The phenomenal growth of basic and applied research on organizational politics in the last 30 years attests to the importance of organizational politics from both the theoretical and practical perspectives (Miller, Rutherford, & Kolodinsky, 2008). Organizational politics is a pervasive workplace phenomenon and reflects the political climate in an organization (Cropanzano, Kacmar, & Bozeman, 1995; Ferris & Kacmar, 1992). Regarded as a situational variable, perceptions of organizational politics (POP) involve “an individual’s attribution to behaviors of self-serving intent, and is defined as an individual’s subjective evaluation about the extent to which the work environment is characterized by co-workers and supervisors who demonstrate such self-serving behavior” (Ferris, Harrell-Cook, & Dulebohn, 2000: 90). Not surprisingly, rich empirical evidence verifies POP’s destructive impact on organizational and employee outcomes. In particular, studies have identified the negative relationships of POP with affective commitment, job satisfaction, and job performance as well as positive relationships with burnout, job stress, and turnover intentions (Chang, Rosen, & Levy, 2009; Cropanzano, Rupp, & Byrne, 2003; Miller et al., 2008; Vigoda-Gadot & Talmud, 2010). POP research on Chinese organizations is particularly important because acting in one’s self-interest is strongly regarded as unethical behavior in Chinese societies (Resick, Martin, Keating, Dickson, Kwan, & Peng, 2011). Hence, Chinese employees could be very sensitive and reactive to the destructive influences of POP.

Moreover, we know little about the effects of organizational politics on employees’ voice behavior, a type of citizenship behavior that highlights the expression and communication of constructive challenges aimed at improving a given situation (Van Dyne & LePine, 1998). Research on voice behavior is particularly important in Chinese societies, where employees have a high degree of power distance orientation that represses the exercise of voice with regard to work-related issues (Zhang, Zhou, Wang, & Cone, 2011). It has also been suggested that voice influences organizational effectiveness in dynamic environments (Van Dyne & LePine, 1998). As China is currently experiencing an economic transformation that has resulted in a dynamic environment, the encouragement of voice is particularly critical for Chinese organizations looking to enhance their competitive advantage (Zhang et al., 2011). One way to encourage voice is to wipe out hampers. Because POP may provide destructive effects on voice, understanding the relationship between POP and voice is theoretically and practically important to facilitate voice. To explore such relationship, this study sought answers to the following two questions. Is POP related to employees’ voice behavior? If so, what is the mediating mechanism that can explain the relationship between the two?

To explain why POP is related to voice behavior, this study examines the mediating role of psychological safety, the belief that exhibiting risky behaviors such as voice will not cause personal harm (Edmondson, 1999). Although studies have applied strain, job satisfaction, and affective commitment to link organizational politics with work behaviors (Chang et al., 2009), such mediating mechanisms cannot fully explain the impact that organizational politics on employees’ voice behavior. Unlike other kinds of citizenship behavior that promote cooperation spirit, voice behavior may create destructive forces to the organization and cause high personal costs for performers, thus leading employees to be scared to speak out (Turnley & Feldman, 1999). This argument indicates that the key to facilitating voice is to lead employees to perceive safety, which is not captured by strain, satisfaction, and commitment. Research suggests that employees speak out when they are comfortable with interpersonally threatening actions (Edmondson, 1999). Informed by such affect-laden cognitive perspective (Detert & Burris, 2007), we believe that a psychological safety model is readily applicable to organizational politics’ effects on voice behavior.

Furthermore, the organizational politics framework suggests that individuals may have a buffer to alleviate the negative impact of organizational politics; consequently, organizational politics does not influence employees equally (Ferris, Adams, Kolodinsky, Hochwarter, & Ammeter, 2002). Hence, we address the question, for what types of employees are POP’s effects on psychological safety and voice behavior buffered? We argue that perceived insider status may provide this buffering effect. Perceived insider status refers to one’s perception of insider (versus outsider) status within the organization, focusing on the relationship between employees and the organization, and reflecting the extent to which individuals have acquired a personal space and acceptance in the organization (Masterson & Stamper, 2003). This study highlights that perceived insider status neutralizes the adverse impact of POP on psychological safety and voice behavior. Taken together, we present a moderated mediation model considering why POP is related to voice behavior by studying the mediating role of psychological safety and how the relationship emerges by investigating the moderating role of perceived insider status.

The present study contributes to the literature on organizational politics and voice behavior in several major ways. First, from the affect-laden cognition perspective, this investigation extends the POP research to include voice behavior, which provides a theoretical framework for assessing a key political climate antecedent of voice. The examination of psychological safety can generate knowledge about how POP impacts employee voice, thereby increasing scholars’ and managers’ understanding of the underlying mechanism through which employees assess their highly political environment and then adjust their voice behavior. Second, we explore the moderating effect of perceived insider status. Investigating how moderators affect voice associated with POP is critical as organizational politics is difficult to avoid (Ferris & Kacmar, 1992); thus, understanding its boundary conditions may offer a suitable way to alleviate the destructive impact of POP. Finally, the current research applies a time lagged and multiple-source design in China in response to a call to address the concerns caused by cross-sectional data regarding voice (Tangirala & Ramanujam, 2008). The following discussion reviews relevant literature and presents specific hypotheses before reporting the results of a multi-wave, multi-source study.

Theoretical foundation and hypotheses

Voice behavior highlights the expression and communication of constructive challenge aimed at improving the situation (Van Dyne & LePine, 1998), thereby creating continuous improvement and disruptive forces to the organizational status quo (Nemeth & Staw, 1989). However, voice requires substantial social-political effort to promote new ideas or concerns and the perception of personal influence over other members (Detert & Burris, 2007). Voice also demands that employees persuade others to accept new ideas, which may disrupt the original routine and be viewed as a threat to organizational harmony or a challenge to authority (Milliken, Morrison, & Hewlin, 2003). Such disruptive forces may lead employees to suffer from backfire including the loss of trust, respect, promotion or other career opportunities as their suggestions may harm the benefits of and offend other people (Milliken et al., 2003). Because the targets of upward voice own reward and coercive power, employees who face uncertainties and fear substantial losses from voice are likely to remain silent (Milliken et al., 2003). Organizational environment characteristics tend to be particularly critical cues that employees use in evaluating whether voice is harmful to their career (Tangirala & Ramanujam, 2008). Apply the affect-laden cognition perspective to voice behavior, we contend that employees who have intentions to voice are likely to scan the environment, interpret environmental cues, evaluate the benefits and costs of voice, and finally make the decision to speak out or keep silent.

Because employees evaluate personal costs before speaking out, the organizational environment that directs employees to predict the voice outcome strongly influences employee voice behavior (Turnley & Feldman, 1999). Not all organizations clearly define the specific environment in which employees work; consequently, rules for the exchange and cues for the environment may be unclear in some organizations, particularly in highly political organizations where employees adopt a competitive and self-serving style as well as band into small groups to acquire extra personal resources (Cropanzano, Howes, Grandey, & Toth, 1997). Because rewards are distributed on the basis of power, the rules may change frequently. Such a volatile and unpredictable environment diminishes employees’ confidence that their energy and efforts will result in positive consequences (Cropanzano et al., 1997). Because POP facilitates uncertainty in terms of the relationship between performance and rewards, it is an important constraint that often discourages employees from taking risks (Hochwarter, Witt, & Kacmar, 2000) or allocating extra individual resources, such as expressing concerns or ideas, to the workplace (Cropanzano et al., 1997). In addition, POP makes employees experience difficulties with understanding the work environment, thereby creating a threat. To scrutinize POP within a theoretical framework considering an underlying process through which POP affects voice, we draw on the mediating role of psychological safety to investigate POP’s effects.

POP, psychological safety, and voice behavior

As previously mentioned, psychological safety is the belief that exhibiting risky behaviors, such as voice, will not cause personal harm (Edmondson, 1999). Research has suggested that psychological safety can exist at both the individual (Ashford, Rothbard, Piderit, & Dutton, 1998; Detert & Burris, 2007) and group levels (Edmondson, 1999; Walumbwa & Schaubroeck, 2009). As our model focuses on personal costs associated with voice, we only examine individual-level psychological safety in the current study.

Because highly political organizations are full of uncertainty, thus preventing employees from obtaining key cues for understanding the causal mechanism between actions and future outcomes, the behavior-outcome relationship appears to be a “black box” characterized by incomplete information. Consequently, employees are less likely to acquire meaningful cues to unravel the complexities and lose their prediction abilities to link their actions and future consequences and, in turn, categorize this situation as a negative and uncontrollable threat (Jackson & Dutton, 1988). Facing this threat, employees will perceive that a high risk exists in honest communication. Research has contended that low access to information is likely to increase insecurity and defensiveness (Edmondson, 1999).

Moreover, employees who lack the courage are unlikely to challenge the organization when they perceive that their organization is unwilling to accept their opinions (Hornstein, 1986). When organizations are highly political and full of uncertainties and conflicts, employees will perceive that high risk exists in honest communication. Conversely, when organizations are characterized by certainty and harmony, employees will perceive low risk and conclude that the organization constitutes a safe work environment, ultimately provoking their psychological safety. A meta-analytic study has indicated that POP leads to psychological strain (Chang et al., 2009). Hence, we propose:

Hypothesis 1

POP is negatively related to psychological safety.

Psychological safety is influential on voice behavior because speaking out is relatively unsecured based on the fact that voice is not included in the formal reward system and can cause negative career outcomes, such as decreased promotion opportunities (Milliken et al., 2003). Unlike in-role job behaviors in which employees need to maintain a certain level to secure their jobs, voice—similar to other extra-role behaviors—is discretionary such that employees have the freedom to adjust their performance without putting themselves in a risky situation (Chen, Hui, & Sego, 1998). Employees may thus withhold their usual voice when they believe that their efforts on voice become a risky investment in highly political organizations where the voice of employees may spur higher opportunities for the employees to suffer backfire. Conversely, employees with high levels of psychological safety perceive little risk to their own interests in demonstrating voice behavior, ultimately facilitating their voice behavior (Walumbwa & Schaubroeck, 2009).

Moreover, psychological safety represents trust and attachment to the organization, which triggers employee voice behavior (Walumbwa & Schaubroeck, 2009). Social exchange theory suggests that individual behaviors are highly influenced by the quality of the relationship between the individual and the organization (Blau, 1964). When employees perceive their organizations as being psychologically safe, they are likely to think about their relationship with the organization in terms of social exchange rather than economic exchange, thereby reciprocating for such high quality treatment as voice behavior. This suggests a positive relationship between psychological safety and voice behavior.

Further, the present study hypothesizes that the relationship between POP and voice behavior is mediated by psychological safety. Research on POP has long argued that the POP–outcome nexus should be examined in its entirety, including mediating effects, in order to unravel the complexities in POP’s effects (Ferris et al., 2002). Hence, in advancing a framework that links POP and voice behavior, we look at the mediating role of psychological safety.

Organizational environment characteristics are particularly critical for employees deciding if they should engage in voice behaviors (Tangirala & Ramanujam, 2008). Highly political organizations are likely to lead individuals to emphasize the negative and uncontrollable aspects of an issue (Ferris et al., 2002), and have substantially less psychological safety. Consequently, in a less psychologically safe work environment, individuals tend to perform behaviors associated with the conservation of resources and tight control mechanisms, which are related to action restriction and manifested in the maintenance of the status quo (Staw, Sandelands, & Dutton, 1981). In other words, employees who lack psychological safety resulting from a political work environment are less likely to challenge the organization when they perceive low psychological safety. As such, POP becomes an important constraint that discourages employees from expressing their concerns and other ideas via psychological safety. Evidence for the mediating role of psychological safety has been accumulating. For example, psychological safety was found to mediate the effects of change-oriented leadership (Detert & Burris, 2007) and ethical leadership (Walumbwa & Schaubroeck, 2009) on follower voice behavior. However, these studies focus on positive leadership but have not highlighted the effects of unfavorable organizational climate on psychological safety and voice. In the context of organizations, politics is pervasive and undermines employees’ psychological safety. Hence, POP may represent the most theoretically relevant construct for linking an unfavorable work environment with reduced psychological safety and voice. Hence, we propose:

Hypothesis 2

Psychological safety mediates the relationship between POP and voice behavior.

The moderating role of perceived insider status

The current research proposes that perceived insider status may provide a boundary to POP’s effects. Research has suggested that organizations can increase employees’ perceptions of control and alleviate the negative effects of uncertainty that are associated with POP (Miller et al., 2008). Along with this line, we argue that perceived insider status can help employees to have perceptions of identifying decision rules and holding extra resources to mitigate the uncertainty associated with unfavorable situations (Masterson & Stamper, 2003), thereby buffering POP’s negative effects on psychological safety and voice. Organizations facilitate a differentiation of their employees (insiders versus outsiders) by using rewards or inducements to send signals to certain employees that they have acquired insider status (Stamper & Masterson, 2002). Perceived insider status can lead individuals to trust the authority and control the environment. In addition, using a good relationship with the organization as protection, insiders can obtain extra organizational support (Stamper & Masterson, 2002), which acts as a buffer by upholding employees’ further risky behavior such as voice. Research has provided evidence for the claim that social support and trust in fellow workers buffer the destructive effects of POP on job outcomes (Vigoda-Gadot & Talmud, 2010).

To sum up, when employees experience reduced uncertainty by perceiving themselves as insiders, POP may not be that much of a threat, thwarting the detrimental effects of POP on their psychological safety and voice. It is thus expected that the negative links of POP to employees’ psychological safety and voice will be alleviated when perceived insider status is high, creating a relatively favorable context in which employees perceive safety to speak out. Conversely, the negative links will be stronger when perceived insider status is low, providing a context in which employees feel risky to voice. Thus, we propose:

Hypothesis 3

Perceived insider status moderates the negative relationship between POP and psychological safety as well as the negative relationship between POP and voice, such that the relationships are weaker for employees with high levels of perceived insider status than for those with low levels of perceived insider status.

The prior arguments represent an integrated framework in which psychological safety mediates the negative relation between POP and voice and insider status moderates the relation between POP and psychological safety as well as the relation between POP and voice. Based on this notion, insider status may also moderate the strength of the mediating mechanism for psychological safety in the relation between POP and voice—a moderated mediation model (Edwards & Lambert, 2007). The perceptions of insider status are important for depressing the indirect effect of psychological safety because insiders are inclined to have relevant knowledge and social support, thereby provoking psychological safety and, ultimately, voice behavior. It is thus expected that the indirect links of POP to employees’ voice behavior via psychological safety should be alleviated when perceived insider status is high. Conversely, when employees perceive themselves as outsiders, POP is more likely to hamper psychological safety and voice behavior in that outsiders have less knowledge and fewer other insiders to interpret the political situation than insiders do. Consequently, the indirect effect of psychological safety on voice should be stronger. Taking these arguments together, we propose:

Hypothesis 4

Perceived insider status moderates the mediating effect of psychological safety on the POP–voice relationship, such that the mediating effect is weaker when the level of perceived insider status is high rather than low.

Methods

Sample and procedures

This study was conducted using a nine-month lagged design in six electronic companies located in a major city of North China. Voice is very important in these knowledge-intensive companies because employees are expected to generate ideas to improve products as well as production procedures; organizations can take advantage of the ideas to enhance their sales and effectiveness. In designing our data collection time points, we reviewed the literature on employee psychological safety and voice behavior and found that prior longitudinal or lagged field studies revealed a wide disparity in time intervals for data collection, ranging from two weeks (e.g., Burris, Detert, & Chiaburu, 2008) to 10 months (e.g., Detert & Burris, 2007). We tested a moderated mediation model that involves both process and contingent effects; therefore, we followed Detert and Burris’s (2007) study and adopted a relatively long time interval (9 months) in order to allow respondents to observe, evaluate, make a decision, and take action.

In the first-wave survey, the subordinates provided information in relation to their demographics, POP, perceived insider status, and other control variables. In the second-wave survey, the subordinates rated psychological safety while their voice behavior was rated by their immediate supervisors. Each supervisor was asked to rate one to six subordinates. Because the independent variable (POP), mediator (psychological safety), and outcome variable (voice behavior) were collected from different sources and/or at different times, the potential problem of common method bias was largely alleviated (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003).

With the assistance of the human resource managers, questionnaires were administered to 338 randomly selected subordinates and their 128 immediate supervisors. Both subordinate and supervisor questionnaires had a code number that was used to match supervisor–subordinate dyadic data. Only the research team had access to the codes and individual responses. Respondents were informed that the survey aimed to examine human resource practices in the surveyed companies and were assured of confidentiality. Each respondent placed his/her completed survey in a sealed envelope and dropped it into a box set up in the human resource department.

At time one, 306 subordinate questionnaires were obtained. In the second-wave survey, questionnaires were distributed to these 306 subordinates and their 117 supervisors. This time, 290 subordinates and 114 supervisors responded to our survey. After we removed the unmatched and/or missing cases, the final sample comprised 283 matched supervisor–subordinate dyads, including 283 subordinates and 112 supervisors, representing valid response rates of 83.7 % and 87.5 % for subordinates and supervisors, respectively. On average, each supervisor rated 2.53 subordinates.

Of the 283 subordinates, 67.1 % were male; 27.9 % were between 31 and 35 years old, and 23.3 % were between 36 and 40 years old. The average organizational tenure was 7.71 years (SD = 7.05). Of the 112 supervisors, 72.3 % were men; 29.5 % were between 31 and 35 years old, and 29.5 % were between 36 and 40 years old. The average organizational tenure was 12.8 years (SD = 7.34).

Measures

The measures of POP, perceived insider status, psychological safety, and other control variables (except for demographics) used a seven-point Likert-type scale while the measure of voice behavior used a five-point Likert-type scale. We applied the five-point scale to the supervisors because, during the pilot study, some supervisors indicated that their time was limited for participating in this survey and a simple scale could save their time. Past research has shown that the validity of measurements is not affected by minor alterations to response formats (Jacoby & Matell, 1971; Wong, Peng, Shi, & Mao, 2011). Following the commonly used back-translation procedure, the scales were first translated from English into Chinese by one of the authors and then back-translated into English by two independent bilingual individuals to ensure the equivalency of meaning (Brislin, 1980). The results of back-translation satisfactorily showed that all important words were included in the back-translated version.

Perceptions of organizational politics (POP)

A 12-item scale developed by Kacmar and Ferris (1991) was used to measure employees’ POP. Response options ranged from 1 (“strongly disagree”) to 7 (“strongly agree”). A sample item was “Favoritism rather than merit determines who gets ahead in this organization.” Cronbach’s alpha was .91.

Perceived insider status

We used a six-item scale developed by Stamper and Masterson (2002) to measure perceived insider status. Response options ranged from 1 (“strongly disagree”) to 7 (“strongly agree”). A sample item was “I feel I am an ‘insider’ in my work organization.” Cronbach’s alpha for the measure was .82.

Psychological safety

We used a three-item scale developed by Detert and Burris (2007) to measure psychological safety. Response options ranged from 1 (“strongly disagree”) to 7 (“strongly agree”). A sample item was “It is safe for me to speak up around here.” Cronbach’s alpha was .80.

Voice behavior

We used a six-item scale developed by Van Dyne and LePine (1998) to measure employees’ voice behavior. Response options ranged from 1 (“strongly disagree”) to 5 (“strongly agree”). A sample item was “This particular follower develops and makes recommendations concerning issues.” Cronbach’s alpha was .80.

Control variables

We controlled for subordinate demographics (age, gender, and organizational tenure), leader–member exchange (LMX), job satisfaction, and positive affectivity that may be associated with psychological safety and/or voice behavior (e.g., Botero & Van Dyne, 2009; Burris et al., 2008; De Cremer, Cornelis, & Van Hiel, 2008; Janssen, de Vries, & Cozijnsen, 1998; LePine & Van Dyne, 1998; Van Dyne, Kamdar, & Joireman, 2008). Age was coded as “1” = under 26, “2” = between 26 and 30, “3” = between 31 and 35, “4” = between 36 and 40, “5” = between 41 and 45, “6” = between 46 and 50, “7” = between 51 and 55, “8” = between 56 and 60, and “9” = above 60. Male was coded as “0” while female was coded as “1.” Organizational tenure was self-reported in years. LMX was measured using a seven-item scale developed by Graen, Novak, and Sommerkamp (1982). Cronbach’s alpha for this measure was .87. Job satisfaction was measured using a four-item scale developed by Cammann, Fichman, Jenkins, and Klesh (1983). Cronbach’s alpha for this measure was .80. Positive affectivity was measured using a 10-item scale developed by Watson, Clark, and Tellegen (1988). Cronbach’s alpha for this measure was .91. In addition, because our data were collected from six companies, we created and controlled for five dummy variables to rule out the potential company effect.

Results

Nesting effect analysis

As 112 supervisors provided ratings of voice behavior for 283 subordinates, which may cause a nesting effect (i.e., the supervisor’s rating of one subordinate may influence his/her rating of another one), we tested this potential effect by performing a one-way ANOVA using supervisor code as the independent variable and voice behavior as the dependent variable. According to the results, the one-way ANOVA was not significant for voice behavior, indicating that supervisor ratings were relatively independent and did not significantly influence the results of the study. Therefore, we applied hierarchical multiple regression to test our hypotheses.

Attrition analysis

We followed Goodman and Blum’s (1996) approach to test whether systematic differences existed between the first and second data collection times. First, a multiple logistic regression was conducted using survey time as the dependent variable and employee demographics, LMX, job satisfaction, positive affectivity, POP, and perceived insider status as independent variables. The results of the multiple logistic regression showed that all logistic regression coefficients were non-significant. Moreover, t-tests were performed to determine whether significant mean differences existed in demographics, LMX, job satisfaction, positive affectivity, POP, and perceived insider status at Time 1 and Time 2. The results of the t-tests indicated no significant mean difference for these variables. Taken together, these results suggest that the respondents randomly dropped out of the study.

Confirmatory factor analyses

Confirmatory factory analyses (CFA) were conducted to evaluate the validity of the key variables using AMOS 17.0. Given the small sample size (283) relative to the measurement items (48), we adopted procedures used in previous research (Mathieu & Farr, 1991) by creating three indicators for each construct with more than three items. Specifically, based on the factor analysis results, the items with the highest and the lowest loadings for each construct were combined first, followed by items with the next highest and lowest loadings, until all the items had been assigned to one of the indicators.

We first examined a seven-factor CFA model including LMX, job satisfaction, positive affectivity, POP, perceived insider status, psychological safety, and voice behavior. The overall model’s chi-squared, the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA) were used to assess the model fit. The results indicated an acceptable model fit: χ 2 (168) = 320.74, p ≤ .01; CFI = .95, TLI = .94; RMSEA = .057. In addition, all the factor loadings for the latent constructs were significant, thereby confirming convergent validity.

Next, the discriminant validity of the seven suggested constructs was tested by contrasting the seven-factor model against several alternative models. As shown in Table 1, the fit indexes revealed that the hypothesized seven-factor model fit the data considerably better than any of the alternative models. Thus, the distinctiveness of the key constructs in the study was supported. Given these results, all proposed seven constructs were applied in further analyses.

Table 1 Results of confirmatory factor analyses
Full size table

Descriptive analyses

Table 2 presents the means, standard deviations, and zero-order Pearson correlations of all key variables.

Table 2 Means, standard deviations, and correlations
Full size table

Tests of hypotheses

We conducted hierarchical multiple regression analyses to test our hypotheses. Hypothesis 1 predicted that POP is negatively associated with psychological safety. As shown in Table 3, POP was negatively related to psychological safety (β = –.20, p ≤ .01, Model 2); thus, Hypothesis 1 was supported.

Table 3 Hypothesis testing
Full size table

Hypothesis 2 predicted that psychological safety mediates the relationship between POP and voice behavior. According to Baron and Kenny (1986), full mediation is supported if four conditions are met: (1) the independent variable (i.e., POP) is significantly related to the mediator (i.e., psychological safety); (2) the independent variable is significantly related to the dependent variable (i.e., voice behavior); (3) the mediator is significantly related to the dependent variable; and (4) when the mediator is present, the relationship between the independent and the dependent variable becomes non-significant. The first condition was met when we examined Hypothesis 1. The results in Table 3 demonstrate that POP was negatively related to voice behavior (β = −.21, p ≤ .01, Model 6), supporting condition 2. In addition, psychological safety was positively related to voice behavior (β = .37, p ≤ .01, Model 7), supporting condition 3. When psychological safety was present, the relationship between POP and voice behavior became non-significant (β = −.11, n.s., Model 10), supporting condition 4. Thus, Hypothesis 2 received support.

Hypothesis 3 predicted that perceived insider status moderates the relationship between POP and psychological safety as well as the relationship between POP and voice. As shown in Table 3, the interaction between POP and perceived inside status was positively related to both psychological safety (β = .14, p ≤ .05, Model 4) and voice (β = .16, p ≤ .05, Model 9). Thus, Hypothesis 3 was supported. We plotted the interactive effect using Aiken and West’s (1991) procedure—namely, computing slopes one standard deviation above and below the mean of the moderating variable as well as using the control variables to calculate the intercept. Figures 1 and 2 show that the interaction patterns are consistent with Hypothesis 3. Specifically, POP was negatively related to psychological safety when perceived insider status was low (r = −.34, p ≤ .01), but was unrelated to psychological safety when perceived insider status was high (r = −.06, n.s.). Similarly, the relationship between POP and voice was negative when perceived insider status was low (r = −.37, p ≤ .01), but was non-significant when perceived insider status was high (r = −.05, n.s.).

Fig. 1
figure 1

The interactive effect of POP and perceived insider status on psychological safety

Full size image
Fig. 2
figure 2

The interactive effect of perceived organizational politics and perceived insider status on voice behavior

Full size image

To examine the moderated mediation in Hypothesis 4, which predicts that perceived insider status moderates the indirect effect of psychological safety, we performed a moderated path analysis, bootstrapping 1,000 samples to compute bias-corrected confidence intervals (Edwards & Lambert, 2007). Our results, which are summarized in Table 4, show that the size of the difference in the indirect effect of psychological safety was .04 with the 95 % confidence intervals computed using bootstrap estimates excluding zero. Specifically, the indirect effect of psychological safety on the relation between POP and voice was significantly weaker at a high level of perceived insider status, supporting Hypothesis 4.

Table 4 Results of the moderated path analysis
Full size table

Discussion

Although organizational politics is prevalent (Ferris et al., 2002) and most organizations recognize the necessity of upward voice as well as perceive organizational silence as a barrier to change and development (Morrison & Milliken, 2000), evident holes in the literature must be addressed. In particular, although it is widely assumed that unfavorable organizational climate will impact employee voice (Tangirala & Ramanujam, 2008), to date studies linking POP to voice behavior, or examining boundary conditions of POP’s effects, are in short supply. In addressing this gap in the literature, our findings not only provide evidence for a linkage between POP and voice, but also outline the mechanism through which the relationship operates (psychological safety) as well as depict a boundary condition (perceived inside status). The inclusion of mediators and moderators in a single model is also consistent with the past studies’ approach to specify a predictive model of voice behavior (e.g., Detert & Burris, 2007).

Contributions

This article makes several contributions to the growing literature on POP and voice behavior. First, the findings of the present research support the notion that psychological safety and voice behavior are more than a function of LMX, job satisfaction, and positive affectivity. Although these three control variables are significantly related to psychological safety and/or voice behavior in the current study, all proposed hypotheses were supported, suggesting that the consideration of POP and perceived insider status provided an additional explanation for psychological safety and voice behavior.

The strength of our psychological safety model of organizational politics is that it provides a generative model for future research that focuses on the outcomes of organizational politics. As previously noted, psychological safety represents an important underlying mechanism with numerous insights into organizational behavior that can be readily applied to future organizational politics research. For example, psychological safety leads employees to challenge status quo (Walumbwa & Schaubroeck, 2009), potentially leading to other unexplored work consequences resulting from organizational politics such as innovative behavior. Hence, aside from facilitating employees to speak out, psychological safety represents another critical mechanism through which one might exhibit other work behaviors.

Second, the moderating effects suggest a complex picture of the ways in which perceived insider status impacts the POP–psychological safety–voice relationship. We predicted that insiders would be less sensitive to POP because they are more likely to obtain inducements such as advice, information, training, and power to identify the rules of reward allocation and decision making. Consequently, they are more likely to have interpretative perspectives to guide their future behaviors. Insiders are thus inclined to have relatively high levels of psychological safety and are more likely to speak out in the highly political organization than outsiders are. To the best of our knowledge, our study is the first attempt to use the psychological safety perspective to explain the link between organizational climate and voice by considering the boundary condition of insider status.

Third, our study sought to engage in theory building by positioning psychological safety as a mediator (cognitive mechanism) of the relationship between POP and voice within a more broadly moderated mediation model. We demonstrated that employee psychological safety is a fundamental psychological route to voice. Thus, by focusing on mediation and moderation together, our model helps better explain how POP relates to voice as well as for whom POP will have the largest relationship with psychological safety and voice. In doing so, our study not only provides rationales and evidence for the claims that organizational climate have a unique relationship with voice, but also extends our understanding of how such a relationship works.

Finally, the current investigation replicates past findings in the Western context (e.g., Detert & Burris, 2007) regarding the relationship between psychological safety and voice behavior in the Chinese context, thereby promoting the external validity of voice research across cultures.

Strengths and limitations

This study has two strengths that are worth noting. First, we collected the data in two phases to provide solid evidence of the causal relationships, which cannot be achieved using a cross-sectional research design. Second, the assessment of employee voice behavior by immediate supervisors rather than by employees themselves helped alleviate the problem of common method variance (Podsakoff et al., 2003).

However, despite these strengths, our investigation has several limitations that are important to note. First, the present findings may be contaminated by common method variance as the data related to POP and psychological safety came from the same source. However, as suggested by Podsakoff et al. (2003), we collected those data at different times and controlled for demographic variables to attenuate such bias. Hence, we believe that common method variance had little or no effect on our empirical results. Second, a constraint on the generalizability of the present findings was the use of manufacturing enterprises. Future research should compare the findings from manufacturing with other industries. Third, this study adopted a general voice behavior measure, although some researchers divide voice behavior into two dimensions: promotive and prohibitive voice behavior (Liang, Farh, & Farh, 2012). It is possible that POP and psychological safety are related to these two voice dimensions in different ways. However, the findings of Liang et al. (2012) indicate that psychological safety is positively related to temporal changes in both types of voice behavior with a similar degree of magnitude, thereby alleviating concerns that it exerts different effects on promotive and prohibitive voice. Nevertheless, future research is needed to examine the effects of POP on both types of voice. Finally, this investigation was conducted in China, resulting in the concern that the significant findings may not be generalized to Western countries. As Chinese people have a high degree of collectivism and focus on mutual relationships, the effects of perceived insider status on the Chinese may be stronger than on the Western people. Hence, it is desirable to replicate our results using Western samples.

Practical implications

In practical terms, we believe that the results reported in this study are important to organizations because they expand our understanding of the predictors of voice. Our findings reveal that voice is a result of situational factors (i.e., POP, perceived insider status) via a psychological mechanism (i.e., psychological safety). Thus, our work provides several paths by which managers and organizations can propel the occurrence of employee voice behavior. The first path suggested by our investigation is to take steps to discourage POP. As individuals respond to organizational politics depending on their perceptions, organizations may need to foster a certain environment, transparent procedures, two-way information flow, open communications and a fair reward system to minimize POP (Ferris et al., 2002).

The second path to increasing voice is to promote employees’ perceptions of insider status. Managers should become aware of the differential perceptions of insider (versus outsider) status and find ways to direct employees to experience insider status (Lapalme, Stamper, Simard, & Tremblay, 2009). For example, managers can promote mentoring and training programs to facilitate employees’ understanding of their work environment and strengthen their perception of insider status (Stamper & Masterson, 2002). In addition, our findings offer implications for selection. Prior research shows that individuals who are high in proactive personality tend to have high perceived insider status (Kim, Hon, & Crant, 2009). Hence, organizations should recruit employees with a high degree of proactive personality.

Finally, the finding that psychological safety is a mediator suggests that organizations can cultivate voice by facilitating psychological safety. This implication is important because POP is prevalent and not all employees can regard themselves as insiders. By bolstering psychological safety, managers can mitigate the negative relationship between POP and psychological safety through other means. For example, research shows that ethical leadership and change-oriented leadership are related to higher psychological safety levels (Detert & Burris, 2007; Walumbwa & Schaubroeck, 2009). This suggests one way in which organizations can attempt to counteract any decrease in psychological safety levels—namely, to ensure that adequate ethical leadership and change-oriented leadership exist.

Conclusion

Voice behavior is important for organizational effectiveness (Detert & Burris, 2007). The present investigation has provided insights into crucial issues regarding POP and voice behavior, revealing perceived insider status as a key moderator and psychological safety measure for being an important mediator of the relationship between POP and voice. This study has critical implications for the organization because it suggests several ways that the organization can follow to depress the destructive impact of POP and encourage employees to speak up. As such, this research can serve as a springboard for future research into additional variables and the underlying processes that facilitate voice.