Employee norm-violations in the service encounter during the corona pandemic and their impact on customer satisfaction

In this study, an experiment was used to examine the effects of employee norm-violations in the service encounter with respect to what was considered appropriate behavior (e.g., social distancing) during the 2020 corona pandemic. The participants were exposed to a grocery store employee whose behavior was manipulated (norm-violating vs. norm-confirming). Norm-violating behavior resulted in lower perceived employee warmth, lower perceived employee competence, higher disgust, and more dehumanization of the employee. These responses mediated the impact of employee behavior on customer satisfaction, so that satisfaction was attenuated when norms were violated. The mediators, however, typically also instill a hostile, avoidance-seeking mindset for those who are subject to norm-violations, which is likely to result in problems when transgressors are to be persuaded to change their behaviors.


Introduction
During the 2020 coronavirus pandemic, health organizations and governments issued several recommendations for the publicsuch as "wash your hands", "stay home if you do not feel well", "keep the distance to other persons" and "cover your mouth and nose if you cough or sneeze". Given the lethal nature of the virus, which was also clearly and continuously communicated during the pandemic, it seems as if the majority in most countries followed the recommendations.
Indeed, they quickly became social norms. Such norms are shared sentiments about what is appropriate behavior (Buckholz and Marois, 2012) or, in other words, rules that guide behavior without the force of law (van Kleef et al., 2015). Previous research on social norms stress that they play a fundamental role for us humans. For example, we humans are highly interdependent, in the sense that we need to cooperate for many purposesand social norms serve the purpose of making cooperation work (Buckholz and Marois, 2012;Fehr and Fishbacher, 2004). Therefore, we seem to be hardwired to quickly judge whether an encountered person is a social norm-violator (van Kleef et al., 2015;van Leeuwen et al., 2012). Moreover, given that another person is indeed viewed as a norm-violator, we typically react negatively in affective terms (van Kleef et al., 2015;Ohbuchi et al., 2004), also if we are not personally victimized (van Kleef et al., 2015;Tomasello, 2014). This appears to be a fundamental and unique human reaction pattern. Chimpanzees, for example, retaliate against those who steal food from them, but they do not punish individuals who steal food from a third party (Tomasello, 2014). In any event, during the 2020 corona pandemic, many shared the experience of being upset when witnessing norm-violations. Hugging and high-fiving spring breakers at Florida beaches, after-ski party participants in the Alps, and backpacker tourists in Australia are examples of groups whose behaviors were condemned. This is not so surprising, however, because all known human societies punish violators of social norms (Spitzer et al., 2007).
The present study examines customers' reactions to employee behavior in the service encounter with respect to corona-related norms; that is to say, a setting in which the customer interacts on a face-to-face basis with an employee who is representing a firm (cf. Bitner et al., 1990). Such service encounters are indeed social encounters and thus they are subject to several norms (also when there is no pandemic). Examples of norm-related employee behaviors in service encounters are greetings, eye-contact, and listening when the customer speaks.
Given the sinister implications of getting the highly contagious and potentially lethal COVID-19, it is expected that employee normviolations with respect to corona-related recommendations would have a negative impact on customer satisfaction. In conceptual terms, however, several customer responses may contribute to this effect. Therefore, the specific purpose of the present study is to examine a set of customer responses that are likely to mediate the impact of employee norm-violations on customer satisfaction: perceptions of an employee in terms of general characteristics (warmth and competence), disgust, and dehumanization. These specific responses were selected not only because they are likely to be affected by norm-violations (and are likely to influence customer satisfaction); they also represent responses that, if they occur, would pose challenges for those who would want to persuade a transgressor to change his or her behavior.
To this end, an experimental approach was used in which participants were randomly allocated to an in-store encounter with an employee who either was violating or conforming to norms for appropriate social behavior during the corona pandemic (e.g., ignoring social distancing vs. respecting social distancing). The experiment was conducted with UK-based participants (n ¼ 220) in the midst of the corona crisis; at the point of the data collection, mid-April 2020, some 13,000 UK residents had died in COVID-19, the prime minister was hospitalized in an intensive care unit, and many parts of the society were subject to a lockdown.

Theoretical framework and hypotheses
If an employee in the service encounter violates social norms regarding what is appropriate behavior during a pandemic, it is assumed that this would, in a first step, influence (a) the customer's perceptions of employee characteristics and (b) the customer's emotional state.
With respect to the customer's perceptions of the employee, it has been argued that warmth and competence are two universal characteristics of a target person. Warmth comprises aspects such as friendliness and helpfulness; competence has to do with the target person's ability in terms of intelligence, skills, and efficacy (Fiske et al., 2007;Fiske, 2018). Both characteristics have been examined in the specific social context in which the target person is an employee (and the perceiver is a customer), and the typical outcome is that both characteristics contribute to boosting the customer's evaluation of the employee (e.g., S€ oderlund and Berg, 2019). Here, in the present study, it is assumed that norm violations by an employee would result in attenuated levels of employee warmth and competenceand that this outcome is particularly likely when the violated norms have to do with preventing the spread of pathogens. More specifically, given that norms regarding appropriate corona-related behaviors serve the purpose of protecting others from a dangerous disease, it is assumed that employee norm-violations signal to customers that the employee is unfriendly and unhelpful and thus low in terms of warmth. With respect to competence, and given massive doses of information to citizens regarding what was appropriate social behavior during the corona pandemic, it is assumed that employee norm-violations signal low employee competence. The following, then, is hypothesized: H1. In a service encounter in which employee behavior violates social norms regarding appropriate corona-related behaviors, the employee will be seen as less warm and less competent compared to when employee behavior is norm-conforming Moreover, violations of social norms typically produce negative emotions such as fear and anger (van Kleef et al., 2015;Ohbuchi et al., 2004). In the present study, however, focus is on another negative emotion, namely disgust. It is typically experienced as a feeling of revulsion and nausea, and its basic and universal function is to protect us humans from disease by signaling danger and instigating withdrawal and avoidance (Curtis et al., 2011;Hodson and Costello, 2007;Kavaliers et al., 2018;Oaten et al., 2009). Consequently, a person with visible signs of infection typically produces disgust (Oaten et al., 2009). Our disease-based protection system, however, is very sensitive, because false rejections of a threat can be costly (Oaten et al., 2011). Therefore, objects that may increase the risk of being contaminatedsuch as feces, urine, vomit, blood and salivacan also produce disgust responses (Oaten et al., 2009). Indeed, the system is so sensitive that it has evolved to react with disgust also when exposed to people's social conduct (Curtis et al., 2011). This is sometimes referred to as socio-moral disgust (Horberg et al., 2009). It means that disgust can be triggered by norm-violating behaviors with only a loose, indirect or symbolic link to the risk of being contaminated by disease. Researchers have found this link for incest and necrophilia (Gutierrez and Giner-Sorolla, 2007), sexual intercourse with food and kissing siblings passionately on the mouth (Horberg et al., 2009) as well as racism, hypocrite, welfare cheating, and abandoning elderly parents (Haidt et al., 1997). This, then, suggests that the violation of social norms per se may contribute to disgust. If so, the violation of social norms designed to prevent the spreading of a disease can have dual potential to boost disgust. To date, however, not much is known about the extent to which various elements in commercial services evoke customer disgust (although it has been reported, for example, that hotels that do not change linens between guests can be disgust-evoking; Rozin et al., 1994). Nevertheless, the following is hypothesized in the case of employee behavior in service encounters: H2. In a service encounter in which employee behavior violates social norms regarding appropriate corona-related behaviors, the employee will evoke a higher level of disgust compared to when employee behavior is norm-conforming In the next step of the customer's information processing activities, it is assumed that the responses above (perceptions of employee warmth, perceptions of employee competence, and disgust) would affect the customer's perceptions of employee humanness. In general, perceived humanness is the extent to which an entity (e.g., a human, an animal, or a machine) is perceived to have characteristics that are typical for humans (Haslam, 2006;Haslam et al., 2008). This dimension of social perception thus comprises a view of humanness as a variable, and empirical studies show that humans do not automatically receive "full" humanness scores in terms of perceived humanness just because they are members of the species Homo sapiens (e.g., Bruneau et al., 2018). Presumably, one reason is that several aspects of being human, such as having an inner life (e.g., intentions, goals and emotions), are not fully accessible for a perceiver (Haslam et al., 2005).
In any event, it has been suggested that warmth (Haslam and Bain, 2007) and competence-related abilities (Johnson et al., 2006;Martini et al., 2016) represent typical human characteristics, so it is expected here that a perceived lack of warmth and competence attenuates perceived humanness. Furthermore, it has been suggested that also disgust evoked by a target person can reduce the perceived humanness of the person (Harris and Fiske, 2011;Valtorta et al., 2019). This reaction appears to be contingent on the association between disease and disgust that was discussed above. For example, homeless people and drug addicts typically elicit disgust (Harris and Fiske, 2011), which may be caused by beliefs that they come in contact with potential sources of disease, and these groups are typically subject to dehumanization. Moreover, disgust elicited by a person is expected to result in harsh moral judgements about the person (Horberg et al., 2009). Given that morality (i.e., the ability to distinguish what is right from what is wrong) is a central aspect of being human (Gray et al., 2007), it is expected that a target person who evokes disgust would be denied humanness also because of a perceived lack of morality. Hence the following is hypothesized for an employee in a service encounter: H3. In a service encounter in which employee behavior violates social norms regarding appropriate corona-related behaviors, perceived employee humanness will be lower compared to when employee behavior is norm-conforming Then, as final step in the customer's responses to the employee, it is expected that perceived employee humanness is positively associated with evaluations of the employee. In the present study, such evaluations are conceptualized as customer satisfaction in relation to the employee. One main reason behind a positive perceived humanness-evaluation association is that we humans need other humans for many purposes (Epley, 2018), which means that it makes sense, from an evolutionary point of view, to hardwire humans with an innate liking for other humans. In empirical terms, several studies have identified a positive association between the perceived humanness of a target and evaluations of this target (e.g., Aggarwal and McGill, 2007;Bruneau et al., 2018). Conversely, a negative association between dehumanizing a target and the attitude towards the target has also been identified (Hodson and Costello, 2007). Given, then, that norm violating behavior results in reduced warmth and reduced perceived competence (i.e., Hypothesis 1), and increased disgust (Hypothesis 2), and given also that these responses attenuate perceived humanness (Hypothesis 3), the following is hypothesized for customer satisfaction: H4. In a service encounter in which employee behavior violates social norms regarding appropriate corona-related behaviors, customer satisfaction will be lower compared to when employee behavior is normconforming Taken together, H1-H4 suggest that employee norm violations in the service encounter have an indirect influence on customer satisfaction, in the sense that perceived employee warmth, perceived employee competence, disgust, and perceived employee humanness are expected to be underlying mechanisms (i.e., mediating variables). In the subsequent analysis, and in addition to testing H1-H4, this potential for mediation will be explored.

Stimuli, participants, and data collection
A between-subjects experimental design was used in which the participants (UK residents, n ¼ 220, M age ¼ 32.30; 53 men, 165 women, 2 non-binary) were randomly allocated to one of two versions of a description of employee corona-related behavior (low norm-compliance vs. high norm-compliance) in a grocery store service encounter (see the Appendix). The participants were recruited from Prolific, an online platform, and thus they were exposed to the experimental treatments online. One hundred and ten participants received the low normcompliance version; 110 received the high norm-compliance version. After exposure to employee behavior, the participants responded to a set of questions designed to measure the response variables in the hypotheses.

Measures
A 10-point scale were used for all measures. As a manipulation check, the following adjective pairs were used as indicators of employee norm-conforming behavior: "norm-violating-norm-conforming", "immoral-moral", "bad manners-good manners", and "breach with what is acceptable-aligned with what is acceptable" (Cronbach's alpha ¼ .90). Employee warmth was measured with the adjective pairs "cold-warm", "unfriendly-friendly", and "impolite-polite" (alpha ¼ .91), and employee competence was measured with the adjective pairs "incompetent-competent", "unprofessional-professional", and "inexperienced-experienced" (alpha ¼ .94). Similar items have been used by Fiske et al. (2002). Disgust in relation to the employee's behavior was captured with the items "not at all disgusting-very disgusting", "not at all distasteful-very distasteful", "not at all repulsive-very repulsive", and "not at all nauseating-very nauseating" (alpha ¼ .98). Similar items appear in Dens et al. (2008). For perceived humanness, these items were used: "The employee behaved very much as a human", "The employee was humanlike", and "The employee acted like humans typically do" (scored as 1 ¼ do not agree at all, 10 ¼ agree completely; alpha ¼ .91). Finally, customer satisfaction was measured with Fornell's (1992) three satisfaction items adapted to the employee as the person to be evaluated: "How dissatisfied or satisfied are you with the employee?" (1 ¼ very dissatisfied, 10 ¼ very satisfied), "To what extent did the employee meet your expectations?" (1 ¼ not at all, 10 ¼ totally), and "Imagine an employee in a grocery store that is perfect in every respect. How near or far from this ideal did you find the employee?" (1 ¼ very far from ideal, 10 ¼ cannot get any closer). Cronbach's alpha was .96 for this scale.

Manipulation check
The manipulation check showed that the employee norm-violating condition produced a lower level of perceived norm-conformity (M ¼ 3.49) than the employee norm-conforming condition (M ¼ 6.32). This difference was significant (t ¼ 10.50, p < .01). Thus the manipulation performed as intended.

Testing the hypotheses
The outcomes for the response variables in the hypotheses are presented in Table 1. All group differences were significant, so H1-H4 were supported.

Mediation analysis
As indicated in the theory section, the influence of employee normviolating behavior on customer satisfaction can be seen as mediated. Three main routes of mediated influence were assumed: (1) employee behaviorwarmthperceived humannesssatisfaction, (2) employee behaviorcompetenceperceived humannesssatisfaction, and (3) employee behaviordisgustperceived humannesssatisfaction. The zero-order correlations between the response variables are presented in the Appendix, and they indicate the possibility of mediation along the assumed lines. In an attempt to explicitly assess the routes, the Hayes (2012) procedure for analysis of sequential mediation was used (Model 6; 5000 bootstrap samples; employee behavior coded as 1 ¼ norm-violation, 2 ¼ norm-compliance). The analysis showed that that each of the routes was subject to a significant indirect effect (route 1: b ¼ 0.29; route 2: b ¼ 0.21; route 3: b ¼ 0.83; p < .05 in each analysis) and thus that sequential mediation was at hand. For each route, however, there was also a significant direct effect, which indicates complementary mediation (cf. Zhao et al., 2010). That is to say, other mediators than those that were included in the analysis are likely to have contributed, too.

Norm-violations and their implications
During the 2020 coronavirus pandemic, social norms regarding appropriate behavior to prevent spreading of the disease were diffused rapidly, and many individualsin many countriesfollowed the norms. Fear of the coronavirus seems to have been the main explanation (Harper et al., 2020). Everyone, however, did not follow the norms. In the present experiment, an attempt was made to explore what would happen when norms are violated in a commercial setting, a grocery store, which was accessible for many customers during the corona pandemic despite various level of lockdown. In the experiment, employee norm-violations affected each of the included customer response variables.
First, it reduced perceived employee warmth and competence. This effect is consonant with prior person perception research in the sense that we humans typically need only thin slices of information regarding one particular characteristic or behavior of a person in order to ascribe the person other characteristics (which we may know very little about). This is sometimes referred to as a halo-effect; thus one attribute of a person "informs" us about the person's other attributes (Forgas, 2011). However, the experiment did not only produce lower levels of perceived employee warmth and competence in the norm-violation condition. This condition also produced low such levels in an absolute sense. In other studies of perceived warmth and competence of target persons, the combination of low warmth and low competence has been shown to characterize, for example, welfare recipients, the poor, and homeless people (Fiske et al., 2002). Typically, people with this low-low combination are subject to contemptuous prejudice, which comprises anger, contempt, hate, and resentment (Fiske et al., 2002). Given that a state of mind characterized by these emotions is not likely to produce the clearest thoughts, it therefore seems as if those who are subject to norm-violations of the type examined in the experiment are not in a good position to communicate with a transgressor about what is appropriate behavior. It may be noted that the typical service encounter represents a setting in which the customer is alone with the employee; there is no supervisor who can intervene in real time and call for corrections. Thus in a service encounter the customer may have a unique role in correcting norm-violators, yet this role is likely to become undermined if the customer is angry and hateful. Second, employee norm-violations enhanced disgust. This outcome is not surprising, because disgust can be seen as part of a highly sensitive disease-protection system for us humans. Indeed, previous research has shown that far less threatening stimuli that what was used in the present experiment can easily evoke disgust. Typically, disgust is associated with avoidance and rejection of a disgust-evoking object (Curtis et al., 2001;Hodson and Costello, 2007;Oaten et al., 2011). This creates additional problems with respect to attempts to influence the behavior of transgressors (in real time). For example, if a customer reacts with disgust to a transgression, and when the customer's natural reaction is to withdraw from the situation, such withdrawal may not provide the transgressor with feedback calling for behavioral adjustment. Moreover, disgust has been shown to provide fuel for prejudices at the group levelparticularly with respect to so-called outgroups. For example, alleged disgusting behavior in ethnic outgroups can be viewed by in-group members as the cause of a pandemic, or as something that allows it to spread further, which in turn can foster xenophobic reactions (Faulkner et al., 2004). Similarly, anti-Semitism and negative attitudes towards sexual minorities are frequently based on what is perceived as disgusting behavior (Faulkner et al., 2004). Given a general strong sensitivity to disgust, it therefore seems possible that what is perceived as disgusting behavior of one particular member of a group may generalize to the group level so that the group itself becomes associated with disgust. If this group is, say, store employees from one particular firm, or store employees in general, and if disgust at the group level is followed by avoidance and prejudices at the same level, one must assume that this creates problematic consequences during a pandemic.
Third, the norm-violating employee in the experiment was subject to dehumanization in the sense that this employee received lower perceived humanness scores than the norm-conforming employee. Prior research has identified dehumanization (particularly with respect to out-groups) as a malicious form of bias, in the sense that is has been associated with many anti-social attitudes and behaviors (e.g., Bastian et al., 2013;Boudjemadi et al., 2017;Cameron et al., 2016). It should be noted, however, that also the norm-complying employee in the present experiment was dehumanizedin the sense that the humanness scores for this employee (M ¼ 7.51) was less than the "full" humanness score (i. e., 10). Presumably, one reason is that employees in service and retail settings often are means to an end for consumers, and this can result in that they are seen as relatively mindless and thus as less human (Epley, 2018). Another reason is that the minds of others are never fully accessible for us. In any event, and with respect to the norm-violating condition in which the employee was even more dehumanized than in the norm-conforming condition, it should again be noted that in a service encounter setting the customer is typically the only party who is exposed to employee behaviors. Thus, again, the customer has a central role when it comes to the potential to correct behaviors that are not appropriate. Given what typically follows when someone views another person as less than human (i.e., various forms of aggression and anti-social behaviors), however, it seems unlikely that a customer who views an employee in dehumanized terms would be in a good position to provide useful feedback about what is appropriate behavior. Moreover, if customers leak information to the employeewithin the frame of a service encounter in which norm-violations occurthat he or she is seen as less than a "full" human, it is likely that the employee may retaliate by dehumanizing the customer (Kteily and Bruneau, 2017). A vicious circle of reciprocal dehumanization, however, is unlikely to foster customer satisfaction and job satisfaction.
Fourth, the norm-violation condition attenuated customer satisfaction, which was conceptualized as the participants' overall evaluation of the employee. In a service encounter, however, the employee typically is the firm from the customer's point of view (Bitner et al., 1990), so the evaluation of an employee often informs, in a valence-congruent way, the customer's view of the firm in which the employee works. Previous research indicates that customer satisfaction at the firm level has high causal potency; it is likely to influence the willingness to come back and word-of-mouth (Anderson and Sullivan, 1993). Such behaviors have implications for the firm's cost and revenues and, ultimately, for the firm's profitability. A low level of customer satisfaction, then, as a result of employee norm-violations, is likely to attenuate profitability.
Taken together, the experiment showed that the general human sensitivity to norm-violations did materialize also in a commercial setting comprising a grocery store. The reduced level of customer satisfaction in the norm-violation condition indicates that it is not good for business if employees ignore norms regarding what is appropriate behavior during a pandemic. The other responses, however, which can be seen as general responses to norm-violations in social settings, indicate that it would be challenging to establish the calm and collected mindset that is needed for those whose task it is to influence the behavior of transgressors. And, again, in a service encounter, the customer is in unique position to do so. Most likely, in many service encounters, no one else except the customer is there.

Limitations and suggestions for further research
One limitation of the present experiment is that it lumped together several types of behaviors (e.g., not paying attention to social distancing and coughing into the air) that were subject to norms during the 2020 corona pandemic. More precision in the effects of norm-violations would thus be obtained with a design in which each norm-violating behavior is manipulated as a discrete factor. In addition, the sample of participants was limited to UK residents. At the time of the data collection, the UK was hit hard by the corona pandemic in terms of the number of deaths, loss of jobs, and the level of lockdown. This may have made the participants extra sensitive to norm-violations, so further research is needed to establish if similar effects would occur also for participants in other locations.
Moreover, the experiment did not allow the participants to respond in real time to the norm violations when it comes to their own behavior vis-� a-vis the employee. Instead, it was speculated that encountering a target person who is producing perceptions of low warmth, low competence, disgust, and dehumanization would instill an avoidancefocused mindset that is unlikely to encourage victims to make attempts to persuade a transgressor to change his or her behavior. This, however, was speculations, and further research is needed to examine how employee norm-violations in service encounters affect customers' behavior towards transgressors. Indeed, further research is also needed on what would be fruitful ways for a customer (who has been subject to a norm-violation) to influence the transgressor to adopt more appropriate behaviors.
Another and related aspect, which the present study says little about, is how employee norm-related behaviors may influence the customer's behavior in relation to the same norms. Thus, if social norms can be contagious, which they appear to have been during the 2020 corona pandemic, one would expect that what employees do and do not do in relation to norms can influence customers' own norm-related behaviors. One would also expect that a lockdown situation, in which the (normcompliant) customer interacts with only a limited number of persons every day, can enhance the salience of the behavior of the few employees that the customer meets. In this type of situation, the behavior of an employee in one particular service encounter may not be particularly influential. But employee behavior in the aggregate may work in the same way as, say, advertising has been assumed to do (cf. Pollay, 1986). That is to say, prolonged exposure to several employees' behaviors, in which there is a common theme, may produce influential spill-over effects on customer behavior (S€ oderlund, 2015). To examine if this is indeed so, however, researchers would need to move beyond the standard experimental approach in which effects are measured immediately after exposure to one single treatment.