The Role of Occupational Stress in the Association between Emotional Labor and Mental Health : A Moderated Mediation Model

This study investigated whether occupational stress factors moderate the effect of emotional labor on psychological distress in call center employees. A cross-sectional and descriptive study using anonymous paper-based survey methods was conducted in a sample of 283 call center employees in South Korea. Participants completed the Emotional Labor Scale, the Depression Anxiety Stress Scale, and the Korean Occupational Stress Scale. Moderated mediation analyses were conducted using the PROCESS macro in order to investigate the relationship among variables. The results showed that the association between surface acting while having emotional labor and psychological distress was mediated by emotional dissonance. The mediated effect of emotional dissonance was moderated by discomfort in occupational climate, suggesting that improving the occupational environment can lessen the level of psychological distress among emotional workers, and that more attention should be devoted to the development of an intervention at the organizational level in order to prevent mental health problems in this population.


Introduction
With the growth of the service industry, emotional labor and its consequences among front-line service employees have been receiving a lot of attention in the field of public mental health.Emotional labor is defined as the process by which workers have to control their feelings in accordance with the organizational demands and their occupational roles [1][2][3].Emotional labor seems to play a critical role in the development of mental health problems.Previous studies have reported the association of emotional labor with depression, suicidal thoughts, anxiety disorder, and somatization [4][5][6][7][8][9].Additionally, a recent study investigated the effect of emotional labor on the symptoms of post-traumatic stress disorders (PTSD) among firefighters and found that emotional damage from emotional labor while on duty had a moderating effect on the association between recent traumatic exposure and the level of PTSD symptoms [10].
There are two strategies of emotional labor: surface acting and deep acting.Deep acting indicates that employees try to create feeling that must be expressed, while surface acting refers to their merely putting on a mask [11,12].When there was a discrepancy between felt emotions and displayed emotions, employees suffered from emotional dissonance, which is an important component of emotional labor [11].The previous findings showed that emotional dissonance is associated with burnout and mediates between the association between emotional labor and employees' well-being [13,14].Therefore, it is possible that emotional dissonance mediates the association of emotional labor with psychological distress.
The deleterious effect of emotional labor on mental health may be accelerated by work-related stress in an occupational environment.It has been known that work-related stress is an occupational hazard with a critical role in the development of mental disorder [15][16][17][18], as well as physical health problems [19][20][21].In the field of research on occupational stress, the Job Demand-Control-Support (JDCS) model predicts that social support can moderate the negative effect of high-strain (high demand-low control) jobs on the well-being of employees [22].According to the JDCS model, occupational support could moderate the association between emotional labor and mental health among employees who have to do high-strain emotional labor.In fact, previous research has reported that perceived organizational support influences employees' performances during emotional labor and moderates the association of emotional labor with job satisfaction, performance, and emotional exhaustion [23][24][25][26].Besides organizational support, a variety of stress factors at work, such as the demands of the job, organizational injustice, and discomfort in the occupational climate, may influence the consequences of emotional labor.However, few investigations have assessed whether occupational stress factors affect the association between emotional labor and psychological distress, including depression and anxiety, or find how this correlation occurs.
The aim of the current study is to investigate whether various components of occupational stress moderate the effects of emotional labor on the psychological distress including depression, anxiety, and stress among employees in service work.Based on the findings of previous studies and theoretical background, we hypothesized that emotional dissonance mediates the association of surface acting while having emotional labor with psychological distress and that occupational stress factors moderates the mediating effect.

Study Design and Ethic
This was a cross-sectional study among employees recruited from a call center of a credit card company in South Korea, where about 1000 employees work.All participants gave their informed consent.Approval for the collection and analysis of the data was obtained from the Ethics Committee of a research institution (IRB #:1044396-201601-HR-005-01).

Study Population
The study involved a secondary analysis of quantitative data that originally were collected to investigate the mental health status of the employees of the call center.The initial finding from these data, which were published earlier [6], focused on the potential predictors of the status of mental health among the workers at the call center.These initial findings raised new questions concerning why some employees appeared to be mentally stable while others did not.Thus, we reinvestigated these data from the perspective of occupational stress since the environmentally-protective factors and the related stressors affected the employees in different ways.The ages of the participants in the study ranged from 18 to 55 (Mean = 36.05,SD = 8.17).The participants were 90.8% (n = 257) females and 9.2% (n = 26) males.High school graduates and those with less education were 33.6% (n = 95) of the total, 32.5% (n = 92) were junior college graduates, and 30.74% (n = 83) were college graduates (Table 1).Note: N = total sample; M = mean; SD = standard deviation.

The Korean Occupational Stress Scale
We used the culture-specific, 24-item, Korean Occupational Stress Scale (KOSS) to measure the participants' levels of occupational stress in the workplace [27].The KOSS was developed and validated with a nationwide random sample of over 10,000 Korean employees to measure Korean-specific, job-related stress with seven subscales, i.e., job demands (4 items), insufficient job control (4 items), interpersonal conflict (3 items), job insecurity (2 items), organizational injustice (4 items), lack of rewards (3 items), and discomfort in occupational climate (4 items).Each item on the questionnaire was rated on a 4-point Likert scale from 1 (not at all) to 4 (very much).Scores for each subscale were calculated based on the scoring method provided by the developers.Possible ranges for each subscale was 0 to 100, with higher scores representing higher levels of job-related stress.The Cronbach's alpha values ranged from 0.51 (occupational climate) to 0.82 (organizational injustice) when they were determined for Korean employees nationwide [27].Reliability coefficients in this study ranged from 0.60 to 0.86.

The Emotional Labor Scale
We used the Emotional Labor Scale (ELS) to measure participants' levels of emotional labor.The scale consists of nine items with three subscales, i.e., surface action, deep action, and emotional dissonance [12,28,29].Each of the surface action and deep action subscales consisted of three items that asked participants to indicate how they managed their emotions to meet their job demands [28].The emotional dissonance subscale (three items) measures the extent to which the participants feel apart from the emotions they express to their customers on the phone [12]; the response categories for all emotional labor items ranged from 1 to 5, i.e., "not at all" to "a lot."Summary scores and subscale scores were calculated by adding the scores for each item.Possible total score of the ELS ranged from 9 to 45, with higher scores indicating greater emotional labor.Cronbach's alpha for the ELS was found to be in the range of 0.85 (emotional dissonance) to 0.90 (deep action) for the employees of the call center [29].In this study, the value of Cronbach's alpha for the total score of the ELS ranged from 0.86 to 0.89.

The Depression Anxiety Stress Scale (DASS)
The levels of psychological distress of call center employees in this study were assessed with the 21-item DASS [30], which measures levels of depression, anxiety, and stress.It consists of seven items in three different subdomains, each of which has self-rating statements of mental health on a 4-point severity/frequency Likert scale (0 = did not apply to me at all or never to 3 = applied to me very much or almost always).Based on the instructions in the scoring guidelines, the scores for the subdomains were calculated by adding the scores for relevant items and doubling those scores.Then, the scores of the three domains were combined to indicate an overall level of psychological distress.
The possible range for the total score on the DASS was 0 to 126.Higher scores represent higher levels of psychological distress.In this study, the Cronbach's alpha ranged from 0.88 to 0.89.

DATA Analyses
Descriptive statistics were used to evaluate sample characteristics and the distribution of main variables (psychological distress, emotional labor, and occupational stress factors).One-way analysis of variance, and an independent samples t-test were used to evaluate the differences in the psychological distress based on demographic variables.For the distribution of main variables, we examined mean with standard deviation, observed score range, skewness, and kurtosis.The analyses of the normality of the distribution of main variables were carried out using the Kolmogorow-Smirnow test.The distribution of measures of all scales differed from normal distribution.The correlations between the main variables were examined using Spearman correlation coefficients.
We performed the mediation analysis to investigate the association of psychological distress with two factors of emotional labor, surface acting, and emotional dissonance, which were significantly correlated with the DASS total score in the results of the correlation analysis.A simple mediation analysis proposed by Hayes [31] was conducted to examine mediating effects of emotional dissonance during emotional labor on the association between surface acting and psychological distress (Figure 1).We used 5000 bootstrap samples and determined the mediating effect of the 95% confident interval.
Sustainability 2019, 11, x FOR PEER REVIEW 4 of 11 possible range for the total score on the DASS was 0 to 126.Higher scores represent higher levels of psychological distress.In this study, the Cronbach's alpha ranged from 0.88 to 0.89.

DATA Analyses
Descriptive statistics were used to evaluate sample characteristics and the distribution of main variables (psychological distress, emotional labor, and occupational stress factors).One-way analysis of variance, and an independent samples t-test were used to evaluate the differences in the psychological distress based on demographic variables.For the distribution of main variables, we examined mean with standard deviation, observed score range, skewness, and kurtosis.The analyses of the normality of the distribution of main variables were carried out using the Kolmogorow-Smirnow test.The distribution of measures of all scales differed from normal distribution.The correlations between the main variables were examined using Spearman correlation coefficients.
We performed the mediation analysis to investigate the association of psychological distress with two factors of emotional labor, surface acting, and emotional dissonance, which were significantly correlated with the DASS total score in the results of the correlation analysis.A simple mediation analysis proposed by Hayes [31] was conducted to examine mediating effects of emotional dissonance during emotional labor on the association between surface acting and psychological distress (Figure 1).We used 5000 bootstrap samples and determined the mediating effect of the 95% confident interval.(B) depicts that the effect of surface acting on psychological distress is mediated by emotional dissonance.Interaction indexes, a refers to the direct effect of the predictor on the mediator, b refers the direct effect of the mediator on the outcome variable, c refers to the direct effect of the predictor on the outcome, and c' refer to the direct effect of a predictor after controlling the indirect effect of predictor through the mediator on the outcome.
Next, moderated mediation analyses proposed by Hayes [31] were conducted to investigate whether occupational stress factors moderate the indirect effect of surface acting on psychological distress through emotional dissonance.Again, 5000 bootstrap samples and a confidence interval of 95% were selected.Once more, surface acting was entered as predictor, psychological distress as outcome, and emotional dissonance as mediator.In addition, each of four occupational stress factors, which were significantly related to psychological distress in the results of a preliminary correlation analysis, was added as a moderator.Figure 2 illustrates this moderated mediation model.
The data were analyzed using IBM SPSS statistics 22 software (SPSS, Inc., Chicago, IL, USA) and PROCESS macro v3.3 [31] for SPSS.Statistical significance was defined as a two-tailed p-value of < 0.01.Next, moderated mediation analyses proposed by Hayes [31] were conducted to investigate whether occupational stress factors moderate the indirect effect of surface acting on psychological distress through emotional dissonance.Again, 5000 bootstrap samples and a confidence interval of 95% were selected.Once more, surface acting was entered as predictor, psychological distress as outcome, and emotional dissonance as mediator.In addition, each of four occupational stress factors, which were significantly related to psychological distress in the results of a preliminary correlation analysis, was added as a moderator.Figure 2 illustrates this moderated mediation model.
The data were analyzed using IBM SPSS statistics 22 software (SPSS, Inc., Chicago, IL, USA) and PROCESS macro v3.3 [31] for SPSS.Statistical significance was defined as a two-tailed p-value of < 0.01.

Preliminary Analysis
The characteristics of the level of psychological distress, emotional labor, and occupational stress factors in the sample are presented in Table 2.There were no significant differences in the DASS scores based on demographic variables, such as gender and the level of education (p > 0.05).Table 3 provides the Spearman correlation coefficients between the DASS total score with the KOSS-SF and the ELS scores.The DASS total scores had a positive correlation with the levels of surface acting and emotional dissonance due to emotional labor and the occupational stress factors, such as the demands of the job, job insecurity, organizational injustice, lack of rewards, and discomfort in the occupational climate (p < 0.001).

Preliminary Analysis
The characteristics of the level of psychological distress, emotional labor, and occupational stress factors in the sample are presented in Table 2.There were no significant differences in the DASS scores based on demographic variables, such as gender and the level of education (p > 0.05).Table 3 provides the Spearman correlation coefficients between the DASS total score with the KOSS-SF and the ELS scores.The DASS total scores had a positive correlation with the levels of surface acting and emotional dissonance due to emotional labor and the occupational stress factors, such as the demands of the job, job insecurity, organizational injustice, lack of rewards, and discomfort in the occupational climate (p < 0.001).

Mediation Model of the Association between Emotional Labor and Psychological Distress
The results of the mediation analysis are shown in Figure 3.The level of surface acting was positively related to emotional dissonance, and the emotional dissonance was positively associated with psychological distress.When statistically controlling for emotional distress, the level of surface acting was not significantly associated with psychological distress, which indicated that the direct effect of surface acting on psychological distress disappeared.The bootstrap confidence interval confirmed that the indirect effect of surface acting on psychological distress through emotional dissonance (Table 4).These results indicated that the relationship between surface acting and psychological distress was completely mediated by emotional dissonance.

Mediation Model of the Association between Emotional Labor and Psychological Distress
The results of the mediation analysis are shown in Figure 3.The level of surface acting was positively related to emotional dissonance, and the emotional dissonance was positively associated with psychological distress.When statistically controlling for emotional distress, the level of surface acting was not significantly associated with psychological distress, which indicated that the direct effect of surface acting on psychological distress disappeared.The bootstrap confidence interval confirmed that the indirect effect of surface acting on psychological distress through emotional dissonance (Table 4).These results indicated that the relationship between surface acting and psychological distress was completely mediated by emotional dissonance.Notes: SE = standard error; LL95%CI = lower level of the 95% confidence interval; UL95%CI = upper level of the 95% confidence interval.

Moderated Mediation Models
Table 5 shows the results of the moderated mediation analyses treating emotional dissonance as the mediator when each of the four occupational stress factors was used as the moderator in the relationship between emotional dissonance and psychological distress: Job demand was a moderator in the model 1; job insecurity in the model 2; organizational injustice in the model 3; lack of reward   Notes: SE = standard error; LL95%CI = lower level of the 95% confidence interval; UL95%CI = upper level of the 95% confidence interval.

Moderated Mediation Models
Table 5 shows the results of the moderated mediation analyses treating emotional dissonance as the mediator when each of the four occupational stress factors was used as the moderator in the relationship between emotional dissonance and psychological distress: Job demand was a moderator in the model 1; job insecurity in the model 2; organizational injustice in the model 3; lack of reward in the model 4; and discomfort in occupational climate in the model 5. Emotional dissonance appeared to be significant mediators, which have already been reported in the previous paragraph.In the model 1, the main effect of job demand on psychological distress was significant (p < 0.001), but interaction between job demand and emotional dissonance was not significant (p > 0.01).In the model 2, job insecurity did not have significant main and moderating effects on psychological distress (p > 0.01).In the model 3, the main effect of organizational injustice was significant for psychological distress (p < 0.001), while moderating effect of organizational injustice was not significant (p > 0.01).In the model 4, the main effect of lack of reward was significant (p < 0.001), while moderating effect of lack of reward was not significant (p > 0.01).Finally, in the model 5, the main effect of discomfort in occupational climate was significant (p < 0.008) and the interaction between emotional dissonance and discomfort in occupational climate was significant in the model 5 (p = 0.010).Thus, these findings indicate that the mediation of emotional dissonance on the association between surface acting and psychological distress was moderated by discomfort in occupational climate.
The association between emotional dissonance and psychological distress was plotted when the levels of discomfort in the occupation climate was 1SD below and 1SD above the mean.This represents the simple effect of emotional labor at two levels of discomfort in the occupational climate (as a moderator) [32].Figure 4 depicts emotional dissonance moderated by discomfort in occupational climate.To evaluate the conditional indirect effects of surface actin on psychological distress via emotional dissonance, as a function of different levels of the discomfort in the occupational climate, we used the bootstrap method for analysis.Indirect effects at levels of the discomfort in the occupational climate (1SD above the mean, at the mean, and 1SD below the mean) were examined by using the 95% CIs of the bootstrap method.As shown in Table 6, the mediating effect of emotional dissonance changed according to the level of the discomfort in occupational climate and was weakest at 1SD below the mean of it.These indicate that emotional laborers suffering from emotional dissonance were susceptible to psychological distress when they felt discomfort in their occupational climate.Notes: SE = standard error; LL95%CI = lower level of the 95% confidence interval; UL95%CI = upper level of the 95% confidence interval.

Discussion
The current study investigated the relationship among emotional labor, occupational stress, and psychological distress using the moderated mediation analyses.The questions addressed by this study were whether emotional labor negatively influences mental health via emotional dissonance, and whether occupational stress factors moderates the negative effect of emotional labor on mental health.The main finding of the study was that emotional labor was related to psychological distress such as depression and anxiety when employees simulate emotions that are not actually felt (surface acting) and that the relationship was mediated by emotional dissonance which is the discrepancy between felt emotion and displayed emotion.Furthermore, we found that stress from an occupational environment moderated the mediating effect of emotional dissonance on the association between emotional labor and psychological distress.To the best of our knowledge, this is the first study to investigate the moderating role of occupational stress in the relationship between emotional labor and psychological distress among Korean service workers.
In this study, we obtained two findings about the association between emotional labor and psychological distress.First, surface acting was positively associated with psychological distress, while deep acting was not significantly related to psychological distress.Second, the effect of surface To evaluate the conditional indirect effects of surface actin on psychological distress via emotional dissonance, as a function of different levels of the discomfort in the occupational climate, we used the bootstrap method for analysis.Indirect effects at three levels of the discomfort in the occupational climate (1SD above the mean, at the mean, and 1SD below the mean) were examined by using the 95% CIs of the bootstrap method.As shown in Table 6, the mediating effect of emotional dissonance changed according to the level of the discomfort in occupational climate and was weakest at 1SD below the mean of it.These indicate that emotional laborers suffering from emotional dissonance were susceptible to psychological distress when they felt discomfort in their occupational climate.Notes: SE = standard error; LL95%CI = lower level of the 95% confidence interval; UL95%CI = upper level of the 95% confidence interval.

Discussion
The current study investigated the relationship among emotional labor, occupational stress, and psychological distress using the moderated mediation analyses.The questions addressed by this study were whether emotional labor negatively influences mental health via emotional dissonance, and whether occupational stress factors moderates the negative effect of emotional labor on mental health.The main finding of the study was that emotional labor was related to psychological distress such as depression and anxiety when employees simulate emotions that are not actually felt (surface acting) and that the relationship was mediated by emotional dissonance which is the discrepancy between felt emotion and displayed emotion.Furthermore, we found that stress from an occupational environment moderated the mediating effect of emotional dissonance on the association between emotional labor and psychological distress.To the best of our knowledge, this is the first study to investigate the moderating role of occupational stress in the relationship between emotional labor and psychological distress among Korean service workers.
In this study, we obtained two findings about the association between emotional labor and psychological distress.First, surface acting was positively associated with psychological distress, while deep acting was not significantly related to psychological distress.Second, the effect of surface acting on psychological distress was mediated by emotional dissonance.It was repeatedly reported that there was the association of emotional labor with burnout and emotional exhaustion [28,33].Moreover, a significant partial mediation role of emotional dissonance in the relationship between emotional labor and emotional exhaustion was revealed in a previous study [34].Our results incorporate and confirm the previous findings, and also include new findings that emotional labor influences mental health through the mediation of emotional dissonance.Also, the finding surface acting was negatively associated with mental health is consistent with previous findings that there are more negative outcome associated with surface acting such as turnover intentions, withdrawal, and job dissatisfaction in comparison to surface acting [35].
Furthermore, by using moderated mediation models, we found that the level of emotional labor has a positive association with the severity of psychological distress and that its effect on psychological distress varies according to the level of discomfort in the occupational climate.In KOSS-SF, which is the scale assessing occupational stress in this study, the discomfort in the occupational climate is defined as inconvenience caused by the degree of collectivism among workers, and it also may generate from dining out after work hours, inconsistency of job orders, an authoritarian culture, and gender discrimination [27,36].Thus, the discomfort in occupational climate assessed in this study could be similar to the lack of organizational support which is perceived by employees in occupational environments.To our knowledge, our finding is the first to show the moderating effect of occupational stress on the negative association between emotional labor and mental health.This study suggests that alleviating discomfort in the occupational climate by constructing a less collectivistic work culture may make service workers less vulnerable to the psychological distress.Also, it suggests that the role of lower or middle managers in charge of these cultural factors seems to be important since subjective discomfort in the occupational climate may be different by the sub-cultures that make up the informal sector of the company.
In addition, we found that occupational stress factors such as high job demand, organizational injustice, lack of reward, as well as discomfort in organizational climate, were associated with mental health problems among call center employees.Occupational stress factors such as lack of support and discomfort in occupational stress have been found to be the risk factors for depression [36].Also, job strain, organizational injustice, and the imbalance between effort and rewards were found to be related to suicidal ideation among workers [37].According to the models for occupational stress factors, such as the demand-control-support model and the effort-reward imbalance model, high workload, lack of reward, and lack of organizational support could have a significant impact on psychological distress [38].Considering the previous findings and hypothetical models of occupational stress, our finding of significant main effects of various occupational stress factors on psychological distress were expected.
Limitations should be considered when generalizing the results of this study.First, the cross-sectional design of the study limited our ability to infer causal relationships among emotional labor, occupational stress factors, and psychological distress.Future studies need to employ a longitudinal approach to investigate the causal relationships among the risk factors and psychological distress among emotional laborers.Second, this study collected data from self-report assessments, which could be consisted of response bias that affected the results.In future studies, conducting standardized interviews of employees to determine the states of their mental health and occupational stress would be better option to provide more accurate and detailed information regarding the mental health problem among emotional laborers.More than 90% of the workers who participated in this study were women, which was indicative of the fact that mainly women are working in the customer service industry in South Korea.The findings of this study cannot be generalized to call centers that have larger percentages of male employees.

Conclusions
In this study, we found that surface acting while having emotional labor was significantly associated with psychological distress via emotional dissonance.Furthermore, the discomfort in occupational climate turned out to be an important moderator of the association between surface acting and psychological distress.These findings suggest that improving the occupational environment can lessen the level of psychological distress among emotional workers, and that more attention should be devoted to the development of an intervention at the organizational level in order to prevent mental health problems in this population.Given the increases in the service industry and the accumulating findings of emotional labor as a risk factor of psychological distress, there is a need to investigate protective factors that can mitigate the negative effects of emotional labor on mental health problems in service workers.

Figure 1 .
Figure 1.The mediation model.(A) depicts the direct effect of surface acting on psychological distress.(B)depicts that the effect of surface acting on psychological distress is mediated by emotional dissonance.Interaction indexes, a refers to the direct effect of the predictor on the mediator, b refers the direct effect of the mediator on the outcome variable, c refers to the direct effect of the predictor on the outcome, and c' refer to the direct effect of a predictor after controlling the indirect effect of predictor through the mediator on the outcome.

Figure 1 .
Figure 1.The mediation model.(A) depicts the direct effect of surface acting on psychological distress.(B) depicts that the effect of surface acting on psychological distress is mediated by emotional dissonance.Interaction indexes, a refers to the direct effect of the predictor on the mediator, b refers the direct effect of the mediator on the outcome variable, c refers to the direct effect of the predictor on the outcome, and c' refer to the direct effect of a predictor after controlling the indirect effect of predictor through the mediator on the outcome.

Figure 3 .
Figure 3. Mediation model showing that the effect of surface acting on psychological distress is mediated by emotional dissonance.Change in beta weight when the mediator is present is highlighted in bold.

Figure 3 .
Figure 3. Mediation model showing that the effect of surface acting on psychological distress is mediated by emotional dissonance.Change in beta weight when the mediator is present is highlighted in bold.

Figure 4 .
Figure 4. Moderating effect of the discomfort in occupational climate on the association between emotional labor and the severity of psychological distress.

Figure 4 .
Figure 4. Moderating effect of the discomfort in occupational climate on the association between emotional labor and the severity of psychological distress.

Table 2 .
Descriptive statistics for the DASS, the ELS, and the KOSS-SF (N = 283).

Table 2 .
Descriptive statistics for the DASS, the ELS, and the KOSS-SF (N = 283).

Table 4 .
Bootstrap results for indirect effect.

Table 4 .
Bootstrap results for indirect effect.

Table 5 .
Moderated mediation analysis when assuming each of occupational stress factors as a mediator (outcome variable = psychological distress).
Notes: B = regression coefficient; SE = standard error of regression coefficient.

Table 6 .
Conditional indirect effect at specific levels of the moderator when treating emotional dissonance as a mediator.

Table 6 .
Conditional indirect effect at specific levels of the moderator when treating emotional dissonance as a mediator.