Linking attachment theory to abusive supervision

Jennifer L. Robertson (DAN Department of Management and Organizational Studies, University of Western Ontario, London, Canada)
Angela M. Dionisi (Sprott School of Business, Carleton University, Ottawa, Canada)
Julian Barling (Smith School of Business, Queen’s University, Kingston, Canada)

Journal of Managerial Psychology

ISSN: 0268-3946

Publication date: 12 March 2018



The purpose of this paper is to explore the impact of leaders’ attachment orientation and social self-efficacy on the enactment of abusive supervision.


Data were obtained from a sample of leader-subordinate dyads (n=114), and were collected using a Panel Service.


The results show that a Close/Depend attachment orientation was negatively associated with abusive supervision, while an Anxious attachment orientation was positively associated with abusive supervision. Social self-efficacy mediated these relationships.

Research limitations/implications

The results generate a deeper understanding of the etiology of destructive leadership. Applying attachment theory to the study of abusive supervision also offers a new theoretical perspective on potential precursors of this behavior.

Practical implications

The findings suggest organizations might benefit from attempts to alter leaders’ destructive attachment orientations, and by extension, reduce their abusive behavior. It may also be possible to reduce the occurrence of abusive supervision by implementing leadership development initiatives aimed at enhancing leaders’ confidence in their social skills.

Social implications

By identifying several potential precursors to abusive supervision, this study highlights possible points of intervention to combat a form of leadership that is linked with employee suffering. Thus, the findings can be used to help improve the working lives of those who are affected by this destructive workplace behavior.


Until now, research has not considered leaders’ attachment orientation as an antecedent to abusive supervision, nor has it explored the meditational role of social self-efficacy. The use of leader-follower dyads in this study also helps reduce issues related to social desirability biases and common method variance.



L. Robertson, J., Dionisi, A. and Barling, J. (2018), "Linking attachment theory to abusive supervision", Journal of Managerial Psychology, Vol. 33 No. 2, pp. 214-228.

Download as .RIS



Emerald Publishing Limited

Copyright © 2018, Emerald Publishing Limited

Abusive supervision—defined as the extent to which supervisors engage in ongoing displays of verbal and non-verbal (non-physical) hostility (e.g. public ridicule, rudeness; Tepper, 2000)—is costly to employees and organizations. Previous research reveals that abusive supervision predicts stress, psychological and physical health problems (e.g. Schyns and Schilling, 2013; Tepper, 2000), increased turnover intentions (e.g. Harvey et al., 2007), burnout (e.g. Yagil, 2006) and negative job-related attitudes (e.g. Schyns and Schilling, 2013) among followers. Moreover, harm to the organization has also been shown; supervisor mistreatment is related to subordinate deviance (e.g. Thau et al., 2009; Tepper et al., 2009), and decreases in employee (e.g. Peng et al., 2014) and organizational performance (e.g. Schyns and Schilling, 2013). When consequences such as these are coupled with estimates that 14 percent of US employees are victims of abusive supervision (Schat et al., 2006), the importance of addressing this workplace problem is evident.

To mitigate the occurrence of this issue, research on the antecedents of abusive supervision has been increasing. Some of this research has linked this type of leadership to contextual (e.g. organizational structure, workplace climate, national culture; Aryee et al., 2008; Luthans et al., 1998; Mawritz et al., 2012), dispositional (e.g. stress, depressive symptoms, authoritarianism, Machiavellianism; Aryee et al., 2008; Burton et al., 2012; Byrne et al., 2014; Kiazad et al., 2010) and developmental (e.g. histories of family undermining and aggression; Garcia et al., 2014; Kiewitz et al., 2012) variables. Further, others have focused on how abusive supervision may be a product of social learning, identity threat or self-regulatory impairment (Tepper et al., 2017). However, of the research that has been generated on the antecedents of abusive supervision, the role of leaders’ cognitive differences in their approach to relationships, such as their working relational mental models (i.e. attachment orientation), has been virtually ignored. Similarly, the impact that a leader’s self-doubt may have on the enactment of abusive behavior has not garnered attention. Accordingly, drawing on attachment theory (Bowlby, 1982; Hazen and Shaver, 1987), social cognitive theory (Bandura, 2001) and the social competencies and interpersonal processes (SCIP) model (Mallinckrodt, 2000), it is the goal of the current study to explore the impact of leaders’ attachment orientation (i.e. close, depend, anxious) on subordinates’ reports of abusive supervision, as well as the role of leaders’ social self-efficacy in helping to explain the connection between attachment and this toxic form of leadership.

We believe that exploring these relationships is important for several reasons. Examining how attachment impacts abusive supervision could provide new theoretical insights into the etiology of this leadership behavior. While previous work investigating the antecedents of abusive supervision has argued that these leaders are ones who feel threatened (identity threat), believe behaving abusively is appropriate (social learning), or are unable to control themselves (self-regulation; Tepper et al., 2017), by applying attachment theory to the study of abusive supervision, we address the question of destructive leadership from a cognitive, perceptual point of view and potentially highlight the cognitive basis of this abuse of power. If we are able to show that leaders’ internal working models of the self and others serve as antecedents to this damaging form of leadership, support for the theory that abusive supervision is a product of how leaders conceptualize their relational world, will be gleaned. Moreover, our examination of social self-efficacy as the mechanism through which attachment affects leaders’ hostile behaviors provides an additional explanation for why leaders engage in abusive supervision, and highlights how in some cases, destructive leadership may be partly attributable to the self-defeating cognitions that leaders hold about themselves. Finally, from a practical perspective, if we are able to show that the ways leaders perceive the self, others (i.e. attachment orientation) and their own capabilities (i.e. social self-efficacy) are linked to this counterproductive workplace behavior, targeted interventions designed to help leaders restructure these damaging perceptions in more positive ways, may be developed.

Theoretical background

Attachment theory

John Bowlby (1982) developed attachment theory, postulating that infants are born with an innate tendency to seek out and maintain close proximity to their primary caregiver. Bowlby theorized that when young children are in close proximity to their caregivers, they feel safe, secure and confident to explore their surroundings. However, when infants are separated from their caregivers, they engage in various “attachment behaviors” (e.g. monitoring, searching, crying) in an attempt to regain proximity to them. The way in which a primary caregiver responds to these attention-seeking behaviors shapes a child’s cognitive working models of the self and others. Extending Bowlby’s original ideas, Ainsworth et al. (1978) established three attachment patterns infants develop given the (un)success of their attachment-seeking behaviors: secure, anxious-ambivalent and avoidant.

Based on the continuity of internal working models—the notion that models of the “self” and “others” developed in the context of infant-caregiver interactions carry forward into adulthood, subsequently guiding expectations, perceptions and behavior in new relationships (Collins, 1996)—Hazen and Shaver (1987) translated Ainsworth et al.’s (1978) typology into three attachment descriptions (i.e. secure, avoidant and anxious) to study the role of attachment in adult romantic relationships (see Feeney, 2008 for a review). Since then, various categorizations and dimensional models of adult attachment have been developed (see Ravitz et al., 2010). In the current study, we rely on Collins and Read’s (1990) three primary underlying attachment dimensions: close, depend and anxiety. The close dimension reflects how comfortable individuals are with closeness and intimacy, the depend dimension concerns the extent to which one trusts others and feels others are available when needed, while the anxious dimension reflects one’s fear of rejection, abandonment and not being loved. We draw on this perspective given the belief that such an approach is preferable to using attachment categories. Compared to categorizations that isolate individuals into one attachment category (despite the fact they may still exhibit some features of another category), the underlying dimensional approach more accurately detects individual differences in scores for the various attachment types (Ravitz et al., 2010). Further, Collins and Read (1990) contend that measuring attachment through their underlying dimensions provides a better understanding of the contents of the attachment mental models that drive cognitive, emotional and behavioral responses in adult relationships. Thus, dimensional measures have been shown to be better predictors of relevant outcomes than categorical measures (Collins and Read, 1990), and have been argued to “be of great utility in research” (Ravitz et al., 2010, p. 428).

Attachment dimensions and social self-efficacy

Defined as the belief in one’s own ability to execute and regulate the important actions in their lives, self-efficacy is a foundational self-perception that takes on domain specific forms (Bandura, 1997). To this end, social self-efficacy beliefs pertain to one’s confidence in their ability to engage in positive social interactions—to form and maintain social relationships, work cooperatively with others and manage different types of interpersonal conflicts (Bandura et al., 2001; Gecas, 1989). A central tenant of self-efficacy theory is that self-efficacy beliefs are developed through enactive mastery experiences and feedback about one’s performance (Bandura, 2001). Thus, our interpersonal experiences with attachment figures, and the views of the self and others that are engendered from these experiences (i.e. our mental models), may in fact shape our perceptions concerning our relational capabilities. In particular, it is believed that the basis of our social self-efficacy beliefs is formed by how caregivers respond to attachment-seeking behaviors (Corcoran and Mallinckrodt, 2000).

According to the SCIP model (Mallinckrodt, 2000), a secure attachment to caregivers results in the generation of important social competencies—dispositions and skills that are necessary for recruiting and maintaining supportive, close relationships, of which social self-efficacy is one (Mallinckrodt, 2000; Mallinckrodt and Wei, 2005)—whereas insecure attachment leads to deficits of these social competencies. Moreover, in line with attachment theory, the SCIP model holds that the positive or negative models of self and others emanating from early attachment relationships are largely responsible for engendering (or not) these interpersonal capabilities.

On the one hand, as the attachment-seeking behaviors of anxiously attached individuals are met with unresponsiveness and unavailability from caregivers (Mikulincer and Shaver, 2016), these children come to see themselves as unacceptable and unworthy of comfort and care (i.e. a negative working model of the self; Cassidy, 2000)—a self-view that they carry with them into adulthood, subsequently shaping their understanding of, and behavior within, social relationships. As a result, anxious adults go on to fear rejection and abandonment (Pietromonaco and Barrett, 2000), show signs of distress when engaged in dyadic conversations (e.g. Guerrero, 1996) and in clinical samples demonstrate high levels of social anxiety (e.g. greater fear of social interactions, negative evaluations and being scrutinized by others in the course of everyday activities) and social withdrawal (e.g. greater avoidance of feared situations; Eng et al., 2001). Adults with an anxious attachment style may thus have difficulty trusting their own ability to engage in positive social interactions because their working models lead them to expect negative outcomes from close relationships, and feel relatively powerless and unable to bring about desired relationship outcomes (Cassidy, 2000; Tronick, 1989).

In contrast, individuals who are comfortable with closeness and dependency have had attachment figures that are responsive, reliable and available when needed. In receiving the response sought from caregivers in times of distress, these children develop a representation of the self as acceptable and worthy, and also come to see others as accessible, reliable and trustworthy (Cassidy, 2000). These positive working models of the self and others promote a healthy capacity for intimacy and positive relationships later in life, influence how these individuals subsequently interpret social events, and shape the expectations they come to hold about future interpersonal situations (Pietromonaco and Barrett, 2000). As securely attached individuals do not worry about others rejecting them (Bohlin et al., 2000; Hazen and Shaver, 1987), display high levels of social skills (e.g. DiTommaso et al., 2003) and are characterized by a psychological sense of “felt” security in the context of relationships (Bretherton, 1985; Pietromonaco and Barrett, 2000), such interpretations and expectations should positively affect their self-perceived competence in social relationships.

Accordingly, when taken together we suggest that while anxious leaders may doubt their capability to initiate social contact and develop positive relationships, those characterized by attachment closeness and dependence will feel confident in their interactions with others, and will believe they are capable of fostering positive relationships with those around them:


The anxious attachment dimension will be negatively associated with social self-efficacy.


The close and depend attachment dimensions will be positively associated with leaders’ social self-efficacy.

Social self-efficacy and abusive supervision

According to social cognitive theory, the self-efficacy belief system is the foundation of human motivation, ultimately influencing one’s thought patterns, emotions and actions (Bandura, 2001). Moreover, social cognitive theory maintains that self-efficacy determines the way that people interact with their social environment and arrange their social relationships. Importantly, leadership is by definition, a social phenomenon (Graen and Uhl-Bien, 1995; Uhl-Bien, 2006), and a prominent mark of effective leadership is the ability to motivate, inspire and develop positive relationships with one’s followers (e.g. Bass, 1990; Bass and Riggio, 2006). However, in line with social cognitive theory, leaders who doubt their ability to engender these interpersonal outcomes (i.e. those with low social self-efficacy) will not only be less motivated to enact these high-quality leadership behaviors, but may ultimately engage in more transgressive conduct within this interpersonal domain (e.g. interpersonal breaches, lying, destructiveness; verbal assaults; Bandura et al., 2001).

According to Bandura et al. (2001, p. 125), a strong sense of efficacy can prevent transgressive behavior “in large part, by promoting prosocialness, curtailing the propensity to disengage moral self-sanctions from socially alienating and harmful conduct, and countering ruminative and vengeful affectivity.” Extending this to social self-efficacy in particular, Bandura et al.’s (2001) longitudinal research shows a connection between lower levels of social self-efficacy, lower levels of positive relationship behaviors (e.g. kindness and cooperativeness), higher levels of hostile and retaliatory thought patterns, and ultimately the enactment of anti-social conduct. In a study capturing the behavioral responses of participants to instances of ambiguous provocation, Erdley and Asher (1996) further find that compared to those who either withdraw or undertake problem solving in the face of harmful acts, those who had an aggressive response were less efficacious about their pro-social skills. Moreover, research highlights the effectiveness of bullying intervention programs that target the social self-efficacy beliefs of those who display aggressive behavior (e.g. DeRosier, 2004), further suggesting the role of this self-perception in the enactment of aggression. When applied to the current context, this research supports the conjecture that leaders low on this positive self-perception will be prone to behave in destructive ways toward their subordinates, as they lack the key capacity to exercise control over anti-social thought processes, affect and actions that this mechanism of personal agency affords:


Leaders’ social self-efficacy will be negatively related to abusive supervision.

The mediating role of social self-efficacy

Consistent with prior research on attachment styles and leadership (see Yip et al., 2017 for a review), the pattern of relationships outlined above points to the potentially far reaching effects of leaders’ attachment orientation, such that the close, depend and anxious attachment dimensions may indirectly (through social self-efficacy) influence leaders’ hostile leadership behavior. This is in line with Collins and Read’s (1994) framework for how attachment affects behavioral responses. According to this framework, the self/other schemas that make up attachment working models have a direct impact on one’s cognitions (e.g. interpretations of social situations; one’s self-image), and it is these cognitions that then determine how one will choose to behave in a given situation. Moreover, a central tenant of the SCIP model is that key social competencies emanating from attachment relationships mediate the direct effects of attachment on outcome variables, ultimately transmitting the impact of attachment to outcomes (Mallinckrodt, 2000; Mallinckrodt and Wei, 2005). Drawing upon these frameworks, we thus suggest that the working models that have developed from leaders’ interpersonal experiences with attachment figures will directly influence their social self-efficacy beliefs, which in turn, will affect how aggressively they behave:


Leaders’ scores on the anxious attachment dimension will be positively and indirectly related to abusive supervision through social self-efficacy.


Leaders’ scores on the close and depend attachment dimensions will be negatively and indirectly related to abusive supervision through social self-efficacy.


Sample and procedure

We recruited a sample of 114 leader-subordinate dyads working in a variety of industries in the US through Clear Voice Panel Services. To recruit the dyads, we first had Clear Voice invite full-time employees (i.e. subordinates) who report directly to, and interact with a leader on a weekly basis, to complete an online survey. As part of this online survey, we asked the subordinates to send a recruitment e-mail (drafted by the researchers) to the leader to whom they report directly. We asked that subordinates send the recruitment e-mail to their direct leader to reduce the possibility that subordinates who had multiple (indirect) leaders would send the recruitment notice to a person (s)he would be prone to rate more favorably.

Leaders and subordinates completed separate online questionnaires. Leaders answered questions about their attachment orientation and their social self-efficacy. To avoid issues associated with common method variance (Podsakoff et al., 2012), subordinates rated their leaders’ levels of abusive supervision. Since attachment orientation and social self-efficacy were self-reported by leaders, we sought to further alleviate concerns about common method variance that could arise from social desirability and acquiescence biases by assuring participants that their survey responses were anonymous (Podsakoff et al., 2003). Also, because leaders rated attachment orientation and social self-efficacy on different Likert scales, concern that some co-variation among studied variables is attributed to similar scale format and anchors is mitigated (Podsakoff et al., 2003)[1].

In the subordinate sample (Mage =39.42 years, SD =10.41), 56.1 percent were male, most were Caucasian (73.7 percent), and most had been working for their organizations for an average of 8.25 years (SD =6.83). In all, 38.6 percent had a college diploma or less, 47.4 percent had an undergraduate university degree and 14.1 percent had a professional/master’s degree or doctorate. On average, subordinates had been working with their direct leader for 2.75 years (SD =2.65).

The mean age and organizational tenure of the leader sample was 45.15 (SD =9.53) and 11.13 (SD =7.41) years, respectively. Most (62.3 percent) leaders were male, 78.1 percent were Caucasian, 29.8 percent had a college diploma or less, 43 percent had an undergraduate university degree, while 27.2 percent had a professional/master’s degree or doctorate.


Adult attachment

We used the Adult Attachment Scale (Collins and Read, 1990) to assess the close, depend and anxious attachment dimensions. Although this measure is one of the most widely used to assess adult attachment, there have been some questions raised about its psychometric properties (see Kurdek, 2002). Therefore, we used the eight items recommended by Kurdek (2002) to measure the close and depend dimensions (four items measure each dimension). Sample items include “I find it relatively easy to get close to others” (close dimension; α=0.73) and “I am comfortable depending on others” (depend dimension; α=0.69). Following others (e.g. Brennan et al., 1998; Collins, 1996; Davila et al., 1998), we combined the close and depend items into an overall measure (i.e. close/depend dimension; α=0.78). The primary reason for doing so was conceptual, as these two dimensions operate similarly to describe Hazen and Shaver’s (1987) original model of adult attachment (Collins and Read, 1990), and because together, they underpin the avoidance dimension of the two-dimensional approach to attachment (Brennan et al., 1998; Mikulincer and Shaver, 2016)[2].

Because Kurdek failed to isolate valid and reliable items to measure anxiety, we used the four items (e.g. “I often worry that others won’t want to stay with me”) from Wu and Parker’s (2017) shortened Adult Attachment Scale to measure this dimension (α=0.65). This short-form measure has been used in previous studies (e.g. Wu and Parker, 2012, 2017) and has demonstrated good construct validity and reliability (Wu and Parker, 2017). All items assessing adult attachment were ranked on a scale of 1 (not at all like me) to 5 (very much like me).

Social self-efficacy

We used five items from Sherer et al.’s (1982) social self-efficacy measure (e.g. “If I see someone I would like to meet, I go to that person instead of waiting for him or her to come to me”). The response scale ranged from 1 (strongly disagree) to 7 (strongly agree). Following the recommendation of Cristobal et al. (2007), we examined the item total correlations of this scale, and identified one item with a value that fell below a minimum cutoff of 0.3 (r=0.01). As this datum suggests that this item is not clearly representative of the construct under study (Wieland et al., 2017), we thus deleted it. In so doing, the reliability coefficient α of this scale improved. The final scale demonstrated good internal consistency (α=0.67).

Abusive supervision

To assess abusive supervision, subordinates responded to Tepper’s (2000) 15-item measure (α=0.98) on a scale of 1 (strongly disagree) to 5 (strongly agree). Sample items include “My leader invades my privacy” and “My leader ridicules me.”

Quality check items

Following DeSimone et al.’s (2015) recommendation, several quality assurance questions meant to verify that participants were paying adequate attention (e.g. “please select strongly disagree”) were included, and were inconspicuously placed throughout both subordinates’ and leaders’ questionnaires.

Control variables

To rule out plausible alternative explanations for our findings, we considered several variables as potential controls. First, to account for the extent to which subordinates were positively disposed to their leader, we asked subordinates to rate how much they enjoyed working with their leader. Moreover, because females are less likely to engage in aggressive behaviors than males (Baron et al., 1999), we also considered controlling for leader gender in our analyses. Likewise, we considered the number of hours leaders work per week as a relevant control. Typically, the more individuals work, the higher their levels of stress (Meurs and Perrewe, 2011), and with stress often comes abusive supervision (Burton et al., 2012). Finally, we contemplated controlling for the length of the relationship between subordinates and their leader, given that subordinates with less exposure to their leader may be less accurate in rating abusive supervision (Wu and Parker, 2017).

Data analysis

Given recent recommendations (e.g. Becker et al., 2016; Bernerth and Aguinis, 2016), we determined whether controlling for the aforementioned variables was necessary by examining the bivariate correlations (see Table I) between these variables and our mediator and dependent variables. Neither leader gender nor tenure with leader was significantly related to social self-efficacy or abusive supervision. We also analyzed our hypothesized model with and without the control variables to determine if the results differed (Becker, 2005). Doing so revealed that the difference in the standardized parameter estimates for all specified relationships in our model was less than 0.1, thereby meeting Becker et al.’s (2016) cutoff. Consequently, to maximize statistical power and enhance the interpretability of parameter estimates (Becker et al., 2016; Bernerth and Aguinis, 2016), we did not include any control variables in our hypotheses tests.

We analyzed our hypothesized model as an observed variable path model using maximum likelihood estimation as implemented in AMOS 25. We compared this model to two alternative partial mediation models in which an additional parameter was added between the anxious attachment orientation and abusive supervision (Model A) and the close/depend attachment orientation and abusive supervision (Model B). All analyses were based on the covariance matrix. Prior to testing these models we checked to ensure that our data are appropriate for maximum likelihood estimation (i.e. are multivariate normal) by obtaining the overall scales’ skewness and kurtosis indicators as well as the critical ratio for multivariate kurtosis through AMOS. Results indicated that the data are multivariate normal as the skewness and kurtosis values were within acceptable ranges of 2 and 7, respectively (West et al., 1995), with skewness values ranging from −0.04 to 0.99 and kurtosis values ranging from −1.26 to −0.25. Further, the critical ratio value (0.44) for multivariate kurtosis was within Bentler’s (2005) cutoff criteria (i.e. <5).


Descriptive data and intercorrelations for all study variables appear in Table I.

The hypothesized model provided a good fit to the data: χ2 (df =2)=2.57, ns; χ2/df =1.29; CFI =0.99; NFI =0.98; PNFI=0.33; RMSEA=0.05; PCLOSE=0.37. These fit indices meet levels considered to indicate good fit by Hu and Bentler (1999) and MacCallum et al. (1996). Standardized parameter estimates and R2 values for the proposed model are shown in Figure 1. As predicted, the anxious attachment dimension was negatively related to social self-efficacy (β=−0.48, p<0.01), supporting H1. In contrast, close/depend was positively related to social self-efficacy (β=0.33, p<0.01), thereby supporting H2. In support of H3, social self-efficacy was negatively related to abusive supervision (β=−0.47 p<0.01). Standardized indirect effects =0.23 (anxious) and −0.16 (close/depend). It is important to note that given the low reliability of some of our measures, it is possible that model relationships are attenuated.

Finally, support emerged for H4 and H5 as comparing our hypothesized fully mediated model to the two alternative partially mediated models revealed that, while each model also provided a good fit (see Table II), neither provided a significantly better fit to the data: Model A, Δχ2 (df =1)=1.74, ns; Model B, Δχ2 (df =1)=1.44, ns. Further, in both models the standardized direct effects between the anxious attachment dimension and abusive supervision (β=0.15) and between the close/depend attachment dimension and abusive supervision (β=−0.13) were not significant (p>0.05). As a result, we retain the fully mediated model, which suggests the anxious and close/depend attachment dimensions indirectly effect abusive supervision through social self-efficacy.


Understanding the antecedents of abusive supervision is critical to preventing its occurrence, and thus, consequences. Accordingly, we sought to expand our current understanding of the cognitive and perceptual precursors that predispose some leaders to behave abusively. Specifically, based on a sample of leader-subordinate dyads working in a variety of industries, we explored how leaders’ attachment orientation affects their engagement in abusive supervisory behavior through their social self-efficacy beliefs. Consistent with our hypotheses, our results suggest that leaders who are high on the close/depend attachment dimensions believe that they have the ability to cultivate successful relationships (i.e. demonstrate higher levels of social self-efficacy), and in turn, are less likely to enact abusive supervision. In contrast, leaders who are high on the anxious attachment dimension doubt their abilities to develop successful relationships (i.e. demonstrate lower levels of social self-efficacy), and consequently, engage in higher levels of abusive supervision. Taken together, these results bring us one step closer to understanding the cognitive basis of abusive supervision.

Theoretical implications

To date, research has shown that the antecedents of abusive supervision operate through three overarching theoretical mechanisms: social learning, identity threat and self-regulatory impairment (Tepper et al., 2017). While this research provides insight into the etiology of abusive supervision, an understanding of how leaders’ cognitive representations of their relational worlds may predispose some to engage in abusive behavior, is missing. This omission not only precludes a promising point of intervention (as discussed below), but also stifles a more holistic understanding of the leaders who engage in this type of mistreatment. By applying an attachment framework to abusive supervision, our research provides an alternative conceptualization of the origins of this workplace problem, thus contributing to theory-building around this issue. As a result, we gain a deeper understanding of who abusive supervisors are, how they view themselves and others, and how these perceptions can impact their behavior.

This study also extends the leadership literature more broadly by supporting the application of an attachment perspective to the study of leadership. Although attachment theory has been used to understand the occurrence of a handful of leadership styles (e.g. Davidovitz et al., 2007; Doverspike et al., 1997; Popper, 2002; Popper et al., 2000; Richards and Hackett, 2012), to date, the number of empirical studies on this topic has been relatively small (Wu and Parker, 2017). Our study thus contributes to this larger research stream.

Finally, our findings on the mediational role of social self-efficacy also contribute to the theoretical underpinnings of abusive supervision, by providing an explanation for how one’s attachment orientation impacts leadership behavior. Specifically, these findings demonstrate that one’s working models of the self/others can influence toxic leadership by shaping how confident leaders are in their ability to positively interact with or form relationships with others. As a result, we gain a deeper understanding of the abusive supervisor as one who is insecure about his/her relationship abilities, and more broadly, about how one’s relational cognitions lead to powerful ability-based self-perceptions, which in turn, impact workplace conduct.

Practical implications

Organizations may be able to use the knowledge produced by this study to positively affect leadership in two ways. First, drawing upon Bowlby’s (1982) notion that attachment schemas can be influenced by new experiences, organizations might benefit from attempts to alter leaders’ destructive attachment orientations, and by extension, reduce their abusive behavior. For instance, insecurely attached employees may become more secure when they experience supportive leadership (Wu and Parker, 2017). Extending this, we suggest that organizations might well profit from providing anxiously attached leaders with emotional support, making it less likely that they will enact abusive supervisory behavior. It might also be possible for organizations to assist leaders in recognizing how they perceive the self and others, and then take steps to counteract any negative perceptions that they hold (e.g. cultivating leaders’ sense of self-worth). Moreover, given the mediating role of social self-efficacy in our study, it may be possible to reduce the frequency and occurrence of abusive supervision by implementing leadership development initiatives aimed at enhancing one’s confidence in their social skills. For example, by providing leaders with cognitive-behavioral counseling, organizations could combat the social insecurities found in some leaders, and consequently, their destructive behavior.

Limitations and future research

Several limitations in our study that should be addressed by future research warrant mention. First, the cross-sectional nature of our data precludes us from making causal inferences. Future research should thus replicate our findings using longitudinal or experimental methodologies. Second, because our study examined American leader-subordinate dyads, the generalizability of our findings to other countries and cultures remains to be demonstrated. Therefore, it is important that future research replicate our findings amongst culturally diverse samples. Third, our research relies on a single subordinate’s perception of abusive supervision. Future studies would benefit from multiple ratings of abusive supervision, especially given possible variance among followers in the experience of mistreatment. Fourth, although we used a variety of procedural and statistical remedies to prevent and test for common method variance, given that the independent variables and our mediating variable were self-reported by leaders, we cannot totally mitigate this concern. We thus recommend that future research use temporal separation of measurement (Podsakoff et al., 2012) of attachment orientation and social self-efficacy.

Fifth, the possibility that our sample consists of relatively non-abusive supervisors remains (i.e. subordinates only invited their leaders if they display little abusive behavior, or mostly non-abusive leaders responded to the invitation to participate). However, as suggested by Rogelberg and Stanton (2007), we mitigate this concern by comparing the mean rating of abusive supervision from our sample to that of population estimates (n=23,791) reported in a recent meta-analysis (Mackey et al., 2017). The mean level of abusive supervision in our study (M=1.92, SD =1.08) was significantly higher than the population mean (M=1.68, SD =0.78), t(23,902)=−3.27, p<0.01. Thus, while our mean level of abusive supervision is low, it is still higher than national norms, and is consistent with Tepper’s (2000, 2007) conclusion that abusive supervision has a low base rate. Further, uncovering effects despite low levels of abusive supervisory behavior represents a conservative test of our hypotheses; we suspect the effects found may be even stronger among more abusive leaders. Nevertheless, we encourage future research to replicate our findings with a sample containing higher levels of abusive supervision.

Several other avenues for future research that arise from our study also warrant mention. First, considering what additional variables—and in particular, those pertaining to leaders’ self-perceptions—may mediate the relationship between attachment and abusive supervision is important. One possibility is leaders’ self-efficacy for emotional regulation (Wang et al., 2011). Individuals who score high on the anxious attachment dimension may be low in this perceived competency given failed early experiences with affect regulation among caregivers (Shaver and Mikulincer, 2007), and thus by extension, may also be more likely to aggressively direct their negative emotions at followers (Deng et al., 2017; Wang et al., 2011).

Second, future research may benefit from exploring potential moderators to the relationships of concern. Specifically, while leaders may possess detrimental attachment orientations given experiences with primary caregivers, the existence of more positive bonds later in life may help to circumvent the negative effects discovered in this study. For example, the levels of support that leaders’ receive from colleagues, their own supervisors or their spouse, may curtail the negative impact of insecure attachment on leadership.

Third, future research might benefit from a simultaneous focus on subordinates’ attachment. In line with victim precipitation models of workplace aggression (Aquino, 2000), subordinates characterized by anxious attachment may behave in ways (e.g. routine displays of anger, hostility) that increase the likelihood a supervisor will behave aggressively toward them. This in no way absolves supervisors from inappropriate behavior, or invalidates subordinate suffering. However, should subordinate attachment prove to be a significant moderator, more support will be found for the conjecture that a combination of leader and follower characteristics is responsible for eliciting destructive leadership (Aquino, 2000).


The implications of abusive supervision highlight the need to fully understand the precursors of this destructive form of leadership. The goal of the current research was to contribute to the growing body of literature on the antecedents of abusive supervision by investigating the predictive role of attachment orientations and social self-efficacy. In so doing, this study speaks to the cognitions and perceptions underlying the emergence of destructive leadership. While the need for longitudinal, multi-level replications remains, our findings provide initial insights into how leaders’ views of the self, others and their own capabilities contribute to their abusive behavior, ultimately expanding our current view of those who enact abusive supervision.


Standardized parameter estimates and R2 values for the hypothesized model

Figure 1

Standardized parameter estimates and R2 values for the hypothesized model

Means, standard deviations and correlations of studied variables

Mean SD 1 2 3 4 5 6 7 8
1. Enjoy working with leader 4.00 0.92
2. Leader gender 0.38 0.49 −0.02
3. Leader hours 45.04 7.84 −0.02 0.02
4. Tenure with leader (in years) 2.75 2.65 −0.04 0.01 −0.15
5. Anxious 2.57 0.88 −0.04 −0.13 −0.13 −0.04 (0.65)
6. Close/Depend 3.21 0.76 0.31** 0.01 −0.03 0.14 −0.54** (0.78)
7. Social self-efficacy 4.85 1.08 0.23* 0.07 0.00 0.08 −0.67** 0.60** (0.67)
8. Abusive supervision 1.92 1.08 −0.34** −0.14 −0.25** −0.05 0.40** −0.38** −0.47** (0.98)

Notes: n=109-114 dyads pairwise deletion. Reliabilities are on the diagonal. *p⩽0.05; **p<0.01

Fit indices for the three models

Hypothesized model 2.57 2 0.99 0.98 0.33 0.05 0.37
Alternative Model A 0.834 1 1.00 0.99 0.17 0.00 0.42
Alternative Model B 1.13 1 0.99 0.99 0.17 0.04 0.35

Notes: n=109 dyads (listwise deletion). CFI, comparative fit index; NFI, normed fit index; PNFI, parsimonious normed fit index; RMSEA, root-mean-square error of approximation; PCLOSE, test of RMSEA significant



We tested whether leaders’ responses to attachment orientation and social self-efficacy were affected by common method variance using Harmon’s single factor approach. To do so, we conducted an exploratory factor analysis (EFA) wherein items measuring the close, depend and anxious attachment dimensions, along with items measuring social self-efficacy, were included in the EFA; the number of factors extracted was constrained to one. We then examined the unrotated factor structure to determine if the one factor explained the majority (i.e. >50 percent) of variance. This test revealed that one factor did not explain the majority of variance (34.42 percent), suggesting that the data likely are not affected by common method variance.


Post hoc analyses were conducted to provide empirical support for combining the close and depend dimensions. Specifically, we computed a series of simple regressions in which the close, depend and anxious dimensions were simultaneous predictors of social self-efficacy and abusive supervision, and each attachment dimension was a single predictor of each criterion variable (i.e. only one attachment dimension was included in the regression). Results from these analyses (which can be obtained from the first author) revealed that when analyzed separately, each dimension was significantly related to social self-efficacy and abusive supervision. However, once all three attachment dimensions were included as simultaneous predictors, the depend dimension no longer made a significant contribution to each model. These results suggest the construct space is shared by the depend and close dimensions, as the close dimension almost completely subsumes the variance explained in social self-efficacy and abusive supervision.


Ainsworth, M.D.S., Blehar, M.C., Waters, E. and Wall, S. (1978), Patterns of Attachment: A Psychological Study of the Strange Situation, Erlbaum, Hillsdale, NJ.

Aquino, K. (2000), “Structural and individual determinants of workplace victimization: the effects of hierarchical status and conflict management style”, Journal of Management, Vol. 26 No. 2, pp. 171-193.

Aryee, S., Sun, L.Y., Chen, Z.X.G. and Debrah, Y.A. (2008), “Abusive supervision and contextual performance: the mediating role of emotional exhaustion and the moderating role of work unit structure”, Management and Organization Review, Vol. 4 No. 3, pp. 393-411.

Bandura, A. (1997), Self-Efficacy: The Exercise of Control, Freeman, New York, NY.

Bandura, A. (2001), “Social cognitive theory: an agentic perspective”, Annual Review of Psychology, Vol. 52 No. 1, pp. 1-26.

Bandura, A., Caprara, G.V., Barbaranelli, C., Pastorelli, C. and Regalia, C. (2001), “Sociocognitive self-regulatory mechanisms governing transgressive behavior”, Journal of Personality and Social Psychology, Vol. 80 No. 1, pp. 125-135.

Baron, R.A., Neuman, J.H. and Geddes, D. (1999), “Social and personal determinants of workplace aggression: evidence for the impact of perceived injustice and the type A behavior pattern”, Aggressive Behavior, Vol. 25 No. 4, pp. 281-296.

Bass, B.M. (1990), “From transactional to transformational leadership: learning to share the vision”, Organizational Dynamics, Vol. 18 No. 3, pp. 19-31.

Bass, B.M. and Riggio, R.E. (2006), Transformational Leadership, 2nd ed., Lawrence Erlbaum Associates Publishers, Mahwah, NJ.

Becker, T.E. (2005), “Potential problems in the statistical control of variables in organizational research: a qualitative analysis with recommendations”, Organizational Research Methods, Vol. 8 No. 3, pp. 274-289.

Becker, T.E., Atinc, G., Breaugh, J.A., Carlson, K.D., Edwards, J.R. and Spector, P.E. (2016), “Statistical control in correlational studies: 10 essential recommendations for organizational researchers”, Journal of Organizational Behavior, Vol. 37 No. 2, pp. 157-167.

Bentler, P.M. (2005), Structural Equations Program Manual, Multivariate Software Inc., Encino, CA.

Bernerth, J.B. and Aguinis, H. (2016), “A critical review and best‐practice recommendations for control variable usage”, Personnel Psychology, Vol. 69 No. 1, pp. 229-283.

Bohlin, G., Hagekull, B. and Rydell, A.M. (2000), “Attachment and social functioning: a longitudinal study from infancy to middle childhood”, Social Development, Vol. 9 No. 1, pp. 24-39.

Bowlby, J. (1982), Attachment and Loss Attachment, 2nd ed., Vol. 1 No. 1, Basic Books, New York, NY.

Brennan, K.A., Clark, C.L. and Shaver, P.R. (1998), “Self-report measurement of adult attachment: an integrative overview”, in Simpson, J.A. and Rholes, W.S. (Eds), Attachment Theory and Close Relationships, The Guilford Press, New York, NY, pp. 46-76.

Bretherton, I. (1985), “Attachment theory: retrospect and prospect”, Monographs of the Society for Research in Child Development, Vol. 50 Nos 1/2, pp. 3-35.

Burton, J.P., Hoobler, J.M. and Scheuer, M.L. (2012), “Supervisor workplace stress and abusive supervision: the buffering effect of exercise”, Journal of Business and Psychology, Vol. 27 No. 3, pp. 271-279.

Byrne, A., Dionisi, A., Barling, J., Akers, A., Robertson, J., Lys, R., Wylie, J. and Dupré, K. (2014), “The depleted leader: the influence of leaders’ diminished psychological resources on leadership behaviors”, Leadership Quarterly, Vol. 25 No. 2, pp. 344-357.

Cassidy, J. (2000), “Adult romantic attachments: a developmental perspective on individual differences”, Review of General Psychology, Vol. 4 No. 2, pp. 111-131.

Collins, N.L. (1996), “Working models of attachment: implications for explanation, emotion and Behavior”, Journal of Personality and Social Psychology, Vol. 71 No. 4, pp. 810-832.

Collins, N.L. and Read, S.J. (1990), “Adult attachment, working models, and relationship quality in dating couples”, Journal of Personality and Social Psychology, Vol. 58 No. 4, pp. 644-663.

Collins, N.L. and Read, S.J. (1994), “Cognitive representations of attachment: the structure and function of working models”, in Bartholomew, K. and Perlman, D. (Eds), Attachment Processes in Adulthood, Jessica Kingsley, London, pp. 53-92.

Corcoran, K.O.C. and Mallinckrodt, B. (2000), “Adult attachment, self‐efficacy, perspective taking, and conflict resolution”, Journal of Counseling & Development, Vol. 78 No. 4, pp. 473-483.

Cristobal, E., Flavián, C. and Guinaliu, M. (2007), “Perceived e-service quality (PeSQ) measurement validation and effects on consumer satisfaction and web site loyalty”, Managing Service Quality: An International Journal, Vol. 17 No. 3, pp. 317-340.

Davidovitz, R., Mikulincer, M., Shaver, P.R., Izsak, R. and Popper, M. (2007), “Leaders as attachment figures: leaders’ attachment orientations predict leadership-related mental representations and followers’ performance and mental health”, Journal of Personality and Social Psychology, Vol. 93 No. 4, pp. 632-650.

Davila, J., Bradbury, T.N. and Fincham, F. (1998), “Negative affectivity as a mediator of the association between adult attachment and marital satisfaction”, Personal Relationships, Vol. 5 No. 4, pp. 467-484.

Deng, H., Walter, F., Lam, C.K. and Zhao, H.H. (2017), “Spillover effects of emotional labor in customer service encounters toward coworker harming: a resource depletion perspective”, Personnel Psychology, Vol. 70 No. 2, pp. 469-502.

DeRosier, M.E. (2004), “Building relationships and combating bullying: effectiveness of a school-based social skills group intervention”, Journal of Clinical Child and Adolescent Psychology, Vol. 33 No. 1, pp. 196-201.

DeSimone, J.A., Harms, P.D. and DeSimone, A.J. (2015), “Best practice recommendations for data screening”, Journal of Organizational Behavior, Vol. 36 No. 2, pp. 171-181.

DiTommaso, E., Brennen-McNulty, C., Ross, L. and Burgess, M. (2003), “Attachment styles, social skills, and loneliness in young adults”, Personality and Individual Differences, Vol. 35 No. 2, pp. 303-312.

Doverspike, D., Hollis, L., Justice, A. and Polomsky, M. (1997), “Correlations between leadership styles as measured by the least preferred co-worker scale and adults’ attachment styles”, Psychological Reports, Vol. 81 No. 3, pp. 1148-1150.

Eng, W., Heimberg, R.G., Hart, T.A., Schneier, F.R. and Liebowitz, M.R. (2001), “Attachment in individuals with social anxiety disorder: the relationship among adult attachment styles, social anxiety, and depression”, Emotion, Vol. 1 No. 4, pp. 365-380.

Erdley, C.A. and Asher, S.R. (1996), “Children’s social goals and self‐efficacy perceptions as influences on their responses to ambiguous provocation”, Child Development, Vol. 67 No. 4, pp. 1329-1344.

Feeney, J.A. (2008), “Adult romantic attachment: developments in the study of couple relationships”, in Cassidy, J. and Shaver, P.R. (Eds), Handbook of Attachment: Theory, Research and Clinical Applications, Guilford, New York, NY, pp. 456-481.

Garcia, P.R.J.M., Restubog, S.L.D., Kiewitz, C., Scott, K.L. and Tang, R.L. (2014), “Roots run deep: investigating psychological mechanisms between history of family aggression and abusive supervision”, Journal of Applied Psychology, Vol. 99 No. 5, pp. 883-897.

Gecas, V. (1989), “The social psychology of self-efficacy”, Annual Review of Sociology, Vol. 15, pp. 291-316.

Graen, G. and Uhl-Bien, M. (1995), “Relationship-based approach to leadership: development of leader-member exchange (LMX) theory of leadership over 25 years: applying a multi-level multi-domain perspective”, The Leadership Quarterly, Vol. 6 No. 2, pp. 219-247.

Guerrero, L.K. (1996), “Attachment-style differences in intimacy and involvement: a test of the four-category model”, Communication Monographs, Vol. 63 No. 4, pp. 269-292.

Harvey, P., Stoner, J., Hochwarter, W. and Kacmar, C. (2007), “Coping with abusive bosses: the neutralizing effects of ingratiation and positive affect on negative employee outcomes”, The Leadership Quarterly, Vol. 18 No. 3, pp. 264-280.

Hazen, C. and Shaver, P. (1987), “Romantic love conceptualized as an attachment process”, Journal of Personality and Social Psychology, Vol. 52 No. 3, pp. 511-524.

Hu, L.T. and Bentler, P.M. (1999), “Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives”, Structural Equation Modeling: A Multidisciplinary Journal, Vol. 6 No. 1, pp. 1-55.

Kiazad, K., Restubog, S.L.D., Zagenczyk, T.J., Kiewitz, C. and Tang, R.L. (2010), “In pursuit of power: the role of authoritarian leadership in the relationship between supervisors’ Machiavellianism and subordinates’ perceptions of abusive supervisory behavior”, Journal of Research in Personality, Vol. 44 No. 4, pp. 512-519.

Kiewitz, C., Restubog, S.L.D., Zagenczyk, T.J., Scott, K.D., Garcia, P.R.J.M. and Tang, R.L. (2012), “Sins of the parents: self-control as a buffer between supervisors’ previous experience of family undermining and subordinates’ perceptions of abusive supervision”, The Leadership Quarterly, Vol. 23 No. 5, pp. 869-882.

Kurdek, L.A. (2002), “On being insecure about the assessment of attachment styles”, Journal of Social and Personal Relationships, Vol. 19 No. 6, pp. 811-834.

Luthans, F., Peterson, S.J. and Ibrayeva, E. (1998), “The potential for the ‘dark side’ of leadership in post-communist countries”, Journal of World Business, Vol. 33 No. 2, pp. 185-201.

MacCallum, R.C., Browne, M.W. and Sugawara, H.M. (1996), “Power analysis and determination of sample size for covariance structure modeling”, Psychological Methods, Vol. 1 No. 2, pp. 130-149.

Mackey, J.D., Frieder, R.E., Brees, J.R. and Martinko, M.J. (2017), “Abusive supervision a meta-analysis and empirical review”, Journal of Management, Vol. 43 No. 6, pp. 1940-1965, doi: 0149206315573997.

Mallinckrodt, B. (2000), “Attachment, social competencies, social support and interpersonal process in psychotherapy”, Psychotherapy Research, Vol. 10 No. 3, pp. 239-266.

Mallinckrodt, B. and Wei, M. (2005), “Attachment, social competencies, social support, and psychological distress”, Journal of Counseling Psychology, Vol. 52 No. 3, pp. 358-367.

Mawritz, M.B., Mayer, D.M., Hoobler, J.M., Wayne, S.J. and Marinova, S.V. (2012), “A trickle-down model of abusive supervision”, Personnel Psychology, Vol. 65 No. 2, pp. 325-357.

Meurs, J.A. and Perrewé, P.L. (2011), “Cognitive activation theory of stress: an integrative theoretical approach to work stress”, Journal of Management, Vol. 37 No. 4, pp. 1043-1068.

Mikulincer, M. and Shaver, P.R. (2016), Attachment in Adulthood: Structure Dynamics and Change, 2nd ed., Guildford Press, New York, NY.

Peng, A.C., Schaubroeck, J.M. and Li, Y. (2014), “Social exchange implications of own and coworkers’ experiences of supervisory abuse”, Academy of Management Journal, Vol. 57 No. 5, pp. 1385-1405.

Pietromonaco, P.R. and Barrett, L.F. (2000), “The internal working models concept: what do we really know about the self in relation to others?”, Review of General Psychology, Vol. 4 No. 2, pp. 155-175.

Podsakoff, P.M., MacKenzie, S.B. and Podsakoff, N.P. (2012), “Sources of method bias in social science research and recommendations on how to control it”, Annual Review of Psychology, Vol. 63 No. 1, pp. 539-569.

Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y. and Podsakoff, N.P. (2003), “Common method biases in behavioral research: a critical review of the literature and recommended remedies”, Journal of Applied Psychology, Vol. 88 No. 5, pp. 879-903.

Popper, M. (2002), “Narcissism and attachment patterns of personalized and socialized charismatic leaders”, Journal of Social and Personal Relationships, Vol. 19 No. 6, pp. 797-809.

Popper, M., Mayselss, O. and Castelnovo, O. (2000), “Transformational leadership and attachment”, Leadership Quarterly, Vol. 11 No. 2, pp. 267-289.

Ravitz, P., Maunder, R., Hunter, J., Sthankiya, B. and Lancee, W. (2010), “Adult attachment measures: a 25-year review”, Journal of Psychosomatic Research, Vol. 69 No. 4, pp. 419-432.

Richards, D.A. and Hackett, R.D. (2012), “Attachment and emotion regulation: compensatory interactions and leader-member exchange”, The Leadership Quarterly, Vol. 23 No. 4, pp. 686-701.

Rogelberg, S.G. and Stanton, J.M. (2007), “Introduction: understanding and dealing with organizational survey nonresponse”, Organizational Research Methods, Vol. 10 No. 2, pp. 195-209.

Schat, A.C., Frone, M.R. and Kelloway, E.K. (2006), “Prevalence of workplace aggression in the US workforce: findings from a national study”, in Kelloway, E.K., Barling, J. and Hurrell, J.J. Jr (Eds), Handbook of Workplace Violence, Sage Publications Inc., Thousand Oaks, CA, pp. 47-89.

Schyns, B. and Schilling, J. (2013), “How bad are the effects of bad leaders? A meta-analysis of destructive leadership and its outcomes”, The Leadership Quarterly, Vol. 24 No. 1, pp. 138-158.

Shaver, P.R. and Mikulincer, M. (2007), “Adult attachment strategies and the regulation of emotion”, in Gross, J.J. (Ed.), Handbook of Emotion Regulation, Guildford Press, New York, NY, pp. 446-465.

Sherer, M., Maddux, J.E., Mercandante, B., Prentice-Dunn, S., Jacobs, B. and Rogers, R.W. (1982), “The self-efficacy scale: construction and validation”, Psychological Reports, Vol. 51 No. 2, pp. 663-671.

Tepper, B.J. (2000), “Consequences of abusive supervision”, Academy of Management Journal, Vol. 43 No. 2, pp. 178-190.

Tepper, B.J. (2007), “Abusive supervision in work organizations: review, synthesis, and research agenda”, Journal of Management, Vol. 33 No. 3, pp. 261-289.

Tepper, B.J., Simon, L. and Park, H.M. (2017), “Abusive supervision”, Annual Review of Organizational Psychology and Organizational Behavior, Vol. 4 No. 1, pp. 123-152.

Tepper, B.J., Carr, J.C., Breaux, D.M., Geider, S., Hu, C. and Hua, W. (2009), “Abusive supervision, intentions to quit, and employees’ workplace deviance: a power/dependence analysis”, Organizational Behavior and Human Decision Processes, Vol. 109 No. 2, pp. 156-167.

Thau, S., Bennett, R.J., Mitchell, M.S. and Marrs, M.B. (2009), “How management style moderates the relationship between abusive supervision and workplace deviance: an uncertainty management theory perspective”, Organizational Behavior and Human Decision Processes, Vol. 108 No. 1, pp. 79-92.

Tronick, E.Z. (1989), “Emotions and emotional communication in infants”, American Psychologist, Vol. 44 No. 2, pp. 112-119.

Uhl-Bien, M. (2006), “Relational leadership theory: exploring the social processes of leadership and organizing”, The Leadership Quarterly, Vol. 17 No. 6, pp. 654-676.

Wang, M., Liao, H., Zhan, Y. and Shi, J. (2011), “Daily customer mistreatment and employee sabotage against customers: examining emotion and resource perspectives”, Academy of Management Journal, Vol. 54 No. 2, pp. 312-334.

West, S.G., Finch, J.F. and Curran, P.J. (1995), “Structural equation models with non-normal variables: problems and remedies”, in Hoyle, R.H. (Ed.), Structural Equation Modeling: Concepts, Issues and Applications, Sage, Thousand Oaks, CA, pp. 56-75.

Wieland, A., Durach, C.F., Kembro, J. and Treiblmaier, H. (2017), “Statistical and judgmental criteria for scale purification”, Supply Chain Management: An International Journal, Vol. 22 No. 4, pp. 321-328.

Wu, C.H. and Parker, S.K. (2012), “The role of attachment styles in shaping proactive behaviour: an intra-individual analysis”, Journal of Occupational and Organizational Psychology, Vol. 85 No. 3, pp. 523-530.

Wu, C.H. and Parker, S.K. (2017), “The role of leader support in facilitating proactive work behavior: a perspective from attachment theory”, Journal of Management, Vol. 43 No. 4, pp. 1025-1049.

Yagil, D. (2006), “The relationship of abusive and supportive workplace supervision to employee burnout and upward influence tactics”, Journal of Emotional Abuse, Vol. 6 No. 1, pp. 49-65.

Yip, J., Ehrhardt, K., Black, H. and Walker, D.O. (2017), “Attachment theory at work: a review and directions for future research”, Journal of Organizational Behavior, Vol. 39 No. 2, pp. 185-198, doi: 10.1002/job.2204.


This work was supported by a grant from the Social Science and Humanities Research Council of Canada (Grant No. 435 2012 1737).

Corresponding author

Jennifer L. Robertson can be contacted at: