Does it pay off to smile even it is not authentic? Customers’ involvement and the effectiveness of authentic emotional displays

Research has shown that while the authenticity of positive emotional displays plays an important role in service encounters, it has not yet demonstrated a universally positive effect on customers’ behavior. This study, which is the first to present customer involvement as a contextual factor that influences customers’ processing, provides a deeper understanding of the effectiveness of the authenticity of emotional displays. The model is based on expectation disconfirmation theory and emotional contagion theory and is validated in a field experiment and two laboratory experiments that use video stimuli with actors in real-world contexts. The results show that even inauthentic displays can meet customers’ expectations depending on their involvement and that high-involvement customers adapt to employees’ authentic emotions more strongly than low-involvement customers do. In summary, the presented model strengthens the understanding of the role of authentic displays and provides an approach to improve the effectiveness of emotional labor strategies.

affect the outcomes of customer-employee interactions is the use of emotional labor. Hochschild (2003) describes emotional labor as control over emotions to generate a desired facial and bodily display. Therefore, employees are expected to exhibit positive emotions when dealing with customers (i.e., smiling). Two techniques are available for generating desired emotional displays. Deep acting enables an individual to experience a desired emotion by placing himself or herself in a situation that elicits the actual emotion (i.e., authentic displays). Surface acting is a technique in which the desired emotion is displayed but not experienced (i.e., inauthentic displays) (Grandey 2000).
However, research on the effects of emotional labor strategies has led to inconsistent findings (Chi et al. 2011). Thus, discussion has arisen regarding the effects of authentic and inauthentic emotional labor strategies (Houston III et al. 2018). While some studies have attempted to explain these inconsistencies based on the characteristics of employees and service encounters (Chi et al. 2011), there is a lack of research that considers customer-related factors (Lechner and Mathmann 2020).
This study contributes to research by considering customer expectations regarding emotional facial displays and involvement. As suggested by Golder et al. (2012), customer expectations are critical to both the quality experience and the quality evaluation process. Our study shows that customer involvement influences expectations of employees' emotional displays. In addition to customers' cognitive reaction to displayed emotions (van Kleef 2014), customers also experience an affective reaction. We show that affective processing is also contingent on involvement. Moreover, we demonstrate that short-term changes in involvement can influence customer processing. Finally, we classify the effects of emotional labor into further emotional displays.
Practitioners can benefit from our findings in several ways. We show when authenticity is important in service encounters and how emotional labor strategies' effectiveness can be increased. Furthermore, we incorporate our findings into the stages of the customer decision process and develop recommendations for service managers to help them tailor emotional labor strategies to low-vs. high-involvement customers. Moreover, we present recommendations to companies that provide service training programs for service employees.
2 Conceptual framework 2.1 Expectations regarding authentic emotional displays Customer expectations are crucial for quality evaluations because they serve as a personal benchmark (Golder et al. 2012). Researchers found that black customers were less bothered by inauthentic smiles because they had lower service expectations due to a "history of differential service experiences" (Houston III et al. 2018, p. 85). These authors focused on what customers had previously experienced; however, expectations are not only based on historical encounters. Helson (1959) traces expectations to contextual factors as well as to the psychological and physiological characteristics of individuals. Customer involvement is a concept that is related to these factors because it reflects an increased motivational state due to highly perceived personal relevance (Celsi and Olson 1988). In accordance with Zaichkowsky (1985), this motivation derives from three dimensions: personal relevance is related to personal values and needs, physical relevance is related to the characteristics of an object, and situational relevance refers to a temporary opportunity.
To date, there is limited understanding of the effect of involvement on customers' expectations of employee behavior. Only one study by Varki and Wong (2003) revealed that high-involvement customers expect to be treated more fairly by sales staff because they expect greater relational efforts.
We assume that involvement also affects customers' expectations of employees' authenticity. This assumption is based on two considerations. First, businesses and employees attempt to provide appropriate employee behavior that is perceived as an effort to meet customers' needs (Hennig-Thurau et al. 2006). Therefore, we equate the authenticity of emotional displays with the degree of emotional effort by salespersons that is increasingly desired by high-involvement customers (Varki and Wong 2003). Second, higher involvement should increase perceived risk and encourage customers to process information more deeply (Celsi and Olson 1988). At the same time, as Houston III et al. (2018) have shown, authentic emotional displays should result in higher trust. Thus, we conclude that high-involvement customers oppose inauthentic service personnel behavior more than low-involvement customers do. Consequently, customers' expectations regarding the authenticity of employees' emotional displays should depend on their involvement. H1. The higher customers' involvement is, the more customers expect authentic emotional displays.
To explore the effect of unfavorable or favorable deviations from customers' expectations, we rely on the concept of loss aversion (Kahneman and Tversky 1979). We assume that customers' expectations regarding employees' emotional displays depend on customers' involvement (H1). Because quality disconfirmation depends on customers' expectations as a reference level of service performance (Golder et al. 2012), it follows that the negative effect of inauthentic emotional displays should be stronger for high-involvement customers. By contrast, the positive effect of authentic emotional displays should be weaker for low-involvement customers. In summary, involvement should strengthen the effect of authentic emotional displays on customers' evaluation. Because expectation confirmation is a vital part of quality evaluation and thus influences customer behavior intention (Golder et al. 2012;Houston III et al. 2018), we expect that the interaction effect of the authenticity of emotional displays and involvement influences loyalty intention through cognitive processing.
H2a. Involvement moderates the positive effect of authentic emotional displays on exceeded expectations such that higher involvement strengthens this effect. H2b. Exceeded expectations mediate the interaction effect of authentic emotional displays and involvement on loyalty intention.

Affective reaction to authentic emotional displays
Emotional contagion theory explains the effect of observed facial expressions on felt emotions based on the idea that people imitate the behavior observed in other people.
The imitated behavior then induces the corresponding emotions (Hatfield et al. 1993). According to facial feedback theory, the facial mimicry of observed authentic expressions leads to stronger emotions, unlike the facial mimicry of inauthentic expressions (Soussignan 2002). Hennig-Thurau et al. (2006) show that these effects are valid in the context of service encounters. We hypothesize that involvement affects the transfer of emotions for the following reasons. First, involvement increases information processing (Zaichkowsky 1994). Emotional contagion assumes social comparison processes wherein the observer uses emotions as social information (Hennig-Thurau et al. 2006). Therefore, it can be assumed that involvement-induced information processing enhances the positive effect of authentic emotional displays. Thus, when the observer puts himself or herself in the mood of the observed person, he or she experiences the corresponding emotions more intensely. Second, involvement heightens felt emotions (Mano and Oliver 1993). Since emotional contagion is characterized by the imitation of observed behavior and the resulting emotions, it can be assumed that involvement-induced emotional experience strengthens the effect of authentic emotional displays. Consequently, intensified imitation should result in stronger emotionalization. Hence, we assume that involvement enhances the positive effect of authentic emotions on the transfer of emotions. Because customers' positive emotions increase their positive evaluation and loyalty intention (Golder et al. 2012;Hennig-Thurau et al. 2006), we hypothesize that the interaction effect of the authenticity of emotional displays and involvement influences loyalty intention through affective processing.
H3a. Involvement moderates the positive effect of authentic emotional displays on positive affect such that higher involvement strengthens this effect. H3b. Positive affect mediates the interaction effect of authentic emotional displays and involvement on loyalty intention.

Study 1
In this study, we analyzed how involvement affects customers' expectations regarding the authenticity of displays of positive emotions (H1).

Method
In a field experiment, we examined what facial expressions customers expect depending on their involvement. We surveyed 139 customers (M age = 38.4, SD = 18.2; 54.7% female) before a service interaction in 25 stores across a variety of service contexts. We surveyed participants in front of the stores they were about to enter. First, we described positive emotional facial displays of employees (see Appendix 1). Therefore, we manipulated the description of emotional displays (i.e., inauthentic vs. authentic). The participants were randomly assigned to one of the two descriptions. The participants indicated their expectations regarding the positive emotional displays described. Finally, they rated their involvement. Complete measure details are provided in the Web Appendix.

Results
To test H1, we conducted a moderation analysis using PROCESS (model 1, 5,000 bootstrapping samples), in which the description of positive emotional displays (i.e.,  inauthentic or authentic) was the independent variable, expectation was the dependent variable, and customer involvement was the moderator. The results showed a significant interaction effect (β = .53, SE = .14, p < .01). We depict this interaction in Fig. 2a and analyzed the effect using the Johnson-Neyman technique (Fig. 2b). In line with H1, the findings show that the expectation of authentic emotional displays increases with involvement. More specifically, low-involvement customers (< 1.89) expect inauthentic rather than authentic emotional displays, while high-involvement customers (> 4.27) expect authentic rather than inauthentic emotional displays.

Study 2
In Study 2, we tested our hypotheses that the authenticity of emotional displays affects both cognitive (H2) and affective (H3) processing contingent on involvement.

Method
By using a 2 (inauthentic vs. authentic positive emotional displays) × 2 (low vs. high involvement) video-stimulated online experiment, we ensured internal validity. We recruited a female actor and recorded two videos that showed a service encounter in a supermarket. The filming took place in a real supermarket, and the videos represented a typical checkout situation. Each showed the scene from a first-person perspective and took the same amount of time (approximately 30 s; see Web Appendix). All dialogue was the same (see Appendix 2). In line with prior research (Lechner and Mathmann 2020), only the facial expression of the employee was manipulated (i.e., inauthentic vs. authentic), and body characteristics were held constant. For this manipulation, we instructed the actor to use surface acting or deep acting (Wang et al. 2017) and ensured that marks of authenticity matched (i.e., inauthentic (authentic): without (with) wrinkles around eyes; Houston III et al. 2018).
To manipulate involvement, we used two different scenarios. To account for the multidimensionality of involvement (Zaichkowsky 1985), we manipulated the reason for purchase (i.e., personal relevance), the products (i.e., physical relevance), and the relevance of the product to the reason for purchase (i.e., situational relevance) (see Appendix 3). The results of manipulation checks for perceived facial displays of the employee and participants' involvement are provided in the Web Appendix.

Measures
The following explanations are presented in the same order as in the experiment. In Study 1, participants reported their involvement using a scale developed by De Wulf et al. (2001). However, this scale does not fully capture the affective and cognitive involvement components. To address this issue, we used the Personal Involvement Inventory (α = .94) developed by Zaichkowsky (1994). To confirm our authenticity manipulation, the participants responded to three items ("The smile of the employee was natural," "The smile of the employee was genuine," and "The smile of the employee was authentic"; 1 = strongly disagree, 7 = strongly agree; α = .93; Houston III et al. 2018). To measure exceeded expectations, we used two items ("The service encounter was…" and "The service employee was…"; 1 = much poorer than expected, 7 = much better than expected; α = .78; Oliver and Burke 1999). To measure customers' positive affect, we used the original valence dimension of the Self-Assessment Manikin Scale (1 = negative, 5 = positive; Kulczynski et al. 2016). Finally, to measure loyalty intention, we used three items ("I will gladly visit this retailer in the future," "I will say positive things about this retailer," and "I can recommend this retailer without hesitation"; 1 = strongly disagree, 7 = strongly agree; α = .9; Hennig-Thurau et al. 2006).

Results
To test our hypotheses, we conducted a regression-based analysis using bootstrap resampling in AMOS 25 that considers all variables simultaneously. The results revealed significant interaction effects of emotional displays and scenarios on exceeded expectations (b = .52, SE = .16, p < .01) and positive affect (b = .51, SE = .17, p < .01). In accordance with H2a, simple slope analysis showed that in the high-involvement condition, inauthentic emotional displays had a more negative effect on exceeded expectations (b = −1.04, SE = .11, p < .01) than the positive effect of authentic emotional displays in the low-involvement condition (b = .52, SE = .11, p < .01). Furthermore, and in line with H3a, simple slope analysis showed that authentic emotional displays had a stronger effect on positive affect in the high-involvement condition (b = 1.06, SE = .12, p < .01) than in the low-involvement condition (b = .55, SE = .12, p < .01).
To test H2b, we calculated the indirect effects of emotional displays on loyalty intention via exceeded expectations. The results showed a stronger indirect effect in the high-involvement condition (b = .49, SE = .09, p < .01) than in the low-involvement condition (b = .24, SE = .06, p < .01). Calculating the index of moderated mediation supported H2b (b = .24, SE = .08, p < .01). To test H3b, we calculated the indirect effects of emotional displays on loyalty intention via positive affect. The results showed a stronger indirect effect in the high-involvement condition (b = .38, SE = .07, p < .01) than in the low-involvement condition (b = .2, SE = .05, p < .01). Calculating the index of moderated mediation supported H3b (b = .18, SE = .07, p < .01).

Study 3
In Study 3, we validated the results from Study 2 in a different scenario. In addition, we considered a short-term change in involvement. While customer involvement was determined before the service encounter in Study 2, in Study 3, we investigated whether the change in involvement during the interaction had an effect on customer processing. Furthermore, we broadened the set of emotions considered. Since research shows that emotional labor can lead to emotional exhaustion (Gaucher and Chebat 2019), we included frustrated negative emotional displays. As a baseline, we also considered a neutral expression.

Method
We used a 4 (inauthentic positive vs. authentic positive vs. frustrated negative vs. neutral emotional displays) × 2 (constant vs. increased involvement) video-stimulated online experiment. The methodological procedure was the same as in Study 2. This time, the videos (approximately 35 s; see Web Appendix) showed a bank service encounter in which the customer wanted to invest a certain amount of money and was served by a male employee (see Appendix 4).
To manipulate emotional expression, the actor again applied corresponding emotional labor techniques (see Study 2). To manipulate the participants' involvement, we followed Shao et al. (2004) and used two different investment options (see Appendix 5). Initially, the customer is interested in a low-involvement product. During the service encounter, he or she either maintains his or her original decision (i.e., constant involvement) or chooses a high-involvement product (i.e., increased involvement). The results of manipulation checks for the perceived emotional expression of the employee and participants' involvement are provided in the Web Appendix.

Measures
To confirm our manipulation of an increase in involvement, we used the same 10-item scale as in Study 2. We measured once based on the customer's original intention (α = .94) and once based on the customer's final decision (α = .96). To calculate the increase in involvement, we subtracted the second value from the first (Hennig-Thurau et al. 2006). To confirm our authenticity manipulation, the participants responded to the same items as in Study 2 (α = .9). To measure emotional exhaustion, we used 2 items ("The employee felt emotionally drained by his job," "The employee felt frustrated by his job"; 1 = strongly disagree, 7 = strongly agree; α = .86; Gaucher and Chebat 2019). To measure customers' positive affect, we used 3 items from Hennig-Thurau et al. (2006) (α = .84). Finally, to measure exceeded expectations (α = .94) and loyalty intention (α = .97), we used the same scales as in Study 2.

Results
To validate our findings from Study 2, we conducted a regression-based analysis using bootstrap resampling in AMOS 25, including the positive emotional displays condition. The results revealed positive interaction effects of emotional displays and scenarios on exceeded expectations (b = .46, SE = .16, p < .01) and positive affect (b = .46, SE = .23, p < .05), confirming H2a and H3a.
To validate H2b, we calculated the indirect effects of emotional displays on loyalty intention via exceeded expectations. The results showed a significant indirect effect in the increased-involvement condition (b = .34, SE = .08, p < .01) and a non-significant effect in the constant-involvement condition (p > .1). Calculating the index of moderated mediation confirmed H2b (b = .25, SE = .09, p < .05). To test H3b, we calculated the indirect effects of emotional displays on loyalty intention via positive affect. The results showed a significant indirect effect in the increased-involvement condition (b = .41, SE = .12, p < .01) and a non-significant effect in the constant-involvement condition (p > .1). Calculating the index of moderated mediation confirmed H3b (b = .29, SE = .15, p < .05).
A four-by-two ANOVA (emotional displays condition by scenario condition) on loyalty intention revealed a significant group difference between positive emotional displays (M inauthentic positive = 3.36, M authentic positive = 3.87) and frustrated negative and neutral emotional displays (M frustrated negative = 2, M neutral = 1.81) of 1.71 (SE = .09, p < .01). However, there was neither a significant main effect of emotional displays (p > .1) nor a significant interaction effect of emotional displays and scenarios (p > .1) when comparing frustrated negative and neutral emotional displays. In addition to classifying the effects of emotional labor into further emotional displays, the results show that H2b and H3b cannot be transferred to negative and neutral emotional displays.

Discussion
Our research provides a deeper understanding of the effects of emotional labor strategies on customer loyalty intention. We find that involvement influences customers' expectations of the authenticity of employees' emotional displays. More precisely, high-involvement (low-involvement) customers expect authentic (inauthentic) emotional displays. Thus, we demonstrate from the perspective of cognitive processing the importance of expectation (dis)confirmation with regard to the authenticity of emotional displays. We show that even inauthentic emotional displays can meet customers' expectations depending on customers' involvement. Thus, our results show that loyalty intention depends more strongly on employees' authentic behavior for high-involvement customers than for low-involvement customers. From the perspective of affective processing, we deepen the understanding by identifying the reinforcing effect of involvement. Finally, we show that the identified relationships apply not only to service encounters in which involvement is determined in advance but also to dynamic encounters in which involvement increases during the interaction.
The findings of this study provide implications for service management. At first glance, it appears that companies would be well advised to train their employees in deep acting techniques (Hennig-Thurau et al. 2006). However, when opportunity costs are taken into account, the question arises of whether this recommendation is always applicable. The present research provides guidelines for (1) when the authenticity of employee behavior should be enhanced and (2) how the effectiveness of authentic emotional displays can be increased.
We provide three different approaches to the first question. First, managerial action should depend on the context of the service. Thus, managers should pay more attention to the use of deep acting in situations with rather high levels of involvement (e.g., high prices, high risk). Second, involvement can also vary depending on the personal situation of the customer (Zaichkowsky 1994). To identify high-involvement customers, employees could offer customers additional background knowledge or highlight additional services or products. Customers who respond to this strategy are likely to be more involved, and deep acting should be used carefully. Third, managers could observe the customer decision-making process. Research has shown that involvement in certain stages (i.e., need recognition, information search, and evaluation) has different origins and consequences (Puccinelli et al. 2009). Depending on the stage, employees can ask specific questions to identify the level of customer involvement. In the need recognition stage, involvement depends on the customer's goals (Celsi and Olson 1988). Customers who are seeking a superior service or product are usually highly involved (classifying question: "Are you looking for an ideal solution or a quick one?"). In the information search stage, involvement determines the extent and depth of customers' information search (Zaichkowsky 1985) (classifying question: "Do you know of the various available opportunities?"). Finally, customers make a judgment during the evaluation stage based on sufficient certainty (Suh and Yi 2006). Therefore, high-involvement customers will have several arguments that justify their decision (classifying question: "How did you come to this decision?").
Training companies that teach deep acting techniques would be well advised to pay attention to endurance training because our findings show that the level of authenticity that is sufficient at the beginning of a service encounter may not be sufficient at the end of the encounter. When involvement increases during the encounter, employees must be able to respond with positive emotional displays that are more authentic. Conversely, employees could increase the effect of authentic emotional displays by increasing customer involvement (e.g., using in-store demonstration; Grewal et al. 2009).
Given the potential risks of surface acting (i.e., emotional exhaustion), our results show that in the case of frustrated negative emotional displays, loyalty intentions are at the same level as in the case of neutral emotional displays regardless of customer involvement. This finding supports the position of continuing to smile even if it is not authentic.
This study also yields future research avenues. Although we considered several different service contexts in Study 1, we used only two service contexts in Study 2 and Study 3. Therefore, it may be premature to generalize the results to all service contexts, and future research should validate our findings in other contexts. For instance, we expect different effects in the area of luxury providers, whose customers have a strong desire to belong to a social group (Amaral and Loken 2016). Luxury brands often promote a "superior mood" in high-end retail stores to differentiate themselves from other brands. Ward and Dahl (2014) demonstrated that the rejection of customers in such a context increases customers' desire to belong to the group to reduce the feeling of exclusion. Thus, inauthentic positive emotional displays could also induce a perception of rejection and could strengthen the desire to belong to the exclusive group. Furthermore, our results show that low-involvement customers expect inauthentic rather than authentic emotional displays. Based on their own low involvement, customers may assume the same for employee and therefore consider employees' inauthentic positive behavior more likely than authentic behavior. This involves "a psychological projection whereby one's own feelings or actions are attributed to others" (Wood and Essien-Wood 2012, p. 985). Future research should elaborate this process more broadly by considering when and with regard to which emotional states projection takes place and how it influences the relationship between customers and employees.
Finally, the present study did not consider the effects of authenticity of emotions displayed in repeated service encounters. In these cases, the relationship between customers and employees could continuously improve. Chi and Chen (2019) demonstrated that in repeated service interactions, relationship strength enhances the positive effect of authentic emotional displays because customers expect higher emotional effort. Therefore, in repeated service encounters, even low-involvement customers could expect authentic employee behavior due to a stronger relationship. Future research should examine the effects of emotional labor strategies in repeated service encounters by considering customer involvement and relationship strength.
Appendix 1. Description of positive emotional displays (Study 1)

Condition: inauthentic/authentic
A not genuine/genuine smile, which does not/does occur naturally and expresses not true/true feelings. How likely do you think it is that the employee will serve you with a smile that does not/does occur naturally and is not genuine/genuine? Appendix 2. Script of service encounter (Study 2)

Condition: inauthentic/authentic
Checkout situation at a supermarket: Employee to customer with an inauthentic smile/ authentic smile: "Hello." Customer replies [mute], "Hello." Employee scans products. Price appears. Employee to customer with an inauthentic smile/authentic smile: "That will be 7.57 EUR." Customer pays cash. Employee provides customer with change and receipt. Employee to customer with an inauthentic smile/authentic smile: "Have a nice day." Customer replies [mute]: "Thank you, you too."

Condition: low involvement
On the way home, you decide to go to a supermarket to make some minor purchases. You do not necessarily need these products but avail yourself of the opportunity to buy them quickly. It is not a serious problem if these few minor products are not available in this supermarket. You can go to another supermarket in the next few days.

Condition: high involvement
This evening, you have invited guests who are very significant to you to your house. Therefore, it is especially important that you cook a delicious meal. You have already bought almost all the ingredients, except for a few important ingredients. The meal will not be successful without the missing ingredients. Therefore, you go to a supermarket again to buy the remaining ingredients.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.