Effects of reciprocal, perceived person–environment fit, and emotional labor on job involvement: Moderated mediation analyses

Abstract Purpose This study examines the relationship between emotional labor and job involvement with convenience store employee, as well as mediating (perceived P–E fit) and moderating (reciprocal P–E fit) variables of this relation by proposing a moderated mediation model. Design/methodology/method The SPSS Program with PROCESS macro (Models 4 and 7) analysis was used with a sample of 230 convenience stores’ employees. Findings Results of this study demonstrated that reciprocal fit and perceived fit had moderated mediation effects between predictor (emotional labor) and criterion variables (job involvement). In addition, perceived fit evidenced positive effects on job involvement. Research limitations/implications We do not investigate the generation differences, futhure research may examine the mediating effects between EL and JI. To fully understand emotional labor, the specific cultural contexts in which it takes place will need to be considered. Practical implications This research has implications for the managers they could conduct social and individual identity to moderate the relationship between emotional workers and negative emotions. Furthermore, HR managers could apply training strategy to mediate the relationship between emotional labor and job involvement. Originality/value In the context of the emotional labor on job involvement, no studies have been conducted that have analyzed the moderated mediation effects especially the relationship between emotional labor and job involvement. This research seeks to build and extend the existing state of research on augmented moderated mediation effects by reciprocal fit and perceived fit.


PUBLIC INTEREST STATEMENT
Taiwan has over 10,000 convenience stores whose employee should be asked keep smiling as a daily routine. To pay your emotion for earning wages is called emotional labor by researchers. Besides, most people like to see smiling face with the frontline service agents. Therefore, even if they meet difficult customers whose smiling face always conceal their angry. This paper describes how to enhance employees' performance, reduce their emotional dissonance and how to make a congruence with their job, organization, supervisor and group. The paper also recommends that human resource managers could apply training strategy to mediate the relationship between emotional labor and job involvement.

Introduction
Focusing on emotional labor as part of intrapsychic processes role often emphasizes personal emotion congruence, such as Fleenor (2014, 2016) who beat the drum supposed emotional demands-abilities fit to distinguish from other fit perceptions. Nowadays, emotion congruence study regarding multiple types of fit exists simultaneously (Chuang, Hsu, Wang, & Judge, 2015), and this research examines reciprocal fit and perceived fit to explore these antecedents, moderator, mediator and job-related outcomes. Few evidence represents the relationship between emotional labor and job involvement (Chi, Grandey, Diamond, & Krimmel, 2011). However, we support Iplik, Topsakal, and Iplik's (2014) argument that emotional labor influences job satisfaction and Lawler and Hall's (1970) argument that one of the variable highly related to job satisfaction is job involvement.
According to Rafaeli and Sutton (1987), emotional labor leads to a positive effect, such as employee posited positive emotion in workplace. Hence, if employees perceived personenvironment fit (P-E fit) in their workplace, this would increase their high degree of job involvement and organizational commitment. Rafaeli and Sutton (1987) described that employees expressed feelings in congruence with experienced emotions that lead to emotional harmony. This is the index of congruence between person needs and work demands. However, employees' emotional abilities fulfill organizational demands and experienced their inner truth feelings, namely, the perspective of positive emotional labor which echoed the perceived emotional demands-abilities fit theory Diefendorff, Greguras, & Fleenor, 2016).
Whether the employees involve their job involvement or not depends on employees' professionals and traits. Moreover, they need to consider the congruence with environment then lead to match their jobs (Goštautaitė & Bučiūnienė, 2010). Therefore, the fit between individual and environment factors is P-E fit. Regarding P-E fit, studies have ignited from 1980s. Several researchers investigated person-job fit and person-organization fit as their research main variables (Kristof- Brown, Zimmerman, & Johnson, 2005). Furthermore, a lot of empirical studies have been done in this field to seek for well P-E fit leading to positive significance correlations among job attitude, behavior, work performance (Cable & Judge, 1996;Kristof-Brown et al., 2005;O'Reilly, Chatman, & Caldwell, 1991) and individual outcome (Downes, Kristof-Brown, Judge, & Darnold, 2017). Kristof-Brown and Billsberry (2013) divided the P-E fit literatures into two categories: one is based on the employees' perspective that emphasizes direct perceived fit (Cable & DeRue, 2002) and the other that examined the difference between person and environment emphasizes indirect fit (Chuang et al., 2015). They posited that both of these studies belong to the concept of compensative, not competitive perspective. We conduct this research from the perspective of personal direct fit to fulfill different environment aspects. No empirical evidence is yet available to confirm the moderated mediation effects with reciprocal fit and perceived fit influencing the relationships between emotional labor and job involvement. In addition, there were few empirical studies concerning the effects of positive emotional labor on job involvement in chain convenience store's employees. Hence, based on the conceptual framework of P-E fit, this research examines the correlations between emotional labor and job involvement.

Research purposes & questions
Previous studies have suggested that perceived emotional labor and employees' healthy are moderated by emotional adaptability and work attitude (i.e., job involvement, organizational identity; Schaubroeck & Jones, 2000). For example, Lyubomirsky, King, and Diener (2005) reported that the employee's attitude toward positive emotion while they experienced positive emotion to result in higher involvement, organizational citizen behavior and performance, which then lead to reduce they experienced dissonance. Therefore, the degree of fits between person (i.e., demand, value, capacity and individual characteristics) and environment (i.e., value, reward, work requirement, culture and material environment) are the major factors affecting employees' attitudes and behaviors are P-E fit.
Research questions in this study include how employees manage their emotional rules in order to fulfill organization in workplace? What are the correlations between emotional labor and job involvement? To what extent is reciprocal fit (person-supervisor fit and person-group fit) moderated effects between emotional labor and perceived fit (person-job fit and person-organization fit)? And to what extent is perceived fit mediator effects to job involvement? The purpose of this paper is twofold: (1) to examine the relationships between emotional labor and job involvement and (2) to draw on both the influence of the moderating role of reciprocal fit and the mediating role of perceived fit on the relationship between emotional labor and job involvement. Figure 1 depicts our conceptual model, which we will discuss in the next sections.

Theoretical development and hypotheses
3.1. Emotional labor, job involvement and person-environment fit Hochschild (1983) offered the perspective of feeling management, surface acting (SA) and deep acting (DA) which prolonged a series of multidimensional concepts of emotional labor, such as Morris and Feldmans' (1996) perspective of control and planning provided four concepts. Grandey (2000) offered two concepts from emotional regulation strategy and Brotheridge and Lees (2003) offered the scale of six concepts. Overall, these concepts provided the conceptual map of emotional labor. In the following, we will show more details, especially the scale which was developed by a Taiwan's scholar, Shang-Ping, Lin. Lin's (2000) EL concept includes the following: (1) basic emotional display, (2) surface emotional control, (3) deep acting, (4) emotional variety degree, and (5) interaction degree.
Most of researchers assumed emotion as a part of work role base on the management perspective-"planning and control", that workers are asked to express appropriate, suitable emotional condition by organization (Hsu, 2015;Lin, 2000). Moreover, Lin's scale indicates the concept of planning and control that can measure the implication of employees' emotional regulation, control and planning actually. Lodahl and Kejners' (1965) perspective of job involvement origins from the point of individual versus ego involvement in work and the point of view from socialist's central life interests. Kanungo (1982) defined job involvement as individual considered the importance of work beyond identity or belief in psychology, namely, individual presents the degree of identity and commitment on their jobs. According to Kanungo's (1982) point of view that the main concepts of job involvement consisted of job identity and job participation.
Job involvement would be affected by individual characteristics, job features (i.e., emotional labor high or low), supervisor variable, and character recognition (Brown, 1996). One of the variables of job features is autonomy that affects employees' emotional labor directly and positively. If employees have high autonomy or better self-control ability, they would not more likely lead to emotional dissonance (Wharton, 1993). Cho and Song (2017) pointed out that autonomy and supervisory support demonstrates their positive influence. Accordingly, we hypothesized that employees with high job involvement will have positive emotion even when they endured high emotional labor.
Regarding positive emotional labor that attributes employees' psychological capital Avey, Luthans, and Youssef (2010) describe that psychological capital is an important individual resource that improves individual development and job involvement in workplace. The results of such studies may have significant theoretical importance. In addition, emotional labor is related to job involvement (i.e., Yoo & Arnold, 2014), with potential moderating influence between them. Notwithstanding the positive emotion more likely enhances employees' job involvement (Yeh, 2018), but higher emotional labor decreases employees' job involvement (Hsu, 2015). Therefore, we propose the following hypothesis: Hypothesis 1: Emotional labor has a direct and positive or negative influence on job involvement.
A lot of P-E fit theories and empirical studies assume person-environment per se with the coincide results, such as job satisfaction, organizational commitment, and job performance (Kristof-Brown et al., 2005;O'Reilly et al., 1991). The major point is interaction between person and environment concerning congruence degree. Recently, P-E fit theory has been classified into two research categories: one emphasizes perceived direct fit from employees' view and the other emphasizes indirect fit, namely, measure fit from person and environment individually (Kristof- Brown & Billsberry, 2013). Based on the perspective of complementary fit, perceived fit, emotional regulation mechanism, and self-regulation theory, we propose P-E fit would mediate the relationship between emotional labor and job involvement. There is no disagreement on this point that Grandey (2000) reported that personal and organizational characteristics may act as main effects on the level of emotional labor performed; for example, employees or those who are supported by supervisors and coworkers may be more likely to display the appropriate emotions as emotional labor.
Most P-E fit theories employ twofold dimensions to measure work related variables: reciprocal fit and perceived fit. With the expectation of measuring double dimensions such as person-group, and person-supervisor fit, researchers use each dimension to investigate fit degree (Herdman & Carlson, 2009). Without expectation, people always interacted with multiple environments, for example, Kristof-Brown et al. (2005) provided individual's fit has a subsequent impact on their attitudes and behavior to illustrate people influence on environment all the time.
Researchers posit that multiple types of fit exist simultaneously. Kristof-Brown et al. (2005) offered a quantity study to explore multiple dimension concepts (Edwards & Billsberry, 2010). In their literature review, it provided that different types of fit including person-group, personsupervision, person-job, and person-organization fits to measure people ante-organization, within-organization, and turnover. This is not to say that these includes the key aspects of the combination all type of fits. However, we propose that both types of fits, reciprocal fit and perceived fit, would be used by employee when they involve emotion demand in workplace. Moreover, empirical study has shown that managers could strive to enhance the perceived fit between the emotional demands and employee abilities (Diefendorff et al., 2016). As such, a subordinate and a supervisor who are attracted to each other on the basis of such similarity are said to "fit" each other (Chuang, Shen, & Judge, 2016). Namely, if employee perceived a high level of fit (high PJ and PO fit), then he/she could enhance job involvement. On the other hand, if employee perceived a high level of emotional demands, then he/she would decrease his/her perceived fit. Therefore, we postulate the following hypotheses: Hypothesis 2: Emotional labor will have a direct and positive or negative influenced on perceived fit.
Hypothesis 3: Perceived fit will have a direct positive influenced on job involvement.

Reciprocal fit as the moderator between emotional labor and perceived fit
P-E fit theory scholars differentiated between two types of fits: reciprocal fit including PS and PG fit and categorized to indirect fit, namely, objective PE fit (Harrison, 1978) or compensative perceptive fit. Further, Kammeyer-Mueller, Schilpzand, and Rubenstein (2013) emphasized that the research of dyadic fit is an important factor to examine the interaction between person and environment, especially, PS and PG fits in the workplace. Both PG and PS fits involved personal relation and personal interaction in the workplaces that were categorized to "relation fit" (Oh et al., 2014) or reciprocal fit. Riordan (2000) pointed out that analyzing from individual interests' phrase which explored the congruence among employee, supervisor, and group was vital. For example, it is critical to measure leader-follower value congruence, supervisor-subordinate personality similarity and manageremployee goal congruence (Van Vianen, 2000). Moreover, respondent to environment means from outside to inner feeling regarding employee's emotional labor demands-abilities fit. For example, when employees felt supervisor and group were matched well than they can perceived the great congruence among individual emotional regulation, job and organization.
Supervisor and coworker supports always promote employees' positive emotion influence to emotional labor (Cho & Song, 2017). Furthermore, emotional labor can facilitate task performance by regulating interaction and precluding interpersonal problems (Ashforth & Humphrey, 1993). We proposed that employee feels that supervisor and colleagues could match well each other or "fitting in" in the workplace then affects the relationship between emotional labor and perceived fit. In sum, reciprocal fit (PS and PG fits) could be the moderator of the relationship between emotional labor and perceived fit.

Perceived fit as the mediator between emotional labor and job involvement
Prior research has shown that employees reported higher perceived fit (Chuang et al., 2016) will have better job performance (Kristof- Brown & Guay, 2011). For example, PO and PJ fits are positively associated with job satisfaction, work performances, organizational citizenship behavior, and proactive behavior (e.g., Greguras, Diefendorff, Carpenter, & Troster, 2014), as well as negatively associated with burnout, stress, and quit intention (e.g., Herdman & Carlson, 2009). Many empirical studies have shown that perceived fit belongs to subjective fit and assumed relation to the coalition between reality and self-assessment (Harrison, 1978). Edwards (1991) proposed the two basic conceptual perspectives to person-job fit: (a) demands-abilities fit and (b) needssupplies fit. Further, Greguras et al. (2014) posited that self-determination theory could influence P-E fit. Morris and Feldman (1996) proposed that the organizational traits are ante-variables affecting emotional labor. In addition, organizational support can reduce negative emotional labor effects (Avey et al., 2010).
In sum, based on the complementarity perceptive of fit, Holland's (1985) theory on occupational interests, and the need-fulfillment paradigm, the present study contains four job fit dimensions: KSAs, personality, interests, and job characteristics. In a hotel setting, Lam, Huo, and Chen (2018) show that PO/PJ fits are the most relevant forms for customer service employees, especially because employees' perceptions of fit with their jobs and with organizational values commonly serve as predictors of their work behaviors (Kristof-Brown et al., 2005). The complementarity or subjective fit reached well match between person and job and organization. We proposed that perceived fit could be a bridge between emotional labor and job involvement. Based on the above discussion, the following hypotheses were proposed.
Hypothesis 4: Employees perceived reciprocal fit that would influence the relationship between emotional labor and perceived fit.
Hypothesis 5: Perceived fit will mediate the influence emotional labor on job involvement.

Research method
We tested our hypotheses in a sample of 230 full-time employees from 60 7-Eleven convenience stores in the middle of Taiwan with convenience sampling method. A total of 152 female (66%) employees and 78 male (34%) employees completed the survey.
A confirmatory factor analysis (maximum-likelihood estimation; Amos 21.0) including all items scales supported the measurement models (see Table 1). The Fornell and Larcker (1981) test provided evidence for discriminant validity. We applied several methods ex ante (before) and ex post (after) to avoid common method variance . Ex ante we used both separation approach of data collecting and design approach of instrument developing. As ex post means of control, the single-factor test was run. An exploratory factor analysis including all indicators reveals that many factors explain 66.247% of the variance. By contrast, one general factor explains 27.016% of the variance only. Hence, this study does not exist common method variance. We adopted the moderated mediation analysis procedures outlined by Preacher, Rucker, and Hayes (2007) to test the proposed moderated mediation effect.  Table 1 shows all variables models of fit goodness index. According to Hu and Bentler's (1999) suggestion that the goodness index standard values, except PPJ fit of NNFI = .858, RMSEA = .107, and SRMR = .059 are out of little range, others reported were goodness. Table 2 shows bivariate correlations of all variables. All correlations were significant. Table 3 shows the results of hypotheses 1 to 3. Hypothesis 1 tested the relationship between emotional labor and job involvement that represented EL to job involvement (β = .428; p < .000). Hypothesis 2 examined the relationship between emotional labor and perceived fit that reported EL to perceived fit (β = .589; p < .000). Hypothesis 3 measured the relationship between perceived fit and job involvement that described perceived fit to job involvement (β = .657; p < .000). All these hypotheses of general models showed significance. The variance inflation factor (VIF) was lower than 5 i.e., the tolerance value is under .2 (Belsley, 1991;Belsley, Kuh, & Welsch, 1980).

Moderated mediation
The test of moderated mediation effect (Hypotheses 4 and 5) with 5,000 bootstrapped samples revealed a conditional indirect effect in the predicted direction (see Table 4). In reciprocal fit, the relationship between emotional labor and perceived fit was positive, as was the relationship between emotional labor and job involvement. The relationship between emotional labor and job involvement was significant mediated by perceived fit (indirect effect = 0.437; 95% bootstrap CI = 0.325-0.557). (index = 0.129; 95% bootstrap CI = 0.041 to 0.220).

Moderated regression analyses
Results of additional moderated regression analyses at each pathway are presented in Tables 5 and 6. The relationship between emotional labor and perceived was moderated by reciprocal fit, B = 0.592, t = 9.388, and p > .05. However, the relationship between emotional labor and perceived fit was moderated by reciprocal fit, B = 0.81, t = 1.728, and p < .05; see Figure 2. Higher levels of reciprocal fit were associated with higher perceived fit in employee, but this effect was relatively stronger than lower reciprocal fit.
For examing the relationship between interaction variables and other variables, We conducted Aldrick and Rodriguez' (2012) approach that counted all moderators' and mediators' mean score higher than one standard deviation belong to high group and lower one belong to low group then conducted to linear regression and plotted two dimensions' interaction figure. We cited the moderators and mediators consisted of PFIT and RFIT to explain their effects (Figure 2). Figure 2 shows that the high RFIT fit group's slop rate is higher than low RFIT fit groups' and they are all presented positive relationship between emotional labor and perceived fit.

Discussion
This study examined the correlation between emotional labor and job involvement in Taiwan convenience store employee and investigated the moderated mediation effects of reciprocal fit and perceived fit between emotional labor and job involvement among employees. The findings of this study indicated that emotional labor was positively correlated with job involvement. This result supported the findings of Fox, Webster, and Casper (2018), Hsu (2015), and Yoo and Arnold (2014). There may be a number of explanations for the role of emotional labor among convenience store employees in promoting job involvement in the Taiwan context. For example, employees with positive or self-regulation emotional labor show more job involvement.
Although previous research has established positive or negative relationship between emotional labor and work-related outcomes, few are known about the relationship between emotional labor and job involvement. Moreover, few studies have divided P-E fit into both perceived fit and reciprocal fits and examine their moderated mediation effects between emotional labor and job involvement.  In sum, in describing perceived fit mediating the relationships between emotional labor and job involvement, reciprocal fit moderating the relationship between emotional labor and perceived fit, we have made the arguments that (a) as Diefendorff, Greguras, & Fleenor (2014) reported that emotional demands-abilities fit could be distinguished from other fit perceptions, we conducted both integration of PE fit theories, such as reciprocal fit would moderated perceived fit and perceived fit prescribed the mediation role between emotional labor and job involvement. (b) emotional labor has a positive influenced on perceived fit by the moderator, reciprocal fit.  When the frontline service employee feel emotional work could get congruence from job and organization which indicated their colleagues and supervisors within emotional harmony in workplace. That is Ashforth and Humphrey (1993) posited that the relationship between emotional workers and negative emotions were moderated by social and individual identity.

Theoretical and empirical contributions
According to Boon, Eckardt, Lepak, and Boselie (2018), strategic HRM focuses on whether and how systems of HR practices help organizations achieve strategic goals and enhance firm performance. HR systems design with suitable reciprocal (PO & PJ fit) and perceived fit (PE fit) to encourage appropriate role emotional labor to promoting employees' job involvement. Our findings suggest that organizational managers should seek to regulate PE fit with employees' emotional labor. Our study supports reciprocal/perceived as moderated mediating conditions to affect emotional labor on job involvement, which provides empirically literature of perceived emotional demands-abilities fit theory. This is a notable departure from the vast majority of emotional labor research, which has concentrated on control and plan reciprocal/perceived fit to influence job involvement.
Our findings show that higher levels of reciprocal fit were associated with higher perceived fit in employee, but this effect was relatively stronger than lower reciprocal fit. According to Kristof-Brown et al. (2005), perceived fit reflects employees' feelings of fitting in, a feeling that may be subject to "cognitive manipulation because the assessment is all done in the head of the respondents" and may differ from objective fit (Cable & DeRue, 2002). Therefore, fit perceptions are better  predictors of employees' attitudes and behaviors than objective fit. Hence, we suggest that frontline service employee would maintain harmony attitudes with PO and PJ fit to resolve the problems of emotional labor than promoting their PE fit.

Limitations and directions for future research
This study has several limitations. All measures were collected from the country area source. Future studies should collect data from additional source (e.g., others metropolis, Taipei or Taichung, etc.).
We do not investigate the generation diffenences, future research may examine the mediating effects between emotional labor and job involvement. Finally, to fully understand emotional labor, the specific cultural contexts in which it takes place will need to be considered (Mesquita & Delvaux, 2013). A fruitful future direction would be thinking over whether different cultures and generation have effects on emotional labor or not.