Introduction

Abusive supervision is a costly social problem for corporations because of subsequent employee turnover, psychological distress, low performance, deviance, productivity losses and health care costs (Zhang et al. 2019; Mackey et al. 2015; Tepper 2007). It is an active form of destructive leadership and has received considerable attention from researchers and practitioners in the last two decades (Fosse et al. 2019; Tepper et al. 2017). Abusive supervision is defined as “subordinates’ perceptions of the extent to which their supervisors engage in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact” (Tepper 2000, p. 178). Examples of such behaviors include belittling, treating rudely, humiliating or ridiculing in front of others, aggressive outbursts, managing with threat, withholding needed information, invading privacy, among others (Aryee et al. 2007; Tepper 2000, 2007).

Job embeddedness has been advanced as a central construct to comprehend why employees stay in their jobs and give the practice a new standpoint to enhance employee retention (Kiazad et al. 2015). To the best of our knowledge, specifically, the relationship between abusive supervision and job embeddedness has not been previously examined. Two recent studies analyzed the moderating role of job embeddedness between abusive supervision and various employee outcomes (Allen et al. 2016; Avey et al. 2015). In addition, regarding a similar factor, Erkutlu and Chafra (2017) reported that leader narcissism has a negative relationship with job embeddedness. This situation is somewhat surprising because abusive supervision has been suggested as a threat to various factors that will potentially enhance job embeddedness. Some of these factors are leader-member exchange (Xu et al. 2012), justice perceptions (Zhang et al. 2019), organization-based self-esteem (Haggard and Park 2018), psychological contract (Kernan et al. 2016) and need-satisfaction (Lian et al. 2012). The present study, therefore, sets to examine the abusive supervision-job embeddedness relationship by focusing on a specific psychological mechanism that will mediate this relationship.

Recent empirical findings indicate that perceived organizational support (POS) will serve as such a mechanism between abusive supervision and job embeddedness. According to organizational support theory, POS carries out an essential role in the employee-organization relationship and offers considerable implications to enhance employee wellbeing and positive orientation towards the organization (Eisenberger et al. 2019; Kurtessis et al. 2015). Moreover, prior research has provided evidence that while abusive supervision is negatively associated with POS (Xu et al. 2018), POS is positively related to job embeddedness (Nguyen et al. 2017; Akgunduz and Sanli 2017). Based on the existence of such relations in the workplace, this study proposes that POS will be an intervening variable transmitting the negative effects of abusive supervision on job embeddedness. This proposition is examined by a research model depicted in Fig. 1. The model is grounded in a theoretical framework drawing on organizational support, organizational justice, social exchange and conservation of resources theories. Abusive supervision has been related to employees’ injustice perceptions, low quality social exchanges and depletion of resources (Zhang et al. 2019). Resource abundance is associated with job embeddedness (Kiazad et al. 2015). As a result, we applied such a framework with multiple theories related to the constructs in question to provide a better insight into the mediating role of POS posited our study.

Fig. 1
figure 1

Research model

Full size image

This study has several intended contributions: First, it adds the emerging theory of research as to why and to what extent abusive supervision influences job embeddedness. Second, it sheds light on the relationship between abusive supervision and job embeddedness by examining the mediating role of POS in this relationship. Third, prior work has been primarily interested in the influence of inherently favorable factors to job embeddedness, including leader-member exchange (Harris et al. 2011), socialization tactics (Allen and Shanock 2013) and organizational identification (Johnson et al. 2010). Research on the influence of negative organizational factors on embeddedness, however, is limited (Holtom et al. 2012; Karatepe 2013). This study addresses this gap by investigating the influence of a form of destructive leadership like abusive supervision on job embeddedness. Finally, with a sample from Turkey, this study answers the research calls made by various researchers for the investigation of abusive supervision and job embeddedness constructs in non-Western contexts (Zhang et al. 2012; Martinko et al. 2013; Peltokorpi et al. 2015).

Abusive Supervision and Perceived Organizational Support

Perceived organizational support (POS) refers to employees’ “general beliefs concerning how much the organization values their contributions and cares about their well-being” (Rhoades et al. 2001, p.825). POS is enhanced by discretionary, sincere and beneficial treatments directed from the organization and/or its agents like supervisors to employees by feeding their perceptions on the organization’s commitment to them (Rhoades et al. 2001; Eisenberger et al. 1986; Armeli et al. 1998; Settoon et al. 1996). Supervisors’ role in this enhancement lies in the fact that the organization directs and evaluates individual employees through them (Shoss et al. 2013; Eisenberger et al. 1986). Supervisors act as interpretive filters for organizational policies and processes and have control over resource allocation (e.g. perks, projects, promotions or expertise) in the workplace (Lian et al. 2012; Mayer et al. 2010; Rhoades and Eisenberger 2002). Moreover, supervisors are like role models and a source of guidance, support and inspiration for their subordinates (Erkutlu and Chafra 2017; Harris et al. 2011). Consequently, a supervisor’s positive behaviors towards employees may increase POS to the degree that such behaviors are attributed to the organization’s own disposition rather than the supervisor’s personal motives (Rhoades et al. 2001). We believe the same logic will apply for negative supervisor behaviors as in abusive supervision, but with an opposite effect on POS.

Employees value favorable treatment in their relationship with the organization (Coyle-Shapiro and Shore 2007; Eisenberger et al. 2001). More specifically, they look to praise, approval, care, and rewards for increased effort (Valle et al. 2019; Eisenberger et al. 2001). Furthermore, except few contexts wherein a supervisor’s undesirable behaviors might be tolerated like military or sport, they expect others, especially supervisors, to refrain from actions threatening their self- and social image (Tepper 2000, 2007). Abusive supervision, conversely, offers employees sustained hostility by supervisors, which will potentially harm their orientation towards the organization and lead a dislike for their jobs (Kernan et al. 2016; Tepper 2000). As such, research has associated abusive supervision with reduced affective commitment (Caesens et al. 2019; Yu et al. 2016), lower organizational citizenship behaviors (Zhang et al. 2019) and increased organization-directed deviance (Valle et al. 2019; Shoss et al. 2013; Bowling and Michel 2011; Mitchell and Ambrose 2007). This study, therefore, proposes that abusive supervision will be in a negative relationship with POS. We draw on organizational support and organizational justice theories to provide a theoretical background for this relationship. Organizational support theory mainly helps clarify why employees blame the organization for abusive supervision. Organizational justice theory sheds further light on their underlying feelings behind this blame.

According to organizational support theory, POS is encouraged by employees’ propensity for “anthropomorphic ascription of dispositional traits to the organization” (Eisenberger et al. 1986, p. 500). That is, employees tend to personify or attribute human-like characteristics to the organization, and view actions of its agents as the organization’s own actions (Rhoades and Eisenberger 2002; Eisenberger et al. 1986). Based on this personification of the organization, moreover, employees may regard the organization’s treatments as an indication of the organization’s supportive or malevolent orientation towards them (Aselage and Eisenberger 2003; Rhoades et al. 2001). From this point of view, we suggest that abusive supervision will damage employees’ POS. This is because, on exposure to it, employees are more likely to feel that not only the supervisor but also the organization mistreats them (Lian et al. 2012). As a result, due to the supervisor’s unfavorable treatments attributed to the organization (Bowling and Michel 2011), they would exhibit low levels of POS. Furthermore, in the eyes of employees, the personified organization is required to cope with abusive supervision by enforcing effective policies, punishing perpetrators, and help victims recover (Tepper et al. 2008; Aryee et al. 2007). Despite that, if the organization fails to adopt the necessary stand against abusive supervision, employees may consider this as an additional signal for low levels of organizational support.

Furthermore, as Rupp (2011) noted, organizational justice may carry out a critical role in the degree to which employees feel valued by their organizations. In parallel, related meta-analyses have revealed that justice is the strongest positive indicator of POS (Kurtessis et al. 2015; Rhoades and Eisenberger 2002). Conversely, previous studies have demonstrated that abusive supervision may damage employees’ organizational justice perceptions and the quality of social exchanges with the organization (Zhang et al. 2019; Park et al. 2017; Lian et al. 2012). As such, research has operationalized employees’ organizational injustice perceptions as a chief mediator of the negative effects of abusive supervision (Wang and Jiang 2015; Aryee et al. 2007; Zellars et al. 2002; Tepper 2000). As a result, taken together with the two constructs’ close but inverse relationship with organizational justice, we propose that abusive supervision will decrease employees’ POS by damaging their justice perceptions. That is, when employees subject to abusive supervision, they are more likely to experience organizational injustice. Next, due to this experience, they would feel that the organization does not value their contributions or care about their circumstances (Tepper 2000). Thus, they would display reduced POS. Considering the dimensions of organizational justice, specifically, such feelings can be argued to arise from employees’ perceptions of interactional injustice (e.g. supervisor’s disrespectful, offensive, and ridiculing behaviors), procedural injustice (e.g. insufficiency or lack of organizational policies and actions to cope with abusers), and/or distributive injustice (e.g. supervisor unfairness in resource allocation, including time, information, expertise, etc.) (Zhang and Liao 2015; Aryee et al. 2007; Tepper 2000).

Consequently, this study reached the following hypothesis.

  • Hypothesis 1: Abusive supervision is negatively related to POS.

Perceived Organizational Support and Job Embeddedness

Job embeddedness is defined as “the combined forces that keep a person from leaving his/her job” (Yao et al. 2004, p.159). It represents a state of inertia or stuckness about employee job change propensities. According to the multi foci view of Kiazad et al. (2015), such a state of inertia is formed by various organizational, occupational and community-based factors that will encourage one to remain with the organization. In the original conceptualization and operationalization of job embeddedness, advanced by Mitchell et al. (2001), these factors were analyzed in three dimensions: links, fit, sacrifice. Links refer to connections to other people, groups, the organization or other institutions. Fit refers to compatibility or comfort with the job, the organization, community and surrounding environment. Sacrifice refers to the cost of any benefits that may be forfeited by quitting a job. Numerous links, strong fit, and/or large sacrifices are assumed to increase one’s embeddedness and prevents him/her from quitting a job (Allen et al. 2016).

The original composite model of job embeddedness was criticized by researchers that it demonstrates serious limitations (Singh et al. 2018; Zhang et al. 2012; Yang et al. 2011; Crossley et al. 2007). For that reason, a global job embeddedness construct was postulated by Crossley et al. (2007). The global construct is distinct from the dimensions of the composite model but still has strong correlations with them (Singh et al. 2018). In addition, it models job embeddedness by addressing one’s general attachment to the organization. More specifically, Crossley et al. (2007) argued that the global construct assesses employees’ general feelings and impressions about their organizational attachment, which are formed after some sort of mental processing. This study also embraces the same approach, and, in line with Singh et al. (2018), considers job embeddedness as a general attachment construct.

Perceived organizational support (POS) is composed of the employees’ perception regarding the supportiveness of the organization’s policies, norms, procedures and actions as they influence them (Eisenberger et al. 2001). Accordingly, POS may inform employees about the organization’s commitment to them and willingness to reward their contributions (e.g. perks or promotion) (Chen et al. 2009; Rhoades and Eisenberger 2002). In addition, it is considered as a source of acceptance and belonging within the organization (Ferris et al. 2009; Rhoades and Eisenberger 2002). POS may also meet employees’ socioemotional needs, including approval, respect, esteem and emotional support (Armeli et al. 1998). Moreover, it may strengthen employees’ affective attachment to the organization (Eisenberger et al. 1986). Employees with high POS, therefore, can be argued to possess favorable feelings about the organization’s orientation towards them, their work environment, and the fulfillment of their needs by the organization (Kurtessis et al. 2015). We suggest that such feelings would also steer them to develop deeper bonds with the organization, better match with the work environment, a higher sense of sacrifice, and thus stronger attachment to the organization (Riggle et al. 2009; Eisenberger et al. 2001; Kurtessis et al. 2015). As a result, POS should be in a positive relationship with job embeddedness. To shed more light on the theoretical background of this relationship, we draw from social exchange and conservation of resources (COR) theories.

POS is a central construct to the social exchange view of (considering the organization as the employer) the employee-organization relationship (Eisenberger et al. 2019). Accordingly, this relationship is a trade of effort and loyalty for material (e.g. pay and perks) and socioemotional resources (e.g. esteem or approval) (Cropanzano and Mitchell 2005; Eisenberger et al. 1986; Blau 1964). POS basically reflects employees’ general judges regarding the quality of this relationship (Settoon et al. 1996). High levels of POS, therefore, indicate a high-quality relationship in which the organization fulfills its exchange obligations to the employee and treat well him or her. Consequently, employees with high POS, based on the norm reciprocity (Cropanzano and Mitchell 2005; Gouldner 1960) are expected to treat the organization in the same manner.

The reciprocity norm leads employees to develop a felt obligation to reciprocate the organization contingent upon the treatment they receive (Cropanzano and Mitchell 2005; Blau 1964). Therefore, while employees tend to positively respond to the organization’s favorable treatments, (Settoon et al. 1996; Baran et al. 2012), they tend to negatively respond to its unfavorable treatments (Lian et al. 2012; Eisenberger et al. 2004). POS is argued to elicit such a felt obligation causing employees to respond to the organization in positive and beneficial ways (Baran et al. 2012). Empirical evidence also exists supporting this argument (Coyle-Shapiro et al. 2006; Rhoades et al. 2001; Kurtessis et al. 2015). Thus, we suggest that employees will reciprocate high POS by increased job embeddedness. That is, when they perceive elevated levels of supportive treatments from the organization, employees are more likely to feel obliged to remain with the organization to repay those treatments (Wikhamn and Hall 2012; Maertz Jr et al. 2007; Eisenberger et al. 1990, 2001; Rhoades et al. 2001; Shore and Wayne 1993). In this way, they would become more embedded in their jobs.

COR theory contributes as well to our understanding of the POS-job embeddedness relationship. The basic principle of COR theory holds that individuals are in a struggle to acquire, protect, and foster those things or resources they value (Hobfoll et al. 2018). The primary cause of such a struggle is that resources help individuals respond to external demands, attain valued goals, and defend from possible losses in the future (Kiazad et al. 2015; Singh et al. 2018). In addition, resource acquisitions and abundance may cause individuals to generate favorable outcomes and have better wellbeing and adaptation (Singh et al. 2018; Harris et al. 2011; Hobfoll 1989, 2002). In line with this reasoning, given the role of POS in the fulfillment of employees’ material and socioemotional resources (Rhoades and Eisenberger 2002), we suggest that high POS will encourage them to stay in the organizations, which would, in turn, increase their job embeddedness. It is because by staying, employees may secure their current resources provided by POS and the resulting favorable outcomes (Kiazad et al. 2015). In addition, they may even have the chance to develop resource surpluses to counterbalance the possibility of future losses (Hobfoll 1989). Moreover, POS signifies the organization’s trustworthiness about fulfilling its exchange obligations to employees (Settoon et al. 1996; Eisenberger et al. 1990). Further, it gives an idea about whether the organization will help them in stressful situations and effective fulfillment of duties (Rhoades and Eisenberger 2002; Armeli et al. 1998). High POS, therefore, has also the potential to signal employees that the organization is not only a reliable basis for current needs but also trustworthy for future rewards and comfort. Consistent with COR theory, such signals suggest POS as a resource signal (Halbesleben et al. 2014) giving information about the organization’s reliability for protecting resources at hand and gaining more in the future, thereby remaining with.

In line with our expectations, moreover, prior research has associated POS with increased organizational attachment (e.g. affective commitment, desire to remain, organizational identification, and lower turnover intention and actual turnover) and improved in-role and extra-role performance (Kurtessis et al. 2015; Allen et al. 2003; Rhoades and Eisenberger 2002; Chen et al. 2009; Ferris et al. 2009; Coyle-Shapiro et al. 2006). Consequently, this study reached the following hypothesis;

  • Hypothesis 2: POS is positively related to job embeddedness.

POS as a Mediator of the Abusive Supervision-Job Embeddedness Relationship

As previously outlined in this study, we propose that the negative effect of abusive supervision on job embeddedness will be indirectly transmitted through POS. That is, abusive supervision would reduce employees’ POS, which would subsequently reduce their job embeddedness.

POS reflects employees’ general beliefs concerning the quality of their exchange relationships with the organization (Settoon et al. 1996). Reduced POS due to abusive supervision, therefore, points out a poor-quality and unbalanced social exchange relationship between the employee and the organization (Eisenberger et al. 2004). This is because, the personified organization has failed to fulfill some of its obligations to the employee, including favorable treatment, justice, and coping with abusers (Coyle-Shapiro and Shore 2007; Eisenberger et al. 2001). Thus, based on the norm of reciprocity (Gouldner 1960), such an employee would no longer feel obliged to reciprocate the organization favorably or beneficially. Instead, to rebalance the relationship, he/she is more likely to negatively reciprocate the organization by unfavorable ways such as reducing his/her loyalty and commitment (Rafferty and Restubog 2011; Mitchell and Ambrose 2007; Eisenberger et al. 2004; Gouldner 1960). In line with this reasoning, we suggest that employees with reduced POS will negatively reciprocate the organization by lowering their embeddedness. That is, because of the organization’s negative treatments toward them, they would no longer feel obliged to stay with the organization. Alternatively, for retaliation, they are more likely to consider withholding their obligations to the organization (e.g. support or performance) or even leaving it (Zhang and Liao 2015; Aryee et al. 2007). In either way, they would feel less attached to the organization, which would, in turn, be less embedded in their jobs.

From a different but complementary perspective, moreover, a poor-quality employee-organization relationship can be argued to end up an insufficiency or lack regarding the valuable resources provided by the organization. It would not be surprising because poor exchanges with the organization will lead to poor gains for employees (Harris et al. 2011). For that reason, reduced POS signifies exhaustion related to these resources, namely organizational resources such as perks, praise, care, and approval. At first glance, such organizational resource exhaustion is contrary to the basic principle of COR theory (Hobfoll et al. 2018). In addition, it threatens employees’ comforts related to the organizational resource abundance, including better wellbeing, increased adaptation and counterbalancing the future resource losses (Hobfoll 1989, 2002). Moreover, as noted before, POS is related to a trust on which the organization fulfills its obligations to employees (Settoon et al. 1996) and helps them in need (Rhoades and Eisenberger 2002; Eisenberger et al. 2001; Armeli et al. 1998). Reduced POS, therefore, may also signal to employees that the organization is no longer trustworthy about protecting and fostering their resources.

As a result, we suggest that reduced POS will discourage employees from staying in the organization, which would, in turn, lessen their job embeddedness. It is because, based on COR theory, the organization no longer promises a fruitful environment for their resources to flourish and thus for themselves to stay. Additionally, as a resource-oriented theory of stress, COR theory posits that individuals experience stress when their resources are threatened, exhausted, or when they fail to acquire new resources following considerable effort (Hobfoll et al. 2018). Perceived and actual resource losses or failure to gain, meanwhile, are anticipated as enough for producing stress (Hobfoll 1989). Moreover, research has shown that employees may experience stress when they lose resources in the workplace (Halbesleben et al. 2014). Thus, considering the related organizational resource exhaustion, reduced POS can be argued to cause stress on employees, thereby weakening further their embeddedness. It is because stress is a major problem for employees leading them to experience various negative mental and physical outcomes, including anger, anxiety, depression, burnout, and emotional exhaustion (Zhang and Liao 2015; Tepper 2007; Hobfoll 1989, 2001). Such outcomes, specifically, may decrease employees’ capability to manage negative circumstances in the workplace and to contribute positively to the organization (Zhang et al. 2019; Zhang and Liao 2015; Aryee et al. 2008). As such, research has associated stress with decreased job satisfaction (Wright and Cropanzano 1998), lessened contextual performance (Aryee et al. 2008), increased counterproductive work behaviors and decreased citizenship behaviors (Zhang et al. 2019).

Consequently, this study proposes a previously untested mediation model in which the effect of abusive supervision on job embeddedness is conveyed through POS. Various studies have similarly suggested POS as an intervening mechanism linking abusive supervision to high levels of turnover intention (Xu et al. 2018; Haar et al. 2016), and to counterproductive work behaviors and low levels of in-role and extra-role performance (Shoss et al. 2013). Consequently, the following hypothesis was formulated.

  • Hypothesis 3: Perceived organizational support mediates the relationship between abusive supervision and job embeddedness.

Methodology

Sample and Procedure

Data were collected in three leading industrial cities of Turkey (İstanbul, Kocaeli, and Tekirdağ). Participants were full-time employees working for privately-owned organizations. Participation was voluntary, and all participants were assured of the anonymity of their responses. Employees with at least one-year tenure participated in the study. Thus, we tried to ensure all participants had adequate time in their organizations to have formed healthier ideas about their organizations, supervisors, and working environments.

Survey questionnaires were distributed in envelopes to organizations who had agreed to contribute to the study. The envelopes also contained a cover letter and a small gift. The cover letter was to explain the purpose of the study to participants; to instruct them not to put their names or any other forms of identification on their surveys; and to encourage them for honest responses as much as possible. Human resource departments of contributing organizations distributed the questionnaires to the participants. To further guarantee honesty and reduce respondents’ evaluation apprehension, we also asked organizations’ contact persons to stress on anonymity and honesty of responses. We also requested from them to emphasize the scientific value and purpose of the study.

Completed surveys were taken back in sealed envelopes. We distributed 1050 questionnaires and received 693 questionnaires. The response rate is 66%. After eliminating incomplete and invalid answers, 644 surveys were used for analysis. We sought to access a large sample from a broad range of sectors to strengthen the generalizability of our findings (Ng and Feldman 2013).

The participants worked in the following industry areas: manufacturing (31.4%), communication (14.9%), services (14.4%), transportation (11.2%), retail (8.5%), banking, finance and insurance (8.1%), construction (6.1%) and healthcare (5.4%) Among the participants, 57% worked for companies with more than 250 employees. In our final sample, the mean age of respondents was 32.22 years (range = 19–63 years), 59% were male, 54.7% were married, and the average tenure with organization was 5.03 years. In terms of their job positions, while most of the respondents were nonmanagerial employees (56.4%), the rest are middle managers (25.2%), first-line managers (14%) and top managers (4.5%).

Measures

Survey items were translated from English to Turkish by following Brislin’s (1980) translation–back-translation procedure to assure the quality of the translation from English to Turkish.

To establish face validity, three academics who have knowledge of organizational behavior checked and revised the translated survey. In addition, 15 employees with more than 5-year tenure were selected from different industries. They evaluated the content, meaningfulness and difficulty of the survey items. Then, the survey was fine-tuned in line with their opinions and suggestions.

Abusive supervision. This study used Mitchell and Ambrose’s (2007) five-item version of Tepper’s (2000) abusive supervision scale. This shortened version scale has acceptable reliability and validity (Mitchell and Ambrose 2007). A sample item is ‘My supervisor ridicules me’. Respondents completed the measures using a 5-point Likert scale (1 = “never” to 5 = “very often”). The reliability of the scale was high (α = .88).

Perceived organizational support. Perceived organizational support was measured with an eight-item scale developed by Eisenberger et al. (1986). All items selected have the highest factor loadings in the original scale. A sample item is “The organization really cares about my well-being.”. Respondents completed the measures using a 5-point Likert scale (1 = “strongly disagree” to 5 = “strongly agree”). The reliability of the scale was high (α = .92).

Job embeddedness. Job embeddedness was measured using six items from Crossley et al.’s (2007) global job embeddedness scale. A sample item is “It would be difficult for me to leave this organization.” In accordance with Ng and Feldman (2012), the item “I feel tied to this organization” was excluded. It is because this item had almost the same meaning as the item “I feel attached to this organization”. Respondents completed the measures using a 5-point Likert scale (1 = “strongly disagree” to 5 = “strongly agree”). The reliability of the scale was high (α = .92).

Control variables. To account for alternative explanations for the findings, the participants’ organizational tenure (measured in years) was controlled for because previous studies reported a strong positive relationship between embeddedness and tenure (Ng and Feldman 2009; Peltokorpi et al. 2015).

Gender, marital status and job position were not controlled because they were not correlated with any of our dependent variables (Becker 2005). Although age was significantly related to job embeddedness, we did not include it because it was highly related to organizational tenure (r = .58, p < .01). Moreover, the inclusion of both could result in biased path estimates (Gavino et al. 2012).

Common Method Variance Assessment

Since the data of this study were collected from a single source, the findings might be subject to common method variance (CMV). We tried to minimize CMV by following the recommendations offered by Podsakoff et al. (2003). At the questionnaire design, the dependent variable was placed after the independent and mediator variable. To obtain more accurate responses, we assured respondents that we would preserve their confidentiality (Podsakoff et al. 2003). And, we shortened the measures to minimize the respondent’s burden (Ng and Lucianetti 2018).

We assessed the influence of CMV by using the unmeasured latent variable approach (Podsakoff et al. 2003). A single latent method factor was added in the CFA model (Podsakoff et al. 2003). That is, all items could load on their theoretical constructs, and on a latent common method variance (CMV) factor. Next, the significance of the factor loadings and factor correlations observed both with and without the latent CMV factor in the model. CMV may influence the findings if any significant (or nonsignificant) changes occur in patterns of results after adding CMV factor (Ng and Feldman 2012). We found that the pattern of the relationships between the constructs was not affected after adding the CMV factor. All the factor loadings remained statistically significant and in the expected direction even after controlling for the effect of the CMV factor. These findings provided evidence that common method variance did not appear to be a problem in the data.

Descriptive Statistics

The means, standard deviations, and correlations among study variables are presented in Table 3. Results reported in Table 3 revealed that abusive supervision was negatively correlated with perceived organizational support (r = −0.40, p < .01.). In addition, perceived organizational support was positively related to job embeddedness (r = 0.63, p < .01.). Results of the correlation analysis also revealed that organizational tenure positively correlated with job embeddedness (r = 0.23, p < .01).

Measurement Model

We used structural equation modeling (SEM) with AMOS 24 to test our research hypotheses. Before testing the hypotheses, we checked to see if the measurement model provided a good fit to the data (Anderson and Gerbing 1988). The measurement model consisted of three latent variables: abusive supervision, POS, job embeddedness and 19 indicators (five items for abusive supervision, eight items for POS and six items for job embeddedness).

The fit of the measurement model was assessed based on the various fit indices recommended by Hu and Bentler (1998): Comparative fit index (CFI should be close to .95), the Tucker–Lewis index (TLI; should be close to .95), the root mean square error of approximation (RMSEA; should be close to .06) and the standardized root mean square residual (SRMR; should be close to .08).

The measurement model demonstrated an acceptable fit and all indices met the respective criteria, (χ2(148, N = 644) =513.29, p < .01 (χ2/df = 3.46; CFI = .95; TLI = .94; RMSEA = .06, SRMR = .04). As shown in Table 1, all items loaded on their respective constructs with (and all loadings are significant at 0.001) factor loadings ranging from 0.64 to 0.89.

Table 1 Confirmatory factor item loadings, construct reliability, and convergent validity
Full size table

To ascertain the distinctiveness of the three constructs, a series of confirmatory factor analyses were conducted. We compared the hypothesized three-factor measurement model with a nested alternative two-factor Model 1 (combining POS and job embeddedness), three-factor Model 2 (combining abusive supervision and perceived organizational support) and a one-factor model (combining all three constructs). Results reported in Table 2 revealed that the hypothesized three-factor model had the best fit. Moreover, the chi-square difference tests confirmed that this model fitted the data significantly (p < .01) better than each of the alternative nested models. The CFA results indicated support for the hypothesized three-factor model and, therefore, provided the construct distinctiveness of the three study variables.

Table 2 Comparison of measurement models
Full size table

Convergent and Discriminant Validity

Following the suggestions of Fornell and Larcker (1981), convergent validity and discriminant validity were assessed by computing the composite reliability (CR) and the average variance extracted (AVE). As shown in Table 1, composite reliabilities of the three latent variables: abusive supervision, POS and job embeddedness ranged from .88 to .92, while the AVE by these constructs ranged from .59 to .65. These results indicate that the CR value of each construct was greater than 0.7 (Bagozzi and Yi 1988). Moreover, the AVE value of each construct was higher than the recommended level of 0.5 (Fornell and Larcker 1981), indicating a good convergent validity.

We also computed the square root of AVE to establish the discriminant validity of the constructs. The discriminant validity is demonstrated if the square root of AVE of each construct exceeds the latent factor correlations between pairs of constructs (Fornell and Larcker 1981). Results presented in Table 3 show that the square root of AVE for each construct was greater than any of the inter-construct correlations, providing support for discriminant validity.

Table 3 Means, standard deviations, and correlations among study variables
Full size table

Hypotheses Testing

Figure 2 represents standardized path coefficients, path significances, and variance explained (R2) by each path for the hypothesized model. In line with Hypothesis 1 abusive supervision was negatively related to perceived organizational support (β = −.43, p < .01). Consistent with Hypothesis 2, perceived organizational support was positively related to job embeddedness (β = .66, p < .01). Finally, it was found that organizational tenure was significantly and positively related to job embeddedness (β = .24, p < .01).

Fig. 2
figure 2

The fully mediated structural equation model with standardized path coefficients (N = 644). **p < .01

Full size image

To test our mediation hypothesis (H3), we tested the relationship between abusive supervision and job embeddedness, leaving out POS. The results demonstrated that the observed covariance matrix provides a good fit into the model (χ2(51) = 164.26, CFI = .97, TLI = .96, χ2/df = 3.22, RMSEA = .05, SRMR = .03). That is, abusive supervision was significantly and negatively related to job embeddedness (β = −.27, p < .01).

We next tested the mediating role of POS in the relationship between abusive supervision and job embeddedness. As shown in Table 4, our proposed fully mediated model exhibited a good fit to the data (χ2(166) = 542.43, CFI = .95, TLI = .94, χ2/df = 3.26, RMSEA = .05, SRMR = .04). We compared this fully mediated model with a partially mediated model using chi-square difference tests. For the partially mediated model, we added a direct path between abusive supervision and job embeddedness. As indicated in Table 4, the partially mediated model also fit to the data well (χ2(165) = 542.33, CFI = .95, TLI = .94, χ2/df = 3.28, RMSEA = .06, SRMR = .04); however, this model does not offer an improvement in fit over the fully mediated model (chi-square difference test: Δ χ2 = 0.10, df = 1, ns). Additionally, abusive supervision was not significantly related to job embeddedness when testing the partially mediated model (β = 0.01, t = 0.32, ns), suggesting full mediation.

Table 4 Comparison of the structural models
Full size table

We also used the bootstrapping procedure recommended by Preacher and Hayes (2008), to further assess the mediation. Bootstrapping is a nonparametric resampling procedure “that does not impose the assumption of normality of the sampling distribution” (Preacher and Hayes 2008, p. 880). To estimate the significance of the mediation, 95% bias-corrected bootstrap confidence intervals (CI) were constructed (with n = 5000 bootstrap resamples) around the estimated indirect effect of abusive supervision through POS. The indirect effect is considered significant if zero does not fall within the confidence interval. Our results indicated that the 95% bias-corrected CIs of the indirect effect of abusive supervision (via POS) on job embeddedness was significant (effect = −0.42; 95% CI: lower limit CI −0.52; upper limit CI −0.34). This is because the zero did not fall between the lower and upper confidence intervals. Thus, hypothesis 3 was supported.

Discussion

In the present study, we developed and tested a research model that addresses the relationship between abusive supervision and job embeddedness. Specifically, we examined the mediating role of perceived organizational support (POS) in this relationship. Our examination draws on organizational support, social exchange, organizational justice, and conservation of resources (COR) theories. Moreover, by a sample from Turkey, this study set out to expand the knowledge about the generalizability abusive supervision and job embeddedness constructs.

Main Findings and Theoretical Implications

To the best of our knowledge, this is the first study linking abusive supervision to job embeddedness by modeling POS as a mediating mechanism. Therefore, the results of our study extended the nomological network of abusive supervision and its harmful impact by showing abusive supervision is negatively and indirectly related to job embeddedness through POS. Furthermore, our results shed more light on the abusive supervision-POS and POS-job embeddedness relationships which have received less attention relatively from researchers. Our main findings and related theoretical implications are as follows:

Abusive supervision was found to be negatively associated with POS, consistent with previous related research (Xu et al. 2018; Haar et al. 2016; Shoss et al. 2013). This result suggests that employees who are regularly exposed to non-physical hostile behaviors of supervisors hold the organization responsible for this situation. Moreover, these behaviors seem to harm their organizational justice perceptions (Zhang et al. 2019). In line with organizational support theory, these findings support the argument that employees tend to personify and attribute supervisor behaviors to the organization (Eisenberger et al. 2001; Rhoades and Eisenberger 2002). In addition, they do not consider the organization as a blameless onlooker at supervisor actions (Shoss et al. 2013). But then, different from the organizational support literature concentrating mainly on favorable supervisor behaviors, our finding supports these arguments for unfavorable supervisor behaviors (Xu et al. 2018). Furthermore, prior research has reported positive associations between favorable leadership, including supervisor support and constructive leadership and POS (Dawley et al. 2010; Kim et al. 2016). Our finding, however, indicates that, as a type of destructive leadership, abusive supervision is in an inverse relation with POS (Shoss et al. 2013).

POS was found to be positively associated with job embeddedness, in accordance with previous related studies (Nguyen et al. 2017; Akgunduz and Sanli 2017). This result proposes that employees’ positive evaluations about the extent of the organizational support they receive constitute a force encouraging them to stay with the organization. Thus, high levels of organizational support, reflected by POS, might be related to increased job embeddedness. Based on the norm of reciprocity (Gouldner 1960), this result points out that employees reciprocate high POS by increased job embeddedness. Research has shown that employees with high POS tend to repay the organization in various favorable ways and thus less likely to seek other employment (Eisenberger et al. 1990; Eisenberger et al. 2001). Our result confirms this argument and specifically suggests job embeddedness as a form of favorable repayment (Nguyen et al. 2017). From a COR theory perspective, this result indicates that employees with high POS choose to stay in the organization to keep enjoying the resource abundance provided by the organization and the resulting favorable outcomes (Hobfoll 1989, 2002). Moreover, a recent study has found that interpersonal forms of social support may have direct or indirect positive effects on employee embeddedness (Singh et al. 2018). Further, this study posited that resources obtained from coworkers are more important than the ones obtained from the organization in general. Our finding, however, emphasizes the importance of resources provided by the organization in the enhancement of one’s embeddedness.

As the major contribution, we found that POS fully mediated the relationship between abusive supervision and job embeddedness. This result suggests that abusive supervision is related to reduced POS, which will, in turn, be related to reduced job embeddedness. To be precise, this result proposes that supervisory hostility ascribed specifically to the organization (Shoss et al. 2013; Bowling and Michel 2011) and reflected in the shape of reduced POS carries out an essential role in reduced job embeddedness. Such a finding is important because it sheds more light on why and to what degree destructive leadership behaviors will harm employees’ job embeddedness (Erkutlu and Chafra 2017). Research has also suggested POS as an intervening mechanism linking abusive supervision to high levels of turnover intention (Xu et al. 2018; Haar et al. 2016).

Consistent with a social exchange perspective, our findings suggest that employees negatively reciprocate (Eisenberger et al. 2004) reduced POS due to abusive supervision by redcued job embeddedness. That is, they seem to display low levels of job embeddedness because they do not want to remain with or contribute to the organization as a response to its negative treatments. For that reason, our findings propose low job embeddedness as a way of retaliation against the organization’s unfavorable treatments.

Retaliation is an expected result of negative reciprocity (Shoss et al. 2013). It is because, as Eisenberger et al. (2019) argued, “people are inculcated with the reciprocity norm as a moral virtue whose violation damages reputations and produces retribution” (p. 1038). By doing so, individuals protect their self-image and prevent exploitation in social relations (Eisenberger et al. 2001; Eisenberger et al. 2004). Nevertheless, subordinates hardly take a stand against an abusive supervisor due to him/her higher status, managerial roles, etc. (Xu et al. 2015). Instead, they tend to orient their retaliations towards the organization (Valle et al. 2019; Xu et al. 2015; Mitchell and Ambrose 2007). Still, it seems problematic to describe job embeddedness as a way of retaliation. It is because job embeddedness defines a state of inertia about employee’s staying in an organization, not an attitude nor behavior (Yao et al. 2004). However, the global job embeddedness (Crossley et al. 2007) construct considers job embeddedness as one’s general attachment to the organization. Further, it measures one’s embeddedness by his/her feelings and impressions regarding this attachment. We, therefore, believe that when these feelings or impressions are negatively affected by employees’ feelings of retribution for the organization, they may reshape unfavorably and give rise to decreased job embeddedness. Such a retaliation, moreover, has the potential to transform into actual turnover such that job embeddedness has been suggested as an important predictor of it (Crossley et al. 2007; Mitchell et al. 2001).

From a COR theory perspective, our findings suggest that, as a social problem (Tepper 2007) and chronic stressor in the workplace (Zhang et al. 2019), abusive supervision is related to the exhaustion of employees’ organizational resources. Further, this exhaustion and resulting stress appear to weaken their attachment to the organization and thus their job embeddedness. That is, employees seem to display low levels of job embeddedness in the face of a negative factor threatening their resources provided by the organization and their wellbeing. This finding is in line with the previous studies proposing abusive supervision as a factor that may deplete one’s cognitive, emotional or social resources and thus leads undesirable outcomes (Zhang et al. 2019; Zhang and Liao 2015; Shoss et al. 2013; Aryee et al. 2008).

In short, the present findings of this study suggest that the perceived quality of the employee-organization relationship, which is reflected by POS (Settoon et al. 1996), is positively associated with job embeddedness. In addition, it mediates the negative relationship between abusive supervision and job embeddedness. Therefore, according to our findings, high-quality relationships with the organization are related to high embeddedness (Nguyen et al. 2017; Akgunduz and Sanli 2017). Whereas, low-quality relations with the organization, resulting from abusive supervision (Xu et al. 2018; Haar et al. 2016; Shoss et al. 2013), is related to low job embeddedness. Based on our theoretical framework, specifically, the negative relationship between abusive supervision and job embeddedness may be explained by employees’ perceptions regarding poor-quality social exchanges with the organization. Moreover, the poor-quality exchanges have the potential to end up an insufficiency or lack regarding the resources provided by the organization. Thus, such organizational resource exhaustion may also help clarify further the abusive supervision-job embeddedness relationship. Similar implications have also been reported by various prior studies for abusive supervision (Zhang et al. 2019; Xu et al. 2015, 2018; Haar et al. 2016; Shoss et al. 2013; Lian et al. 2012; Aryee et al. 2007).

Our analysis also found that organizational tenure was positively related to job embeddedness. A possible explanation for this is that an organization will become more attractive to employees parallel to the increase in what it provides (Singh et al. 2018; Kiazad et al. 2015; Harris et al. 2011). That is, as the experience of employees increases, the resources provided by their organizations (higher salary, respectability, person-job fit, perks, quality relationships, etc.) are more likely to increase as well. Thus, employees with high tenure would be more attached to their organizations.

Finally, to date, most of the job embeddedness and abusive supervision studies have been conducted primarily in the US. However, limited research has been conducted in non-Western samples (Martinko et al. 2013; Tepper 2007; Zhang et al. 2012; Peltokorpi et al. 2015). Through the present study, we explored the applicability and validity of job embeddedness and abusive supervision in Turkey. Our findings support that abusive supervision (Tepper 2000) and global job embeddedness (Crossley et al. 2007) measures are valid constructs in Turkey. Additionally, Turkey has a high-power distance culture (power distance index =66) and power disparity between supervisors and their subordinates is high (Basabe and Ros 2005). Power distance refers to “the extent to which a society accepts the fact that power in institutions and organizations is distributed unequally” (Hofstede 1980: 45). Individuals in high power distance cultures are more tolerant of abusive supervision than individuals in low power-distance cultures because of the respect for authority (Lian et al. 2012; Xu et al. 2015; Martinko et al. 2013). However, despite this fact, our results indicate abusive supervision still leads to negative consequences for employees in Turkey.

Practical Implications

This study offers several implications for organizational practitioners regarding employee retention. Employee retention is a critical issue for employers because when an organization loses valuable employees, it also loses training, experience, talent, and even morale (Zhang et al. 2012). For that reason, organizations should always be in a struggle to keep valuable members (Kiazad et al. 2015; Mitchell et al. 2001).

Our findings provide evidence that abusive supervision influences job embeddedness through the intervening mechanism of perceived organizational support (POS). This finding indicates that supervisors’ abusive behaviors do not only affect the supportiveness of the organization in the eyes of employees, but also their attachment to the organization, suggesting abusive supervision as a threat to employee retention. Thus, to protect their outlook in the eyes of employees and increase the possibility to keep them, organizations should support favorable employee treatment and openly cope with abusive supervisors. In this way, employees can be given the message that their organization is by no means tolerant of abusive supervision and supervisors practicing it are outliers, not representative of the organization (Shoss et al. 2013).

To this end, organizations can train supervisors in interpersonal relationship skills and anger management. In this way, supervisors may have improved interpersonal skills (treatment of subordinates with dignity, respect, and sensitivity) and form more high-quality relationships with their subordinates (Harris et al. 2011; Aryee et al. 2007). Moreover, organizations should establish zero-tolerance policies and develop norms that make clear that abusers will be punished swiftly (Tepper et al. 2008). Given that abused employees tend to keep silent rather than report perpetrators (Haar et al. 2016), organizations are also needed to establish secure channels for victims to reveal abusers (Xu et al. 2015). These practices would eventually encourage an organizational culture espousing supportive employee treatment and attenuate or prevent abusive tendencies in supervisor (Shoss et al. 2013; Peltokorpi 2018).

Our finding also suggests POS as a key instrument to enhance employee retention. It is because it accounted for half of the variance of job embeddedness in our analysis. For that reason, organizations should enhance their members’ POS to keep them. Interestingly, the actions listed above to cope with abusive supervision could serve such a goal. This is because employees tend to consider organizational policies, norms, and culture as part of the organizational support they receive (Eisenberger et al. 2001). Moreover, they expect their organizations to provide them with informational and instrumental support to help them cope with occupational stressors such as abusive supervisors (Baran et al. 2012).

For the enhancement of POS, specifically, managers should first be aware that employees expect care, praise, approval, justice, and rewards for their contributions from the organization and/or its representatives (Eisenberger et al. 1986; Rhoades and Eisenberger 2002). In addition, individuals value the most things given when needed and without any open expectation of the donor (Eisenberger et al. 1986). Next, organizations should focus on providing employees with favorable work experiences (Kurtessis et al. 2015), as well as discretionary, sincere and beneficial treatments (Rhoades et al. 2001; Eisenberger et al. 1986; Armeli et al. 1998; Settoon et al. 1996). Examples for such experiences and practices are as follows: rewards and growth opportunities, financial inducements, involving employees in organizational decision-making, strategic human resource management practices, socialization tactics for newcomers, supervisor support, work-schedule flexibility, and dependent care assistant (Kiazad et al. 2015; Allen and Shanock 2013; Baran et al. 2012; Armeli et al. 1998). Additionally, organizations should make employees feel that their increased effort on behalf of the organization is valued. Thus, the organization should invent ways to notice the increased effort and then sincerely reward those efforts.

Future Research

Future research could first examine other potential mediating mechanisms for the abusive supervision-job embeddedness relationship, including leader-member exchange (Xu et al. 2012), justice perception (Park et al. 2017), self-esteem (Vogel and Mitchell 2015), emotional exhaustion (Xu et al. 2015), blame attributions (Wang and Jiang 2015) and work stress (Zhang et al. 2019).

Furthermore, there is a need to examine potential moderators of the relations in our model. One such variable may be power distance orientation because the relationships discovered in this study may vary based on the levels of power distance. Lin et al. (2013), for instance, have shown that high power distance employees are more tolerant to abuse. Moreover, they have demonstrated that the relationships of abusive supervision with employee mental health and job satisfaction are weaker for employees who are higher in power distance orientation.

Moreover, prior research has suggested that job embeddedness has both positive and negative sides (Allen et al. 2016). As an employee’s current situation is unfavorable, job embeddedness may cause him/her to have undesired attitudes and behaviors towards the organization. And, the reason behind this might be the negative side of job embeddedness leading employees to feel stuck and frustrated (Sekiguchi et al. 2008). In fact, people with negative embeddedness would like to leave but cannot (Lee et al. 2014). Hom et al. (2012) call this group of people as “reluctant stayers.” Consequently, we propose that subsequent research should explore the negative side of embeddedness because extant research is relatively silent about which organizational factors makes employees feel stuck and frustrated.

Limitations

This study has several limitations that should be noted to interpret our results properly. First, all variables were collected using self-reports, which may raise common method variance (CMV). We, therefore, conducted the tests recommended by Podsakoff et al. (2003) to diagnose the effect of CMV on our analyses. The results revealed CMV did not pose a serious threat to our study. Moreover, confirmatory factor analyses showed that the constructs empirically discriminated, which increased our confidence in the results. Additionally, self-report data are said to be more appropriate for evaluating individual perceptions (Lian et al. 2012), including abusive supervision or perceived organizational support.

Second, the mean level of abusive supervision was low in our sample (mean = 1.25). Yet, abusive supervision is described as a low-base-rate phenomenon (Zellars et al. 2002) and this limitation has been encountered in various similar studies. Ranging from low to high, the mean level reported by these studies are as follows: 1.21 (Pundt and Schwarzbeck 2018), 1.25 (Yan et al. 2017), 1.26 (Tepper et al. 2004), 1.30 (Eschleman et al. 2014) and 1.38 (Decoster et al. 2013), to high 1.62 (Lian et al. 2012), 1.82 (Mitchell and Ambrose 2007), 1.87 (Aryee et al. 2008) and 2.36 (Shoss et al. 2013). Furthermore, our findings provided evidence that the mediation effect of POS in the abusive supervision-job embeddedness was significant. We believe that low level of abusive supervision did not appear to be a problem in the data.

Third, our sample was formed by participants from different organizations. It is possible that organizational context effects might influence our findings (Ferris et al. 2009). Diverse sample, however, helps to strengthen the generalizability of the results (Ng and Feldman 2013). Lastly, by design, our data were cross-sectional, it limits to make causal inferences among study variables, but future studies can use a longitudinal design to extend our findings.

Conclusion

This study found that perceived organizational support could serve as a mediator by providing a full explanation for the negative influence of abusive supervision on job embeddedness. This finding suggests that abusive supervision is related to reduced job embeddedness through reduced perceived organizational support. Our findings are important because this is the first study examining the abusive supervision-job embeddedness relationship. Moreover, only a few studies have focused on the relation of a negative organizational factor, like abusive supervision, to job embeddedness (Holtom et al. 2012). Our findings also propose that abusive supervision and job embeddedness measures are valid constructs in the context of Turkey. As the first research exploring why and to what extent abusive supervision influence job embeddedness, we hope our findings encourage other researchers to focus greater attention on this new promising line of research.