How ostracism jeopardizes customers' interests at restaurants: a study in context of COVID-19

Abstract

Purpose

With theoretical underpinnings in the conservation of resources theory, this research aims at understanding the link between workplace ostracism (WPO) and its effects on customers' interests in the context of COVID-19, with the mediation of stress and moderation of self-efficacy (SE).

Design/methodology/approach

The study followed a time-lagged design. A sample of 217 frontline employees working in the food sector of southern Punjab, Pakistan, responded to the study questions using the survey method with structured questionnaires. A Statistical Package for the Social Sciences (SPSS) tool was utilized for data analysis with bootstrapping and PROCESS macro.

Findings

The findings show that an important mechanism by which ostracism translates into customer service sabotage (CSS) is the increase in perceived stress levels of the employees. Additionally, SE was found to be an important personal resource that acts as a moderator in the said relationship.

Practical implications

Employees with high SE sense less workplace stress even during a pandemic. Leadership should consider the stress-alleviating effect of SE for lessening the damaging influence of WPO on customers.

Originality/value

The study fills an important empirical gap in the context of the COVID-19 pandemic, by showing that due to resource loss perceived by employees while being targeted by ostracism, they may decide to transfer their frustration towards organizational customers by sabotaging their service experience.

Citation

Sarwar, A., Abdullah, M.I., Imran, M.K. and Rafiq, N. (2023), "How ostracism jeopardizes customers' interests at restaurants: a study in context of COVID-19", Journal of Asian Business and Economic Studies, Vol. 30 No. 3, pp. 210-225. https://doi.org/10.1108/JABES-12-2021-0215

1. Introduction

Providing excellent customer service has become a norm in the western services sector due to its established link with customer satisfaction (Thuy and Thao, 2019; Templer et al., 2020). The customers are considered the kings (Kim and Aggarwal, 2016). Regrettably, instead of always treating the customers warmly, service providers indulge in counterproductive behaviors (Lee and Ok, 2014), like sabotage of customer service.

Customer service sabotage (CSS) means service personnel's misbehavior that is purposely intended to damage the service experience (Harris and Ogbonna, 2006). Changing the speed of service delivery, avenging actions, being troublesome and demonstrating irritation or aggression to the organizational customers are examples of this CSS behavior. According to a research carried out by Harris and Ogbonna (2002), among the frontline employees, over 85% admitted that they indulge in some kind of CSS behavior, like slowing down the speed of service, dropping sauce on the customer's clothes and spitting in the customer's food. Additionally, they also told that CSS occurs on daily basis in their workplace environment.

Research related to CSS is imperative since customers are crucial for their organizations. Studies have demonstrated that frontline personnel's behavior directly affects customer satisfaction, loyalty, eagerness to come back and the word of mouth phenomenon (Harris and Ogbonna, 2009; Lee and Ok, 2014). This consecutively impacts organizational revenue growth, profitability and shareholder value (Harris and Ogbonna, 2009). Harris and Ogbonna (2012, p. 2043) have called future researchers to “explore further the drivers of service sabotage.”

According to Bradshaw (2020), it has been noted that there is an increase of 11% in takeaways since the outbreak of COVID-19 and during the lockdown. Additionally, it has also been observed that since the outbreak of COVID-19, cases of mistreatment have increased (Ahmed et al., 2021). The statistics related to rising cases of uncivil behavior (Usdaw, 2020) are worrisome as they have the potential to deeply and adversely impact organizational outcomes.

In Pakistan, customers also confront adverse CSS behavior before the outburst of COVID-19. According to Kashif and Zarkada (2015), customers in Pakistan frequently encountered unpleasant experiences. Possibly an important factor contributing to such customer sabotage behavior might be workplace stress, which results in several unwanted actions from employees (Fox and Spector, 2006). The Pakistani hospitality sector is particularly marked by high stress in the presence of various stressors (Nawaz and Sandhu, 2018). According to a study conducted by Nisar et al. (2021), the stress and depression prevailing in employees working in the Pakistani hospitality sector during COVID-19 resulted in several deviant behaviors including verbal abuse, harassment and physical assault.

It is believed that one reason behind stress, in addition to the pandemic, would be the mistreatment faced by employees from co-workers, peers and supervisors. Mistreatment results when employees experience verbal abuse, offensive actions or unfair burden at the workplace (Abubakar et al., 2018). A growing number of researchers have started focusing on workplace mistreatment faced by employees (e.g. Steinbauer et al., 2018). Several studies have shown its damaging effects on personnel's well-being, actions and attitudes (e.g. Jahanzeb and Fatima, 2018).

Yet, it is only recently that scholars (e.g. Sarfraz et al., 2019) have begun to concentrate on the extensively occurring kind of maltreatment, i.e. workplace ostracism (WPO) (Fox and Stallworth, 2005; as cited by Steinbauer et al., 2018), especially in Pakistan (e.g. Ali et al., 2020b). WPO is defined as the degree to which a person gets the impression of being overlooked, isolated or excluded by workplace peers, colleagues and/or supervisors (Ferris et al., 2008). Several researchers have demonstrated an increasing occurrence of ostracism in workplaces (O'Reilly et al., 2014; Fox and Stallworth, 2005).

COVID-19 resulted in feelings of disconnectedness in employees working in the hospitality industry (Nisar et al., 2021) due to the social distancing protocols. In such a situation, when they are faced with WPO, the feelings of isolation would be heightened, which may lead to stress and depression. Earlier research has also shown that when personnel perceive to be socially disconnected and isolated, it increases their depression (Li and Huynh, 2020). Similarly, a recent study (Sarwar et al., 2022a, b) showed that the fear of a pandemic leads to emotional exhaustion in employees, which in turn negatively affects their performance. Therefore, it would be interesting to study how during the days of a pandemic, how ostracized employees would react to depletion in their resource reservoir.

The conservation of resources theory provides an ideal opportunity to understand the consequences of WPO. WPO as a stressor drains worthy resources that are vital for the encouragement of personnel within the organization (Nisar et al., 2021). In this condition, an individual's protective mechanism would be activated. For defending against further loss in resources, employees would experience constant strain and encounter further resource depletion, ensuing in a variety of undesirable work-related consequences (Hobfoll and Shirom, 2000).

Keeping in mind the adverse outcomes of WPO (Jahanzeb and Fatima, 2018; Chung, 2018), it is pertinent to identify the boundary situations in which its harmful influence might be mitigated. It is believed that self-efficacy (SE) as a personal resource might act as a buffering agent and mitigate the harmful effect of WPO on stress as well as job outcomes. SE represents an individual's belief in his or her skills, abilities and control over external incidents (Bandura, 1997). Previous research has demonstrated that highly self-efficacious personnel have a powerful sense of motivation to decipher intricate job issues, and they discover personal delight in such endeavors (e.g. Gist and Mitchell, 1992).

Therefore, the purpose of this research is to answer the questions: (a) whether stress acts as a mediator in the relation between WPO and CSS; and (b) whether SE can moderate the indirect association between WPO and CSS via stress.

2. Hypotheses development

2.1 Direct relationship

Previously, scholars have shown that ostracism is related to negative affect (Williams et al., 2002), depression, anxiety, frustration (Colligan and Higgins, 2006) and emotional exhaustion (Wu et al., 2012). Moreover, Williams (2001) put forward that WPO plays the role of an interpersonal stressor, resulting in stress. This argument is also supported by the COR theory, which posits that stress occurs when key resources are threatened or actually lost (Hobfoll et al., 2018, p. 104).

When an employee is in negative mental state, it results in undesirable job outcomes, like counterproductive work behavior (CWB) (Ferris et al., 2008). Additionally, according to Imran et al. (2021), WPO harms employee performance. Similarly, other researchers (e.g. Sarwar et al., 2020a, b; Abubakar et al., 2018) have recently shown that WPO leads to service sabotage behavior. This might be because according to COR theory, ostracism as a stressor reduces the resources required to fulfill work demands (Wu et al., 2012). Ostracized employees may feel that they are not being equitably rewarded by the organization (Sarwar et al., 2020a, b). In trying to restore resource balance and equity, they decrease work efforts toward organizational success, which is essentially dependent on the customers. Based on these arguments, it is hypothesized that

H1a.

There is a positive relationship between WPO and CSS.

H1b.

WPO is positively related to stress.

Stress adversely influences efficiency, effectiveness and an individual's quality of performance (Savery and Luks, 2001). People with higher levels of stress experience cognitive exhaustion, which leads to draining a person's energy that he or she requires for task completion. Moreover, it is pertinent to mention here that recently COVID-19 has also played a part in affecting employees stress' and job performance level (Sarwar et al., 2021). According to Hafeez et al. (2022), psychological state of employees has the capacity to translate into service sabotage behavior towards the customers.

Stress arising from WPO would result in CSS behavior because in accordance with COR theory (Hobfoll, 1989, 2001), people are inclined to preserve resources when facing stress. To conserve resources, employees would hesitate to spend further time and efforts in better service quality and would therefore enact service sabotage. Hence,

H1c.

There is a positive relationship between stress and CSS.

2.2 The mediating role of stress

Contemporary research on WPO has observed a positive association between WPO and deviant behavior (Ferris et al., 2008). Segregated individuals concentrate on the present and have diminished views of future aims and objectives (Twenge et al., 2003). Subsequently, these people are incapable of concentrating on their long-term career targets and ambitions, e.g. promotions resulting from extending better customers services.

Particularly, when employees come across challenging encounters with consumers while facing segregation at workplace and being in distressful cognitive conditions, they may perceive physical and mental exhaustion, diminished levels of energy and unwarranted weariness. Therefore, they would be depleted of emotive resources to handle the ongoing challenges (Lee and Ok, 2014) related to better customer services. In order to cope with related perceptions of resource drainage and for maintaining and defending existing resources, employees may elect to withdraw efforts exerted in good customer services and therefore might engage in CSS behavior.

Consequently, the current research proposes stress as a mediator such that WPO decreases the quality of consumer service experience because of the increasing levels of perceived stress by personnel by following the lead of Sarwar et al. (2020a, b). In line with the groundings of conservation of resources theory (Hobfoll, 1989), personnel who sense loss in resources, like social belongingness because of workplace segregation and isolation, would perceive more stress. For upholding their resource preservation course, they would limit the use of resources in pleasant customer service and consequently display CSS behavior (Hobfoll, 2001). Hence,

H2.

Stress acts as a mediator between the positive relation of WPO and CSS.

2.3 The dual moderation of self-efficacy

The extant literature demonstrates that employees with high levels of SE draw from constructive feelings of achievement when completing their job duties (Lee and Akhtar, 2007). This results in a reduced sense of stress at the job. Self-efficacious employees can identify several ways of doing their job (i.e. providing better customer services) because they are capable of utilizing a broader knowledge base for accomplishing their job (Bandura, 1997) even in the presence of undesirable circumstances including WPO (Sarwar et al., 2019). This results in an increased probability of performing better which may reduce the level of stress that they experience (Schwarzer and Hallum, 2008) while facing ostracism.

SE enhances employees' capability of performing positively in a stressful and ostracizing workplace environment (Sarwar et al., 2019). Employees who strongly believe in their abilities and skills are disposed to achieve their task duties (i.e. better customer services) more frequently as their resource pools would be less weakened by negative spirits and stress ascending due to social isolation. Hence, self-efficacious people do not need to reserve resources during a workday (Hobfoll, 2001).

According to Shao and Skarlicki's (2014) research on the application and effect of stress on the employee's customer services behavior during service delivery, a positive link was found between stress and employee behavior. However, Fida et al. (2015) showed that employees who are more self-efficacious have a lower propensity to act counterproductively. They demonstrated that SE moderates the association of stressors and counterproductive behaviors. Generally, past literature has shown that SE promotes positive behaviors (Bandura et al., 2003). Such a role of SE in preventing counterproductive behavior is also in line with the studies that report that employees with positive core self-evaluations exhibit less counterproductive behavior (e.g. Chang et al., 2012).

The COR theory posits that personnel who are able to extract worthy personal resources feel a desire to indulge in substantial performance-boosting behaviors since they expect such actions to be a source of generating additional resource expansion in the future (Hobfoll and Shirom, 2000) (for instance, better pay and chances of promotion for quality service endowment to customers). Hence, it is proposed that

H3.

SE acts as a moderator between the indirect relation of WPO and CSS via stress in a way that the association will be weaker for highly self-efficacious personnel.

Figure 1 shows the integrated moderated-mediation model.

3. Methodology

3.1 Sample design, size and procedure

The sample was drawn from restaurants operating in the southern Punjab region of Pakistan through the convenience sampling technique. Research has demonstrated that a growing number of food outlets have commenced operations in Pakistan, with a projected increase of 1.17% consumption between 2015 to 2019 (Ahmed et al., 2020).

In this study, eight restaurants from Bahawalpur city, nine from Multan city, three from Lodhran city, four from Khanewal city and two restaurants from Vehari city participated the study. A renowned formula of the 10:1 rule, i.e. 10 participants for every question of the scale was used to compute the required size set of the sample (Kline, 2013; Bentler and Chou, 1987), which produced the needed sample size of 300. In behavioral research, the response rate yielded from nonmanagerial employees is generally 60% (Anseel et al., 2010). In order to get 300 responses, around 510 forms were distributed at T1 amongst the full-time restaurant employees. From these, 286 forms were completed at T1; 286 forms were distributed again at T2; and 242 questionnaires were returned back. The questionnaires from respondents at T1 and T2 were matched by using a code (name initials and birth month) produced by following written guidance on the questionnaires.

First, the participants answered questions regarding their demographic characteristics and WPO at time T1. Next, after two weeks of the time gap, the respondents answered the statements about perceived stress and SE at time T2. This approach was adopted to decrease the chances of common method variation (Podsakoff et al., 2012).

The participant's customers were asked to fill in questionnaires regarding CSS at T2. In total, 79 customers were selected on the spot in the restaurants to fill in the questionnaires regarding the person they interacted with the most during their transaction (after making sure that the respective employees had responded to other questions). After matching the responses at T1 and T2 and those of employees and their customers, the responses left in analyzable form were 217 (151 males and 66 females). These figures align with the overall labor force statistics from Pakistan, which demonstrate that workplaces and restaurants are dominated by male and young staff members (Islam et al., 2021).

During the data collection phase, COVID-19 lockdown restraints were somewhat relaxed, and a smart lockdown was prevailing in major cities of Pakistan (Islam et al., 2021). The data collection only started after each respondent gave their informed consent and was aware of the ethical aspects considered regarding data confidentiality. For improving openness in responses, the participants ascertained that their responses would not be shared with anyone on any platform. All this ensured the reduction of common method variance (Podsakoff et al., 2012).

3.2 Instruments

The participants had five response options from 1 (“strongly disagree”) to 5 (“strongly agree”) for all scales unless otherwise mentioned.

3.2.1 Workplace ostracism

WPO was measured by using a five-point Likert scale developed by Ferris et al. (2008). It consists of ten items. Sample items are “You noticed others would not look at you at work” and “Others left the area when you entered.” The scale showed a good reliability value (Cronbach's alpha = 0.73).

3.2.2 Stress

Stress was measured by a scale created by Galinsky et al. (1998). The responses were made up of a five-point Likert scale, which had options from never to very often. The scale comprises seven items. The questions asked the respondents about the frequency of how they felt about a particular state during the last three months. The sample statements included in the questionnaire were “frustrated by work” and “nervous or stressed.” This scale showed that it is reliable. The reliability values came out to be satisfactory (i.e. Cronbach's alpha = 0.76).

3.2.3 Self-efficacy

SE was assessed by utilizing the scale created by Chen et al. (2001). The scale comprises eight items The respondents answered statements like “I will be able to successfully overcome many challenges” and “I am confident that I can perform effectively on many different tasks.” The scale showed a good reliability value (Cronbach's alpha = 0.81).

3.2.4 Customer service sabotage

CSS was measured by an instrument created by Chi et al. (2015) who have adapted it from Harris and Ogbonna (2006) and Chi et al. (2013). This instrument has six items. The response options were made up of a Likert scale of five points, with the choices from 1 (Never) to 5 (Always). A sample from this scale is how often that (specific) employee indulges in these kinds of actions with you during your transactions; for example, “Behaves negatively with you” and “Tries to take revenge.” This instrument demonstrated a good reliability value (Cronbach's alpha = 0.79).

3.2.5 Control variables

Age and gender were included as control variables. Earlier researchers (e.g. Sarwar and Imran, 2019; Sarwar et al., 2022a, b) have shown that career-related outlooks, attitudes and behavior of Pakistani women within contemporary organizations vary as compared to men. Therefore, scholars have used such covariates (Sarwar et al., 2020b) related to WPO and burnout (e.g. Chung, 2018).

4. Data analysis and results

First, the data were analyzed to ensure that it is appropriate with respect to descriptive statistics, correlation values and variance. This was done with Statistical Package for the Social Sciences (SPSS) and AMOS. Afterward, the moderated-mediation model was tested by utilizing model 58 of Hayes's PROCESS macros. Data were set for resampling at 5000, resamples.

The demographic characteristics of the participants are depicted in Table 1 which shows that 70% of the participants were male and fell in the age bracket of 26–30 years (40%). Most of the employees were graduates, and 50% of the participants had 6–10 years of work experience.

4.1 Preliminary analyses

Before moving towards the main hypotheses testing, some preliminary analyses were conducted to describe notable characteristics of the sample. Table 2 reflects the standard deviation, means and correlation coefficients of the study variables along with control variables. The results showed a positive correlation between WPO and stress (r = 0.33, p < 0.01); and WPO and CSS (r = 0.45, p < 0.01). Further, as predicted, there is a negative association between WPO and SE (r = −0.39, p < 0.05) that indicates the importance of SE characteristics of personality. People who have SE on the higher side may not be much affected by ostracism faced in the workplace. In addition, SE was negatively associated with stress and CSS (r = −0.56, p < 0.01 and r = −0.29, p < 0.05; respectively). Moreover, as predicted, stress and CSS has witnessed a moderate positive association (r = 0.48, p < 0.01).

In order to ensure the absence of common method bias, Harman's single factor test was applied. The total variance was verified by utilizing a single-factor model through exploratory factor analysis. The maximum variance explained by the model was 28.1%, which is less than 50%. This strengthens our confidence that the data were free from common method bias (Harman, 1976; Podsakoff and Organ, 1986).

4.2 Confirmatory factor analysis (CFA)

Confirmatory factor analysis (CFA) was carried out by utilizing AMOS (analysis of a moment structures) software. CFA is important to generalize the attained outcomes of a study (Hoyle, 1991). McArdle's (1996) rules were used to carry out CFA as it is appropriate to come up with the model fit values step by step on the basis of the principles available in the work carried out by Kline (2006). The values of model fit were obtained by using Byrne's (2013) method. The values of CFA confirmed that all constructs were reasonably operationalized and measured what they aimed to measure in this study. The results demonstrated suitable figures lying within conventional ranges, i.e. chi-square/df = 2.49 < 3.00, Root Mean Square Error of Approximation (RMSEA) = 0.052 < 0.08, goodness of fit index (GFI) = 0.92, Bentler Bonnet Nonnormed Fit Index (BBNNFI) = 0.94, Comparative Fit Index (CFI) = 0.91 and incremental fit index (IFI) = 0.96 > 0.90 (see Table 3).

Moreover, Table 3 also shows the average variance explained (AVE). This indicator is used to ensure convergent validity. All the results surpassed the value of 0.50, which is an acceptable criterion. Additionally, for checking the reliability, the alpha coefficient was calculated, which showed satisfactory values higher than 0.70. The values of other indicators were found to be within acceptable range, i.e. WPO (α = 0.73, AVE = 0.59), stress (α = 0.76, AVE = 0.65), CSS (α = 0.79, AVE = 0.59) and SE (α = 0.81, AVE = 0.61). These tests ensured the constructs have convergent validity. In order to check the discriminant validity, average shared variance (ASV) was calculated, which also shows acceptable values. ASV values ensured the unique capability of the studied construct, as evident from Table 3. Further, the values of ASV and MSV are lower than the AVE, which ensures the convergent and discriminant validity of the constructs.

4.3 Test of direct and mediation relationship

The first hypothesis proposed the direct relationships between WPO, stress and CSS as H1 (a, b and c), whereas the second hypothesis predicted that stress will mediate the relation between WPO and CSS. To analyze these predictions, a four-step technique was used to test mediation relations as elaborated by Baron and Kenny (1986). This research also assessed the parameters of the mediation relationship with Model 4 of PROCESS macro (Hayes, 2017).

Regression results are shown in Table 4. It consists of three models to clarify the values obtained from the mediation test. Model 1 (Step 1) indicated that WPO is significantly related to CSS behavior (β = 0.34, t = 9.42, p < 0.01) and around 18% variation found in CSS through WPO (R² = 0.18, F = 29.98, p < 0.01). Further, Step 2 explained in Model 2 showed a positive association between WPO and stress (β = 0.36, t = 8.13, p < 0.01). In third model, after controlling for WPO, stress positively impacted CSS (β = 0.44, t = 15.74, p < 0.01). Lastly, the Model 4 of Haye's (2017) PROCESS macro was utilized as biased-corrected percentile bootstrap method. The results demonstrated that conditions for indirect route of WPO to CSS through stress was fulfilled, ab = 0.11, SE = 0.02, 95% (0.07, 0.13). The H1 (a, b, c) and H2 were satisfied as all four criteria were fulfilled to establish mediation influence of stress between WPO and CSS.

4.4 Test of moderated-mediation relationship

The third hypothesis of this research proposed that SE acts as a buffering agent between the mediating link of WPO and customer sabotage behavior through perceived stress. To examine this relationship, model 58 of Hayes (2017) was used, and two regression models were drawn: first, to test the moderating role of SE in the association of WPO and stress and second, to test the interactive effect of SE on the association between stress and CSS.

In Table 5, Model 1, statistics elaborated a positive association amongst WPO and stress (β = 0.31, t = 6.42, p < 0.01). This influence was weakened by SE (β = −0.27, p < 0.05). Further, in Model 2, a significant relation between stress and CSS was established (β = 0.26, t = 4.22, p < 0.5), and importantly, the stated association was again moderated by SE (β = −0.16, p < 0.05).

For the descriptive purpose, the simple slops of the moderated relationships were plotted as explained in Models 1 and 2 with moderating variables at the standard deviation values of +1 and −1. The results elaborated in Figures 2 and 3 indicate the collaborative effect of SE in the association between (a) WPO and stress and (b) stress and CSS. Hence, based on the above statistics, it is confirmed that H3 was fully supported.

5. Discussion

This study has shown that workplaces have a negative impact on mental state of hospitality sector's employees'. Earlier researchers have demonstrated a positive association between WPO and negative affect (Williams et al., 2002), anxiety (Ferris et al., 2008) and emotional exhaustion (Wu et al., 2012). This study has added to these research studies by not only demonstrating that WPO is positively related to stress but have also extended them by demonstrating WPO's link to detrimental organizational consequences in the form of CSS and boundary condition of SE in the Pakistani hospitality sector's context.

Silent treatment and isolation can produce negative self-perceptions that cause a sense of resource loss (e.g. belongingness) in individuals. Support from others has been established as an important resource to confront tense circumstances (Hobfoll, 1989). When personnel face WPO, it indicates a dearth of social support from workplace peers; hence, a loss in resources is perceived. For protecting themselves from more resource loss, employees engage in negative behavior (Sarwar et al., 2020a, b) and in undesirable actions, which in this research appeared as CSS.

Moreover, another major contribution of this research is the finding that SE acts as an extenuating agent on the harmful influence of WPO. Personal factors, like SE are underexplored in the current literature on WPO. In line with COR theory, the personal trait of SE can serve as a useful resource in lessening the negative effects of colleagues' disrespectful workplace acts. Because self-efficacious people can complete their tasks effectively (Bandura, 1997), they are more confident about their competencies and therefore, are less likely to sense the feelings of being overburdened by a negative mental state. Such individuals can conserve, maintain and use their energy towards performance-improving actions (Hobfoll and Shirom, 2000). Similarly, as the results of this study demonstrate because such self-efficacious employees have ample resources to maintain their task performance, they, therefore, can maintain healthy service relations with clients and indulge less in CSS behavior even under stressful situations triggered by WPO. This in turn not only helps them in gaining a wide customer base and career advantage but also helps the organization in the form of satisfied customers. Consequently, SE is influential in enabling the person to concentrate more on work and less on distractions created by an antisocial work context.

The interactive influence of SE and WPO on stress as well as the collective effect of SE and stress on CSS behavior, therefore, aligns with the premises of COR theory in that SE as a personal resource helps personnel in utilizing their skill base to meet their work demands simultaneously, especially when there are resource-draining work circumstances involved.

5.1 Theoretical contributions

WPO has become a major problem in contemporary organizations (Fatima et al., 2021), with detrimental effects for not only employees (Imran et al., 2021) but the organizations as well. Such effects include anxiety, emotional exhaustion, turnover and deviant behaviors (Sarwar et al., 2019). This research broadens the past research on WPO by demonstrating its association with employees' mental health in the form of stress. Earlier, though stress has been linked to harmful consequences (Chung, 2018), its association with WPO has not often been investigated from the stress outlook (Wu et al., 2012), especially in the context of a global pandemic, making this research an important contribution to the existing knowledge.

The moderated-mediation model proposed by this study was embraced by the tenets of COR theory in the background of Pakistan's food industry and COVID-19, with WPO as an antecedent, stress as a mediating variable, SE as a moderator and CSS as a dependent variable.

This study demonstrates that during a global pandemic, even WPO becomes a challenging problem confronted by frontline staff working in the food industry. COVID-19 is a demanding condition that puts extra conservational demands on workers employed in the food and beverages sector (AHDB, 2020). Additionally, the epidemic has shaped a worrying situation due to strong service pressures linked with well-prepared and well-timed delivery of hygienic food. The study highlights that this pandemic might have increased the prevalence of ostracism in employees due to heightened demands of the pandemic and drenched their resource reservoir resulting in CSS behavior.

Additionally, SE as a positive personal trait and resource was also studied within the premises of COR theory as it plays a mitigating role on the harmful influences of resource loss perceived from WPO, which results in stress and CSS. This research has shown that people with high SE would be least affected by harmful effects of WPO and will involve lesser in the incidences of CSS behavior as compared to those with low levels of SE even under the pandemic situation. People with high SE focus more on their capabilities rather than on others' behavior and therefore utilizes their skills well in line with COR theory.

5.2 Practical implications

The results of this study provide several important implications. As evident from the results, WPO results in stress; therefore, management should try to lessen the occurrence of ostracism. But because WPO is difficult to detect, some form of this covert mistreatment will continue to persist in organizations. Since employees' fear being tagged as vulnerable (Estes and Wang, 2008), they would hesitate to report incidences of behavior, like ignoring, isolating or excluding. In these circumstances, the leadership should look for innovative ways of spotting these behaviors and analyze their origins, like work burden, unwanted role models or ways of communication (Pearson and Porath, 2004). Appropriate leadership styles can be utilized to alter employees' behavior, motivation and creativity level (Minh-Duc and Huu-Lam, 2019; Ali et al., 2020a).

Self-efficacious employees might be encouraged during recruitment, selection, compensation and rewards distributions to inspire employees to use their abilities and skills and find suitable ways of reducing exposure to ostracizing behavior. SE might be boosted with training sessions. This would be advantageous for the whole organization (Jacobsen and Bogh Andersen, 2017). Generally, any personal resource that inspires employees' confidence towards work completion should be principally beneficial in the lessening of ostracism since such resources incline personnel to utilize stress handling tactics to complete their work responsibilities when they sense unfavorable situations.

5.3 Limitations and calls for future researchers

This study is focused on the moderating role of a single specific personal resource, SE; therefore, future researchers could add more moderating variables. Moreover, because WPO inflicts negative outcomes on worklife, there is a reason to believe that it might also affect family life. Hence, similar relationships might be explored with the addition of nonwork-related consequences, like work–family conflict.

6. Conclusion

The current study has extended the existing knowledge base by enlightening the association between WPO and CSS under the premises of COR theory and in the context of a global pandemic. The moderated-mediation model studied in this research has revealed that WPO results in stress which in turn gives rise to CSS behavior, but SE as a moderator, being a positive personality trait and an important resource, can alleviate the strength of this association at both stages. Self-efficacious employees tend to behave unfavorably towards customers less frequently.

Figures

Figure 1

The proposed integrated model of moderated mediation

Figure 2

Moderating influence of SE (self-efficacy) on the association between WPO (Workplace Ostracism) and ST (Stress)

Figure 3

Moderating influence of SE (self-efficacy) on the relation between ST (stress) and CSS (Customer Service Sabotage)

Table 1

Demographic details of the participants

Particular (N = 217)	Frequency	Percentage
Gender
Male	151	70
Female	66	30
Age
Up to 25	66	30
26–30	86	40
31–35	43	20
36–40	10	5
41–45	7	3
45–50	3	1
51 and above years	2	1
Experience
up to 5	77	35
06–10	108	50
11–15	10	5
16–20	11	5
21–25	5	2
26 and above years	6	3
Academic qualification
Graduation	165	76
Master	25	12
MS/M. Phil and above	27	12

Table 2

Descriptive statistics and correlation analysis

Constructs	Mean	SD	A	G	WPO	ST	Self-Efficacy	CSS
Age (A)	2.14	1.07	1
Gender (G)	1.30	0.46	−0.19**	1
Workplace ostracism (WPO)	3.30	0.65	0.23**	0.17*	1
Stress (ST)	3.13	0.75	−0.11	0.22*	0.33**	1
Self-efficacy	3.26	0.82	0.09	0.04	−0.39*	−0.56**	1
Customer service sabotage (CSS)	2.36	0.80	−0.10	−0.05	0.45**	0.48**	−0.29*	1

Note(s): N = 217, age and gender are control variables, *ρ < 0.05 and **ρ < 0.01

Table 3

Confirmatory factor analysis and scale reliability

Construct Description	Chi-square/df	RMSEA	GFI	CFI	BBNNFI	IFI	CR	AVE	MSV	ASV	AR
Fit indices	2.49	0.052	0.92	0.91	0.94	0.96
WPO							0.81	0.59	0.41	0.24	0.73
Stress							0.85	0.65	0.44	0.23	0.76
CSS							0.82	0.59	0.42	0.20	0.79
Self-efficacy							0.83	0.61	0.41	0.19	0.81

Note(s): Acceptable range of indices chi-square/df < 3.0, GFI-CFI-BBNNFI-IFI > 0.90, RMSEA < 0.08, CR/AR > 0.7, CR= composite reliability, AVE = average variance extracted, MSV = maximum shared variance, ASV = average shared variance, AR = alpha reliability/Cronbach alpha, WPO = workplace ostracism, CSS = customer service sabotage and SE= self-efficacy

Table 4

Testing the direct and indirect effect of workplace ostracism on CSS

Predictor	Model 1		Model 2		Model 3
	(CSS)		(Stress)		(CSS)
	β	t	β	t	β	t
Workplace ostracism	0.34	9.42**	0.36	8.13**	0.17	4.61*
Stress					0.44	15.74**
R2	0.18		0.24		0.28
F	29.98**		41.15**		49.96**

Note(s): Each column is a regression model that predicts the criterion at the top of the column

*p < 0.05, **p < 0.01, N = 217 and bootstrapping at 5,000

Table 5

Testing the moderated mediation effect of workplace ostracism on CSS

Predictor	Model 1		Model 2
	(Stress)		(CSS)
	β	t	β	t
Workplace ostracism	0.31	6.42**	0.29	5.84**
Self-efficacy	−0.18	−4.49*	−0.17	−4.41*
WPO*SE	−0.27	−5.61**
Stress			0.26	4.22*
ST*SE			−0.16	−2.68*
R2	0.35		0.32
F	55.67**		49.78**

Note(s): The beta values are standardized coefficients; thus, they can be compared to determine the relative strength of different variables in the model. *p < 0.05, **p < 0.01, N = 217 and bootstrapping at 5,000