Responsible service employee behaviour and its impact on customer satisfaction during the coronavirus crisis

ABSTRACT

The purpose of this study is to explore a set of mechanisms that mediate the influence of the impact of responsible service employee behaviour on customer satisfaction during the Covid-19 pandemic. A questionnaire was distributed online to UK residents who were instructed to recall and focus on either one very dissatisfying or one very satisfying face-to-face service encounter with an employee during the Covid-19 pandemic. A structural equation modelling approach was used to analyse the associations between the hypothesized response variables. The main finding was that the impact of responsible employee behaviour in service encounters on customer satisfaction was sequentially mediated by perceptions of employee morality and perceived employee humanness. A more parsimonious mediation model comprises only employee morality as a mediator. The attributions of morality to employees are important in a setting in which new social norms guide interpersonal behaviour and in which the violation of such norms can have serious health implications. This study adds empirical evidence to the emerging discourse in the service and retail literature on the implications of the Covid-19 pandemic. It also contributes to the literature on customers’ moral reactions in commercial settings, and to the literature in which perceived humanness is seen as a relevant characteristic of human employees. The results imply that responsible employee behaviour should be encouraged not only from the perspective of the well-being of customers and employees, but also from a business point of view.

KEYWORDS:

• Responsible employee behaviour
• covid-19
• morality
• humanness
• customer satisfaction

1. Introduction

Today, many firms are engaged in activities related to corporate social responsibility (CSR). So far, however, most existing studies have examined this on the corporate level (Aguinis and Glavas 2012; Hejjas, Miller, and Scarles 2019; Lee 2008), particularly with respect to effects of CSR communications (e.g., Kim, Kang, and Mattila 2012). In contrast, an emerging stream of research acknowledges that the behaviour of the firm’s frontline employees can also be seen in social responsibility terms (Aguinis and Glavas 2012; Edinger Schons et al. 2019; Korschun, Bhattacharya, and Swain 2014): responsible employee behaviour can be defined as the actions of the employee that contribute to or detract from organizational-level performance in each of the key areas of CSR (Wiernik et al. 2019). The CSR literature thus underlines that employee behaviours can be seen in terms of social responsibility, and the present study has been inspired by this view in its attempt to examine effects of employee behaviours in service encounters on customers during the Covid-19 pandemic. One main reason why a focus on social responsibility aspects of employee behaviours is warranted is that the notion of ‘social’ took on new meanings during the pandemic, in the sense that many customers became imprisoned in their homes, haunted by fear and boredom, and mindful that contact with other people could be devastating (cf. Krastev 2020). Moreover, the pandemic brought along new norms for safety behaviours of frontline employees, such as keeping social distance and using face masks. In the present study, it is the employee’s responsible behaviours in terms of such safety-related behaviours that is focused. The main thesis is that these safety behaviours can influence the well-being of the customer and that they are capable of setting in motion sense-making among customers in such a way that several variables, including customer satisfaction, are influenced.

There are already several attempts in the service marketing literature to discuss implications of the Covid-19 pandemic for the service field (Berry et al. 2020; Voorhees, Fombelle, and Bone 2020; Söderlund 2020; Chen et al. 2021). Empirical real-life studies of how customers react to employees’ pandemic-related behaviour, however, have so far been relatively rare. The present study continues the discussion in Söderlund (2020) about the impact of the employee’s norm-respecting behaviour on customer satisfaction. In contrast to Söderlund (2020), in which a scenario-based experimental approach was used; however, the present study is based on critical incident data about customers’ real experiences of service encounters during the pandemic.

The specific purpose of the present study is to explore a set of mechanisms that mediate the influence of employees’ corona-related behaviours in service encounters on customer satisfaction. That is to say, in the midst of the pandemic in which hundreds of people died every day, and in which people who did not follow recommendations were subjected to various social sanctions, it is not surprising that employees’ corona-related behaviour in service encounters would influence customer satisfaction. The present study, however, deals with how this influence manifests itself. More specifically, this study assesses in empirical terms the extent to which this influence is (sequentially) mediated by the customer’s attribution of moral capability and humanness to employees. An examination along those lines makes several contributions. It adds empirical evidence to the discourse in the service and retailing literatures on the implications of the Covid-19 pandemic, and it also contributes to the literature on customers’ moral reactions in commercial settings and to the literature in which perceived humanness is seen as a relevant characteristic not only of synthetic agents (Söderlund and Oikarinen 2021) but also of human employees (e.g., Söderlund 2020).

2. Theoretical framework and hypotheses

2.1 Responsible employee behaviour and perceived morality

In this study, the focus is on the service employee’s consumer protection (safety) aspect of responsibility (see Chung et al. 2015) during the Covid-19 pandemic. Several specific behaviours that belong to this category evolved as an interplay between the behavioural recommendations provided by international and national health organizations and governments, such as keeping social distance, frequent and careful hand washing, staying home if you are ill, and wearing a face mask.

The employee’s behaviour in service encounters is subject to copious social norms, and the Covid-19 pandemic added several new norms for the employee – such as those that originated from behavioural recommendations from health organizations and governments. In general, social norms, which represent shared sentiments related to what is considered appropriate behaviour, are prerequisites for cooperation and prosocial behaviour (Buckholtz and Marois 2012). Presumably, it is the importance of social norms for human interaction that has made most people highly sensitive to norm violations (van Kleef et al. 2015; van Leeuwen, Park, and Penton-Voak 2012) so that a violation is typically negatively charged (van Kleef et al. 2015; Ohbuchi et al. 2004) and subject to various forms of punishments in all known human societies (Spitzer et al. 2007; Buckholtz and Marois 2012). Moreover, the extent to which a person violates or conforms to a social norm typically elicits inferences about the individual’s moral standing (Haidt 2001).

In the present study, the focus is on one specific moral variable, namely morality – a person’s ability to distinguish between what is right and what is wrong and to do what is right (Gray, Young, and Waytz 2012). Morality is an advanced capability, because it requires both agency and emotionality – two main characteristics of having a mind (Gray, Gray, and Wegner 2007; Gray, Young, and Waytz 2012). Morality, then, is a general capability needed for specific moral-related reactions to an event, such as the ability to detect a norm-violating event, moral judgment regarding one particular event, assessments of the moral character of an agent (e.g., his or her trustworthiness and honesty), and ex post reasoning about an agent’s intent and guilt (Haidt 2001; Malle 2016).

From a perception point of view, and given that employees’ pandemic-related behaviour in a service encounter are subject to social norms, and given a general tendency to attribute morally charged characteristics to those who violate and conform to social norms, it is assumed in the present study that customers who interact with employees in service encounters use cues from the employee’s pandemic-related behaviour to generate inferences about the employee’s morality in the following way:

H1: Perceived responsible behaviour of an employee is positively associated with perceived employee morality

2.2 Morality and perceived humanness

Perceived humanness is a general dimension referring to the extent to which a subject in a social setting is perceived as typically human (Haslam and Bain 2007; Haslam et al. 2008), and it has been used in studies of both non-human objects (e.g., brands, virtual assistants, and robots) and real humans. With respect to the latter, a typical study assesses the extent to which individuals or groups who are subject to various negative stereotypes are dehumanized in perceptual terms. Examples include drug addicts (Cameron, Harris, and Payne 2015), older people (Boudjemadi, Demoulin, and Bastart 2017) and women in general (Rudman and Mescher 2012). Recently, however, it has been shown that service employees can also be subject to dehumanization (Söderlund 2020). In the present study, it is assumed that the perceived morality affects the perceived humanness. The main reason for this is that morality signals the presence of a mind (Gray, Gray, and Wegner 2007; Gray, Young, and Waytz 2012), and perceptions of an agent’s mind are crucial when an agent is ascribed humanness (Söderlund and Oikarinen 2021). Hence, the following is hypothesized for customers who interact with employees during a pandemic:

H2: Perceived employee morality is positively associated with perceived employee humanness

2.3 Perceived humanness and customer satisfaction

In a social perception situation, it is expected that an agent’s perceived humanness affects the overall evaluation of the agent. One main reason is that humans are subject to a ‘person-positivity bias’ in that what is perceived as typically human is positively charged (Sears 1983). Presumably, this bias has evolutionary roots; there are several fitness advantages of humans liking each other, such as those related to mating, caring for offspring and improved cooperation in groups. Additional reasons behind this form of ‘what is human is good’ heuristic is that (1) a human who ascribes humanness to another human implies perceived similarity, and similarity typically boosts evaluations, and (2) we humans have well-developed schemes for making sense of other humans and the application of such schemes enhances processing fluency – and high levels of fluency typically generate positive affect, which can colour evaluations in a valence-congruent way (Söderlund and Oikarinen 2021).

In a service encounter, then, it is expected that perceptions of employee humanness would have a positive influence on overall evaluations of the employee. Here, in the present study, customer satisfaction with the employee is used as the evaluation variable, and it is thus expected that perceived employee humanness would be positively associated with customer satisfaction. Empirical results that point in this direction have been provided in a study by Söderlund and Oikarinen (2021). The following, then, is hypothesized for customers’ perceptions of employees in service encounters:

H3: Perceived employee humanness is positively associated with customer satisfaction

2.4 Mediation issues

The reasoning involved in H1-H3 implies that the influence on customer satisfaction of the employee’s pandemic-related behaviour in service encounters is sequentially mediated by perceptions of the employee’s morality and humanness. To test this explicitly, the following is hypothesized

H4: The impact of perceived employee responsible behaviour on customer satisfaction

is serially mediated by perceived employee morality and perceived employee humanness

However, several rival routes of influence on customer satisfaction are possible. First, the influence on customer satisfaction of the employee’s pandemic-related behaviour in a service encounter may be mediated by perceived morality alone. This route should be seen in the light of the role of moral-related variables as antecedents to evaluations in perception situations; several studies suggest that the assessment of a person in moral terms is likely to have a direct influence on the overall evaluation of the person (Goodwin, Piazza, and Rozin 2014; Goodwin 2015). For this alternative route, then, perceived humanness does not contribute as mediator. Second, it is possible that the impact of the employee’s pandemic-related behaviour on customer satisfaction may be mediated by perceived humanness alone (and not by perceived morality). The possibility of this route should be seen in the light of the fact that a perceiver has no direct access to other peoples’ minds (Gray, Young, and Waytz 2012) and that advanced mind-related capabilities, such as morality, may be hard to assess – particularly in service encounters, in the sense that many such encounters are relatively brief. Third, the influence of the employee’s pandemic-related behaviour on customer satisfaction may be direct or mediated by other variables than perceived morality and perceived humanness. The extent to which such rival routes are at hand must be assessed empirically, and the test of H4 will be conducted so that it allows for an identification of rival routes. In any event, Figure 1 provides an overview of the hypotheses developed in this section.

Figure 1. Perceived service employee responsible behavior and its impact on customer satisfaction during the coronavirus crisis.

3. Methodology

3.1 The context for the data collection

The data for the present study were collected from UK residents in April, 2020. The geographical and time aspects of this context can be assumed to have affected the participants responses, in the sense that it must have been nearly impossible to be ignorant of the pandemic. For example, the UK had declared a lockdown on March 23, and 20 percent of the working-age-population was furloughed in April. Moreover, the retail and wholesale sectors were hit hard; it was estimated that 1.7 million such jobs were at risk (Mckinsey 2020). The seven-day peak of the Covid-19 deaths in the UK regarding this wave of the pandemic was reached on April 13 (942 deaths), and on April 26 it was still at a high level (787 deaths) (https://github.com/CSSEGISandData/ COVID-19). In terms of psychological reactions, the coronavirus worried people especially with respect to their ability to make plans, their own well-being and personal travel plans, and those who still had a job were worried about their health and safety at work (Statistical Bulletin 2020). As for the global context, and in terms of the pandemic’s effects on deaths in all countries, the data were collected during the very first peak of the pandemic (41,239 deaths were recorded when the data collection was completed on April 27). Three additional (and more severe peaks) were to follow before 2021 ended with the sinister result of more than 5 million deaths (WHO Coronavirus).

3.2 Data collection approach and participants

A critical incident-based approach was used (cf. Bitner, Booms, and Tetreault 1990), in the sense that the participants were instructed to recall either one very dissatisfying or one very satisfying face-to-face service encounter with an employee that they had experienced in their roles as customers during the Covid-19 pandemic. The participants were informed that ‘The interaction with an employee should have the following characteristics: it happened during the corona pandemic and the employee’s behaviour made you think about coronavirus-related issues. You may select either a dissatisfying or satisfying face-to-face interaction of this type. But the interaction should be of the type that you can recall it, and it should be about what happened at one particular occasion’. A critical incident approach of this type was assumed to be suitable for the present study, because the participants were able to define what is a relevant incident themselves (Gremler 2004), which facilitates access to the participants’ own perspectives (Chell and Pittaway 1998).

The participants were asked to write a story of what happened in the selected encounter, and to answer a set of open-ended and closed-ended questions about it (e.g., when it happened, the task they engaged in during the encounter, and the type of firm the employee represented). The variables in the hypotheses were measured with standardized questionnaire items (see below).

As already indicated, the participants were UK residents when the data were collected (April 25–26, 2020), and they were recruited from Prolific, an online panel developed for research purposes (cf. Palan and Schitter 2018). Three attention check questions (‘trap questions’) were used to screen out participants with low attention levels (see Jones, House, and Gao 2015). Moreover, as an additional screening activity, the stories provided by the participants were read and the participants who provided stories that did not comprise face-to-face employee–consumer interactions (e.g., business-to-business interactions and phone-based interactions) were excluded. This resulted in a sample of 229 participants (M_age = 46.35; SD = 15.17; 126 women and 103 men) for the analysis.

3.3 The service encounters in the sample

The face-to-face interactions with employees included in the sample happened 0–48 days before the data collection (M_{days earlier} = 10.31, SD = 8.94). The zero-order correlations between the time that had passed between the selected encounter and the time of the data collection and (a) customer satisfaction, (b) employee morality, and (c) employee humanness were not significant. The zero-order correlation between the time that had passed between the selected encounter and the time of the data collection and (d) perceived responsible employee behaviour was negative and significant (r = −0.141, p = 0.033), thus indicating that more recently selected encounters comprised a higher level of perceived employee responsibility. This may reflect that employees became more strict in their corona-related behaviour (or that their behaviour became more salient for the participants) as the pandemic advanced.

The employees the participants had interacted with represented grocery stores (39.3%) or supermarkets (23.6%) as well as courier and online shopping deliveries (6.6%), pharmacies (6.1%), post offices (2.6%), small shops, local shops and small supermarkets (2.6%), other stores and shops (unspecified shops, discount budget stores, charity shops) (2.6%), banks (2.2%), take away and fast food (1.7%), DIY stores (1.7%), and butchers and fish shops (1.3%). Moreover, the interactions included several tasks, such as shopping, parcel or food delivering, paying for shopping, and waiting in the queue when entering a store.

Of the 229 encounters in the sample, 139 comprised a satisfying encounter, while 90 comprised dissatisfying encounters. Examples of such stories are provided in Table 1. To capitalize on the benefits of having a full-range satisfaction variable, the hypothesis testing was made with a pooled sample (i.e., n = 229 in the analysis).

Table 1. Examples of satisfying and dissatisfying service encounters in the sample

High satisfaction

I was out shopping and following the guidelines when another customer bumped into me from the back. I politely asked them to get behind the marked area of the shop floor, but he ignored me. The supervisor of the store (a really lovely lady), was in the isle we were in, and proceeded to gently ask the gentleman to move back to the previous point he had just been in. The man was rather rude and began to curse at her. I couldn’t read the expression on the supervisor’s face, as she was wearing a mask, but her eyes spoke volumes of concern and a little bit of, perhaps anger or disappointment … she ended up removing the man from the store. All I could do was thank her in earnest, and after sanitizing my hands, I got her and her staff a card and some chocolates to, hopefully, show them that they are supremely appreciated. (female, 36 years) I had what I felt was very satisfying interaction with an employee at my local butchers’ shop. This is a small town shop which generally uses an old fashioned ‘sit up and beg’ bicycle outside the shop to advertise and display the origins of their meat on an attached chalk board. The bike is very large and has a basket on the front. To the amusement and admiration of all customers the bicycle has been moved inside the shop in front of the counter. This only leaves room for one customer at a time in the shop (good thing), with the remainder queuing outside. This enforced distancing has become quite a talking point in the shop and has made a potentially tense situation very relaxed and even quite numerous. (female, 70 years)

Low satisfaction

I visited a supermarket to do some shopping. I made sure I was clear and covered and kept my distance, but I wasn’t wearing a face mask but was more than 2 meters away when I approached an employee and asked him how he was and asked for help locating the eggs, he then took a few steps back with an angry look, I could tell through his eyes as he was wearing a mask and said not to speak with him as I do not have a face mask, but I wasn’t displaying any symptoms, and there were others around that didn’t have a face mask either, I felt very sad, angry and unclean. (male, 64 years) During lockdown there was a knock at the door and I opened it as I was expecting a parcel. Instead, there was a man standing right on the doorstep, saying he worked for a firm that laid driveways and asking if I needed any work doing. I was furious – he was right in my face and had been round all the other houses on the street. This firm was trying to capitalise on people being at home but completely disregarding social distancing advice and putting people at risk. I would not now use this company under any circumstances. (female, 45 years)

3.4 Measurement

The variables in the hypotheses were measured with multi-item scales scored on a 10-point scale. The endpoints were labelled 1 = do not agree at all and 10 = agree completely for all items (except the satisfaction items, for which the Fornell (1992) endpoints were used). SmartPLS 3.0 was employed to assess the measures, and the outcome is presented in Table 2 in terms of the factor loadings, Cronbach’s alpha (CA), composite reliability (CR), and average variance extracted (AVE). For a discriminant validity assessment, the Fornell-Larcker criterion was used.

Table 2. The properties of the multi-item measures

Variables	Items	Loadings	CA	CR	AVE	M	SD
Perceived employee responsible behaviour	Th e employee kept a social distance. The employee was concerned with how close the customers were in relation to each other. The employee wore protective items (gloves etc).	0.86 0.84 0.75	0.75	0.86	0.67	6.93	2.86
Perceived employee morality	The employee was capable of telling right from wrong. The employee could understand negative and positive consequences of her/his behaviour. The employee appeared to have had a sense of what is fair. The employee could distinguish between what is good and what is bad.	0.92 0.96 0.96 0.98	0.97	0.98	0.91	7.06	2.97
Perceived employee humanness	The employee behaved very much like a human being. The employee was humanlike. The employee acted like human beings typically do.	0.97 0.97 0.92	0.95	0.97	0.91	7.67	2.69
Customer satisfaction	How dissatisfied or satisfied are you with the employee? To what extent did this employee meet your expectations? Imagine an employee that is perfect in every respect. How near or far from this ideal did you find this employee?	0.99 0.98 0.97	0.98	0.99	0.96	6.46	3.52

Table 2 shows that Cronbach’s alpha was higher than 0.7 for all the variables, which indicates that the scale reliabilities had adequate and stable measurement properties for preliminary research (Peterson 1994). Internal consistency reliability was also confirmed for all the variables through the assessment of composite reliability, which produced higher values than the common threshold of 0.7 (Wong 2013; Bagozzi and Yi 1988). Moreover, convergent validity was confirmed by an assessment of average variance extracted AVE, in the sense that AVE > 0.5 or higher indicates acceptable convergent validity (Bagozzi and Yi 1988). For a discriminant validity assessment, the Fornell–Larcker method showed that no inter-variable correlations were problematic (Fornell and Larcker 1981; Wong 2013).

The perceived responsible behaviour of an employee was measured with three items, which focused on the responsible behaviour in terms of the consumer protection aspect of responsibility (see Chung et al. 2015): ‘The employee kept a social distance,’ ‘The employee was concerned with how close the customers were in relation to each other,’ and ‘The employee wore protective items (gloves etc.).’ To explicitly assess if this would be perceived by the participants in responsibility terms, they were given the statement: ‘The employee acted in socially responsible ways.’ The measure of perceived employee responsible behaviour was positively and significantly associated (r = 0.65, p < 0.01) with the explicit responsibility item, thus suggesting that the former tap into perceived responsibility.

Perceived employee morality was measured with four items focusing on morality as perceptions of a general capability. These were: ‘The employee was capable of telling right from wrong,’ ‘The employee could understand negative and positive consequences of her/his behaviour,’ ‘The employee appeared to have had a sense of what is fair,’ and ‘The employee could distinguish between what is good and what is bad’. A similar scale has been used by Söderlund and Oikarinen (2021).

The perceived employee humanness was measured with three items: ‘The employee behaved very much like a human being,’ ‘The employee was humanlike,’ and ‘The employee acted like human beings typically do.’ Similar items have been used, for example, by Söderlund and Oikarinen (2021) and Söderlund (2020).

Three items from adapted from Fornell (1992) were used to measure customer satisfaction, namely ‘How dissatisfied or satisfied are you with the employee? (1 = very dissatisfied, 10 = very satisfied),’ ‘To what extent did this employee meet your expectations? (1 = not at all, 10 = totally),’ and ‘Imagine an employee that is perfect in every respect. How near or far from this ideal did you find this employee? (1 = very far from, 10 = cannot get any closer).’ In an attempt to assess the validity of this measure, and given that customer satisfaction is typically positively associated with various intentions and behavioural expectations (Altunel and Erkurt 2015; Gronholdt, Martensen, and Kristensen 2000; Curtis et al. 2011; Söderlund and Colliander 2015), these two questions were asked: ‘How likely is it that you would interact with the same employee again? (1 = very unlikely and 10 = very unlikely),’ and ‘How likely is it that you would recommend this employee to people you know? (1 = very unlikely and 10 = very unlikely).’ Since the zero-order correlations between the satisfaction measure and these two intention measures were positive and significant (r = 0.71, p < 0.01; r = 0.88, p < 0.01), it can be contended that the satisfaction measure exhibited acceptable predictive validity.

4. Analysis and results

4.1 Hypothesis testing approach

A Partial Least-Squares Structural Equation Modeling (PLS-SEM) approach was used. The main rationale behind an SEM-based approach given the set of hypotheses in the present study is that it allows for testing of all the hypotheses simultaneously (rather than assessing individual associations in isolation) and it does so by taking measurement errors into account (Sarstedt et al. 2020). The variables and their associations in Figure 1 served as the proposed model. This was assessed using the software SmartPLS 3.0 (SRMR = 0.048). Thus, all hypothesized associations (H1-H4) and the rival associations regarding the mediation aspects were assessed within the frame of the same model. The path coefficients are reported in Table 3.

Table 3. Path coefficients for the associations in the proposed model

Association	Path coefficient	t	p
Employee responsible behaviour – Employee morality (H1)	0.54	11.44	<0.01
Employee morality – Employee humanness (H2)	0.79	15.41	<0.01
Employee humanness – Customer satisfaction (H3)	0.22	3.97	<0.01
Employee morality – Customer satisfaction	0.58	9.23	<0.01
Employee responsible behaviour – Employee humanness	−0.04	0.55	0.59
Employee responsible behaviour – Customer satisfaction	0.25	6.23	<0.01

4.2 Testing the hypotheses

For the first hypothesis, the findings indicate a significant positive association between perceived responsible behaviour of the employee and perceived employee morality (b = 0.54, p < 0.01). Thus, H1 was supported. Given that such behaviours are functions of social norms, and given people’s general sensitivity to such norms, this result is in tune with previous research indicating that behaviour subject to social norms can easily elicit moral-related inferences about an agent (Haidt 2001).

Regarding the second hypothesis, the results show that there was a significant positive association between perceived employee morality and perceived employee humanness (b = 0.79, p < 0.01), which means that H2 was supported. This thus suggests that perceived employee morality was used for inferences about the humanness of the employees. The positive association between these two variables is consonant with findings from some previous empirical studies (Söderlund and Oikarinen 2021) and, in the light of morality as an advanced mind-related capability, it reflects the importance of mind perceptions as indicators of what is human (and humanlike).

With respect to the third hypothesis, there was a significant positive association between perceived employee humanness and customer satisfaction (b = 0.22, p < 0.01). This finding thus reflects the existence of an ‘what is human is good’ heuristic (Söderlund and Oikarinen 2021).

In addition to the hypothesized associations, Table 3 reveals that two other associations were significant: employee morality was positively associated with customer satisfaction (b = 0.58, p < .01), and employee responsible behaviour was positively associated with customer satisfaction (b = 0.25, p < .01).

For the mediation hypothesis (H4), and for the assessment of the rival mediation routes, the procedure advocated by Nitzl, Roldan, and Cepeda (2016) and Sarstedt et al. (2020) was followed. Mediation was assessed within the frame of a proposed SEM-based model by examining the significance of indirect paths while direct effects are controlled for. This analysis indicated a significant indirect influence (b = 0.09, t = 4.02, p < 0.01) of responsible behaviour of employees on customer satisfaction with perceived morality and perceived humanness as mediators. Thus, H4 was supported.

When it comes to the rival routes, the first route, comprising responsible employee behaviour – morality – customer satisfaction, was also significant (b = 0.43, t = 9.07, p < 0.01). However, the second route, comprising responsible employee behaviour – humanness – customer satisfaction was not significant (b = −0.007, t = 0.55, p = 0.58). Moreover, the third route, a direct influence of responsible employee behaviour on customer satisfaction, was significant (b = 0.25, p < 0.01). Taken together, then, the results indicate support for H4, but (a) the significant indirect effect in the first rival route, and the magnitude of its coefficient, and (b) the non-significant second route, suggest that employee morality served as a more potent determinant of satisfaction than employee humanness. This is perhaps not so surprising, given that attribution of moral-related characteristics to a person typically affects the overall impression of the person (Goodwin, Piazza, and Rozin 2014; Goodwin 2015). At the same time, the direct significant influence of responsible employee behaviour on customer satisfaction indicates the potential for also other (unmeasured) meditating variables. In other words, and given the significant direct effect, the mediation effects identified in the proposed model should be seen as indications of complementary mediation (Zhao, Lynch, and Chen 2010).

5. Discussion

5.1 Summary of main findings

Overall, the present study shows that the impact on customer satisfaction of responsible employee behaviour in service encounters during the Covid-19 pandemic was sequentially mediated by perceptions of employee morality and perceived employee humanness. A more parsimonious mediation model, however, comprises only employee morality as a mediator. In any case, the results indicate the important role of attributions of morality to employees in a setting in which new social norms guide interpersonal behaviours (and in which the violation of such norms can have serious health implications).

5.2 Theoretical contributions

This study is focused on the individual employee’s CSR-related activities, which represents an emerging research stream compared to macro-level studies (Aguinis and Glavas 2012). The present study contributes to the CSR literature by highlighting the role of frontline employees’ responsible behaviours as a way to enhance customer satisfaction. Thereby, the present study adds evidence that CSR-related activities can indeed contribute to customer satisfaction (Gao and Mattila 2014; Chung et al. 2015; Alhouti, Wright, and Baker 2021).

The present study also contributes to the literature on service encounters and, more specifically, to the literature dealing with employee behaviour and characteristics as antecedents to customer satisfaction (e.g., Bitner, Booms, and Tetreault 1990; Keh et al. 2013; Wu and Liang 2009). One of the main findings was that attributions of employee morality were positively associated with customer satisfaction, and this contributes to the service encounter literature in the sense that morality-related aspects have typically been covered in studies comprising service failures (e.g., in terms of perceived justice). Most service encounters, however, do not result in failures, as evidenced by the fact that customer satisfaction is generally skewed in empirical studies; that is to say, most customers are satisfied rather than dissatisfied after interacting with their suppliers (cf. Fornell 1992). In any event, the role of the attributions of employee morality as an antecedent to customer satisfaction is consonant with earlier studies (outside the service encounter area), which emphasize how the perceptions of another person’s morality play a prominent role in forming overall evaluative impressions of the person (Brambilla and Leach 2014).

Moreover, although perceived humanness had a less prominent role compared to morality in the present study (as an antecedent to customer satisfaction), perceived humanness was nevertheless positively and significantly associated with customer satisfaction. This provides further evidence of the existence of a ‘what is human is good’ heuristic in social perception settings (Söderlund and Oikarinen 2021). Its relatively low causal potency as an antecedent to satisfaction in the present study (i.e., in relation to employee morality) may be seen in the light of the pandemic in which social interactions were frequently stressed as potentially problematic from a health point of view.

Finally, the present study offers contributions to the discussions in the service marketing field about the implications of the Covid-19 pandemic (e.g., Voorhees, Fombelle, and Bone 2020; Söderlund 2020; Berry et al. 2020; Chen et al. 2021). More specifically, the present study represents one of the hitherto relatively few empirical studies of the impact the effects of pandemic-related employee behaviours on customer satisfaction.

5.3 Managerial implications

Obviously, during a pandemic, face-to-face service encounters between customers and employees can contribute to the spreading of disease. To respond to this, many firms have transformed their services by developing innovative and safe services (Berry et al. 2020). Many firms have also adopted the recommendations for interpersonal interactions (e.g., social distancing and wearing a face mask) that were issued during the pandemic. The results of the present study indicate that employee behaviour that was perceived as responsible in the light of the new social norms that emerged during the pandemic were rewarded with higher customer satisfaction. Thus, the results imply that responsible employee behaviour should be encouraged not only from the point of view of the well-being of customers and employees, but also from a business point of view, given the impact of customer satisfaction on the firm’s profitability (Fornell 1992).

5.4 Limitations and suggestion for further research

A pandemic, by definition, is characterized by its global scope. The Covid-19 pandemic, however, affected different countries in different ways. Additionally, the countries’ responses were subject to variation. The present study employed data from customers in the UK, which was hit particularly hard by the coronavirus during April 2020 in terms of a high number of deaths, unemployment, and the level of lockdown. This means that the results in the present study can have been influenced by the severe Covid-19 pandemic situation in the UK. Thus, further research, comprising data from customers in other countries, is needed for a richer understanding of the consequences of employee responsible behaviour during a pandemic. It should also be noted that the UK is not homogenous; regulations and restrictions in different parts of the UK (England, Scotland, Wales, and Northern Ireland) varied during the pandemic, and so did probably also employee behaviours, but the present study did not examine this as a discrete source of impact. Moreover, the Covid-19 pandemic influenced personnel-related aspects in several ways and it did so differently in different waves; for example, the staffing problems created by employee illness and employee work-related worries may have been particularly pronounced when the present study was carried out. This may have influenced employee behaviours – and their effects – in the present study in idiosyncratic ways (i.e., in ways that may not represent what would happen in other waves of the Covid-19 pandemic and in other pandemics).

Another limitation of this study is the use of the CIT-derived data mainly from the standardized questionnaire items. The qualitative data generated from the participants (i.e., the stories they told) could not be used to test the hypotheses in the present study, but such stories could indeed be used for other purposes – such as creating a participant-derived typology of employee behaviours with an impact on customer satisfaction in the same way as, for example, Bitner, Booms, and Tetreault (1990).

It should also be underscored that the present study focused on employee responsibility behaviours in terms of safety aspects (during a pandemic). However, there are several other aspects of social responsibility (Chung et al. 2015) that can be assessed when it comes to employee behaviours (and assessed as antecedents to customer-related outcomes such as satisfaction), so this is another task for further research.

The results (particularly the significant direct effect of responsible employee behaviour on customer satisfaction) also indicate that additional mediators than those that were used in the present study (morality and humanness) would add to our understanding of the effects of responsible employee behaviour in service encounters. For example, it has been argued that morality requires two mind-related capabilities, agency and emotionality (Gray, Young, and Waytz 2012), and they represent additional variables that may provide more details to the understanding of the process by which a person is ascribed morality.

Further research is also needed to examine moderating variables – particularly in terms of individual differences between consumers. It is likely, for example, that people’s fear of Covid-19 (Ahorsu et al. 2020) may boost the causal potency of employee morality and thus strengthen the impact of morality on customer satisfaction for those who are subject to high levels of fear. Moreover, as the Covid-19 pandemic unfolded, it became clear that not everyone was satisfied with the health-related recommendations initiated by governments (e.g., social distancing and wearing face masks) and followed by many service firms. This indicates that customers are likely to be subject to variation when it comes to the extent to which such recommendations are seen as legitimate, and one may assume that those who oppose the recommendations would be less likely to be satisfied when they encounter employees who are exhibiting norm-compliant behaviour. This latter aspect needs particular attention if the Covid-19 pandemic is seen in the same way as Žižek (2020) sees it: it is only a test for even more serious crises that are underway – such as global warming.

Acknowledgments

The authors are grateful to Richard Jones, Tanja Lepistö and Jaana Tähtinen for constructive feedback and comments to early version of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Eeva-Liisa Oikarinen

Eeva-Liisa Oikarinen is an Assistant Professor at the Department of Marketing, Management and International Business at Oulu Business School, University of Oulu, Finland. Her current research interest is human-to-human marketing, and transformative implications of human capabilities (humour, humanness, and emotions) in digitalized business interactions (recruitment advertising, service encounters and innovation management). She has published in European Journal of Marketing, Journal of Service Management, Journal of Business Research, Industrial Marketing Management, Journal of Retailing and Consumer Services and Corporate Reputation Review, among others.

Magnus Söderlund

Magnus Söderlund is Professor of Marketing and head of Center for Consumer Marketing at Stockholm School of Economics in Sweden and Senior Fellow at Hanken School of Economics, Finland. His main research interest is how consumers react to various marketing-related objects and activities.