-modeling techniques in Mplus (Muthén & Muthén, 2017). To examine the interaction effect of abusive supervision and organizational concern, both variables were standardized (Cohen et al., 2003). To test for mediation and moderated mediation, we inspected confidence intervals of (conditional) indirect effects (Hayes, 2013). We estimated all indirect effects using a parametric bootstrap (Bauer et al., 2006) that employs a Monte Carlo simulation with 20,000 replications to estimate a bias-corrected confidence interval for the indirect effect. The results are shown in Table 2.
In Hypothesis 1a, we expected stress to mediate the negative relationship between abusive supervision and extra-role behavior. As shown in Table 2, abusive supervision was positively related to stress (γ = .20, p < .01), which again was negatively related to extra-role behavior (γ = −.26, p < .05). As the indirect effect was negative and significant (estimate = −0.05, p < .05, 95% CI [–0.116, −0.004]), we found support for Hypothesis 1a.
In Hypothesis 2, we posited the relationship between abusive supervision and stress to vary as a function of organizational concern. Model estimations revealed a significant interaction effect of abusive supervision and organizational concern on stress (γ = .17, p < .05). To facilitate interpretation of this effect, we plotted the interaction (Figure 2) and conducted simple slope analyses at different values of organizational concern (−1 SD, +1 SD). In line with Hypothesis 2, abusive supervision was positively related to stress when organizational concern was high (γ = .35, p < .01). However, when organizational concern was low, the slope was nonsignificant (γ = .03, ns).
In Hypothesis 3a, regarding the full moderated mediation, it was expected that organizational concern also strengthens the indirect effect of abusive supervision on extra-role behavior via stress. The indirect effect of abusive supervision on helping behavior through burnout was significant with high organizational concern (estimate = −0.09, ns, 95% CI [–0.200, −0.010]). The indirect effect diminished with low organizational concern (estimate = −0.01, ns, 95% CI [–0.059, 0.040]). Accordingly, we found full support for Hypothesis 3a.1
Study 2
Method
Sample and Procedure
As in Study 1, data were collected in Germany and students recruited study participants from different organizations as partial fulfillment of course requirements. For Study 2, we collected multilevel, three-source data from employee–coworker–supervisor triads. Focal employees rated abusive supervision, organizational concern, and stress; coworkers rated focal employees' extra-role behavior, whereas supervisors rated focal employees' in-role behavior. After respondents had agreed to participate, we distributed survey packages that contained coded questionnaires for employees, coworkers, and supervisors. In total, we received 232 completed employee–coworker–supervisor matched surveys, nested within 78 teams. All employees (52% female) worked full-time, were on average 37.81 years of age, and had an average organizational tenure of 6.14 years.
Measures
Abusive supervision was measured with the same 10 items from Tepper (2000) as in Study 1. The coefficient α was .82.
Organizational concern was measured with the same 10 items from Rioux and Penner (2001) as in Study 1. The coefficient α was .91.
Stress was assessed with the eight item irritation scale by Mohr et al. (2006). Irritation captures a state of mental impairment resulting from a perceived discrepancy between a given situation and an important personal goal (Mohr et al., 2006). Consistent with our COR perspective, the experience of abusive supervision collides with individuals' striving toward protecting, advancing, and regaining personal resources. This then positions irritation as an adequate measure of stress in terms of COR. There is empirical evidence applying COR to predict effects on and of irritation (Bormann & Diebig, 2020; Van Doorn & Hülsheger, 2015). Sample items were “I have difficulty relaxing after work,” “Even at home I often think of my problems at work,” and “I get irritated easily, although I don't want this to happen.” We applied a 6-point Likert-type scale from 1 (strongly disagree) to 6 (strongly agree). The coefficient α was .84.
Extra-role behavior was measured using the full 20-item OCB questionnaire from Staufenbiel and Hartz (2000). Again, ratings came from coworkers. We applied a 5-point Likert-type scale from 1 (strongly disagree) to 5 (strongly agree). The coefficient α for this scale was .90.
In-role behavior was assessed using Williams and Anderson's (1991) six-item job performance measure. Sample items were “This employee adequately completes assigned duties” and “This employee fulfills responsibilities specified in job descriptions.” We applied a 6-point Likert-type scale from 1 (strongly disagree) to 6 (strongly agree). The coefficient α for this scale was .87.
Construct Validity of the Proposed Measurement Model
To test the discriminant validity of the study measures, we estimated multilevel CFAs to account for the nested structure of our data. However, we faced convergence issues. Thus, we performed CFAs where we did not encounter any convergence problems. As we did in Study 1, we created three parcels for each variable (i.e., a combined 15 parcels for five study variables) to maintain a favorable indicator-to-sample-size ratio (Little et al., 2002). The target model (χ2 = 116.34, df = 80) consisted of five factors (abusive supervision, organizational concern, stress, extra-, and in-role behavior). RMSEA = .04, SRMR = .04, and CFI = .98 all indicated acceptable model fit. An alternative model in which all indicators loaded on a single factor showed inferior model fit (Δχ2 = 1,234.78, p < .01, Δdf = 10; RMSEA = .25; SRMR = .16; CFI = .40), as did other models in which extra- and in-role behavior (Δχ2 = 292.09, p < .01, Δdf = 4; RMSEA = .13; SRMR = .11; CFI = .85) and extra- and in-role behavior and organizational concern (Δχ2 = 661.32, p < .01, Δdf = 7; RMSEA = .19; SRMR = .12; CFI = .67) were collapsed into one factor, respectively.
Results
Table 3 shows the descriptive statistics and correlations for all variables of Study 2.
Given the nested structure of the data, we estimated our model using TYPE = TWOLEVEL in Mplus. We followed Preacher et al. (2011) in the setup of our model with all variables measured at the individual level. ICC (1) scores for the study variables ranged between 0.10 and 0.45, which indicates that there remains substantial variance in each construct at the within (i.e., individual) level. To facilitate interpretation of the interaction effect of abusive supervision and organizational concern on stress, simple slope analyses were conducted as suggested by Preacher et al. (2006). Testing the full hypothesized model of moderated mediation, we estimated the indirect effects of abusive supervision on extra- and in-role behavior via stress at different values of organizational concern (−1 SD, +1 SD). All indirect effects were estimated using a parametric bootstrap (Bauer et al., 2006) that employs a Monte Carlo simulation with 20,000 replications. Abusive supervision and organizational concern were centered around the respective grand mean.
Table 4 shows all relevant estimates. Regarding Hypothesis 1, abusive supervision was positively related to stress (γ = .37, p < .01). Stress was negatively related to extra-role behavior (γ = −.08, p < .05) but unrelated to in-role behavior (γ = −.09, ns). As the indirect effect of abusive supervision on extra-role behavior via stress was significant and negative (estimate = −0.03, p < .05, 95% CI [−0.069, −0.000]), we found support for Hypothesis 1a. In contrast, as the indirect effect of abusive supervision on in-role behavior via stress was nonsignificant (estimate = −0.03, ns, 95% CI [−0.078, 0.004]), Hypothesis 1b had to be rejected.
Multilevel results also indicated a significant interaction effect of abusive supervision and organizational concern on stress (γ = .15, p < .05). In support of Hypothesis 2, the plot of this interaction (Figure 3) and the simple slope analysis revealed that when organizational concern was high, abusive supervision increased stress (γ = .53, p < .01); when organizational concern was low, however, this relationship disappeared (γ = .22, ns).
Organizational concern also affected the indirect relationship between abusive supervision and extra-role behavior through stress. When organizational concern was high, the indirect effect was negative and significant (estimate = −0.04, p < .05, 95% CI [−0.095, −0.001]). When organizational concern was low, the indirect effect was nonsignificant (estimate = −0.02, ns, 95% CI [−0.048, 0.002]). These results provide support of Hypothesis 3a. Regarding in-role behavior and Hypothesis 3b, the indirect effect was nonsignificant regardless of whether organizational concern was high (estimate = −0.05, ns, 95% CI [−0.105, 0.007]) or low (estimate = −0.02, ns, 95% CI [−0.053, 0.006]). Thus, Hypothesis 3b had to be rejected.2
Discussion
In this article, we respond to calls for more work that sheds light on the conditions that ameliorate or exacerbate the psychological and behavioral effects of abusive leadership (Tepper, 2007; Zhang et al., 2019). In two studies and drawing on COR theory, we were able to replicate the negative indirect relationship between abusive supervision and extra-role work behavior via increased stress responses. The indirect effect on in-role behavior where activities are often externally regulated was not significant. The logic underlying the mediation models was contingent upon the felt concern for the organization. Subordinates who were exposed to mistreatment but cared strongly about the organization (and its agents) were more likely to experience stress (i.e., resource depletion) than were those who experienced abuse and felt low levels of organizational concern.
Theoretical Implications
Our study has several important implications for several organizational-behavioral theories. First, we show that employees who care about (i.e., emotionally connected to) the organization put themselves at risk. It would appear that these individuals are particularly vulnerable to and affected by mistreatment from the organization – or its agents (e.g., leaders). Previous studies have shown the efficacy of self-control strategies (Yuan et al., 2018) and emotion-regulation/cognitive reappraisal (Chi & Liang, 2013) in handling experienced abuse. We contribute to this literature by showing how psychological and emotional detachment from the organization may have a protective function. In both our studies, the relationship between abusive supervision and follower stress disappeared entirely when organizational concern was low. Moreover, the fact we observed this effect in two separate samples with different measures of stress further supports the robustness and generalizability of our argumentation.
Second, our study sheds light on the darker side of organizational concern. So far, research has focused almost entirely on its beneficial side. For example, it has been found to be a positive predictor of OCB (Kim et al., 2013; Rioux & Penner, 2001) and knowledge sharing (Lu et al., 2019). Still, our analyses also indicate that there appears to be a second side of the coin with situations where individuals who care little for the organization are able to minimize resource losses which their highly concerned counterparts are less able to avoid. The latter are more susceptible to abusive supervision as they place more personal value on their work and being a member of the organization. In this way, not being able to distant the self from a threat of resource loss at work presents a loss not only for focal individuals themselves but with the subsequent decrease in OCBs also for coworkers and an organization as a whole. Bringing in commitment theory (Meyer, 2016; Meyer & Allen, 1997), we imagine that it will be harder for abused employees to detach emotionally from the organization when affective commitment is high than when their commitment mindset is dominated by economic-based (continuance) and/or obligation-based (normative) forms of commitment. While counterproductive behaviors such as production deviance and withdrawal have already been shown to function as protective resources (Krischer et al., 2010), we extend this line of work toward low organizational concern. These results have important implication for the field as they challenge the literature's more favorable view of concern at work. A study like the present one illustrates how important a detailed look at boundary conditions is to understand the role of motives at work.
Third, we also extend our understanding of COR theory by identifying organizational detachment as a protective means by which individuals can conserve resources in the face of threats. This complements prior research that had so far focused almost exclusively (Qin et al., 2018) on desired behaviors and attitudes such as voice or citizenship as means to protect and replenish resources (Koopman et al., 2016). The notion that organizational detachment might have dispositional roots extends recent applications of classic attachment theory (Bowlby, 1982; Fraley & Waller, 1998) to organizational foci (Feeney et al., 2020). An interesting question for future research would be whether the resource losses associated with leader abuse depend on the nature of one's dominant attachment style, either as secure, anxious, or avoidant attachment.
Finally, we did not find evidence for stress to decrease in-role behavior. While this was in contrast to our predictions, this is somewhat in line with our COR-based framework. When individuals are stressed, they try to protect their remaining resources. Doing so, they primarily cut back in areas of work that are nonmandatory rather than in areas of their job duties. This should, however, not indicate that employee in-role behavior is unaffected by abusive supervision. Consistent with previous work (Tepper et al., 2017), we found abusive supervision to reduce supervisor-rated in-role behavior. In our case, the effect was robust above and beyond stress and organizational concern.
Limitations and Avenues for Future Research
Although we were able to collect two-source data for Study 1 and multilevel (three-source) data for Study 2, the designs were still cross-sectional. Across studies, independent and mediator variables were measured using the same source raising the issue of common method variance. Thus, we temper our claims of causal primacy among the concepts under investigation. Our findings are largely consistent with theory, whereby causation can be implied rather than explicitly proven. Yet, common method bias should be less of an issue with regard to the core contribution of our paper, which is the interaction of abusive supervision and organizational concern on stress. According to Siemsen et al. (2010), interaction effects cannot be artifacts of common method variance; moreover, they tend to be deflated in such study settings. Still, we hope future endeavors follow-up on this research by considering longitudinal approaches and experimental designs.
As expected, the base rate for abusive supervision in both studies was rather low (Study 1: M = 1.48, SD = 0.47; Study 2: M = 1.63, SD = 0.60). Thus, when we refer to high abusive supervision in this paper, it is high in relative terms. In absolute terms, the level of reported abuse in both studies was low–medium at best, not unlike levels that are typically reported in the literature (Klasmeier et al., 2021; Mawritz et al., 2014; Mitchell & Ambrose, 2007; Tepper, 2000). We suspect that response-style biases such as impression management or social desirable responding patterns might contribute to an under-reporting of leader abuse. As such, conventional studies on abusive supervision are likely to chronically underestimate the true effects in the population. With lower mean values comes restricted variance in the focal measure, which makes it more difficult to find empirical relationships. Thus, readers should interpret our findings as lower-bound estimates of the phenomena in question. To more fully examine the effects of high abusive supervision (in absolute terms), future research may be well advised to turn to experimental designs that model encounters of low, moderate, and high supervisory abuse.
We note that abusive supervision and OCBs were unrelated in Study 1. Although some readers might find the lack of a zero-order correlation surprising, the establishment of a significant effect of the independent variable on the dependent one is not a necessary condition for testing indirect effects (for a current view of mediation analysis, see Hayes, 2017). Moreover, low correlations between primary variables might indicate the presence of unmeasured moderators or possibly suppressor variables. Accordingly, future research is needed to further deepen our understanding how the effects of abusive supervision on employee attitudes and behavior vary as a function of one or more characteristics of the situation. Consistent with our COR approach, it may be that resource losses through abusive supervision may be offset by concurrent resource gains through increased support from coworkers or the organization (Kurtessis et al., 2017). In light of our studies' findings, especially identifying ways for highly engaged individuals to be less affected by abusive supervision through increased social support, future research may pave the way for interesting, theory-based moderated mediation models.
Our study positions organizational concern motives as a beneficial resource. This should, however, not deemphasize the potential for collateral, long-term damage for both the organization and the focal individual that is linked to low organizational concern. As argued above, low organizational concern is likely to also contribute to decreases in organizational effectiveness due to reduced knowledge sharing or OCB. From the perspective of the focal individual, there may be even more collateral damage that is beyond our study's scope but should still be discussed at this point. Hobfoll (1989) suggests that reevaluating resources to combat resource losses can be problematic when they challenge basic notions concerning the self and the world. In our case, employees devalue the worth of work. However, as we live in a society that highly esteems success at work, processes of psychologically detachment from work may present substantial threats to self-esteem and self-worth. This should be considered when interpreting the results and implications of our study. So, to reiterate an earlier claim, we need to consider all – potentially even double-edged – effects tied to an organizational phenomenon such as organizational concern motive to understand the full range of implications for organizations.
References
(2002). The effects of item parceling on goodness-of-fit and parameter estimate bias in structural equation modeling. Structural Equation Modeling: A Multidisciplinary Journal, 9(1), 78–102. 10.1207/S15328007SEM0901_5
(2006). Conceptualizing and testing random indirect effects and moderated mediation in multilevel models: New procedures and recommendations. Psychological Methods, 11(2), 142–163. 10.1037/1082-989X.11.2.142
(2020). Following an uneven lead: Trickle-down effects of differentiated transformational leadership. Journal of Management. Advance online publication. 10.1177/0149206320931584
(1982). Attachment and loss: Retrospect and prospect. American Journal of Orthopsychiatry, 52(4), 664–678. 10.1111/j.1939-0025.1982.tb01456.x
(2010). The relationship between stressors and creativity: A meta-analysis examining competing theoretical models. Journal of Applied Psychology, 95(1), 201–212. 10.1037/a0017868
(2013). When do subordinates' emotion-regulation strategies matter? Abusive supervision, subordinates' emotional exhaustion, and work withdrawal. The Leadership Quarterly, 24(1), 125–137. 10.1016/j.leaqua.2012.08.006
(2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Erlbaum.
(2003). The relationship of emotional exhaustion to work attitudes, job performance, and organizational citizenship behaviors. Journal of Applied Psychology, 88(1), 160–169. 10.1037/0021-9010.88.1.160
(1999). Does stress lead to a loss of team perspective? Group Dynamics: Theory, Research, and Practice, 3(4), 291–302. 10.1037/1089-2699.3.4.291
(2020). Organizational attachment: Conceptualization, measurement, and incremental prediction of work attitudes and outcomes. Journal of Personnel Psychology, 19(3), 113–124. 10.1027/1866-5888/a000252
(2006). Dispositional predictors of organizational citizenship behavior: Motives, motive fulfillment, and role identity. Social Behavior and Personality: An International Journal, 34(6), 603–616. 10.2224/sbp.2006.34.6.603
(1998). Adult attachment patterns: A test of the typological model. In J. A. SimpsonS. R. Rhodes (Eds.), Attachment theory and close relationships (pp. 77–114). Guilford Press.
(2001). A new look at nurse burnout: The effects of environmental uncertainty and social climate. The Journal of Nursing Administration, 31(2), 91–96. 10.1097/00005110-200102000-00009
(2006). Sources of social support and burnout: A meta-analytic test of the conservation of resources model. Journal of Applied Psychology, 91(5), 1134–1145. 10.1037/0021-9010.91.5.1134
(2010). Organizational concern, prosocial values, or impression management? How supervisors attribute motives to organizational citizenship behavior. Journal of Applied Social Psychology, 40(6), 1450–1489. 10.1111/j.1559-1816.2010.00625.x
(2014). Getting to the “COR.” Journal of Management, 40(5), 1334–1364. 10.1177/0149206314527130
(2015). Engineering exchanges: Daily social identity threat predicts burnout among female engineers. Social Psychological and Personality Science, 6(5), 528–534. 10.1177/1948550615572637
(2017). Leadership and stress: A meta-analytic review. The Leadership Quarterly, 28(1), 178–194. 10.1016/j.leaqua.2016.10.006
(2007). An investigation of abusive supervision as a predictor of performance and the meaning of work as a moderator of the relationship. The Leadership Quarterly, 18(3), 252–263. 10.1016/j.leaqua.2007.03.007
(2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press.
Hayes, A. F. (2017). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (2nd ed.). Guilford Press.
(1987). Fine-tuning some resistant rules for outlier labeling. Journal of the American Statistical Association, 82(400), 1147–1149. 10.1080/01621459.1987.10478551
(1989). Conservation of resources: A new attempt at conceptualizing stress. American Psychologist, 44(3), 513–524. 10.1037/0003-066X.44.3.513
(2001). The influence of culture, community, and the nested‐self in the stress process: Advancing conservation of resources theory. Applied Psychology: An International ReviewApplied Psychology, 50(3), 337–421. 10.1111/1464-0597.00062
(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. 10.1080/10705519909540118
(2003). Resources, coping strategies, and emotional exhaustion: A conservation of resources perspective. Journal of Vocational Behavior, 63(3), 490–509. 10.1016/S0001-8791(02)00033-7
(2013). Why and when do motives matter? An integrative model of motives, role cognitions, and social support as predictors of OCB. Organizational Behavior and Human Decision Processes, 121(2), 231–245. 10.1016/j.obhdp.2013.03.004
(2021). On the destructiveness of laissez-faire versus abusive supervision: A comparative, multilevel investigation of destructive forms of leadership. European Journal of Work and Organizational Psychology. Advance online publication. 10.1080/1359432X.2021.1968375
(2016). Integrating the bright and dark sides of OCB: A daily investigation of the benefits and costs of helping others. Academy of Management Journal, 59(2), 414–435. 10.5465/amj.2014.0262
(2010). Can counterproductive work behaviors be productive? CWB as emotion-focused coping. Journal of Occupational Health Psychology, 15(2), 154–166. 10.1037/a0018349
(2005). The Copenhagen Psychosocial Questionnaire-a tool for the assessment and improvement of the psychosocial work environment. Scandinavian Journal of Work, Environment & Health, 31(6), 438–449. 10.5271/sjweh.948
(2017). Perceived organizational support: A meta-analytic evaluation of organizational support theory. Journal of Management, 43(6), 1854–1884. 10.1177/0149206315575554
(2002). To parcel or not to parcel: Exploring the question, weighing the merits. Structural Equation Modeling: A Multidisciplinary Journal, 9(2), 151–173. 10.1207/S15328007SEM0902_1
(2019). Facilitate knowledge sharing by leading ethically: The role of organizational concern and impression management climate. Journal of Business and Psychology, 34(4), 539–553. 10.1007/s10869-018-9555-8
(2017). Abusive supervision: A meta-analysis and empirical review. Journal of Management, 43(6), 1940–1965. 10.1177/0149206315573997
(2013). A review of abusive supervision research. Journal of Organizational Behavior, 34(S1), S120–S137. 10.1002/job.1888
(1982). Burnout: The cost of caring. Prentice Hall.
(2017). Is consistently unfair better than sporadically fair? An investigation of justice variability and stress. Academy of Management Journal, 60(2), 743–770. 10.5465/amj.2014.0455
(2014). Hostile climate, abusive supervision, and employee coping: Does conscientiousness matter? Journal of Applied Psychology, 99(4), 737–747. 10.1037/a0035863
(Ed.). (2016). Handbook of employee commitment. Edward Elgar Publishing.
(1997). Commitment in the workplace: Theory, research, and application. Sage.
(2007). Abusive supervision and workplace deviance and the moderating effects of negative reciprocity beliefs. Journal of Applied Psychology, 92(4), 1159–1168. 10.1037/0021-9010.92.4.1159
(2006). The assessment of psychological strain in work contexts. European Journal of Psychological Assessment, 22(3), 198–206. 10.1027/1015-5759.22.3.198
(2017). Mplus user's guide (8th ed.). Muthén & Muthén.
(1988). Organizational citizenship behavior: The good soldier syndrome. Issues in organization and management series. Lexington Books.
(1990). The motivational basis of organizational citizenship behavior. In B. M. StawL. L. Cummings (Eds.), Research in organizational behaviour: An annual series of analytical essays and critical reviews (pp. 43–72). JAI Press.
(1997). Organizational citizenship behavior: It's construct clean-up time. Human Performance, 10(2), 85–97. 10.1207/s15327043hup1002_2
(2005). Stress and burnout: The significant difference. Personality and Individual Differences, 39(3), 625–635. 10.1016/j.paid.2005.02.009
(2009). Individual- and organizational-level consequences of organizational citizenship behaviors: A meta-analysis. Journal of Applied Psychology, 94(1), 122–141. 10.1037/a0013079
(2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63(1), 539–569. 10.1146/annurev-psych-120710-100452
(2006). Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31(4), 437–448. 10.3102/10769986031004437
(2011). Alternative methods for assessing mediation in multilevel data: The advantages of multilevel SEM. Structural Equation Modeling: A Multidisciplinary Journal, 18(2), 161–182. 10.1080/10705511.2011.557329
(2018). The short-lived benefits of abusive supervisory behavior for actors: An investigation of recovery and work engagement. Academy of Management Journal, 61(5), 1951–1975. 10.5465/amj.2016.1325
(2001). The causes of organizational citizenship behavior: A motivational analysis. Journal of Applied Psychology, 86(6), 1306–1314. 10.1037/0021-9010.86.6.1306
(2010). Common method bias in regression models with linear, quadratic, and interaction effects. Organizational Research Methods, 13(3), 456–476. 10.1177/1094428109351241
(1993). Basic research and practical problems: The promise of a "functional" personality and social psychology. Personality and Social Psychology Bulletin, 19(3), 251–264. 10.1177/0146167293193001
(1995). Psychological empowerment in the workplace: Dimensions, measurement, and validation. Academy of Management Journal, 38(5), 1442–1465. 10.5465/256865
(2000). Organizational Citizenship Behavior: Entwicklung und erste Validierung eines Meßinstruments. Diagnostica, 46(2), 73–83. 10.1026//0012-1924.46.2.73
(2015). Too many motives? The interactive effects of multiple motives on organizational citizenship behavior. Journal of Applied Psychology, 100(4), 1239–1248. 10.1037/apl0000001
(2019). Breaking the cycle of abusive supervision: How disidentification and moral identity help the trickle-down change course. Journal of Applied Psychology, 104(1), 164–182. 10.1037/apl0000360
(2000). Consequences of abusive supervision. Academy of Management Journal, 43(2), 178–190. 10.2307/155637510.5465/1556375
(2007). Abusive supervision in work organizations: Review, synthesis, and research agenda. Journal of Management, 33(3), 261–289. 10.1177/0149206307300812
(2017). Abusive supervision. Annual Review of Organizational Psychology and Organizational Behavior, 4(1), 123–152. 10.1146/annurev-orgpsych-041015-062539
(2015). What makes employees resilient to job demands? The role of core self-evaluations in the relationship between job demands and strain reactions. European Journal of Work and Organizational Psychology, 24(1), 76–87. 10.1080/1359432X.2013.858700
(1991). Job satisfaction and organizational commitment as predictors of organizational citizenship and in-role behaviors. Journal of Management, 17(3), 601–617. 10.1177/014920639101700305
(2018). Resource depletion perspective on the link between abusive supervision and safety behaviors. Journal of Business Ethics, 162(1), 213–228. 10.1007/s10551-018-3983-2
(2019). Why abusive supervision impacts employee OCB and CWB: A meta-analytic review of competing mediating mechanisms. Journal of Management, 45(6), 2474–2497. 10.1177/0149206318823935
1Given the low base rates of abusive supervision, it may be possible that single outliers may have driven the effects of abusive supervision. We performed additional analyses to address this issue. We scanned for outliers and found one case to be more than 1.5× the interquartile range (IQR) above the 75% percentile (no cases were 3 IQR's above; Hoaglin & Iglewicz, 1987). We removed this case and reran our analyses. Results remained qualitatively identical.
2As in Study 1, we conducted supplementary analyses to inspect whether outliers affected our results regarding abusive supervision. After removing 10 cases that were 1.5 IQR's above the 75% percentile, all direct effects of our research model remained qualitatively identical, but the interaction of abusive supervision and organizational concern on stress was not significant. However, Hoaglin and Iglewicz (1987) suggest that the 1.5× is inaccurate approximately 50% of the time and thus recommended higher thresholds. In a second step, we used 3 IQR's above the 75% percentile, which led to the removal of only one case. The results for all focal relationships were qualitatively identical with the results reported in the main analysis.
