1 Introduction

The attention to technology industry has been accompanied by an increased focus on job performance and turnover intention, because having superior employees and retaining them as important human resources are equally important under today’s serious competition within this fast-changing industry (Ahn and Ok 2017; Huang and Lin 2017). Defined as a deliberate and conscious willingness to quit a job from a firm within a foreseeable future (Ertureten et al. 2013; Lin 2017a, b), turnover intention is often a neglected but important issue for business practitioners (Jones et al. 2007; Kashyap and Rangnekar 2016). It is thus critical for human resource managers to learn why employees develop an intention to leave the organization for good, because strong turnover intention in an organization can substantially harm organizational morale and cause additional costs on recruiting and training new employees. Similar to turnover intention, job performance is a key factor for a firm’s survival, because it represents a set of behaviors for achieving the job goals of an organization.

Albeit job performance and turnover intention have been researched from multiple lenses in the literature (e.g., the perspectives of job satisfaction, organizational commitment, and so on), they have been relatively understudied from the simultaneous aspect of positive affect and perceived benevolent climate, consequently resulting in a research gap for this study to fill. The aim of this study is to address the research gap by investigating what roles positive affect and perceived benevolent climate play respectively in the development of turnover intention and job performance. This study defines positive affect as a pleasant feeling state or a good mood (Estrada et al. 1994) and perceived benevolent climate as individuals’ perceived organizational normative system emphasizing the interests of others in the organization (e.g., Victor and Cullen 1987). Previous literature has suggested that human emotion (e.g., positive affect) and organizational climate (e.g., perceived benevolent climate) should be simultaneously taken into account to explain people’s behavior in business organizations (Edwards et al. 2009; Melgoza et al. 2009). Without understanding positive affect and perceived benevolent climate in a single model setting, our knowledge about how to improve turnover intention and job performance can be highly delimited, and the management initiatives taken to reduce turnover intention and increase job performance will remain ineffective based on the subjective misconception of traditional management practices. For example, when managers focus solely on positive affect but ignore benevolent climate, employees often disregard the consequences of their actions and decisions both within and outside their organization (e.g., Blome and Paulraj 2013), consequently causing serious side-effects of their performance in a long run.

Research on positive affect is growing due to its potential influence on job outcomes (e.g., Arvey et al. 1998; Lyubomirsky et al. 2005). Previous literature has provided evidence that positive affect (or mood) leads to improved job performance (Kaplan et al. 2009). In addition, the literature suggests that positive affect (e.g., happiness) reduces turnover intention (Trakulmaykee and Mohd-Yusoff 2013; Wright and Bonett 2007). Therefore, scholars and practitioners have called for more research to clarify precisely how positive affect fosters or hinders job outcomes such as job performance and turnover intention. At the same time, it has been argued that the organization’s ethical climate (e.g., perceived benevolent climate) is a situational moderator in the development of an individual’s perceptions and job outcomes (Liu et al. 2018; Treviño et al. 1998). The moderation of perceived benevolent climate is theoretically justifiable by the ethical climate theory, indicating that individuals tend to work harder for organizations they perceive to have a “benevolent climate”, or one that focuses on the welfare of others (Hollingworth and Valentine 2014).

This study differs from previous research in two important ways. First, while previous research has mainly focused on positive affect as exogenous predictor for performance (Juravich and Babiak 2015; Knight and Eisenkraft 2015; Peñalver et al. 2017) or turnover intention (Bouckenoogh et al. 2013; Hui et al. 2007; McNall et al. 2015), this study complements the previous research by exploring the mediating mechanism of positive affect. Second, although perceived benevolent climate has been recognized as an important factor based on concern for others and can facilitate employees’ decent behavior expected by the organization (Ashman and Winstanley 2007; Biswas and Varma 2007; Cullen et al. 2003), its moderating role in the development of job performance and turnover intention among servicing professionals is still unknown. In other words, despite its importance as a core element of work environment, perceived benevolent climate has been rarely examined regarding how it interacts with servicing workers’ psychological and emotional states to jointly influence turnover intention and job performance. Collectively, this study develops a mediation model of turnover intention and job performance by taking into account positive affect and benevolent climate, substantially complementing the literature.

2 Theories and development of hypotheses

Drawing upon the conservation of resources (COR) theory and ethical climate theory, this study develops a research model to explain the formation of turnover intention and job performance. In the model, this study hypothesizes that turnover intention and job performance are both related to job stress directly and indirectly through the mediation of positive affect. The model paths related to turnover intention or job performance are hypothetically moderated by perceived benevolent climate. Job stress is examined as a key predictor in this study, because its main effect on turnover intention (or job performance) is highly sensitive and subject to changes in ethical climate (e.g., perceived benevolent climate) (Mulki et al. 2008).

2.1 The COR theory and ethical climate theory

As a specific type of stress theory, COR theory provides an overarching theoretical framework to understand the nature of stress as a human phenomenon that is tied to employees’ job experiences in workplaces (Hobfoll and Freedy 1993). According to the COR theory (Hobfoll 1988, 1989, 1998), employees who perceive job stress are likely driven to evaluate their psychological resource burden and loss that make them emotional with reduced positive affect. Specifically, the COR theory argues that positive affect can be conceived as an employee’s personal resource that is negatively influenced by job stress (Janssen et al. 2010).

In considering potential resource burden and loss at work, employees are motivated to assess whether the organization they work for comprises benevolent climate that maximizes the interests of employees (Elçi and Alpkan 2009) to alleviate the impact of job stress. For that reason, perceived benevolent climate becomes a key moderating factor in the stress—performance linkage. Previous explanations of ethical climate theory (Victor and Cullen 1987, 1988) have given details on benevolent climate as a specific type of ethical climate based on concern for others. According to ethical climate theory (Goldman and Tabak 2010; Martin and Cullen 2006; Rothwell and Baldwin 2007; Victor and Cullen 1987, 1988), this specific kind of ethical climate (i.e., benevolent norms) in a firm can be attuned to “humanity as a whole” and intervene employees’ behavior (Cullen et al. 2003; Krinsky 2016). Employees perceiving benevolent climate in their organization view their organization as having a sincere interest for their well-being (Cullen et al. 2003; Wimbush and Shepard 1994). Therefore, perceived benevolent climate plays an important moderating role in the development of turnover intention and job performance. Following the above-discussed two theories, this study then moves on to justify its hypotheses in the following.

2.2 Development of hypotheses

As employees’ working lives have been increasingly stressful in today’s competitive markets, job stress has become an important factor influencing their job outcomes, such as turnover intention and job performance (Chen et al. 2011). Job stress is defined as the mind and body arousal resulting from physical and/or psychological demands related to a job (Siu and Cooper 1998). Previous literature has indicated that stressful work environments can lead to detrimental effects on employees’ emotion, intention, and behavior (e.g., Bohle and Quinlan 2000; Greenberg and Baron 2003). Employees who experience high degrees of job stress are likely to be unhappy and poorly motivated and have a weaker intention to remain at work (Arshadi and Damiri 2013), suggesting the impact of job stress respectively on positive affect and turnover intention.

The literature has suggested an indirect effect of job stress on the intention of leaving a job (Hossain et al. 2015; Yin-Fah et al. 2010). Such indirect effect is theoretically justifiable because of the existence of multiple potential drivers related to the intention of quitting not only the job but also the organization. For example, a framework developed by Curbow (1990) has indicated the extent to which job stress influences employees’ psychological well-being (e.g., emotion, affective tone) and consequently the rates of worker job turnover (Esmaeilifar et al. 2018). Collectively, the larger the amount of job stress is for employees, the greater their turnover intention will develop (Chen et al. 2011; Applebaum et al. 2010). At the same time, job stress can cause serious emotional detrimental effects on employees, because it enhances negative reactions such as anxiety and anger (Hodapp et al. 1988), suppressing the development of positive affect. According to the COR theory, one plausible mechanism through which job stress may affect job outcomes is emotion (e.g., positive affect) (Matthews 2008). Garg and Dhar (2014) have found that exaggerated job stress negatively influences workers’ emotional and physical health (e.g., positive affect, happiness) and thus hamper their job performance in a long run. Collectively, aside from its direct impact on turnover intention, job stress can influence turnover intention indirectly through positive affect. Consequently, the mediating role of positive affect can be stated as below.

H1

Job stress positively relates to turnover intention directly and indirectly through the mediation of positive affect.

Following the above-discussed influence of job stress on positive affect, this study further hypothesizes the mediation of positive affect between job stress and performance (Azmi et al. 2016; Jex 1998; Motowidlo et al. 1986), which can be explained by two reasons. First, job stress can disrupt employees’ equilibrium and result in them deviating from their normal behavioral patterns, suggesting a direct and negative relationship between job stress and job performance (e.g., Chen et al. 2011; Westman and Eden 1996). Second, positive psychology notes that employees who possess positive affect (i.e., delightful positive mindset or happiness) perform better at job challenges than others, which has been recognized as the “happiness advantage” (Achor 2012). Employees with high degrees of positive affect tend to actively engage in the behavior of developing a better way to do things for performance improvement (Gong and Zhang 2017; Griffin et al. 2007; Verner-Filion and Vallerand 2016), suggesting a direct and positive relationship between positive affect and job performance.

The indirect relationship between job stress and job performance has been discussed in previous research. For example, a study by Soran et al. (2014) has found that emotional intelligence plays a mediating role between job stress and job performance. They further suggest that some personality traits or affective tone may be mediators between job stress and performance. Specifically, previous research has suggested that positive affect helps at synchronizing job demands and individual capabilities to improve job performance (Lin et al. 2014). Collectively, positive affect or emotion can mediate the indirect effect of job stress on job performance. Such indirect effect is understandable because job performance is likely driven by multiple behavioral pathways such as job embeddedness, commitment, and attitude, etc. (e.g., Coetzee and Chetty 2015). For instance, Noermijati and Primasari (2015) suggest that job stress indirectly and significantly influence job performance through employees’ job satisfaction. To sum up, the indirect relationship between job stress and job performance is derived as below.

H2

Job stress negatively relates to job performance directly and indirectly through the mediation of positive affect.

According to the ethical climate theory, perceived benevolent climate can be an important moderator (Kuenzi and Schminke 2009) that influences the impact of positive affect on job outcomes. A stronger benevolent climate signals appealing virtues based on which employees are more convinced of remaining in the firm (Stewart et al. 2011), regardless of their emotion. As a result, their turnover intention is less likely influenced by their positive affect, given a stronger benevolent climate. Analogously, positive affect has a weaker influence on job performance when perceived benevolent climate is stronger. Employees who perceive a stronger benevolent climate tend to progress from concern for self, to concern for the organization, and then to concern for others and social communities (Cullen et al. 1993; Kohlberg 1984). Those who perceive they are employed in an environment with greater benevolent or altruistic values strive more strongly for their job duties (e.g., Sims and Keon 1997) even if their positive affect may not remain at a high level. In summary, a stronger perceived benevolent climate is expected to weaken the relational strength between positive affect and turnover intention and between positive affect and job performance. Two hypotheses are therefore derived as below.

H3

Perceived benevolent climate negatively moderates the negative relationship between positive affect and turnover intention.

H4

Perceived benevolent climate negatively moderates the positive relationship between positive affect and job performance.

Previous studies (Blome and Paulraj 2013; Nguyen and Leclerc 2011; White 2005) have called for more research relating benevolence as a moderator to pertinent organizational factors. The climate of benevolence in an organization acts as a kind of job demand (e.g., concern for others) and can thus intervene in the development of job performance, which is exercised through the interactions between benevolence and job stress.

Whereas job stress is referred to as the tense feeling due to the formal requirements of employees’ job, benevolence refers to those actions performed by employees that are outside the boundaries of the formal job description and are performed by employees under their own discretion (Nguyen and Leclerc 2011). Specifically, job stress can be considered as an in-role component of employees’ job demands, while benevolence can be seen as an extra-role component of the job demands. Given that they are highly related to each other in terms of job demands, they are likely to have positive interactions in the development of turnover intention and job performance, which means a positive moderation of perceived benevolent climate between job stress and turnover intention and between job stress and job performance. In other words, turnover intention is more likely enhanced and job performance is more likely to drop when benevolence and job stress both become greater. This is understandable, because benevolence and job stress are both role components of job demands (Lapierre 2007; Leung 2008), and thus the effect of job stress can be amplified with the increase of benevolence due to their synergistic influence. Collectively, the hypotheses regarding the positive interaction of job stress and perceived benevolent climate can be stated as below.

H5

Perceived benevolent climate positively moderates the positive relationship between job stress and turnover intention.

H6

Perceived benevolent climate positively moderates the negative relationship between job stress and job performance.

In summary, the research model that demonstrates the above hypothesized relationships between research constructs in this study is provided in Fig. 1.

Fig. 1
figure 1

Conceptual model

Full size image

3 Methodology

3.1 Subjects and procedures

The hypotheses of this study were empirically verified using a two-wave survey of service personnel from international business firms in high-tech industry in Taiwan. As high-tech industry often has a relatively high turnover rate due to serious competition and market turbulence (Chien and Chen 2007), studying the issue of turnover intention in this industry is highly important for both scholars and practitioners. A total of six large high-tech firms in a well-known science park in northern Taiwan we contacted agreed to help with our data collection. This study attempted to survey all the service professionals in these firms without sampling, resulting in an estimated sample frame of 500 service professionals in total. The human resource (HR) departments of these firms helped distribute questionnaires directly to their employees. Eventually, employees who received our questionnaires decided on their own to choose whether to participate in the survey or not. Of the questionnaires distributed to 500 servicing professionals twice with 1 month part, 299 usable matched questionnaires were returned for a questionnaire response rate of 59.8%. In our sample, 48.49% of the subjects were female (i.e., 145 employees), 64.21% of the subjects were not married (i.e., 192 employees), and 92.64% had a bachelor degree or above (i.e., 277 employees). The age of the sample ranged from 20 to 64 years old, and the average age was 28.79 years old.

This study looks to reduce the threat of common method variances (CMV) with two critical measures. First, this study adopted quality measurement items by drawing from prior literature, refining them based on pilot testing, and then using them in a two-wave survey (i.e., two surveys at two different points of time with 1 month apart). Previous research has indicated that CMV can be reduced through the use of high quality scales and two-wave survey methods (Feng et al. 2010; Helm and Conrad 2015; Podsakoff et al. 2003). Indeed, after our first survey, when the second survey was conducted the same participants were unlikely to have memorized the responses they gave 1 month previously. This reduces the threat of CMV and increases our causal inference. Second, CMV is unlikely to occur in this study due in part to its main focus on the moderation of perceived benevolent climate (i.e., its interaction effects). Testing for moderating effects has the advantage of mitigating the CMV threat (Chen and Lin 2014). Moderations are less susceptible to CMV, because such moderations are nearly inconceivably a part of respondents’ cognitive maps (Chang et al. 2010; Chen and Lin 2013; Tsai et al. 2014). Nevertheless, this study performed Harmon’s single-factor analysis, which tests whether there is significant bias resulting from using a single instrument to collect empirical data (Matusik and Heeley 2005). In this analysis, if common method variances substantially appear in the data, then either one factor will account for most of the covariance among all the factors or a single factor will emerge from the factor analysis (Lin and Bhattacherjee 2008). Exploratory factor analysis for the five factors in this study showed these five factors explaining respectively 26.18, 22.03, 19.84, 17.58, and 14.37% of the total variance. These percentages indicated that the variances were appropriately distributed among multiple factors, revealing that common method biases were unlikely to affect the observed data or subsequent analysis (Lin and Bhattacherjee 2008).

3.2 Measures

The constructs discussed in this study were measured using five-point Likert scales translated and modified from existing literature by an academic focus group familiar with organizational behavior and human resource management. A pilot survey before the actual survey was conducted to improve the quality of our measurement items. The data obtained in the pilot survey from 62 participants were subjected to exploratory factor analysis (EFA) and reliability analysis. These participants were then excluded from our actual survey. Based on the analysis of the pilot data, measurement items with poor loadings, high cross loadings, or low reliability were removed or rephrased. “Appendix” summarizes all the measurement items and literature sources. Note that perceived benevolent climate is studied in this research as a typical type of individual-level psychological climate suggested by Schwepker (2017) because the literature has suggested that individuals’ perceptions concerning values and norms build upon organizational climate are actual influences on their emotional and behavioral responses (Biswas and Varma 2007; Schwepker et al. 1997).

4 Data analysis

The survey data were analyzed firstly using two-step structural equation modeling (SEM) (Anderson and Gerbing 1998) and secondly using moderated regression analyses. Statistical results from SEM or regression analyses are presented below.

4.1 Confirmatory factor analysis

The results of confirmatory factor analysis conducted by this study show that the measurement model of this study fitted very well with its survey data (see Table 1). Specifically, the comparative fit index (CFI), normed fit index (NFI), and non-normed fit index (NNFI) all equaled or exceeded 0.90, the root-mean-square residual (RMR) was smaller than 0.05, and the root-mean-square error of approximation (RMSEA) was smaller than 0.08 (Fornell and Larcker 1981).

Table 1 Standardized loadings and reliabilities
Full size table

Based on the test results in Table 1, convergent validity was statistically confirmed, because: (1) all factor loadings were statistically significant; (2) the average variances extracted (AVE) for all research constructs were larger than 0.50; and (3) the values of Cronbach’s alpha all exceeded 0.80 (Fornell and Larcker 1981). This study uses two methods to double verify the discriminant validity of our data. This study first performed AVE (Average Variance Extracted) analysis to assess the discriminant validity in this study. In Table 2, the square root of the AVE of each construct was larger than the correlation of pairwise constructs in the correlation matrix, supporting discriminant validity. Furthermore, Chi square difference tests were applied to evaluate discriminant validity (see Table 3). As Chi square difference statistics for all pairs of constructs exceeded their critical value based on the experiment-wise error rate at an overall significance level of 0.01, discriminant validity was supported. Finally, the fit indices of the validation models recommended by the literature (e.g., Lin 2010; Saklofske et al. 2003) were analyzed in this study. The results in Table 4 reveal that the research model based on the five factors proposed by this study is the best choice among all alternative and competing models.

Table 2 Correlation matrix and the square root of AVE
Full size table
Table 3 Chi square difference tests for examining discriminate validity
Full size table
Table 4 Fit indices of the validation models of this study
Full size table

4.2 Test results of SEM

The test results of SEM found in this study (see Fig. 2) show that job stress positively related to turnover intention directly and indirectly through the mediation of positive affect (thus, H1 is supported). However, job stress is only related to job performance indirectly through the full mediation of positive affect due to the insignificant relationship between job stress and job performance (thus, H2 is only partially supported).

Fig. 2
figure 2

Test results of SEM

Full size image

With regard to our moderating effects, moderated regression analysis was performed to confirm whether H3-H6 are supported or not (see Table 5). The test result in Model 1 shows that perceived benevolent climate negatively moderated the relationship between positive affect and turnover intention under the coefficient of − 0.17 (p < 0.01), but did not moderate the relationship between job stress and turnover intention. In Model 2, the test result shows that perceived benevolent climate did not moderate the relationship between positive affect and job performance, but indeed moderated the relationship between job stress and job performance under the coefficient of 0.26 (p < 0.01). To double check its hypothesized mediation effects, this study conducted a bootstrapping analysis. The test results in Table 6 showed that 95% confidence intervals did not cover zero, supporting significant mediation effects (Preacher and Hayes 2008). Table 7 summarizes the results of the hypotheses.

Table 5 Moderated regression analysis
Full size table
Table 6 The test results of the mediation using bootstrapping analysis
Full size table
Table 7 Results of hypotheses
Full size table

The unsupported H4 and H5 may suggest that predicting employees’ turnover intention and job performance is a complicated process because employees that fit benevolent climate may or may not show high turnover intention or job performance (Boswell et al. 2008). Hence, an overall in-depth understanding of the complicated process requires the interaction with different factors. These unsupported results for H4 and H5 in this study are somewhat analogous to previous research (Tanner et al. 2015) with mixed results that show the significant moderating effect of ethical climate on the relationship between individuals’ factors and job satisfaction and the insignificant moderating effect on the relationship between individuals’ factors and job performance.

5 Discussion

The statistical results of this study reveal important findings that can substantially complement the literature of job stress and benevolent climate. For example, this study finds that job stress positively relates to turnover intention directly and indirectly through the mediation of positive affect. Moreover, the relationship between positive affect and turnover intention is negatively moderated by perceived benevolent climate, whereas the relationship between job stress and job performance is positively moderated by perceived benevolent climate. Based on these empirical results, this study presents theoretical and managerial implications as follows.

5.1 Theoretical implications

The findings of this study reveal two theoretical implications. First, complementing the conservation of resources theory (COR) (Hobfoll and Freedy 1993; Hobfoll and Shirom 1993), this study offers fresh theoretical insights into turnover intention and how it is influenced by job stress through the mediation of positive affect. From the view of Hobfoll and Freedy (1993), employees who are confronted with job stress are expected to then use various coping strategies (e.g., quitting the job). Previous literature based on COR has suggested that turnover intention is likely to occur when valued resources are lost or are threatened to cause high job stress (Taris et al. 2001). Despite the evident relationship between job stress and turnover intention based on COR, the mediating role of positive affect between job stress and turnover intention has not yet been examined in previous studies. For that reason, the findings of this study help explain how COR can be used to further scrutinize how job stress influences turnover intention directly and indirectly through the mediation of positive affect.

Second, the significant moderating effects of perceived benevolent climate in this study provide further support for what Kohlberg (1984) argued in the theory of moral development—that caring (a proxy of benevolence) is an important type of ethical standard that can moderate the impact of employees’ psychological state on their job outcomes. The roots of the benevolence (or caring) are embedded in the benevolence theory or in terms of moral philosophy (Martin and Cullen 2006). Since perceived benevolent climate encourages employees to make greater efforts towards the interests of others, it represents a force that causes people to act in ways that they may not normally do. By pressing forward with insightful and meaningful research to complement the benevolence theory and moral philosophy, this study reveals that individuals’ job performance can be influenced in part by an overarching concern for the well-being of others in their organization. Hence, the findings of this study help explain how the ethical climate theory (Martin and Cullen 2006) can be effectively applied to further scrutinize how perceived benevolent climate alters the relationship between job stress and job performance.

5.2 Managerial implications

This study provides important managerial implications regarding the mediating mechanism of positive affect and the moderating mechanism of perceived benevolent climate in the development of turnover intention and job performance. Since positive affect fully mediates the relationship between job stress and job performance, managers can view positive affect as an effective sensor for understanding the relative magnitude of job stress and then taking necessary precautious measures to alleviate job stress. As employees’ stress coping ability helps improve job performance indirectly, it is important for managers to recruit quality service workers by carefully evaluating their endurance under certain job pressure. Managers should develop various stress and emotional management courses (e.g., coping strategies for job stress, practices to understand, maintain, and share personal emotions), whereby service workers can clearly understand how job stress and positive affect influence performance achievement.

Consistent with previous research (Cullen et al. 2003), perceived benevolent climate plays a key moderating role for the relationships among individuals’ psychological strains, emotional states, and turnover intention. If service workers perceive strong benevolent climate in the organization, they are likely encouraged by the organization to promote their concerns for each other, consequently facilitating job performance. It is important for managers to learn leveraging benevolent climate effectively (e.g., Sherman et al. 2008) in order to improve turnover intention and job performance. Benevolent leadership styles to enhance a benevolently engaged organizational climate are necessary. Top management should create discussion aimed at promoting benevolent climate and use internal marketing to increase workers’ motivation to sustain core benevolent values. Given the positive moderating effect of perceived benevolent climate on the relationship between job stress and job performance, positive organizational interventions to decrease job stress should be provided in case of high perceived benevolent climate. In other words, under such circumstances, job performance shows highly sensitive to the negative effect of job stress, and thus it is critical to reduce job stress by, for example, increasing employees’ sense of job control through training courses of job development and enrichment, setting clear goals to eliminate ambiguity and conflict in workplaces, facilitating social communications among employees to help develop strong interpersonal relationships, and encouraging exercises and quality rest time so as to maintain an emotional balance (Karimi and Alipour 2011).

In summary, this study discovers the mediation of positive affect and the moderation of perceived benevolent climate in the formation of turnover intention and job performance. A firm that takes our preceding suggestions about positive affect and job stress as guidance can be assured of helping generate better employee retention and boosting their job performance. Broadly speaking, ethical climate can be considered as an organization’s lubricant that smoothes out its holistic operations. For that reason, a firm should gradually develop its ethical climate over the long run via open-minded dialogues between it and its employees.

5.3 Limitations

Despite this study’s great efforts, two limitations should be noticed. The first limitation related to the servicing professionals in high-tech industry investigated by this study. As these servicing professionals in high-tech industry counted heavily on technology to interact with and provide service to customers in a timely manner (Kandampully 2002), the findings of this study might be more generalizable and applicable to the workers using technology to deliver service (e.g., e-banking, e-ticketing, e-tailing, etc.) rather than those relying on face-to-face interaction to deliver service (e.g., hair salon service, pizza delivery, laundering, cleaning, and dyeing). Second, this study did not include variables (e.g., subjective norm, moral identity, ethical judgment, justice) beyond the COR theory and ethical climate theory upon which this study is built. It is likely that these variables may have substantial mediating or moderating effects versus what we have found in this study.

In summary, future research may try to gather longitudinal data by recording the survey subjects’ actual behavior across several long periods of time. Future scholars are advised to include a variety of factors related to traits of individuals and service industries (e.g., Chen et al. 2017; Luan and Tien 2017; Tai 2017; Wang et al. 2017) based on different academic theories (e.g., Bhattacherjee and Lin 2015; He et al. 2014; Lin 2017a, b) so as to investigate turnover intention from broader aspects. Last but not least, the test results herein that disconfirm our hypotheses (i.e., the unsupported ones) may be further evaluated in future research so that the relationship between turnover intention, job performance, and their key predictors and moderator can be authentically verified.