Introduction

Employee voice behaviors – defined as workers’ expression of their constructive suggestions, concerns, or thoughts to the organization they work for (Van Dyne, Ang, & Botero, 2003) – have been considered to be critical to organizational functioning and performance (Grant, 2013; Kim, MacDuffie, & Pil, 2010; Morrison, 2011). Extensive research has shown that employees who exhibit voice behaviors benefit their organizations by helping to create better working environments (Hirschman, 1970), enhance organizational efficiencies (LePine & Van Dyne, 1998), improve operating effectiveness (Morrison, 2011), and reduce employee turnover (McClean, Burris, & Detert, 2013). Due to the importance of voice behaviors for employee and organizational success, most research on voice behaviors is devoted to identifying the factors that can promote voice behaviors (e.g., LePine & Van Dyne, 2001; Morrison, Wheeler-Smith, & Kamdar, 2011; Tangirala & Ramanujam, 2008; Venkataramani & Tangirala, 2010). However, it is still unclear how, why, and under what conditions job characteristics, such as job autonomy, are related to employee voice behaviors.

Several studies have established the important role of voice in organizational success and in preventing harmful events from damaging the organization; however, there are still several significant research gaps in the current voice literature. First, although researchers have examined a wide range of antecedents of voice, such as Chamberlin, Newton, and Lepine’s (2017) identification of contextual and individual factors, there is still a relative lack of understanding of how to enhance employees’ voice behaviors from the viewpoint of psychological factors in the workplace. Specifically, intrinsic motivational factors (e.g., psychological needs) are crucial for human behaviors in the workplace and may potentially predict voice behaviors (Chamberlin et al., 2017; Liang, Farh, & Farh, 2012). In particular, employee voice is a type of extra-role behavior that is related to intrinsic motivation (Liang et al., 2012). However, the link between employee inherent needs and voice behaviors is still unexamined.

Second, the vast majority of voice researchers have operationalized voice as a unitary construct, focusing only on the constructive aspect of voice behaviors (Morrison, 2011; Ng & Feldman, 2012; Van Dyne & LePine, 1998); however, the nature of voice entails not just proactive suggestions for improvement but also inactive suggestions for how to maintain the status quo without damage to the organization (LePine & Van Dyne, 1998; Lin & Johnson, 2015; Song, He, Wu, & Zhai, 2020). Additionally, a recent meta-analysis found that different types of voice behaviors lead to different outcomes (e.g., the promotive voice behavior leads to better job performance than prohibitive voice behaviors) (Chamberlin et al., 2017). Thus, there is a need to more fully understand the effects of psychological factors on promotive and prohibitive aspects of voice.

Finally, given that voice behaviors are of a proactive nature, the perception of the current organization may provide the boundary condition of the effect of employee motivation on voice behaviors. Organizational factors are considered as key contextual factors that influence the link between individual characteristics and employee voice behaviors (Zhou & George, 2001). For example, Venkataramani and Tangirala (2010) found that work-group identification can strengthen the positive effect of personal influence toward group decision on voice behaviors. One contextual factor that likely plays a role in voice relationships that has not yet been examined is person-organization fit (P-O fit). P-O fit refers to the congruence between the values of employees and the values and norms of organizations (Lauver & Kristof-Brown, 2001). High P-O fit relates to enhanced organizational commitment, increased job performance, high job satisfaction, and reduced turnover (e.g., Han, Chiang, McConville, & Chiang, 2015). However, P-O fit as an evaluation of organizations, which may influence the relationship between psychological needs and employee voice behaviors, has been ignored in the voice literature.

Aiming at addressing these research gaps, the present study contributes to the employee voice literature in three primary ways: First, in line with Self-Determination Theory (SDT) (Deci & Ryan, 1985) and the Job Demands-Resources (JD-R) Model (Demerouti, Bakker, Nachreiner, & Schaufeli, 2001), we focus on aspects of an individual’s needs and suggest that autonomy, as a psychological need, is indirectly related to employee voice behaviors. Intrinsic needs in the workplace are related to engagement, and need fulfillment is necessary for humans to effectively function (Deci & Ryan, 2008), which implies a positive impact of need satisfaction on engagement and ultimately employees’ voice behaviors. Hence, we examine the link between psychological needs (i.e., autonomy) and voice behaviors and the role of motivational factors as mechanisms for how the fulfillment of human needs can influence voice behavior enactment. Second, to better understand the influence of intrinsically psychological needs on distinct voice behaviors rather than a sole construct of voice behaviors, we investigate the influence of job autonomy on both promotive and prohibitive voice behaviors. Liang et al. (2012) proposed two different types of voice: promotive voice and prohibitive voice. Promotive voice refers to speaking up with innovative ideas or recommendations to enhance organizational functioning, whereas prohibitive voice refers to expressing concerns with regard to work processes or work-related issues to avoid harmful incidents from organizations. Thus, to help us understand the distal psychological predictors and motivational mechanism for all aspects of voice behaviors, this study adopts Liang et al.’s (2012) promotive and prohibitive behaviors in our proposed model. Last, to help build our understanding of organizational conditions that best support employee voice behaviors and help organizations optimize the effects of work engagement, we examine the role of P-O fit as a supportive factor in the relationship between engagement and employee voice behaviors. Research suggests that perceptions of fit are related to a variety of important organizational factors, such as organizational attitudes, commitment, and task and contextual performance (e.g., Goodman & Svyantek, 1999; Han et al., 2015; Verquer, Beehr, & Wagner, 2003). Thus, this study proposes that person–organization fit (P-O fit) serves as a boundary condition for the relationships between job autonomy, engagement, and employee voice behaviors, which can strengthen the direct effect of engagement and the indirect effect of job autonomy through engagement on employee voice behaviors. Consequently, in the current study, a conditional indirect relationship, depicted in Fig. 1, is proposed.

Fig. 1
figure 1

Moderated mediation model investigating the links between job autonomy, work engagement, person-organization fit (P-O fit), and employee voice behaviors

Full size image

Theoretical Foundations

Self-Determination Theory

The self-determination theory (SDT) developed by Deci and Ryan (1985, 2000) postulates that individuals have intrinsic needs that must be fulfilled to help them develop and grow towards their fullest capacity through accepting challenges, taking responsibilities, and striving to fulfill their own interests. SDT suggests three specific intrinsic needs: (a) the need for autonomy, which refers to the need to make one’s own decisions and behave according to one’s own volition; (b) the need for relatedness, which indicates the need to be mutually respected and engaged in interpersonal relationships; and (c) the need for competence, which refers to the need to complete tasks successfully and to obtain achievement and satisfaction in one’s activities.

Accordingly, a condition or environment that satisfies these three vital psychological needs will facilitate an individual’s personal growth and the development of subsequent positive individual outcomes (e.g., motivation, task and contextual performance, and well-being). Conversely, a condition or environment that hinders these inherent needs will impede personal development which may lead to negative behaviors (e.g., increased absenteeism and turnover intention) and psychological states (e.g. experience of depression or anxiety) (Deci, Connell, & Ryan, 1989). For instance, Benedetti, Diefendorff, Gabriel, and Chandler (2015) found that intrinsic motivation for pursuing work tasks was associated with higher levels of vitality and job satisfaction. This suggests that human autonomy, a component of intrinsic motivation, is related to greater well-being because people engage in workplace activities out of their own will. Within the current study, we focus specifically on the need for autonomy.

Job Demands-Resources Model

In a similar vein, the job demands-resources (JD-R) model provides additional theoretical support for our model and posits that job resources, such as physical, psychological, social, or organizational characteristics of the job (i.e., skill variety, task identity, task significance, job autonomy, and feedback), are related to the level of self-motivation, which in turn influences individual well-being and job performance (Demerouti et al., 2001). Particularly, research applying the JD-R model has found that job autonomy – as a job resource that affords an individual a certain level of freedom (Langfred, 2005) – can influence engagement and subsequent job performance (Bakker & Demerouti, 2007; Mauno, Kinnunen, & Ruokolainen, 2007; Shantz, Alfes, Truss, & Soane, 2013). Similarly, research has indicated that job autonomy is associated with high work productivity and job satisfaction through the induction of intrinsic work motivation (De Cooman, Stynen, Van den Broeck, Sels, & De Witte, 2013). Moreover, P-O fit can be also seen as a job resource which can influence the employee motivation–job performance relationship (Alfes, Shantz, & Alahakone, 2016).

Within the current study, we provide a direct test of the JD-R job resource assertions within the framework of SDT: due to the fulfillment of autonomy-related needs, autonomy is an employee inner need that is related to increased work engagement and ultimately contextual performance (employee voice behaviors in this study), while P-O fit, a function of personal and environmental factors, is a secondary job resource that can work with autonomy to further stimulate employees’ contextual performance.

The Direct Effect of Job Autonomy on Work Engagement

Job autonomy refers to employees’ inherent need to have a choice in initiating and regulating their own actions (Deci et al., 1989). According to SDT, people are likely to be content when autonomous needs are satisfied and when they have volition regarding their behaviors. Moreover, Gagné and Deci (2005) suggested that factors within a work environment, such as job duties and context (e.g., job autonomy as a critical job characteristic), can facilitate autonomous motivation and further enhance employee job performance. That is, job autonomy, which is an important job characteristic and can be considered a vital component of job design, helps to satisfy the need for job-related autonomy and promote autonomous motivation, which can ultimately enhance employees’ work-related outcomes (Morgeson, Delaney-Klinger, & Hemingway, 2005). Consequently, employees who report high levels of job autonomy have high satisfaction of the need for autonomy and tend to participate in expected behaviors such as in-role behavior (Grant, 2008).

Based on SDT, human behavior can be influenced by extrinsic and intrinsic motivations, which are classified based on the level of autonomy. With intrinsic motivation, in contrast to external motivation, behavior arises or a certain activity is performed because it is inherently satisfying. Intrinsic motivation is naturally autonomous, and research has found that high intrinsic motivation is associated with high job performance and increased well-being (Baard, Deci, & Ryan, 2004; Benedetti et al., 2015). Consequently, because high job autonomy represents a high level of self-determination, job autonomy helps to increase autonomous motivation (i.e., intrinsic motivation) (Moran, Diefendorff, Kim, & Liu, 2012). Moreover, the JD-R model posits that job resources in the workplace can increase individuals’ motivation (e.g., engagement with work and commitment to work); therefore, high job autonomy, as a psychological resource, likely further enhances employees’ intrinsic motivation at work (Bakker & Demerouti, 2007).

Employee engagement is a positive, fulfilling, work-related state of mind consisting of three components: (a) vigor, which is illustrated by vitality and persistence at work; (b) dedication, which is described by strong involvement in a job and a sense of enthusiasm, pride, and willingness to accept challenges; and (c) absorption, which is characterized by a pleasant mental state of being fully immersed in a job and the difficulty in detaching oneself from work (Schaufeli & Bakker, 2004). As such, engagement is a motivational construct (Rich, Lepine, & Crawford, 2010), as the employee’s autonomous and volitional investment of physical, emotional, and cognitive resources towards the job (Christian, Garza, & Slaughter, 2011) can be viewed as a type of intrinsic motivation. According to SDT (Gagné & Deci, 2005), autonomy is a way to help meet employees’ inherent needs and is significantly related to engagement. Furthermore, from the viewpoint of the JD-R model (Bakker & Demerouti, 2007), job autonomy as a job resource within an organization can stimulate individuals’ work engagement. We therefore propose the following hypothesis:

  • Hypothesis 1: Job autonomy will be positively related to work engagement.

The Indirect Effect of Job Autonomy on Voice Behaviors through Work Engagement

Research indicated that employees who display higher levels of autonomy are likely to have stronger persistence and involvement in their role tasks compared to those who display lower levels of autonomy (Gagné & Deci, 2005; Grant, 2008). Moreover, autonomy is related to increased intrinsic motivation, which in turn can influence individuals’ behaviors (Ryan & Deci, 2000). In working environments where individual needs are met, employees often exhibit high engagement, which can motivate them to further dedicate themselves to their work and improve their performance (Parker, Jimmieson, & Amiot, 2010). In addition, individuals have more freedom while working and shape their work in ways that allow them to perform at their best, which makes them more willing to exhibit pro-organizational extra-role behaviors beyond formal work requirements (Zhang & Chen, 2013).

More self-determined employees are, therefore, more likely to engage in their work and to be more willing to help coworkers solve problems they encounter and propose constructive suggestions that can improve organizational effectiveness (i.e., they exhibit increased voice behaviors). In a similar vein, employees who feel more autonomous in the workplace are more pro-socially motivated and more likely to volunteer to exhibit discretionary behaviors beneficial to the organization (Gagné & Deci, 2005; Spector, 1986). Recent researchers found that employees with high levels of autonomy (i.e., free choice in how they do their work) identify more with the organization and are more willing to exhibit pro-organizational extra-role behaviors such as organizational citizenship behaviors (Zhang & Chen, 2013). Conversely, if individuals perceive that they are less likely to be able to make decisions in their jobs, they will be less willing to exhibit important extra-role behaviors such as voice behaviors, wherein employees express constructive suggestions and concerns beyond their job duties and responsibilities in the workplace (Ng & Feldman, 2011; Van Dyne et al., 2003).

On the one hand, we argue that self-determined employees are more likely to engage in their jobs and further express new ideas and suggestions on how to improve the organizational functioning (i.e., a promotive voice behavior). Specifically, engaging in a promotive voice behavior involves proposing ways to gain outcomes or achieve ideal goals which can lead to positive outcomes for organizations (Liang et al., 2012). In particular, Parker, Williams, and Turner (2006) found that employees with autonomy are more willing to improve current workplace environments, such as by challenging the status quo via motivational mechanisms. Prior research has also found that individuals with higher job autonomy tend to expand their work to perform extra roles that are beneficial to the organization and regard speaking up to express concern about the organization as one of their responsibilities (Choi, 2007). On the other hand, we also argue that employees with a high level of job autonomy are more willing to devote their efforts and energies to their work and also to express concerns about problematic or inappropriate work-related processes and behaviors (i.e., a prohibitive voice behavior). Prohibitive voice behavior involves avoiding undesired situations and seeking to decrease mistakes so as to prevent possible losses (Liang et al., 2012). Thus, psychological needs may be related to both promotive voice behavior and prohibitive voice behavior (via engagement), because the former focuses on gaining desired outcomes, while the latter focuses on avoiding harmful factors that result in unexpected consequences. Drawing on SDT and the JD-R model, job autonomy as an inner need fulfillment may be able to intrinsically motivate employees like work engagement, which further lead to better pro-organizational behaviors, such as employee voice behaviors (Chamberlin et al., 2017). Consequently, we argue that more self-determined employees will be more willing to perform voice behaviors, including both promotive and prohibitive behaviors. Thus, this study hypothesizes that:

  • Hypothesis 2: Job autonomy will be positively related to (a) promotive voice behavior (b) prohibitive voice behavior indirectly through work engagement.

The Moderating Role of Person-Organization (P-O) Fit

Given that employees tend to satisfy their needs for autonomy and competence, which have significant influences on employee behaviors (Deci & Ryan, 1985), it is important to identify the influence of the congruence between employee needs and organizational attributes. Person-organization (P-O) fit is broadly defined as “the compatibility between people and organizations that occurs when (a) at least one entity provides what the other needs, or (b) they share similar fundamental characteristics, or (c) both” (Kristof, 1996, p. 4). More specifically, three forms of compatibility are identified: (1) an employee’s personal characteristics are congruent with the characteristics of the organization and other employees (i.e., supplementary fit); (2) an employee’s characteristics can fill gaps that are not filled by other employees (i.e., demands–abilities fit); and (3) an employee’s psychological needs can be satisfied by the characteristics of the work environment (i.e. needs–supplies fit). Christian et al. (2011) found, for example, that perceptions of P-O fit (e.g., demands–abilities fit and needs–supplies fit) can have an influence on individuals’ willingness to make an investment in their workplace. Overall, P-O fit, and particularly needs–supplies fit, represents the compatibility between the values, needs, and performance of a person and those of an organization (Cable & DeRue, 2002), which should provide conditional effects of work engagement on employee voice behaviors in our model (Maden-Eyiusta, 2016).

The perception of P-O fit as a moderator may influence the engagement–behavior relationship (Maden-Eyiusta, 2016). Specifically, employees who report high P-O fit may display strong work motivation and high satisfaction with their job because they believe that their values and abilities match with organizational values and job demands (Kristof-Brown, Zimmerman, & Johnson, 2005). Particularly, when employees think that the current organizational attributes fulfill their needs, they are more likely to be motivated by it and tend to put more effort into their job (Maden-Eyiusta, 2016). As such, P-O fit as a contextual factor may strengthen the effect of work engagement on employee behaviors. Moreover, an employee with high levels of P-O fit perceptions demonstrates high organizational identity, high organizational commitment, and low turnover intention (Greguras & Diefendorff, 2009; Hoffman & Woehr, 2006). In addition, employees with higher P-O fit perceptions found self-fulfillment and contentment within their status in the workplace; these individuals are willing to invest more effort and energy into their work and thus are more engaged at work (Gagné & Deci, 2005; Vansteenkiste et al., 2007). That is, high P-O fit perception suggests that a job is meaningful and congruent with the capabilities of the employees; it is also strongly associated with discretionary workplace behaviors that are advantageous to the organization, such as extra-role behavior (e.g., organizational citizenship behavior) (Hoffman & Woehr, 2006).

As such, in a high P-O fit work environment, employees are more likely to involve themselves in their job to produce positive job-related outcomes. These employees are more dedicated to their work and tend to exhibit an increase in voice behaviors concerning work-related issues to help improve the organization; thus, a multiplicative effect between work engagement and P-O fit on their extra-role behaviors such as voice behaviors, which are viewed as a form of constructive communication to improve organizational effectiveness, can be expected. Particularly, highly engaged employees who experience strong P-O fit may tend to value any opportunities to achieve ideal organizational goals and provide future-oriented suggestions (i.e., promotive voice). These employees may have a strong sense of responsibility to avoid any possible risks/mistakes to harm the organization and may provide prudent suggestions (i.e., prohibitive voice). Moreover, P-O fit can be regarded as a job resource, because employees are more likely to be attracted to and stay with an organization in which they feel they fit, and they are also more willing to perform work-related behaviors (Alfes et al., 2016). In line with the JD-R model (Bakker & Demerouti, 2007), when employees receive more resources, they will be more willing to devote their efforts and energies to their job (i.e., work engagement), which in turn should influence their job-related behaviors (e.g., Mäkikangas, Aunola, Seppälä, & Hakanen, 2016). Accordingly, P-O fit is expected to strengthen the positive effect of engagement on both promotional voice behavior and prohibitive voice behavior. Thus, this study hypothesized that:

  • Hypothesis 3: P-O fit perceptions will moderate the relationship between engagement and (a) promotive voice behavior and (b) prohibitive voice behavior. Specifically, the relationship is stronger when the perception of P-O fit is high vs. low.

Given that employees tend to seek an environment where their inherent needs (e.g., the needs for autonomy and competence) can be satisfied (Deci & Ryan, 1985), high perceptions of P-O fit represent a match between individual needs and organizational supplies (i.e., needs–supplies fit). Specifically, P-O fit can influence how psychological need satisfaction impacts employee attitude and behaviors (Malhotra, Sahadev, & Sharom, in press). Moreover, SDT suggests that employees have a need to make job-related decisions on their own. Thus, in a high P-O fit workplace, employees with high job autonomy should report the highest level of psychological need fulfillment and extra-role behaviors (Greguras & Diefendorff, 2009). Therefore, this study expects that self-determined employees with strong P-O fit perceptions will display increased promotive voice behavior and prohibitive voice behavior through work engagement. Thus, this study hypothesized the following:

  • Hypothesis 4: P-O fit will moderate the indirect effect of job autonomy on (a) promotive voice behavior and (b) prohibitive voice behavior (via work engagement). Specifically, the indirect relationship of job autonomy with (a) promotive voice behavior and (b) prohibitive voice behavior via engagement is stronger when the perception of P-O fit is high vs. low.

Method

Participants and Procedure

Data were collected at two time points (two weeks between Time 1 and Time 2) to limit common method variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Participants in the current study were invited to participate in this research project using part-time MBA students at National Chiao Tung University as organizational contacts. These contacts recruited study participants through an e-mail sent to employees at their organization, which clearly described the purpose of the research project and the details of the research procedure and assured potential participants of the privacy and confidentiality of their information. Employees interested in joining this research project could respond to this email. Participants who completed the two-wave questionnaire were compensated with a gift card (about 3 USD). At Time 1, the online questionnaire was distributed to 348 employees, of whom 296 employees completed it (response rate = 85%). At Time 2, the questionnaire was given to 296 employees, of whom 262 employees completed it (retention rate = 89%). Participants were from several industries in Taiwan, including manufacturing (55.73%), service (9.92%), financial (6.49%), construction (5.73%), and education (4.58%).

Of the participants, 48.09% were male and 51.91% were female. Of these, the majority of the participants were between 36 and 45 years old (38.17%) and worked in the technology (28.6%) and manufacturing (28.2%) industries. The participants’ tenures included more than 5 years (42.37%), 3–4 years (17.56%), 1–2 years (24.81%), and less than 1 year (15.26%). Most of the participants (46.18%) possessed a bachelor’s degree.

Measures

Since the questionnaire for this study was distributed in Taiwan, the original English scales needed to be translated into Chinese. To ensure that the translation accurately conveyed the meaning of the original English scale, we followed the back-translation procedure recommend by Brislin (1980). At Time 1, the variables of job autonomy and P-O fit were assessed. At Time 2, two weeks after Time 1, the variables of work engagement and employee voice behaviors (including promotive and prohibitive voice behavior) were assessed. All the study variables were assessed using a five-point scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

Job Autonomy

Job autonomy was measured with a three-item scale developed by Spreitzer (1995). An example item is “I can decide on my own how to go about doing my work.” Past research demonstrated that the scale had good validity and reliability (Aquino, Grover, Bradfield, & Allen, 1999). The Cronbach’s alpha for these items was .89.

P-O Fit

P-O fit was measured with three items adapted from Cable and DeRue (2002). A sample item is “My organization’s values and culture provide a good fit with the things that I value in life.” Previous research has demonstrated the appropriate criterion-related validity of this measure (Van Vianen, Shen, & Chuang, 2011). The Cronbach’s alpha for these items was .94.

Work Engagement

Work engagement was measured with the abbreviated version of the Utrecht Work Engagement Scale (UWES-9) (Schaufeli, Bakker, & Salanova, 2006). It is composed of three dimensions: vigor, dedication, and absorption. A sample item for vigor is “When I get up in the morning, I feel like going to work.” A sample item for dedication is “My job inspires me.” A sample item for absorption is “I get carried away when I am working.” Guarnaccia, Scrima, Civilleri, and Salerno (2018) demonstrated adequate reliability and validity of this measure. The Cronbach’s alpha for these items was .92.

Employee Voice Behaviors

Employee voice behaviors are composed of promotive voice behavior (five items) and prohibitive voice behavior (five items), measured with a scale developed by Liang et al. (2012). A sample item for promotive voice behavior is “I proactively develop and make suggestions for issues that may influence the unit.” A sample item for prohibitive voice behavior is “I advise other colleagues against undesirable behaviors that would hamper job performance.” Guo, Zhu, and Zhang (2020) indicated that the scale had good validity and reliability. The Cronbach’s alpha values for the variables of promotive and prohibitive voice behavior were .94 and .86, respectively.

Control Variables

Gender, age, education, tenure, and type of industry were included as control variables because previous voice research suggests that these demographic characteristics may influence employee voice behaviors (Chan, 2014; Takeuchi, Chen, & Cheung, 2012). In addition, we controlled perceived organizational support as a significant contextual factor which may influence employee voice behaviors (e.g., Tucker, Chmiel, Turner, Hershcovis, & Stride, 2008). In particular, the categorical variables (i.e., gender, education, and type of industry) were measured using continuous scales: gender was coded as 0 (female) and 1 (male); education was coded as 1 (junior high school and below), 2 (senior and vocational high school), 3 (junior college), 4 (university), and 5 (graduate school); and type of industry was coded as 1 (high technology), 2 (manufacturing), 3 (construction), 4 (medicine/public health), 5 (law/accounting), 6 (bank/financial industry), 7 (service industry/ food industry/ tourism industry), 8 (communication/ education), 9 (retail industry), and 10 (others).

Analytic Strategy

In the present study, we applied the latent moderated structural equations (LMS) approach to estimate our model (Klein & Moosbrugger, 2000) using Mplus 8.4. The LMS approach allows us to get more accurate and robust parameter estimates and confidence intervals with latent variables than the general regression with observed variables (Cheung & Lau, 2017). Moreover, because Mplus is not able to provide model fit indices (e.g., RMSEA and CFI) for the LMS approach, we used the loglikelihood difference test to assess the fit of our models (e.g., Breevaart & Bakker, 2018; Wayne, Lemmon, Hoobler, Cheung, & Wilson, 2017). Specifically, we estimated the chi-square difference based on the loglikelihood values and scaling correction factors between the model including the latent interaction term and the model excluding the interaction term (e.g., Breevaart & Bakker, 2018). Moreover, to examine the indirect effect in our model, we adopted Cheung and Lau’s (2017) approach to develop bias-corrected bootstrap confidence intervals (N = 2000) for each estimated parameter.

Results

Common Method Variance (CMV)

Given that all measures were self-reported and some measures were collected at the same time (job autonomy and P-O fit at Time 1; work engagement and employee voice behaviors at Time 2), we proactively executed the CFA marker variable method recommended by Williams, Hartman, and Cavazotte (2010) to evaluate the potential influences of common method variance (CMV) and clarify the concern of whether CMV biased our research results. The test procedure is as follows. First, we chose dispositional self-regulation as the marker variable, which we measured with the five-item scale developed by Luszczynska, Diehl, Gutiérrez-Doña, Kuusinen, and Schwarzer (2004). An example item was “I can concentrate on one activity for a long time, if necessary.” Specifically, we included dispositional self-regulation as the marker variable with other research variables in the confirmatory factor analysis (CFA) model. For the second step, the baseline model, which set a zero correlation between the marker variable and other variables in the study, was examined. Additionally, we set the marker variable with the same factor loadings and variances of error terms as the marker variable in the CFA model in the first step. For the third step, we tested the constrained model (Method-C), which involved setting all factor loadings the same from the marker variable to all the indicators of the CFA model. If the Method-C model was significantly better than the baseline model, then the influences of CMV might be significant. As shown in Table 1, the Method-C model was significantly different from the baseline model, indicating that the CMV potentially biased the research results (Method-C model: χ2[250] = 453.86, p < .001; Baseline model: χ2[251] = 531.21, p < .001; Δχ2[1] = 77.35, p < .001). For the fourth step, the unconstrained (Method-U) model was examined. The Method-U model was different from the Method-C model in that the factor loadings of the marker variable were freely estimated for the Method-U model, and we set equal factor loadings of the marker variable for the Method-C model. The results in Table 1 indicated that the Method-U model fitted the data better than the Method-C model (Method C model: χ2[250] = 453.86, p < .001; Method-U model: χ2[232] = 402.12, p < .001; Δχ2[18] = 51.74, p < .001). For the last step, the Method-R model was estimated; in the Method-R model, the covariance between the study variables is set to be the same as the covariance estimated in the baseline model. If the Method-R model fitted the data worse than the Method-U model, then the effects of CMV might bias the research results. The results showed that the Method-R model was not significantly different from Method-U model (Method-R model: χ2[242] = 408.50, p < .001; Method-U model: χ2[232] = 402.12, p < .001; Δχ2[10] = 6.38, p = .78), indicating that the presence of CMV did not bias our research results.

Table 1 Chi-square, goodness-of-fit Values, and model comparison tests
Full size table

Confirmatory Factor Analysis (CFA)

Before testing our hypotheses, we evaluated the convergent validity and discriminant validity of our measures by conducting a confirmatory factor analysis (CFA) test using Mplus 8.4. We assigned each item loaded on its proper factor, and, based on the theoretical definition and scale structure; specifically, we set work engagement as a second-order latent factor including vigor, dedication, and absorption. The model fit was adequate (χ2 (263) = 419.62, p < .01; CFI = .96; TLI = .95; RMSEA = .05; SRMR = .05; Hu & Bentler, 1999). Moreover, the CFA results showed that all factor loadings were significant (p < .01), all standardized factor loadings were larger than .65, and correlation coefficients among all latent factors were smaller than 1.0.

Moreover, we compare the proposed five-factor model to alternative models. Table 2 presents the results of the CFA model with five factors, including job autonomy, work engagement (second-order latent factor), P-O fit, promotive voice behavior, and prohibitive voice behavior. The five-factor model wherein work engagement was a second-order latent factor demonstrated a significantly better fit to the data than the five-factor model wherein work engagement was a single-order latent factor (χ2 (265) = 567.76, p < .01; CFI = .92; TLI = .91; RMSEA = .07; SRMR = .05); the four-factor model, in which promotive voice behavior and prohibitive voice behavior are loaded on one factor (χ2 (267) = 612.76, p < .01; Δ χ2 (4) = 193.14, p < .001; CFI = .91; TLI = .90; RMSEA = .07; SRMR = .06); the three-factor model, in which job autonomy and P-O fit are loaded on one factor (χ2 (270) = 837.88, p < .01; Δ χ2 (3) = 225.12, p < .001; CFI = .85; TLI = .83; RMSEA = .09; SRMR = .08); the two-factor model, in which job autonomy, promotive voice behavior, prohibitive voice behavior, and P-O fit are loaded on one factor (χ2 (272) = 2135.67, p < .01; Δ χ2 (2) =1297.79, p < .001; CFI = .50; TLI = .45; RMSEA = .16; SRMR = .22); and the one-factor model, in which five factors are loaded on one factor (χ2 (275) = 2464.57, p < .01; Δ χ2 (3) =328.90, p < .001; CFI = .41; TLI = .36; RMSEA = .17; SRMR = .17). Therefore, the CFA results support the convergent validity and discriminant validity of our five-factor model and demonstrate that the psychometric properties of the measures used in the current study are statistically adequate.

Table 2 Results of confirmatory factor analysis
Full size table

Hypotheses Testing

The means, standard deviations, intercorrelations, and reliability estimates of the variables are reported in Table 3. Job autonomy was positively associated with work engagement (r = .25, p < .001), whereas work engagement was positively related to employee voice behaviors, including promotive voice behavior (r = .45, p < .001) and prohibitive voice behavior (r = .36, p < .001), thus providing preliminary support for our Hypotheses.

Table 3 Means, standard deviations, and correlations of the variables
Full size table

Figure 2 displays the unstandardized path coefficients using Mplus 8.4 with maximum likelihood. Specifically, all variables were first-order latent variables except for work engagement, which was a second-order latent variable including three first-order variables: vigor, dedication, and absorption. Moreover, the result of the loglikelihood difference test showed that the model with the latent interaction term (Δloglikelihood = 6.70 (2), p < .05) was better than the model without the latent interaction term (CFI = .95; TLI = .94; RMSEA = .04; SRMR = .05; see Table 4), indicating that the model fit the data well (Wayne et al., 2017).

Fig. 2
figure 2

Unstandardized path coefficients of moderated mediation with latent variables. N = 262. *p < .05; **p < .01

Full size image
Table 4 Results of model comparison between the model including the latent interaction term and the model excluding the interaction term
Full size table

Hypothesis 1, which proposes that job autonomy will be positively related to work engagement (b = .13, SE = .06, p < .05), was supported. Work engagement was also positively related to promotive voice behavior and prohibitive voice behavior (b = .51, SE = .10, p < .01; b = .45, SE = .10, p < .01).

Hypothesis 2 proposes that job autonomy will be indirectly positively related to employee voice behaviors through work engagement. Because the results support the positive relationship between job autonomy and work engagement (i.e., Hypothesis 1) and the positive relationships between work engagement and both promotive voice behavior (b = .51, SE = .10, p < .001) and prohibitive voice behavior (b = .45, SE = .10, p < .001), we tested the direct effects of job autonomy (b = .10, SE = .08, p = .21; b = .05, SE = .08, p = .52), P-O fit (b = .05, SE = .09, p = .57; b = .03, SE = .09, p = .71) and the interaction effect between job autonomy and P-O fit (b = .28, SE = .11, p < .01; b = .15, SE = .09, p = .11) on promotive voice behavior and prohibitive voice behavior. Given the nonsignificant relationships between job autonomy and promotive voice behavior and prohibitive voice behavior, our results demonstrated that work engagement fully mediates these relationships, therefore supporting Hypothesis 2.

Hypothesis 3 proposes that P-O fit perceptions will moderate the relationship between work engagement and (a) promotive voice behavior and (b) prohibitive voice behavior. The results suggested that there is a significant interaction effect between work engagement and P-O fit on promotive voice behavior (b = .19, SE = .09, p < .05) but not prohibitive voice behavior (b = .07, SE = .09, p = .48). Moreover, as shown in Fig. 3, the simple slope test demonstrated that the effect of work engagement on promotive voice behavior will be stronger when P-O fit is high (b = .28, SE = .09, p < .01) versus low (b = − .02, SE = .10, p < .001) (Aiken & West, 1991), therefore indicating support for Hypothesis 3a but not for Hypothesis 3b.

Fig. 3
figure 3

The standardized moderating effect of person-organization fit (P-O fit) on the relationship between work engagement and promotive voice behavior

Full size image

With respect to Hypothesis 4, we proposed that P-O fit will moderate the indirect effect of job autonomy on employee voice behaviors through work engagement. Because the results only supported the moderating effect of P-O fit on the relationship between work engagement and promotive voice behavior, we further examined the conditional indirect effect of job autonomy on promotive voice behavior through work engagement. Our results showed that the index of moderated mediation, which is the product term of the interaction between work engagement and P-O fit on promotive voice behavior and the direct effect of job autonomy on work engagement, was significant (b = .02, SE = .02, p < .05; 95% CI = .000, .066). Following Cheung and Lau (2017), we examined the magnitude and significance of the indirect effect of job autonomy on promotive voice behavior through work engagement at various levels of P-O fit. The bootstrapping results revealed that the conditional indirect effect of job autonomy on promotive voice behavior through work engagement was stronger at the higher P-O fit (estimate = .09; 95% CI = .005, .173) compared to the lower P-O fit (estimate = .05; 95% CI = .002, .134). These results therefore support Hypothesis 4a but fail to support Hypothesis 4b.

Discussion

Although prior research has investigated the antecedents of employee voice behaviors due to the positive impacts of voice behaviors on organizational success, there is still a relative lack of understanding of whether psychological needs, which constitute an intrinsic motivational factor, are associated with employee voice behaviors, and specifically whether employee voice behaviors are related to intrinsic motivation (Liang et al., 2012; Song et al., 2020). Drawing on the SDT and JD-R model, the current study used the LMS approach, which is an appropriate method to examine our research model. Our results provided support for a theoretical model in which job autonomy is positively related to both promotive and prohibitive voice behaviors through work engagement.

Furthermore, given that organizational factors are considered to be key contextual factors that influence employee proactive behaviors such as voice behaviors (Zhou & George, 2001), the perception of the current organization may provide a boundary condition for the effect of work engagement on voice behaviors. Our findings indicated that P-O fit moderated the relationship between work engagement and promotive voice behavior, as well as the indirect relationship of job autonomy on promotive voice behavior through work engagement. Specifically, when P-O fit is high, both the direct effect of work engagement on promotive voice behavior and the indirect effect of job autonomy on promotive voice behavior via work engagement are more positive compared to when P-O fit is low.

Contrary to our expectations, P-O fit did not moderate the relationship between work engagement and prohibitive voice behavior. These findings shed light on the differentiation of the two voice behavior components – promotive and prohibitive – in Liang et al.’s (2012) voice model and suggest that they are influenced by different processes and conditions. Whereas P-O fit strengthened the relationship between work engagement and promotive voice behaviors, P-O fit was not an influential conditional factor on the relationship between work engagement and prohibitive voice behaviors. This is consistent with previous research demonstrating that promotive and prohibitive voice behaviors reflect different constructs and are influenced by different factors (e.g., job satisfaction, organizational commitment, and organizational justice) (Chamberlin et al., 2017). One possible explanation for this difference within the current model may be that the nature of prohibitive voice behavior (unlike promotive voice) is problem focused, where the voice behavior is used to help avoid any possible risks or harm within the organization (Morrison, 2011). This problem focused expression of concerns may arouse negative emotions in others that can cause conflict or defensiveness from other members of the organization (Liang et al., 2012). Because P-O fit represents the compatibility between an employee’ values, needs, and performance of a person and an organization’s (Cable & DeRue, 2002), employees with high P-O fit may be less likely to actively engage in voice behavior that could lead to negative affect or disruptions in social interactions within the organization. Conversely, P-O fit helped strengthen the relationship between engagement and voice behaviors when the voice behaviors were promotive, in that they were growth focused instead of problem focused. It is possible that the differential role of P-O fit on the two type of voice behaviors may be due to the difference in affective responses that voice behaviors can elicit from other organizational members. Expanding the research on employee voice to look at these two voice behaviors separately helps to further our understanding of employee voice and support future research this area. Below, we describe how our findings inform theory and practice.”

Theoretical Implications

This study’s findings offer a significant contribution to the existing literature in three primary ways. First, SDT was applied to support the linkages among job autonomy, work engagement, and voice behaviors (including promotive and prohibitive voice behaviors). A high satisfaction of psychological needs (e.g., job autonomy) is related to increased work motivation (e.g., work engagement) and employee voice behaviors. Moreover, SDT postulates that people are inclined to satisfy their psychological need to have authority over their own decisions; when this need is met, work motivation will be increased, which will then relate to the display of subsequent work behaviors (Deci & Ryan, 1985). Following this theoretical rationale, the current study supports the importance of job autonomy in employee motivation and extra-role behaviors (e.g., voice behaviors). Our research contributes to the employee voice literature by extending the application of SDT on employee extra-role behaviors and identifying the importance of psychological needs.

Second, the current study suggests that job autonomy as a psychological need is related to promotive and prohibitive voice behaviors indirectly through work engagement. That is, to improve both promotive and prohibitive voice behaviors, job autonomy must first support work engagement. This is in line with prior research that shows that basic need satisfaction results in autonomous motivation, which in turn has an impact on the level of work efforts and discretionary pro-organizational behaviors (De Cooman et al., 2013; Zhang & Chen, 2013). Furthermore, according to the JD-R model, work characteristics can serve as a type of job resource that can enhance the relationship between employee motivation and job performance, including both in-role and extra-role behaviors (Mäkikangas et al., 2016). The concept of job autonomy has been consistently identified as one of the most significant job design components in terms of the satisfaction of human psychological needs (i.e., need for autonomy) and has been shown to influence employees’ work motivation and job performance (De Cooman et al., 2013; Morgeson et al., 2005; Schaufeli, Bakker, & Van Rhenen, 2009). The current study builds on the need-behavior relationship by suggesting that job autonomy (an indicator of autonomous need satisfaction) is related to not only higher in-role job performance but also higher extra-role performance (promotive voice behavior in the current study). Our findings also further confirmed that job autonomy could stimulate employee voice behaviors through work engagement, indicating that work engagement (a form of employee motivation) is an important mechanism for understanding why job autonomy is related to employee promotive and prohibitive voice behaviors.

Third, our findings show that the perception of P-O fit is a significant factor which strengthened the relationship between engagement and promotive voice behavior and the indirect relationship between job autonomy and promotive voice behavior (via work engagement). According to SDT (Deci & Ryan, 1985; Ryan & Deci, 2000) and previous research (e.g., Alfes et al., 2016), individuals who perceive a fit between themselves and their social context are more likely to develop greater meaning from work and engage in more positive workplace behaviors. In addition, P-O fit can be viewed as a job resource which can satisfy employees’ basic psychological needs due to the correspondence between job demands and employees’ abilities (i.e., demands–abilities fit and need for relatedness) (Alfes et al., 2016; Ryan & Kristof-Brown, 2003). In a similar vein, prior research found that satisfying human needs can increase the link between work motivation and job behaviors (Gagné & Deci, 2005). Our findings further support the notion that higher P-O fit serves as a job resource that strengthens the indirect relationship between job autonomy and promotive voice behavior.

Accordingly, drawing on past research and theory from the SDT and JD-R model, the current study advances our understanding of work motivation by investigating how, why, and under what circumstances job autonomy is related to employee voice behaviors. Building on the prior research on job autonomy, our results suggest that job autonomy, as a inner need, is related proximally to employee engagement (a motivational factor) and distally to employee voice behaviors. Additionally, as a response to Christian et al.’s (2011) suggestions, this study contributes to a better understanding of the mediating role of engagement and the moderating role of employee fit in the relationship between job autonomy and positive employee behaviors. Finally, we also respond to Chamberlin et al.’s (2017) call for more research focusing on the different forms of voice behaviors (e.g., promotive and prohibitive voice behaviors) to enrich the field of voice literature.

Practical Implications

The current study has three important practical implications. First, our results suggest that job autonomy is beneficial for engagement directly and both promotive and prohibitive voice behaviors indirectly; thus, practitioners should consider job autonomy as an important work-related need that can influence motivation and subsequent contextual performance (e.g., employee voice behaviors). Because job characteristics are known to be related to the satisfaction of psychological needs (e.g., Van den Broeck, De Cuyper, De Witte, & Vansteenkiste, 2010), organizations can help support need satisfaction by designing and promoting jobs with supportive job characteristics. Earlier research also found that specific job characteristics such as job autonomy enable employees to define their work roles more broadly and further regard extra-role behaviors as in-role behaviors; thus, employees engage in more contextual activities that they view as in-role activities in the workplace (Gellatly & Irving, 2001). Based on our findings, we further suggest that managerial levels may be able to support employee motivation and employee voice behaviors by providing appropriate autonomy or authority at work (e.g., the liberty to make decisions).

Second, the results also reveal that work engagement is an important mechanism underlying the relationship between job autonomy and employee voice behaviors (promotive and prohibitive voice behaviors). In particular, our results indicate that the relationship between job autonomy and promotive voice behavior is significant and that the indirect effects of job autonomy on promotive and prohibitive voice behaviors through work engagement are significant. These findings emphasize work engagement as a critical motivational mechanism to stimulate both promotive and prohibitive voice behaviors. In alignment with prior research (e.g., Chamberlin et al., 2017), our results suggest that engagement is a significant antecedent of employee voice behaviors, including promotive and prohibitive voice behaviors. Therefore, engagement is likely an important driver of extra-role behaviors (e.g., employee voice behaviors and organizational citizenship behavior) (Babcock-Roberson & Strickland, 2010; Rurkkhum & Bartlett, 2012) and should serve as a focus for interventions designed to increase promotive and prohibitive voice behaviors. In order to promote employee voice behaviors, practitioners should concentrate on evaluating and enhancing engagement.

Third, our study highlights the importance of P-O fit in the needs–motivation–contextual performance (e.g., employee voice behaviors) relationships. Our findings may be beneficial for practitioners to facilitate high levels of P-O fit. Specifically, during the recruitment and employee selection process, human resource managers and staff could foster P-O fit by working to place job applicants in jobs that match their knowledge, skills, abilities, and personality by applying multiple P-O fit assessment tools such as personality tests and structured interviews. Moreover, human resource practitioners could also assess incumbents’ perceptions of P-O fit and provide them with adequate training to maintain or promote their P-O fit perceptions.

Limitations

Although our findings provide a number of significant contributions to the voice literature, the current study has the following limitations that should be addressed in future research. First, all variables were measured using self-reported questionnaires. This might raise concerns regarding CMV. Although we collected data at two time points, future studies should collect data from multiple sources to further prevent CMV. For example, assessing employee voice behaviors from a supervisor rating or P-O fit using objective fit measures would help strengthen our understanding of the current model and provide support for increased generalization of our findings beyond self-reported perceptions.

Second, although we applied SDT to develop our hypotheses, the present study did not measure other intrinsic needs simultaneously (i.e., the need for relatedness and the need for competence), which precludes us from examining our underlying theoretical model directly. In the current study, we tested the model only by examining job autonomy – a fulfillment of the need for autonomy in the workplace. However, we did not assess actual needs or need satisfaction. Investigating all three needs simultaneously and more comprehensively evaluating the needs might provide more nuanced information on how human needs influence employee motivation and subsequent behaviors. For future research, researchers might consider Gagné and Deci’s (2005) comprehensive research framework of SDT to advance our knowledge by examining social environmental factors and individual factors of autonomous work motivation (i.e., intrinsic motivation), which leads to job performance, psychological well-being, commitment, and job satisfaction.

Third, despite the current study showing that individual psychological needs (e.g., job autonomy) and individual perceptions (e.g., P-O fit) affect promotive and prohibitive voice differently, we do not provide insights into the contextual organizational factors that may produce these different effects. Future research should consider possible contextual organizational factors (i.e., supervisor behaviors and workplace climate) in encouraging employee voice behaviors, because voice behaviors are a relatively risky endeavor for individuals compared to other extra-role behaviors (e.g., organizational citizenship behavior), as it challenges the status quo within the organization.

Conclusion

The current research contributes to the literature on work motivation by applying the SDT and JD-R model to investigate the mediating role of engagement on the relationship between job autonomy and employee voice behaviors and the moderating effect of P-O fit on this relationship. This study found that job autonomy has a direct relationship with promotive voice behavior and an indirect relationship with employee voice behaviors (including promotive and prohibitive voice behaviors) through work engagement. In addition, both the direct relationship between engagement and promotive voice behavior and the indirect relationship between job autonomy and promotive voice behavior via work engagement were stronger when employees reported high (vs. low) levels of P-O fit. Accordingly, appropriately increasing employees’ job autonomy can be an effective intervention to increase their work engagement and enhance subsequent voice behaviors. Increasing the level of P-O fit through proper job matching or appropriate employee training is likely also important for supporting voice behaviors.