Organizational justice is a key concern in the field of organizational behavior. Many previous theoretical and empirical studies show that employees’ perception of the fairness of their reward significantly affects their attitudes and behaviors (Colquitt, Conlon, Wesson, Porter, & Ng, Reference Colquitt, Conlon, Wesson, Porter and Ng2001). Among the types of organizational justice, distributive justice has been a hot issue. Distributive justice refers to the employees’ evaluation of the fairness of the rewards (e.g., pay, promotion) they receive from the organization are consummate with their contribution or performance (Niehoff & Moorman, Reference Niehoff and Moorman1993). If employees perceive that their rewards are unjust compared to coworkers, their positive attitudes or behaviors for organization (e.g., organizational commitment, trust, and organizational citizenship behavior) will be decreased (Colquitt et al., Reference Colquitt, Conlon, Wesson, Porter and Ng2001; Cropanzano, Byrne, Bobocel, & Rupp, Reference Cropanzano, Byrne, Bobocel and Rupp2001).
When distributive injustice occurs, turnover is likely to happen, since turnover is one of the means an employee can take a proactive step to remedy the injustice. In fact, many prior justice theorists have addressed how employees’ perception of distributive justice affects their turnover intention (Aquino, Griffeth, Allen, & Hom, Reference Aquino, Griffeth, Allen and Hom1997; Aryee, Budhwar, & Chen, Reference Aryee, Budhwar and Chen2002). Cohen-Charash and Spector (2002), in their meta-analysis, found that distributive justice, among the three types of organizational justice (e.g., distributive, procedural, and interactional) has the strongest positive influence on turnover intention.
Nonetheless, it is somewhat unclear why and when distributive justice affects turnover intention. First, most of the existing studies on the relationship between organization justice and turnover intention have been focused on procedural and interactional justice rather than distributive justice (e.g., Elovainio et al., Reference Elovainio, van den Bos, Linna, Kivimaki, Ala-Mursula, Pentti and Vahtera2005). Thus, surprisingly, the relationship between distributive justice and turnover intention has not been sufficiently explored. Second, the theoretical explanatory mechanisms for the influence of distributive justice on turnover intention have not been fully examined. As noted above, most previous studies have found a variety of mediators including the perceived organizational support (Masterson, Lewis-McClear, Goldman, & Taylor, Reference Masterson, Lewis-McClear, Goldman and Taylor2000), trust (Masterson et al., Reference Masterson, Lewis-McClear, Goldman and Taylor2000; Aryee, Budhwar, & Chen, Reference Aryee, Budhwar and Chen2002), leader–member exchanges (Karriker & Williams, Reference Karriker and Williams2009) on the relationship between procedural or interactional justice and turnover intention. However, the theoretical mediating mechanisms underlying the relationship between distributive justice and turnover intention is less clear.
In this regard, the purpose of the present study is twofold. First, we try to examine the less-explored relationship between distributive justice and turnover intention. In particular, we pay attention to the mediating effect of job stress on this relationship. Unfairness in the organization can cause job stress, since unfairness can be a stressor that causes employees to doubt their ability to perform well, and also influence their psychological and physical health (Maslach, Schaufeli, & Leiter, Reference Maslach, Schaufeli and Leiter2001; Tepper, Reference Tepper2001; Elovainio et al., Reference Elovainio, van den Bos, Linna, Kivimaki, Ala-Mursula, Pentti and Vahtera2005; Moliner, Martinez-Tur, Peiro, Ramos, & Cropanzano, Reference Moliner, Martinez-Tur, Peiro, Ramos and Cropanzano2005). In fact, although job or role-related factors, such as role ambiguity, role conflict, and workload, have been well examined as determinants of employees’ stress (e.g., Lee & Ashford, Reference Lee and Ashforth1996; LePine, LePine, & Jackson, Reference LePine, LePine and Jackson2004; Newton & Jimmieson, Reference Newton and Jimmieson2008; Hung, Fisher, Gapp, & Carter, Reference Hung, Fisher, Gapp and Carter2012), there is a lack of attention on empirical studies on the effect of distributive justice as a predictor of job stress (Halbesleben & Buckley, Reference Halbesleben and Buckley2004; Judge & Colquitt, Reference Judge and Colquitt2004).
The second purpose is to identify the cross-level moderating effect of organizational climate that is a boundary condition affecting the relationship between distributive justice and its outcomes at the individual level. Employees’ job stress may be influenced by organization or group-level factors (e.g., climate) (Cooper, Cooper, & Eaker, Reference Cooper, Cooper and Eaker1988). In this respect, we can expect that, although employees’ perception of distributive justice affects their job stress at the individual level, this relationship may be different, according to organizational climates. More specifically, we suggest that empowerment climate at the group level affects the relationship between employees’ perception of distributive injustice and job stress. Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004) suggested that an empowerment climate that provides autonomy, information sharing, and team accountability positively relates to employees’ attitudes and behaviors toward their organization. In this study, we consider empowerment climate as a moderator at the work-unit level that influences the relationship between distributive justice and job stress at the individual level. Although a few empirical works (e.g., Seibert, Silver, & Randolph, Reference Seibert, Silver and Randolph2004; Chen, Kirkman, Kanfer, & Allen, Reference Chen, Kirkman, Kanfer and Allen2007a; Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b; Si & Wei, Reference Si and Wei2012) have theorized or tested the role of empowerment climate, the present study on cross-level effects of empowerment climate will contribute to the justice and empowerment literature by identifying the boundary condition affecting the influence of distributive justice on employees’ attitudes.
THEORETICAL BACKGROUNDS AND HYPOTHESES DEVELOPMENT
A variety of justice theories attempt to explain the effect of distributive justice. For instance, a referent cognitive theory, the instrumental model (or self-interest theory), and the group value model have been frequently used for explaining the effects of distributive, procedural, and interactional justice (Brockner & Wiesenfeld, 1996; Cropanzano et al., Reference Cropanzano, Byrne, Bobocel and Rupp2001). However, we think that the theoretical rationales addressing the relationship between distributive justice, job stress, and turnover intention are somewhat different from these prevailing theories. Accordingly, although we basically use the prevailing justice theories mentioned above, we set the hypotheses through simultaneously integrating other justice-related theories (e.g., social comparison theory, uncertainty management theory) as well as job stress-related theories (e.g., conservation resource model, job demand, and resource theory).
The relationship between distributive justice and job stress
Adams’ (Reference Adams1965) equity theory is considered as one of the most popular theories to explain how distributive justice operates in an organization. The equity theory proposes that employees compare the ratio of their own reward and effort to the ratios of reward and effort of others. Through this comparison process, when employees perceive that their ratios are unequal, they will feel relative deprivation or discontent (Martin, Reference Martin1981). This psychological distress can lead to dysfunctional attitudes or behaviors, including turnover, as a way of restoring the inequity.
As one form of psychological distress, job stress can occur when employees feel distributive injustice. Parker and DeCotiis (Reference Parker and DeCotiis1983: 165) defined job stress as ‘the feeling of a person who is required to deviate from normal or self-desired functioning in the work place as the result of opportunities, constraints, or demands relating to potentially important work-related outcomes.’
There are some theories linking distributive justice and job stress. First, uncertainty management theory provides the clue for such relationship (Lind & van den Bos, Reference Lind and van den Bos2002; van den Bos & Lind, Reference van den Bos and Lind2002). Employees usually face uncertainty concerning their incentives, promotions, and employment security. These uncertainties lead employees to anxiety and psychological distress (Judge & Colquitt, 2004). Therefore, employees usually have the strong need and desire to reduce uncertainty. Meanwhile, whether the organization provides a fair and just reward to employees in return for their contributions is an adequate informational signal to employees (Lind & van den Bos, Reference Lind and van den Bos2002). Thus, if employees perceive that their organization is fair and consistent in the distributional sense, their uncertainty will be decreased. In other words, distributive justice can reduce employee uncertainty.
Second, job stress and job burnout literature also provide rationales for our expectation. For instance, the effort–reward imbalance model proposed that the inconsistency between employees’ efforts and rewards for those efforts cause job stress (Cole, Bernerth, Walter, & Holt, Reference Cole, Bernerth, Walter and Holt2010). That is, if employees are fairly rewarded for their efforts, then they would not experience job stress; however, if they receive an unfair reward, they may feel job stress. The theory of conservation of resources (Hobfoll, Reference Hobfoll2001; Schaufeli, Reference Schaufeli2006) also posits that stress and burnout occur when employees perceive threats to their valuable resources. Here, resources are defined as ‘those objects, personal characteristics, conditions or energies that are valued in their own right or that are valued because they act as conduits to the achievement or protection of valued resources’ (Hobfoll, Reference Hobfoll2001: 339). Thus, according to this theory, stress is a result of a threat to resources (e.g., the perception that one might lose his or her job), the actual loss of a resource (e.g., the loss of the job), and insufficient gain of additional resources.
Supposing that fair reward for one's efforts can be a type of resource, when employees receive a low reward (i.e., resources) for their efforts, they may feel stress. In this sense, Maslach, Schaufeli, and Leiter (Reference Maslach, Schaufeli and Leiter2001) and Halbesleben and Buckley (Reference Halbesleben and Buckley2004) proposed fairness as a possible predictor of job burnout. They argued that, even though employees made a great effort for excellent performance and actually achieved desirable results for their organization, if they do not receive a reasonable reward from the organization, they may feel that they cannot do anything to contribute to the outcome. Cole et al. (Reference Cole, Bernerth, Walter and Holt2010) also suggested that employees in an unjust or unfair working environment will have a lack of resources, and thus be more vulnerable to distress. Likewise, the lack of fairness may bring job stress since employees might get upset and feel exhausted.
In addition, some existing empirical studies support the significant relationship between distributive justice and job stress. Tepper (Reference Tepper2000) found that distributive justice was negatively correlated with psychological distress (e.g., anxiety, emotional exhaustion, depression). De Boer, Bakker, Syroit, and Schaufeli (Reference De Boer, Bakker, Syroit and Schaufeli2002) showed that distributive injustice increased employees’ health complaints. Janssen (Reference Janssen2004) found that distributive justice decreased the job-related anxiety and burnout. Walster, Berscheid, and Walster (Reference Walster, Berscheid and Walster1973) also proposed that employees’ perception of an unequal exchange relationship with the organization can cause psychological tension and distress. Taken together, we propose the following hypothesis:
Hypothesis 1: Employees’ perception of distributive justice would be negatively related to their job stress.
The mediating effect of job stress
Many justice theorists have well addressed how employees’ perception of distributive justice affects their turnover intention (Aquino et al., Reference Aquino, Griffeth, Allen and Hom1997; Aryee, Budhwar, & Chen, Reference Aryee, Budhwar and Chen2002; Cohen-Charash & Spector, 2002; Hechanova, Amampay, & Franco, Reference Hechanova, Amampay and Franco2006). According to the justice theories (e.g., Brockner & Wiesenfeld, Reference Brockner and Wiesenfeld1996), when employees are rewarded unfairly compared to their input or contribution, they may think that their organization is unreliable and does not respect them. Accordingly, employees’ need to maintain the membership within the organization would be decreased (Lavelle, Rupp, & Brockner, Reference Lavelle, Rupp and Brockner2007). Colquitt et al. (Reference Colquitt, Conlon, Wesson, Porter and Ng2001) showed in their meta-analysis on organizational justice that distributive justice had the strongest influence on employees’ withdrawal behavior than any other type of justice (e.g., procedural, interactional, and informational justice).
In the literature of job stress, the influence of job stress on employees’ turnover intention has been subjected to much empirical scrutiny (e.g., Lee & Ashford, Reference Lee and Ashforth1996; Cole et al., Reference Cole, Bernerth, Walter and Holt2010). Tepper (Reference Tepper2001) argued that depletion of adequate resources in return for employees’ efforts may bring them to stressful situations, and eventually cause emotional exhaustion. Lee and Ashford (Reference Lee and Ashforth1996), in their meta-analysis, demonstrated that emotional exhaustion had a very high correlation with turnover intention.
Moreover, when seen in terms of social comparison process (Festinger, Reference Festinger1954), employees’ job stress caused by distributive injustice may increase their turnover intention. When compared with others, if employees are less paid or unfairly treated compared to others, this comparison may serve as a stressful situation. In this case, employees who feel injustice in social exchange relationships with the organization may lack the motivation to maintain a relationship with the organization (Maslach, Schaufeli, & Leiter, Reference Maslach, Schaufeli and Leiter2001; van Dierendonck, Schaufeli, & Bunnk, Reference van Dierendonck, Schaufeli and Bunnk2001).
Taken together, distributive injustice, which is a predictor of job stress, may cause employees’ negative attitudes toward their organization. As such, it is expected that employees’ perception of distributive justice will be negatively related to job stress, which in turn will decrease their turnover intention. Cole et al. (Reference Cole, Bernerth, Walter and Holt2010) found that emotional exhaustion mediated the influence of distributive injustice on turnover intention. Schaufeli (Reference Schaufeli2006) also have proposed that, when employees feel distributive injustice, they will withhold their efforts or psychological attachment in their organization to reduce stress and restore unfair treatment by increasing turnover intention. Accordingly, we expect that employees’ perception of distributive justice reduce the job stress, and in turn decrease their turnover intention. This prediction is consistent with a mediation model.
Hypothesis 2: Employees’ job stress would mediate the relationship between distributive justice and turnover intention.
Empowerment climate
Empowerment means the delegation of managers’ authority to the employees, and thus, employees make decisions regarding their work with their own willingness and competence (Conger & Kanungo, Reference Conger and Kanungo1988; Seibert, Silver, & Randolph, Reference Seibert, Silver and Randolph2004). Thus, the empowerment is related to self-determination and intrinsic motivation (Deci & Ryan, Reference Deci and Ryan1985; Conger & Kanungo, Reference Conger and Kanungo1988; Liden & Tewksbury, Reference Liden and Tewksbury1995; Holdsworth & Cartwright, Reference Holdsworth and Cartwright2003; Zhang & Bartol, Reference Zhang and Bartol2010), because individuals who feel a higher level of empowerment may also feel a greater degree of intrinsic control or self-generated motivation to perform well (Spreitzer, Reference Spreitzer1995; Liden, Wayne, & Sparrowe, Reference Liden, Wayne and Sparrowe2000; Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b). Indeed, there have been many empirical studies showing the positive influence of empowerment on employees’ work attitudes and performance. For instance, Spreitzer, Kizilos, and Nason (Reference Spreitzer, Kizilos and Nason1997) found that the empowered employees had lower job stress. Hechanova, Amampay, and Franco (Reference Hechanova, Amampay and Franco2006) showed that employees’ perception of empowerment was positively related to job satisfaction and performance. Moye and Henkin (Reference Moye and Henkin2006) also found that employees who feel empowered in an organization have higher trust in their leaders.
Like this, at the individual level, employees’ psychological empowerment has been well studied. Psychological empowerment is defined as an individual's experience of intrinsic motivation that is based on cognitions about themselves in relation to their work (Spreitzer, Reference Spreitzer1995, Reference Spreitzer2007). However, individuals’ attitudes or behaviors may be influenced by not only empowerment at the individual level but empowerment climate at the group level. That is, empowerment climate may also exert positive signals to individuals, such as enhancing self-determination and intrinsic motivation (Chen et al., Reference Chen, Kirkman, Kanfer and Allen2007a; Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b; Spreitzer, Reference Spreitzer2007; Seibert, Wang, & Courtright, Reference Seibert, Wang and Courtright2011). Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004) have suggested that an empowerment climate is a shared perception regarding the extent to which a group makes use of structures, policies, and practices to support employees’ access to power. According to Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004), the more an organization provides potentially sensitive information on costs, productivity, quality, and financial performance to employees (information sharing), implements the organizational structures and practices that encourage employees’ autonomous behaviors (autonomy through boundaries), and delegates decision-making authority to employees (team responsibility and accountability), the more empowerment climate emerges in groups. As such, empowerment climate means the extent to which a team as a whole has autonomy as well as competence to perform meaningful tasks that can impact important organizational outcomes (Zhou, Wang, & Chen, Reference Zhou, Wang and Chen2012).
Here, although individual's psychological empowerment and empowerment climate are positively related (Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b; Zhou, Wang, & Chen, Reference Zhou, Wang and Chen2012), empowerment climate and psychological empowerment are conceptually distinct (Seibert, Silver, & Randolph Reference Seibert, Silver and Randolph2004; Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b; Spreitzer, Reference Spreitzer2007). According to Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004), first of all, while psychological empowerment is related to individuals’ subjective experiences of empowerment, empowerment climate is relatively related to perception of whether the organization's system, policies, and practices encourage employees’ empowerment. In addition, while psychological empowerment mainly focuses on individuals’ internal psychological states such as meaning, competence, self-determination, and impact in the work roles (Spreitzer, Reference Spreitzer2007), empowerment climate reflects the meaning of organizational structures and practices related to information sharing, autonomy, and accountability.
Cross-level main effect of empowerment climate on job stress
Indeed, the climate of the group in which an employee is embedded critically affects his/her attitude and behavior. As with the mechanism of the effect of empowerment on attitudes or behaviors at the individual level (Holdsworth & Cartwright, Reference Holdsworth and Cartwright2003; Chen et al., Reference Chen, Kirkman, Kanfer and Allen2007a; Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b; Zhang & Bartol, Reference Zhang and Bartol2010), empowerment climate will have a positive influence on employees’ outcomes. For instance, employees working in teams that are more empowered will think that that they have more intrinsically meaningful work and, as a group, have a higher degree of discretion in deciding how they carry out their team tasks; they believe that they have the collective ability to accomplish works or roles, which have an impact or significant importance for their organization (Seibert, Wang, & Courtright, Reference Seibert, Wang and Courtright2011). In this regard, we expect that, after controlling for the effect of distributive justice at the individual level, the empowerment climate will decrease the employees’ job stress, because work groups’ practices regarding empowerment (e.g., participative decision making, empowering leadership) will provide a positive feeling toward work groups or energetic behaviors.
Hypothesis 3a : The empowerment climate will be negatively related to employees’ job stress beyond the effect of distributive justice at the individual level.
Cross-level moderating effect of empowerment climate in the relationship between distributive justice and job stress
Empowerment climate will also moderate the relationship between distributive justice and job stress at the individual level. It is well acknowledged that individuals are embedded within the relationship with a group or organization, and thus they do not act alone. Some previous studies have examined the moderating role of empowerment climate at team level on individuals’ consequences. For instance, Chen, Lam, and Zhong (Reference Chen, Lam and Zhong2007b) found that the climate of empowerment within a team moderated the positive relationship between leader–member exchanges and followers’ negative feedback-seeking behavior. Si and Wei (Reference Si and Wei2012) found that the positive relationship between supervisors’ transformational leadership and subordinates’ creative performance was stronger when the team had a high level of empowerment climate. These previous studies suggest that, although justice perception influences stress at the individual level, this relationship may be affected by climates at the group level (Naumann & Bennett, Reference Naumann and Bennett2000; Liao & Rupp, Reference Liao and Rupp2005; Mayer, Nishii, Schneider, & Goldstein, Reference Mayer, Nishii, Schneider and Goldstein2007). In this regard, we suggest that an empowerment climate at the group level affects the relationship between distributive justice and job stress at the individual level. Two possible rationales for explaining the moderating effect of empowerment climate are driven from theories on job stress.
First, job demand–resource model, as one of theories on job burnout, proposes that burnout is formed by two job-related characteristics – job demand and job resource (Demerouti, Bakker, Nachreiner, & Schaufeli, Reference Demerouti, Bakker, Nachreiner and Schaufeli2001). Job demands are the aspects of the job that require effort; on the other hand job resources are characteristics of the job that assist in achieving work goals or lead to personal growth. Information, autonomy, and accountability, as dimensions of psychological empowerment, can serve as job resources to perform well. Thus, if employees who do not have appropriate information, autonomy, and accountability for their work will feel that they lack resources, and they will thus experience burnout (Maslach, Schaufeli, & Leiter, Reference Maslach, Schaufeli and Leiter2001; Holdsworth & Cartwright, Reference Holdsworth and Cartwright2003; Newton & Jimmieson, Reference Newton and Jimmieson2008). In contrast, an empowerment climate may serve the role of social support from supervisors or the organization (Chen, Lam, & Zhong, Reference Chen, Lam and Zhong2007b). The job demand–resource model emphasizes the important role of social support as a resource that can reduce the likelihood of burnout (Halbesleben & Buckley, Reference Halbesleben and Buckley2004). Therefore, if employees perceive a lack of autonomy accountability and control over the job, they may experience frustration and stress. On the contrary, if the employees’ shared cognition about empowerment is high, even though one perceives unfairness in terms of distribution, this may not easily relate to job stress. Indeed, empirical studies (e.g., Lee & Ashforth, Reference Lee and Ashforth1996; Demerouti et al., Reference Demerouti, Bakker, Nachreiner and Schaufeli2001; Hung et al., Reference Hung, Fisher, Gapp and Carter2012) found that employees who had a feeling of control over their job participated in work process, received feedback, and showed lower emotional exhaustion.
Second, the empowerment climate may reduce the job stress induced by distributive injustice through enhancing self-confidence. When employees are empowered, they will feel that they have the ability to determine work outcomes, feel competent to achieve their goals, and believe that they have an impact on the work environment (Spreitzer, Reference Spreitzer2007). Empowerment signals to employees that the organization trusts their judgment and competence (Erdogan & Bauer, Reference Erdogan and Bauer2009), which may convey to employees that they have high status within the organization, and thus may feel less stress. Likewise, at the group level, when employees in an organization collectively perceive that they are empowered concerning their work, they are likely to have positive emotions (e.g., happiness, joy). As a result, employees in groups with an empowerment climate will work with a higher level of interpersonal trust among team members, share information among coworkers, and receive sufficient support from managers. These positive and favorable environments within the groups will facilitate teamwork, trust in coworkers, and confidence in work, which will reduce the negative impact caused by stress related to unfair rewards. (Figure 1)
Figure 1 The research model
Taken together, we expect that the empowerment climate would influence the negative relationship between distributive justice and job stress at the individual level, such that the negative relationship would be stronger for groups with a higher level of empowerment climate than for groups with a lower level.
Hypothesis 3b: The empowerment climate will moderate the relationship between distributive justice and job stress, such that the negative relationship will be stronger for group with a high level of empowerment climate.
METHOD
Sample and procedures
The sample consisted of 5,834 employees from 115 work groups in South Korea. Human Resource Department managers in each work group served as the contact for our research. Using contact information obtained from the Human Resource managers, we sent the survey questionnaires with a cover letter explaining the purpose of the survey to respondents through an e-mail.
Ultimately, the final usable return of 4,432 surveys from 90 work groups yielded a response rate of 75.9% at the individual level and 78.2% at the work group level. For each group, the average number of respondents was 49 ranging from 25 to 95. Across work groups, the average age and organizational tenure of respondents were 34.9 years (SD = 5.2) and 8.5 years (SD = 6.9), respectively. A total of 78.2% of the respondents were male. For educational level, 86.8% held college education or above, and 13.2% had below college education. In terms of job title, 97.8% were front-line employees and 2.2% were managers. Respondents worked in one of four types of jobs; R&D (29.1%), manufacturing (19.9%), marketing and sales (30.3%), and administrative work (20.7%).
Measures
All the variables were measured by respondents to questions on a 5-point Likert-type scale ranging from 1 = ‘strongly disagree’ to 5 = ‘strongly agree’. The specific measures are described below, along with the Cronbach's α coefficients for each measure.
Distributive justice
We assessed the distributive justice with two items adapted from the five-item measure of Niehoff and Moorman (Reference Niehoff and Moorman1993), which was developed to assess the fairness of various work dimensions such as pay level, work load, and job responsibility. Among these dimension, we assessed the distributive justice by focusing on the pay level. The items are: ‘I think that my level of pay is fair’ and ‘Overall, the rewards I receive here are quite fair.’ Internal consistency reliability was 0.74.
Job stress
Job stress was assessed with six items by adopting the validated 15-item measures developed by Parker and Decotiis (Reference Parker and DeCotiis1983). A higher score means that employees feel a high level of stress in work in terms of time and anxiety. Sample items are ‘There are lots of times when my job drives me right up the wall,’ ‘Working here leaves little time for other activities,’ and ‘I feel like I never have a day off.’ Internal consistency reliability was 0.93.
Turnover intention
We used a single-item measure for assessing the intention to leave by referring the previous measures on turnover intention (e.g., Stahl, Chua, Caligiuri, Cerdin, & Taniguchi, Reference Stahl, Chua, Caligiuri, Cerdin and Taniguchi2009). The item is, ‘I am willing to leave this organization soon.’
Empowerment climate
Empowerment climate was assessed with nine items adopted from the measure of Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004). Originally, Seibert, Silver, and Randolph (Reference Seibert, Silver and Randolph2004) developed the 30-item measure for empowerment climate that consists of three dimensions such as information, autonomy, and responsibility and accountability. Among these items, we adopted nine items for our research purpose. Sample items are: ‘Our organization encourages their people to take risks by trying new things and taking risks themselves,’ ‘Our organization gives opportunities to perform a challenging role or task to develop and tap into my capabilities,’ and ‘People in our organization can confidently voice their opinions without having to worry about how their manager will respond.’ Internal consistency reliability was 0.88.
Control variables
We controlled the age, gender, educational level, tenure, and job title, because these variables might influence the job stress or turnover intentions (e.g., Griffeth, Hom, & Gaertner, Reference Griffeth, Hom and Gaertner2000).
Validity test
Confirmatory factor analyses were conducted to evaluate the convergent and discriminant validity of measures, using AMOS 7.0. We evaluated the goodness of each model by determining whether the values of comparative fit index (CFI) and Tucker–Lewis index (TLI) is >0.90, and the value of the root–mean squared error of approximation (RMSEA) is <0.08 (Lance & Vandenberg, Reference Lance and Vandenberg2002). The confirmatory factor analyses results showed that the four-factor model including distributive justice, job stress, empowerment climate, and turnover intention yielded a better goodness-of-fit index [χ2(df) = 2,908.5(130), CFI = 0.940, TLI = 0.921, RMSEA = 0.069], with a significant change of χ2 of 11,539.8 (Δdf = 5, p < .001) than a single-factor model considering all four measures as a single factor [χ2(df) = 11,630.6(135), CFI = 0.751, TLI = 0.685, RMSEA = 0.139].
Our data may have the possibility of common-method bias because all measures were gathered from the same source in the same questionnaire (Podsakoff, MacKenzie, Lee, & Podsakoff, Reference Podsakoff, MacKenzie, Lee and Podsakoff2003). As a result, it is critical for us to make our best effort in minimizing common method bias through statistical remedies. One is the Harman's one-factor test suggested by Podsakoff et al. (Reference Podsakoff, MacKenzie, Lee and Podsakoff2003). The results of an unrotated principal component factor analysis including all items of measures showed that eight factors were extracted with an eigenvalue >1 and each item was loaded on its appropriate factor, with primary loadings exceeding 0.40. Additionally, the first factor accounted for only 23.2% of the total variance, and hence no general factor appeared in the factor structure.
Aggregation
Following the procedure suggested by Chan (Reference Chan1998), the referent-shift model was used for empowerment climate to aggregate data at the individual level into work group level. To justify the aggregation of empowerment at the individual level to the climate at the work group level, we tested within-group consensus and between-group variance (James, Demaree, & Wolf, Reference James, Demaree and Wolf1993; Chan, Reference Chan1998). First, within-group agreement statistic (rwg) for empowerment climate was calculated. The value of rwg of an empowerment climate was 0.83 with range from 0.75 to 0.99. This rwg value was above the acceptable rwg value of 0.70 (James et al., 1994). Second, we calculated the values of the intraclass reliability index (ICC1) and the reliability of group mean index (ICC2) by following one-way analysis of variance (Bliese, Reference Bliese2000). ICC1 provides an estimate of the extent to which individual-level variability on a given measure is explained by higher-level units and the ICC2 provides an estimate of the reliability of group means. ICC1 and ICC2 of empowerment climate were 0.10 and 0.85 (F = 5.70, p < .001), respectively. These values exceeded the acceptable level and suggested that within-group agreement was sufficient (Castro, Reference Castro2002). Based on these results, we concluded that the aggregation of empowerment at the individual level to empowerment climate at the work group level was justified.
Analytical strategy
We used hierarchical linear modeling (HLM) analyses to test all the hypotheses by using HLM 7 software, since the current model is the effect of distributive justice at the individual level and the cross-level effect of empowerment climate at the work group level (Bryk & Raudenbush, Reference Bryk and Raudenbush1992). HLM is a suitable statistical technique for simultaneously estimating the effects of predictors on dependent variable outcome at different levels, while maintaining appropriate levels of analyses for these predictors (Hofmann, Griffin, & Gavin, Reference Hofmann, Griffin and Gavin2000).
We followed four models using HLM analyses (Bryk & Raudenbush, Reference Bryk and Raudenbush1992). Briefly describing our analytical strategy with HLM, first, we estimated the null models (with no predictors involved) for the dependent variables (turnover intention and job stress) – and found the variance in dependent variables by examining the significant Level 2 variance (τ00). The significance of between-group variance (τ00) is a necessary condition that must be satisfied before further analysis can be undertaken, and demonstrates that there is sufficient between-group variance in job stress.
In the second model, we performed a random-coefficient regression model (Level 1 analysis) to estimate the effects of an independent variable (distributive justice) on mediating (job stress) and a dependent variable (turnover intention) at the individual level. The significance of βij relates to Hypotheses 1 and 2. Additionally, we can estimate whether the relationship between an individual's distributive justice and job stress varied over work groups through the significance of between-group variance (τ66), and whether the absolute amount of job stress varied in accordance with work units through the significance of between-group variance (τ00). Assuming significant between-group variance in both slopes and intercept for individual's distributive justice predicting job stress, in the third model, an intercept-as-outcomes model (Level 2 analysis), in which intercept estimates derived from the Level 1 analysis were regressed on empowerment climate, was performed. The purpose of this model was to test whether empowerment climate could account for the between-group variance in pooled Level 1 intercept from the previous random-coefficient regression model. More specifically, this model tested whether the individual's job stress was affected by not only the individual's distributive justice but also empowerment climate. Accordingly, the third model tests Hypothesis 3a. Finally, the fourth model is a slopes-as-outcomes model (Level 2 analyses) in which slope estimates derived from the Level 1 analysis were regressed on empowerment climate. This model tested for cross-level moderating effects that would reveal whether the relationship between distributive justice and job stress at the individual level varies depending on empowerment climate at the work group level. Thus, this final model tests the Hypothesis 3b. The following equations are our final model for testing cross-level effects of empowerment climate.
$${ {\rm{Level}}\,{\rm{1}}:\,{\rm{Job}}\,{\rm{stres}}{{{\rm{s}}}_{ij}}\, = \,{{{\rm{\beta }}}_{{\rm{0}}j}} + {{{\rm{\beta }}}_{{\rm{1}}j}}\,{\rm{(Age)}} + {{{\rm{\beta }}}_{{\rm{2}}j}}\,{\rm{(Gender)}} + {{{\rm{\beta }}}_{{\rm{3}}j}}\,{\rm{(Education)}}\cr \qquad\qquad\qquad\quad\qquad+ \,{{{\rm{\beta }}}_{{\rm{4}}j}} \ {\rm{(Tenure)}}\, + \,{{{\rm{\beta }}}_{{\rm{5}}j}}\,{\rm{(Title)}}\, + \,{{{\rm{\beta }}}_{{\rm{6}}j}}\,{\rm{(Distributive}}\,{\rm{justice)}}\, + \,{{r}_{ij}} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{0}}j}}\, = \,{{r}_{{\rm{00}}}}\, + \,{}{{r}_{{\rm{01}}}}\,{\rm{(Empowerment}}\,{\rm{climate)}}\, + \,{{U}_{{\rm{0}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{1}}j}}\, = \,{{r}_{{\rm{10}}}}\, + \,{{U}_{{\rm{1}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{2}}j}}\, = \,{{r}_{{\rm{20}}}}\, + \,{{U}_{{\rm{2}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{3}}j}}\, = \,{{r}_{{\rm{30}}}}\, + \,{{U}_{{\rm{3}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{4}}j}}\, = \,{{r}_{{\rm{40}}}}\, + \,{{U}_{{\rm{4}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{5}}j}}\, = \,{}{{r}_{{\rm{50}}}}\, + \,{{U}_{{\rm{5}}j}}{} \cr {\rm{Level}}\,{\rm{2:}}\,{{{\rm{\beta }}}_{{\rm{6}}j}}\, = \,{}{{r}_{{\rm{60}}}}\, + \,{}{{r}_{{\rm{61}}}}\,{\rm{(Empowerment}}\,{\rm{climate)}}\, + \,{{U}_{{\rm{6}}j}} $$
Following Hofmann, Griffin, and Gavin (Reference Hofmann, Griffin and Gavin2000), we grand-mean-centered all Level 1 variables to facilitate interpretation except control variables. The results of HLM did not show the explained variance (R 2), thus we calculated R 2 for each model, according to the formula suggested by Hofmann, Griffin, and Gavin (Reference Hofmann, Griffin and Gavin2000)Footnote 1.
RESULTS
Table 1 shows the means, standard deviations, and intercorrelations. As Table 1 shows, correlations provide initial supports for the Hypotheses. For instance, distributive justice was negatively related to job stress (r = –0.414, p < .001), and job stress was positively related to turnover intention (r = 493, p < .001).
Table 1 Means, standard deviations, and correlations
Notes. n = 4,432 at the individual level, n = 90 at the work group level.
Gender (0 = female, 1 = male), education (0 = below college, 1 = above college), title (0 = manager, 1 = employee).
*p < .05, ***p < .001.
Table 2 presents the results of HLM analyses. First of all, we tested Hypotheses 1 and 2 through performing a random coefficient model, since these hypotheses are related to testing the main effect of distributive justice on turnover intention and the mediating effect of job stress at the individual level with no predictors at the work group level. Model 3 showed that distributive justice was negatively related to turnover intention (γ60= –0.509, p < .001, R 2Level 1 = 0.221) with the increase of explained variance (ΔR 2 for Level 1) of 15.4%. Thus, Hypothesis 1, which expects that distributive justice would be negatively related to turnover intention, was supported.
Table 2 The results of HLM analyses for testing hypotheses
Notes. n = 4,432 at the individual level, n = 90 at the work group level.
*p < .05, **p < .01, ***p < .001.
Hypothesis 2 on the mediating effect of job stress was assessed by Baron and Kenney's (Reference Baron and Kenny1986) three-step procedures. First, as shown in Model 3, distributive justice was negatively related to turnover intention. Second, Model 7 showed that distributive justice had a negative influence on job stress (γ60 = –0.415, p < .001, R 2Level 1 = 0.179). Finally, as seen in Model 4, when both distributive justice and job stress were regressed on turnover intention, while the coefficient of distributive justice (γ60 = –0.400, p < .001) was reduced compared to that of Model 3, the effect of job stress (γ60 = 0.423, p < .001) was significant with the increase of explained variance (ΔR 2 for Level 1) of 10.8%. Taken together, job stress had a partial mediating effect. Thus, Hypothesis 2 was supported.
Hypotheses 3a and 3b relate to the cross-level effects of an empowerment climate. First, as a precondition for testing HLM, a null model without any predictor was tested. Model 5 provided evidence of significant between-group variance in job stress (τ00 = 0.071, χ2 = 526.151, p < .001), thus justifying further cross-level analysis. Moreover, the χ2 test of Model 7 also indicated that there was significant between-group variance in the intercepts (τ00 = 0.046, χ2 = 282.406, p < .001) and slope (τ66 = 0.008, χ2 = 128.411, p < .01) for job stress, after controlling the effect of distributive justice at the individual level. Thus, we can go to the next analysis to test the cross-level main effect of empowerment climate on job stress.
Model 8 as an intercept-as-outcomes model showed that the empowerment climate was negatively related to job stress at the individual level (γ01 = –0.740, p<.001, R 2Level 2 = 0.652). Thus, Hypothesis 3a, which predicts that empowerment climate will have a negative influence on job stress beyond the effect of individual-level distributive justice, was supported.
Model 9 as a slope-as-outcomes model demonstrated that an empowerment climate (γ01 = –0.209, p < .01, R 2Level 2 = 0.250) had a significant cross-level moderating effect on the relationship between distributive justice and job stress at the individual level. Thus, Hypothesis 3b, which expects that the strength of the relationship between distributive justice and job stress will be influenced by an empowerment climate, was supported.
Figure 2 shows the plot for the cross-level moderating effect of an empowerment climate. As expected in our hypotheses, a simple slope analysis (Aiken & West, Reference Aiken and West1991) showed that the negative relationship between an individual's distributive justice and job stress was stronger in work groups with a high empowerment climate (simple slope = –0.462, t = –14.92, p < .001) than in work units with a low empowerment climate (simple slope = –0.296, t = –7.95, p < .001).
Figure 2 The plots for cross-moderating effects of empowerment climate on the relationship between distributive justice and job stress
DISCUSSION
Distributive justice is an important concern for all employees and organizations. Equitable reward is one of the strongest motivators for employees to continuously perform well and to show favorable attitudes for the organization. In contrast, distributive inequity can bring undesirable consequences for employees.
Even though distributive justice may be significantly related to employees’ job stress and turnover intention, it is surprising that this relationship has not been fully investigated empirically and theoretically. We expected that employees’ perception of distributive justice would influence their job stress, and in turn affect turnover intention. In addition, although the relationship between distributive justice and job stress operates at the individual level, the strength of this relationship may be different in accordance with the group-level climate. In this respect, we tried to examine the empowerment climate as a group-level variable with a multi-level research design, and found that an empowerment climate at the work group level reduced employees’ job stress and also affected the negative relationship between distributive justice and job stress at the individual level.
Theoretical and practical implications
The theoretical implications of our study are threefold. First, we empirically tested the possible relationship between distributive justice and job stress which had been suggested by some previous studies (e.g., Maslach, Schaufeli, & Leiter, Reference Maslach, Schaufeli and Leiter2001; Tepper, Reference Tepper2001; Halbesleben & Buckley, Reference Halbesleben and Buckley2004; Judge & Colquitt, 2004). Additionally, most previous studies have implicitly emphasized the role of procedural and interaction justice in the job stress phenomenon, and thus the effect of distributive justice on job stress has been somewhat neglected. However, we found that distributive injustice can be a stressor, and job stress explained the relationship of distributive justice and turnover intention.
Second, our study contributes to the literature of organizational justice and job stress, in that we tried to explore the theoretical explanatory mechanisms such as uncertainty management, social comparison, and job stress and/or burnout theories. In particular, we paid attention to the job resource–demand theory which argues that stress is a result of a mismatch of job demand and job resource (Hobfoll, Reference Hobfoll2001). With this theory, we posited that the reward or outcome employees received from their organization may be job resource and their effort or contribution may be job demand. Thus, when employees do not receive a reward commensurate with their input, they feel that they do not have an appropriate level of resources compared to the job demand, and thus feel job stress.
Third, we showed the cross-level main and interactional effect of empowerment climate on the relationship between distributive justice and job stress at the individual level. Consistent with prior studies, the employees’ job stress was mainly influenced by their own perception of distributive justice; however, empowerment climate also influenced their job stress.
Additionally, our study suggests some practical implications. First, organizations must pay attention to the work group-level climate (Mayer et al., Reference Mayer, Nishii, Schneider and Goldstein2007; Li & Cropanzano, Reference Li and Cropanzano2009). Managers need to understand that empowerment climate is critical to reduce the negative influence of distributive justice on job stress. As our results showed, empowerment can reduce job stress, and can mitigate the effects of distributive injustice on job stress through creating a positive group environment. Thus, it is important that organizations need to provide employees with the opportunity for them to be empowered, such as participation in decision making (Newton & Jimmieson, Reference Newton and Jimmieson2008). In addition, organizations should train managers on how to foster an empowerment climate in their work units.
Limitation and future research
We acknowledge some limitations. First, the possibility of common method bias and reverse causality must be considered when interpreting the results of relationships among variables. Our results may be over-inflated, given that all variables were obtained from a single source and measured by a self-reported questionnaire (Podsakoff et al., Reference Podsakoff, MacKenzie, Lee and Podsakoff2003). In addition, it is possible that job stress may be an antecedent causing the perception of distributive injustice. Therefore, we recommend that future research attempts to avoid the bias caused by a common method by acquiring data from multiple sources.
Second, this study may be limited in its potential to be generally applicable. Because we collected data from South Korean companies, one might question whether our findings can be applied to other nations with different cultures. Hence, future research, with ample data and with different cultures, should determine whether our results would be valid for other cultural contexts.
Third, a cautious interpretation is needed for understanding the results of this study, because we did not control the effects of procedural and interactional justice on job stress as well as turnover intention. Indeed, previous empirical studies, beside distributive justice, procedural, and interactional justice also are significantly related to job stress (Judge & Colquitt, 2004; Cole et al., Reference Cole, Bernerth, Walter and Holt2010) and turnover intention (Colquitt et al., 2000; Hausknecht, Sturman, & Roberson, Reference Hausknecht, Sturman and Roberson2011). Moreover, according to Cohen-Charash and Spector (Reference Cohen–Charash and Spector2001) and Mayer et al. (Reference Mayer, Nishii, Schneider and Goldstein2007), although distributive, procedural, and interactional justices are different constructs, these three types of organizational justice are strongly related. In this respect, in future research, procedural and interactional justice should be controlled to precisely isolate the effect of distributive justice.
Finally, we did not fully consider individual characteristics. For instance, individuals’ justice orientation (Liao & Rupp, Reference Liao and Rupp2005) and psychological contracts such as transactional contract (Rousseau, Reference Rousseau1989) may affect the influence of distributive justice on employees’ attitudes. Thus, it is necessary to further explore individuals’ personality or dispositional traits that might affect the relationship between distributive justice and job stress.
Conclusion
This study attempted to explain the theoretical reasons why an employees’ perception of distributive justice leads to turnover intention and how an empowerment climate at work groups influences. We found that job stress mediated the influence of distributive justice on turnover intention. Hence, organizations must fully understand that employees’ stress can be caused by distributive injustice, and in turn increase turnover intention. Moreover, our study emphasizes the empowerment climate as a useful vehicle for diminishing the negative influence of distributive injustice on job stress. As our results show, when work groups have a high level of empowerment climate, the negative influence of employees’ perception of distributive injustice on job stress was weak. In this regard, organizations have to pay more attention to encourage an empowerment climate in the work groups. Educating empowering leadership or job redesign to make employees feel empowered might serve to achieve an empowerment climate.
Acknowledgements
The authors would like to thank the editor and anonymous reviewers for constructive comments.
