Abstract
The current study draws from contemporary theories of morality to examine moral motives underlying service employees’ interactions with clientele. Specifically, we posit that employees who exhibit strong moral commitments to service equality (MCSE) are likely to make efforts to treat all clients equally—even when differential treatment is externally motivated by economic incentives or workplace frustrations. An analysis of self-report survey data from restaurant servers (n = 963) confirms robust associations between MCSE and various employee behaviors. Specifically, restaurant servers who report strong MCSE are less likely to report treating clients differently, whether for economic reasons (e.g., service discrimination, preferential treatment of regulars, service sweethearting, flirting) or in response to workplace frustrations (e.g., venting, disrespect toward clientele). We conclude with a discussion of the complexity of workers’ moral motivations and the need for further research in this area.
Introduction
After a long lull, there is a recent resurgence in sociological attention to morality (cf. Brueggemann 2014; Fourcade and Healy 2007; Hitlin and Vaisey 2013; Kolb 2014; Stets and Carter 2011). This renewed interest corresponds with a growing cross-disciplinary emphasis on morality in explaining human behavior (cf. Haidt 2001; Wikström et al. 2012). Broadly defined, morality refers to cultural codes that prescribe what is right/wrong, good/bad, or acceptable/unacceptable in and across social settings. Surprisingly, at first glance, morality does not appear to be a central concept in sociology of work literature (see Bolton and Laaser 2013; Darr 2011; Minkler 2004). However, examples of noneconomic (e.g., moral and social) concerns influencing marketplace behaviors are actually quite abundant in this literature. In fact, a core assumption of much work in this domain is that economic activities are embedded in sociocultural systems regulated by norms, beliefs, values, and social relations that often reinforce, mitigate, or supersede instrumentally rational decision-making processes (Darr 2011; Granovetter 1985; Massengill and Reynolds 2010; Sayer 2000).
To date, however, most sociological analyses documenting evidence of value or normative influences on workplace behaviors have relied upon ethnographic or other qualitative methods. In contrast, the current study aims to quantitatively assess whether specific moral motivations are related to frequently studied workplace behaviors among a large national sample of U.S. restaurant employees. Unlike workplace ethnographies, our design allows us to assess the prevalence and scope of morally motivated workplace action while statistically controlling for factors that might confound the relationship between internalized moral beliefs and workplace behaviors. In doing so, this study illustrates the general salience of value-rational or moral motives in predicting a broad array of actions and interactions in workplace settings. The study also adds to a mounting body of research showing that moral motives often dominate decision making even in the presence of countervailing instrumental incentives (e.g., Brewster, Brauer, and Lynn 2015; Wikström et al. 2012). Moreover, the study highlights the potential for core sociological concepts like values, norms, and moral commitments to contribute to rapidly emerging interdisciplinary research aimed at documenting and explaining systematic deviations from economically rational action (cf. Ariely 2008; Thaler 2015).
Specifically, the current study examines whether restaurant servers with strong moral commitments to service equality (MCSE; see Brewster et al. 2015) are less likely to report engaging in tip-driven service discrimination, preferential service to regulars, “service sweethearting” (i.e., giving away free food/drink without permission), and strategic flirting with clients. Institutionalized economic reward structures shape servers’ desires to enhance tip earnings, which, in turn, often directly motivate each of these behaviors in the restaurant setting. However, inspired by contemporary scholarship on morality (e.g., Bolton and Laaser 2013; Etzioni 1988; Haidt 2012; Stets and Carter 2011; Wikström et al. 2012), we argue each of these behaviors also represents a type of disparate treatment of clientele that should be incompatible with strong internalized moral commitments to treat clients equally.
Similarly, we examine whether servers with strong MCSE are less likely to engage in workplace venting and to disrespect restaurant patrons. These behaviors often reflect emotionally reactive responses to common workplace frustrations such as expecting or receiving a poor tip (see Billingsley 2016; Paules 1991). However, they also represent disparate reactions to clientele that potentially threaten the provision of quality service and, as such, should be inhibited by strong internalized moral commitments to treat clients equally.
We begin by describing how the institution of tipping in service industries encourages employees to act in strategic, economically rational ways. We then review research relevant to our predictions about moral motives governing workplace behaviors and test our predictions by analyzing survey responses from a large (n = 963), geographically diverse, national sample of restaurant servers.
Background
Tipped Employees as Economically Rational Actors
The structuring of economic incentives to motivate desired behavior—a hallmark of modern social life—is perhaps no more obvious or complete than in occupations where the practice of tipping is institutionalized as a primary form of employee compensation. In these occupations, U.S. employers are permitted to pay tipped employees below the state minimum wage—typically between $2.13 and $5.00 an hour. However, after withholding taxes, many tipped employees do not receive an employer-issued paycheck. Rather, these employees often depend almost exclusively upon tipped income, or the discretionary (and often uncertain and modest) amounts of money left by consumers after services have been rendered (see Jayaraman 2016). Given their discretionary nature, tip amounts vary substantially across service encounters due to (among other things) differences in bill sizes, servers’ behaviors, and customers’ expectations or moods (see Lynn and McCall 2016) and, as such, tipped employees can never be certain how much they will earn on any given week or month (Jayaraman 2016).
In response to these conditions of economic uncertainty, tipped employees have been shown to intentionally manipulate interactions with their clientele to increase tips and, by extension, to control the reward structure of their employment. In fact, it appears normative for research in consumer settings to implicitly or explicitly depict tipped employees as calculative, utility-maximizing agents whose interactions with customers (and managers; see Paules 1991) are primarily governed by a desire to satisfy their self-interests (e.g., tips) while minimizing costs.
Such depictions can be traced at least as far back as William Whyte’s (1948) classic treatise, Human Relations in the Restaurant Industry, in which he notes that restaurant servers “must appear to subordinate themselves to customers and at the same time learn to manipulate the people and the situation to their own advantage” (p. 120). A good server is one who “does not simply respond to her customers” but rather “acts with some skill to control their behavior” and, in doing so, is able to “get the jump on” the relationship and maximize the chances of receiving an adequate tip (Whyte 1964:39). Subsequent research has further documented the various strategic techniques tipped employees use to “get the jump on” customers or otherwise manipulate the service encounter to “wrangle tips” from their clientele (cf. Brewster and Wills 2013; Deshotels and Forsyth 2006; Gatta 2002; Paules 1991; Sallaz 2002; Thompson 2015). In essence, researchers often portray customers as “material that is processed” by the instrumentally rational tipped employee, with the goal being “the production or extraction of a finished product: the tip” (Paules 1991:34).
Tipped Employees Balance Economic and Value-rational Motives
The (over)emphasis on economically rational motives in research on employee behaviors in tipped occupations is understandable given the uncertainty inherent in tipping as a compensation structure and the routine transformation of social interactions and relationships into (uncertain) economic exchanges in these workplace settings. Nevertheless, unidimensional portrayals of tipped employees as being calculative and governed primarily by a desire to maximize their self-interest are at odds with a long history of sociological scholarship underscoring the limits of Homo Economicus models of workplace behaviors and human action more generally. For instance, in his classic work Economy and Society, Weber (1978) described instrumental rationality as only one type of motive governing human behavior. According to Weber (1978), social action is also governed by value-rational considerations, or what he describes as “a conscious belief in the value for its own sake of some ethical, aesthetic, religious, or other form of behavior, independently of its prospects of success” (p. 24). Empirical examples of marketplace behaviors that are governed by such value-oriented motives are woven throughout the sociological literature, particularly in the area of sociology of work.
Consider the call center employees studied by Sallaz (2015) who, as a result of being evaluated by the overall volume rather than quality of their calls, are encouraged by management to use deception to expedite their dealings with disgruntled customers. “To play the game of call control,” according to Sallaz (2015), “one must suppress any compunction one initially feels about deceiving callers . . . One must come to tolerate, even enjoy, lying” (p. 22). However, motivated by moral edicts, many of Sallaz’s coworkers refused to “play the game” and, as a result, often experienced added workplace difficulties (e.g., long calls with abusive customers) that would eventually lead them to quit. For instance, one devoutly religious woman named Julie refused from the beginning of her employment to deceive her callers but acknowledged that, by not lying, her job was “very difficult” to perform. Unwilling to compromise her values, Julie was unable to expedite her interactions with irate customers and eventually “grew weary of being sworn at, and resigned” (Sallaz 2015, p. 22).
Similarly, operating under severe budget cuts and inadequate resources, the victim-service workers studied by Kolb (2014) were paid modest wages, suffered from fatigue, and were under a considerable amount of strain. Yet, when the advocates and counselors were offered an opportunity to reduce their workloads and restrict the number of services that were offered to their clients, they refused. According to Kolb (2014), the workers’ refusals to attenuate their workloads stemmed from the “moral wages” that they were paid for their efforts (pp. 32–33). By providing help where it is needed, the workers were able to perceive themselves as empathetic and caring individuals who “were doing the right thing, at the right place, and at the right time.” Thus, for these workers and others employed in caring professions, doing the right thing may offer a form of symbolic compensation that offsets modest income, high stress, and sleepless nights (Kolb 2014).
Value-oriented motives also impinge upon the quintessentially economic interactions between tipped employees and their customers (see Erickson 2004; Gatta 2002; Granovetter 1985:489–90; Mulinari 2016). For instance, Erickson (2004:560–61) identified restaurant servers who choose to invest in the emotional demands of their labor by attempting to cultivate meaningful and ongoing connections with their clientele (see also Sosteric 1996). For these servers, customers are not perceived primarily as a means to procure economic ends but, rather, are people who offer opportunities to experience meaningful social interactions while at work. For instance, one of Erickson’s (2004) respondents proclaims that “a good night [at work] is a lot of familiar faces, not a lot of money” (p. 562).
Applying “Morality” Theories to Workplace Behaviors
Each of the aforementioned examples effectively illustrates the limits of unidimensional portrayals of market behavior as strictly “economic” activity and of workers as instrumentally rational actors motivated primarily by economic incentives. These empirical accounts underscore a flourishing moral economy, or a “pattern of work relationships that are rooted in social, moral, and symbolic norms and traditions” (Grint 2005:388; see also Bolton and Laaser 2013; Granovetter 1985). Furthermore, these insights are consistent with modern “morality” theories that posit the primacy of moral motives over instrumental rationality in governing most human behaviors.
Although varied, morality theories tend to underscore human decision-making processes that involve automatic, emotional, or deontological judgments, beliefs, or intuitions about what is right/wrong, worthy/unworthy, good/bad, or just/unjust (e.g., Haidt 2001; Hitlin and Vaisey 2013; see also Durkheim [1925] 2012; Smith [1759] 2009) Furthermore, morality theories often make sharp distinctions between instrumentally rational decision-making processes and normative, value, or moral commitments (e.g., Etzioni 1988; Haidt 2001; Wikström et al. 2012; cf. Weber’s instrumental vs. value rationality). For instance, moral commitments have been posited to influence decision making before the neural processes governing instrumental rationality take effect, such as by filtering out morally prohibited action alternatives from perception and instrumental deliberation altogether (Treiber 2011; Wikström et al. 2012). In other words, moral convictions might effectively define the conditions under which instrumentally rational decision-making processes operate. Other theorists go even further by describing instrumental rationality itself as a causally ineffective by-product of the human tendency to construct post hoc rationalizations for automatic, intuitive, and emotionally governed behavior (Haidt 2001).
Despite differences in specific mechanisms, causal processes, and disciplinary origins, morality theories generally share the view that internalized moral commitments and associated moral emotions are more than merely types of social or emotional costs that people link to certain behaviors (e.g., discriminatory or preferential service) and then factor into an instrumental calculus when making utilitarian decisions (cf. Etzioni 1988:67–87; Kroneberg, Heintze, and Mehkop 2010:263–64; Nagin and Pogarsky 2001). Rather, moral convictions are conceptualized as imperatives that constitute ends in themselves—that is, people decide to act (or to refrain from acting) because they believe that it is the morally right (or wrong) thing to do, regardless of whether the action (or inaction) is perceived as the most efficient way to satisfy selfish interests.
Drawing insights from morality theories (e.g., Graham et al., 2013; Wikström et al. 2012), we assume that moral sentiments vary systematically across individuals and groups and that such differences have important implications for attitudes, actions, and everyday social interactions. For instance, some individuals are expected to place greater emphasis on (i.e., perceive as more salient, experience stronger emotional valence toward, and/or express a deeper moral commitment to) specific moral values such as “equality” relative to economic concerns or other moral values (e.g., care or loyalty; see Graham et al., 2013; Haidt 2012). Likewise, those with strong moral commitments to equality are expected to treat others similarly even in the presence of institutionalized pressures to act in a discriminatory or preferential manner.
In this study, we focus on one particular internalized moral motive—the MCSE—posited to inhibit restaurant servers from engaging in disparate or unequal (e.g., discriminatory or preferential) workplace behaviors. Below, we describe several specific workplace behaviors that we expect employees with strong MCSE will be less likely to engage in than their counterparts with weaker MCSE, even though each of these unequal actions also is frequently motivated by economic concerns stemming from server’s dependence on customers’ voluntary provisions of postservice tips.
Explaining Unequal Service
Discriminatory or Preferential Service
A growing body of empirical evidence indicates that the custom of tipping encourages frontline service employees to discriminate in their service delivery by giving more of their attention to customers known (e.g., regular clientele) or thought to be good tippers at the expense of those known or thought to tip inadequately (e.g., Brewster 2015; Gatta 2002; Sallaz 2002; Sosteric 1996). However, recent research by Brewster et al. (2015) showed that restaurant servers were less likely to report discriminating against their black and Hispanic customers if they exhibited strong internalized moral commitments to treat all clients equally. Notably, this relationship was observed to be strongest when servers perceived Hispanics and blacks to be poor tippers. That is, servers reporting strong MCSE were less likely to report withholding effort from their black and Hispanic customers than those with weaker moral convictions—even when discrimination was presumably perceived as an economically rational response to the prospects of receiving inferior tips from such customers. Likewise, we expect that tipped employees who are morally committed to service equality will refrain from providing preferential treatment to regular customers and from discriminating against clients more generally, regardless of whether these servers perceive such actions to be an economically advantageous response to customers’ known or predicted tipping intentions.
Service Sweethearting
Under some conditions, the economic interests of employees and customers align to encourage alliances against management (Leidner 1993). For instance, service employees may give unauthorized free products or services to their customers with the expectation that their generosity will be reciprocated. This type of theft, or unauthorized “gift” offered at the company’s expense, has been called “service sweethearting.” This form of employee theft is particularly common in tipped occupations, where such reciprocated exchange is facilitated by the institutionalized transfer of service for tips (cf. Brady, Voorhees, and Brusco 2012; Gatta 2002).
Instrumentally rational considerations, such as the prospects of receiving better tips or the risks associated with getting caught, are certainly important for understanding why servers engage in or refrain from service sweethearting. However, a growing body of evidence indicates that moral convictions may be (at least) as important as instrumental motivations in constraining criminal and deviant behaviors, including unlawful theft and corporate crimes (e.g., Antonaccio and Tittle 2008; Kroneberg et al. 2010; Paternoster and Simpson 1996). Likewise, Brady and colleagues’ (2012) recent study found that service employees’ moral identities (“personal ethics”) were negatively associated with providing customers with services/products free of charge. Furthermore, this study found that perceived economic incentives only encouraged this type of service sweethearting among employees with weak personal ethics, thus underscoring the salience of noneconomic motives in governing economically rational actions. Similarly, we expect that restaurant servers with strong MCSE will be unlikely to give away services/products for free, even if doing so would result in greater tips.
Strategic Flirting, Workplace Venting, and Disrespect
Some service employees are motivated by economic concerns to occasionally or frequently flirt with, vent about, or openly disrespect customers. For instance, research shows that tipped employees will strategically use flirtatious communication to make their customers “feel special” and thus willing to relinquish larger gratuities (cf. Brewster and Wills 2013; Deshotels and Forsyth 2006; Ogbonna and Harris 2002). Furthermore, in industries where service providers are economically dependent on customers’ provisions of tips, employees will often emotionally react by venting about a client who has left or is expected to leave an inadequate tip (Billingsley 2016; Gatta 2002; Sallaz 2002). Likewise, while frontline service employees will rarely openly disregard organizational edicts by blatantly disrespecting their customers, it does happen and, like workplace venting, such incidents are often emotionally reactive responses to an inadequate gratuity left at the end of the service encounter (Mulinari 2016; Sallaz 2002; Thompson 2015).
Thus, like economically motivated discrimination, service sweethearting, and preferential treatment for regulars, strategic flirting is a form of unequal service that frequently reflects servers’ economic motives (cf. Lynn and McCall’s 2016 discussion of attractiveness and opposite sex servers). Likewise, servers’ reactive behaviors (e.g., venting and open disrespect) are also indirect or direct forms of unequal service that frequently reflect economic concerns. Certainly, alternative motives are also salient. For instance, servers might flirt with their customers due to (presumably noneconomic) feelings of romantic attraction (Henningsen, Braz, and Davies 2008). Furthermore, servers may talk negatively about and/or disrespect their clientele for numerous reasons aside from concerns about inadequate tips. Nevertheless, each of these actions also represents a form of unequal treatment of clientele and, as such, is inconsistent with an internalized moral commitment to equality. Thus, we expect that service employees with strong MCSE will be less likely to strategically flirt with, vent about, or openly disrespect their clients.
Data and Measures
The predictions outlined above are tested using anonymous survey data from U.S. respondents currently employed or previously employed (in the past year) as restaurant servers. The survey asked about individuals’ attitudes, opinions, experiences, and behaviors as a restaurant server. Participants were passively recruited by posting a link to the questionnaire that remained active between February 27, 2013, and March 14, 2013, on Web sites known to attract restaurant servers. 1
Web-based questionnaires are often used as a cost-effective medium to generate large geo-demographically diverse samples of individuals from occupationally specific or difficult to access target populations (cf. Kissane and Winslow 2016). Despite known limitations, Web-based surveys can produce more reliable estimates than other popular survey delivery modes (see Gosling et al. 2004; Simmons and Bobo 2015). For instance, evidence suggests that Web self-administered surveys are effective in reducing or eliminating response biases attributed to participants tendencies to respond to sensitive questions (e.g., discrimination or theft) in a socially desirable manner (Kreuter, Presser, and Tourangeau 2008; Simmons and Bobo 2015). Furthermore, Web-based surveys, like surveys in general, are more formally replicable than case studies and ethnographic analyses of workplaces.
Despite these shared advantages, this study differs in an important respect from Web surveys that are administered to opt-in Internet panels with known characteristics (e.g., Simmons and Bobo 2015). The current study relies on an unspecified sampling frame, which impedes calculation of true participation rates and assessment of population representativeness. As a result, our sample of restaurant servers may differ in meaningful ways from other U.S. servers who (1) do not have Internet access, (2) refrain from visiting the online server blogs that posted our questionnaire, and/or (3) chose not to click on the posted survey link. Nonetheless, we have no specific theoretical or empirical reasons to believe that such differences would consistently bias the correlational patterns observed between servers’ MCSE and the multiple workplace outcomes examined in this study. 2 Furthermore, our nonprobability sample of servers is demographically diverse and quite large compared with existing quantitative research on full-service restaurant servers, and the descriptive conclusions should be no less valid than those drawn from qualitative ethnographies and other commonly used nonprobability sampling designs frequently found in work and occupational research.
Of the 1,786 individuals who clicked on the survey link, 207 did not reside in the United States or had not worked as a restaurant server in the past year, and thus were omitted from the analysis. Of the remaining 1,579 respondents, an additional 616 were omitted from further analysis due to missing data on one or more of our dependent variables. 3 While less than 6 percent of the remaining 963 cases had missing values on any one of the independent or control variables, multivariate listwise deletion across these variables would result in the loss of an additional 14 percent of cases (n = 139). Thus, multiple imputation was used to estimate values for observations with missing data on each of the independent variables included in this analysis. These procedures resulted in a geographically diverse (48 states and the District of Columbia) final analytic sample of 963 individuals who were currently (n = 864) or recently (n = 99) employed as restaurant servers. Supplemental models using alternative analytic techniques for dealing with missing data, including listwise deletion and full information maximum likelihood, generated consistent results.
Dependent Variables
Tip-driven service discrimination is measured using three items tapping servers’ reports of deliberately varying service in response to predictions of customers’ tipping intentions. Respondents were asked to indicate how much they disagree or agree (0 = strong disagree to 6 = strongly agree) with two statements: “I give better service to customers I expect to be good tippers than to those I expect to be bad tippers” and “I always give my best effort when serving regardless of who or what my customers are” (reverse coded). In addition, one item asked respondents how frequently (0 = never to 4 = all of the time) they “give substandard service to customers expected to be poor tippers.” Principal components analysis indicated all three items loaded on a single factor with an eigenvalue greater than one (1.91), which accounted for 64 percent of the total item variance. Responses were standardized and summed to create an index on which higher values indicate greater levels of tip-related service discrimination (α = .71).
Preferential treatment for regular clientele is measured as servers’ reports of how often (0 = never, 4 = all the time) they “give more attention and care to regular customers known to tip well.” Less than 3 percent (n = 26) of servers reported to never provide preferential service to their regular customers, so “never” and “rarely” categories were collapsed to create an ordinal measure ranging from zero to three. Higher values represent a greater tendency to devote extra attention and care to regular clientele.
Service sweethearting is measured with one item asking respondents how often (0 = never to 4 = all the time) they give “customers free food and/or drinks in order to increase the tip they leave.” As very few respondents reported committing this type of employee deviance “all the time” (n = 18), this category was collapsed with “often” to create an ordinal variable ranging from zero to three. Higher values indicate more frequent provisions of free food or drinks in an attempt to increase tips.
Flirting with customers is measured with one item asking respondents how often (0 = never to 4 = all the time) they flirt with their clientele. As less than 5 percent of our sample reported to do so “all the time” (n = 48), these responses were combined with “often” to create an ordinal variable ranging from zero to three. Higher values indicate more frequent flirting with customers.
Workplace venting (also known as negative discourse) is measured with two items asking respondents how often (0 = never to 4 = all the time) they have “made negative ‘behind the scenes’ comments about customers” or “described customers using derogatory terms or phrases.” Servers’ responses to these items were summed to create an index (r = .54) on which higher values indicate servers more frequently vent about customers to coworkers.
Customer disrespect is measured with one item asking respondents how often (0 = never to 4 = all the time) they have “openly showed disrespect to customers.” Few respondents reported to “sometimes,” “often,” or “always” (n = 56) disrespect customers, so these categories were combined with the “rarely” category to create a binary measure comparing those who report having disrespected their customers (=1) with those who reported to have never done so (=0).
Primary Independent Variable
MCSE is measured as an index of three items that tap various dimensions of employees’ MCSE, including their perceptions of just deserts, moral emotions (i.e., guilt), and moral beliefs. Specifically, respondents were asked how much they agree (1 = strongly disagree to 7 = strongly agree) with following statements: “Bad tippers do not deserve the same quality of service as good tippers” (reverse coded); “I would feel guilty if I treat some customers better than others”; and “I think it is morally wrong to treat some customers better than others.” These items are similar to those frequently used in measures of moral beliefs or emotions in research on morality and crime (cf. Antonaccio and Tittle 2008; Brauer and Tittle 2017; Wikström et al. 2012). Principal components analysis indicated all three items loaded on a single factor with an eigenvalue greater than one (1.88) that accounted for 63 percent of the total item variance. Responses to these items were averaged to create an index (α = .70) on which higher values indicate stronger MCSE.
Control Variables
Our analyses also control for server demographic, personality, and work-related characteristics that might confound the associations between servers’ MCSE and workplace behaviors. First, models control for subject’s sex (female = 1, male = 0), race (white = 1, nonwhite = 0), and respondent’s age in years. Respondents were also asked to indicate the number of years worked as a restaurant server (0 = less than one year to 10 = 10 or more years) and whether they are currently employed as a server (1 = yes).
Personality confounds were also controlled using measures of the Big-Five Personality domains. Each domain—extraversion, neuroticism, conscientiousness, agreeableness, and openness—is measured with a 7-point semantic differential scale on which bipolar trait terms were anchored at opposing ends of each scale (cf. Shafer 1999).
Rather than being motivated by moral convictions, servers may strive to provide equally good service to all of their customers because they anticipate being rewarded with greater tips or punished for unequal service delivery. Thus, participants were asked to indicate how much they agree (1 = strongly disagree to 7 = strongly agree) with the following statement: I find that customers tip better the better the service I provide them.” In addition, respondents were asked to estimate how much effect service quality has on the tips they receive (1 = very small effect to 7 = very large effect). These two items were standardized and averaged to create an index (r = .55) measuring servers’ perceptions of rewards for providing good service. Servers’ perceptions of costs for providing poor service were measured with one item asking respondents how much they agreed (1 = strongly disagree to 7 = strongly agree) with the following statement: “I will get in trouble if I do not treat all customers the same.”
Servers who endorse stereotypes depicting certain customers as poor tippers may be instrumentally motivated to treat clients differently based on these perceptions (Brewster 2015). As such, we control for the server’s agreement with known stereotypes casting members of the following customer aggregates as poor tippers: teenagers, elderly, Christians, Jews, coupon users, foreigners, Asians, blacks, Hispanics, tables with kids, women, all-female dining parties, and women dining alone (very bad tippers = 2, below average tippers = 1, no expectation or average/above average tippers = 0; see Brewster 2015). These items were then summed to create an index on which higher values indicate greater endorsement of negative tipping stereotypes.
Models also control for participants’ service orientations or dispositions toward customer service, which have been linked to multiple employee behaviors (e.g., Kim 2011). Following Gwinner and colleagues (2005), participants were asked how much they agreed (1 = strongly disagree to 7 = strongly agree) with five statements (e.g., “I pride myself in providing courteous service”). participants’ responses to these five questions were averaged to create an index measuring service orientation (α = .84).
Finally, models also control for respondents’ tendencies to adapt their service to meet customer needs. Following Gwinner and colleagues (2005), respondents were asked how much they agreed (1 = strong disagree to 7 = strongly agree) with five statements (e.g., “I believe that each customer requires a unique approach”). Responses to these five statements were summed and averaged to create an index measuring servers’ personalized service provisions (α = .82).
Results
Table 1 provides summary statistics for the analytic sample. Table 2 presents results from multivariate ordinary least squares (OLS; Models 1 and 5), binary logistic (Model 6), and cumulative ordinal (Models 2, 3, and 4) regression models examining associations between MCSE and workplace behaviors. 4
Descriptive Statistics for Variables in Analysis Predicting Employee Behaviors (n = 963).
Note. MCSE = moral commitments to service equality.
Column entries are Pearson’s correlation coefficients representing zero-order associations with MCSE.
p < .05 (two tailed).
Selected Regression Coefficients from Models Predicting Measures of Employee Behaviors (n = 963).
Note. Continuous variables are mean-centered. Values in parentheses are standard errors. All MCSE coefficients remain statistically significant after adjusting for multiple comparisons across six dependent variables using a Bonferroni correction (α / m = .05 / 6 = .0083; p < .0001 for all six MCSE coefficients). MCSE = moral commitments to service equality; OLS = ordinary least squares.
Estimates derived from OLS regression models.
Coefficients are derived from ordinal logistic regression models. Model 2 thresholds: 0 = −2.93 (.338); 1 = −1.13 (.322); 2 = 0.431 (.320). Model 3 thresholds: 0 = −0.925 (.318); 1 = 0.535 (.317); 2 = 2.15 (.328). Model 4 thresholds: 0 = −2.08 (.318); 1 = −0.601 (.311); 2 = 1.02 (.312).
Binary logistic regression estimates.
p < .05 (two tailed).
Model 1 of Table 2 shows a strong, negative linear association between MCSE and servers’ reported tip-driven service discrimination (b = −.859, p < .0001). In Model 2, servers with strong MCSE display significantly lower cumulative log-odds of giving added attention and care to regular customers who are known to be generous tippers (b = −.660, p < .0001). Model 3 shows strong MCSE is associated with lower cumulative log-odds of engaging in service sweethearting (b = −.306, p < .0001). According to Model 4, servers with strong MCSE have significantly lower cumulative log-odds of flirting with customers (b = −.228, p < .0001). Finally, Models 5 and 6, respectively, show that servers with strong MCSE report less frequently making disparaging comments about customers (i.e., venting; b = −.334, p < .0001) and are less likely to openly disrespect restaurant clientele (b = −.278, p < .0001, odds ratio [OR] = .758).
Contextualizing Morality Associations
Results from Table 2 are highly consistent with our predictions. In fact, net of controls, servers’ internalized MCSE are associated in the predicted direction with all six workplace behaviors examined in this study. Table 3 further contextualizes the strength of these documented associations by comparing the relative portion of unique variance accounted for by MCSE with that of other variables (e.g., instrumental attitudes, service aptitude, personality). Specifically, Table 3 shows that even when the unique contributions of other key variables are coupled with conceptually related constructs, these total associations are generally smaller in magnitude—in some cases substantially so—than associations with MCSE. For instance, even when the contributions of service orientations and personalized service behaviors, or of perceived rewards of good service, costs of inequitable service, and negative tipping stereotypes, are considered jointly, these constructs typically account for less than 3 percent of the variation in each outcome and, in most cases, for substantially less (<1 percent). In contrast, moral convictions uniquely account for 23.8 percent of the observed variation in servers’ self-reported tip-driven service discrimination, 16.3 percent in preferential treatment of regular customers known to be generous tippers, 4.2 percent in service sweethearting, and 5.8 percent in servers’ self-reported disparaging comments about restaurant clientele. Overall, these results illustrate the salience of servers’ MCSE to a variety of employee behaviors in U.S. restaurants.
Unique Associations between Key Study Variables and Employee Behaviors (n = 963).
Note. Unique variance calculated by comparing R2 (or Cox and Snell R2) values in models without predictor(s) and models with all predictors. Cox and Snell R2 should not be interpreted as variance explained in the same sense as R2 in traditional OLS models; however, the measure can be interpreted as a general indicator of the association between a dependent variable and the predictors. Comparisons using traditional OLS R2 coefficients and McFadden’s R2 coefficients produced consistent substantive results. Bold indicates predictor or group of predictors accounting for the largest portion of unique variance; hence, moral convictions account for the largest portion of unique variance in five out of six models. MCSE = moral commitments to service equality; OLS = ordinary least squares.
Includes “service orientation” and “personalized service” variables.
Includes “good service rewards,” “poor service costs,” and “negative tipping stereotypes” variables.
Includes measures of subject’s race, gender, age, employment status (e.g., current server), and work experience.
Exploring Boundary Conditions
We conducted exploratory moderation analyses to identify potential boundary conditions of the observed associations between servers’ MCSE and study outcomes. Specifically, we estimated supplemental models predicting each behavioral outcome that separately included a product term between MCSE and each of the study covariates. Eight interaction coefficients (out of 90 total) reached traditional levels of statistical significance (p < .05). Four of these eight interactions were observed in models predicting discriminatory service: respondents’ age (b = .013, p = .004), serving experience (b = .036, p = .005), personalized service (b = .131, p = .005), and good service rewards (b = .10, p = .015) were found to weaken the association between MCSE and economically motivated service discrimination. In addition, the coefficients associated with the interactions between MCSE and neuroticism (b = .058, p = .043) and MCSE and service orientation (b = −.109, p = .027) were significant in models predicting preferential treatment for regular clientele. Finally, a significant interaction term between MCSE and poor service costs (b = −.045, p = .011) was observed in a model predicting negative comments. Only the interaction between MCSE and serving experience was observed across multiple outcomes. Specifically, the associations between MCSE and tip-driven service discrimination and between MCSE and the likelihood of openly disrespecting clientele (b = .038, p = .038) were strongest among less experienced servers.
Considering the number of interactions tested and the lack of robust patterns across outcomes, many (or all) of the eight observed interactions likely reflect Type I errors. This is because, when conducting a large number of tests without a priori hypotheses, the Type I error rate increases as a function of the number of tests conducted. For instance, at the conventional alpha level (.05), there is a 99 percent chance that the null hypothesis would be incorrectly rejected for at least one of the 90 interaction effects that we tested (p[at least one significant result] = 1 – (1 – .05)^90 = .99). One way to guard against an incorrect rejection of the null hypothesis in this situation is by adjusting the family-wise error rate using a Bonferroni correction (see Armstrong 2014). None of the interaction coefficients were statistically significant after using this adjustment for unplanned multiple comparisons (α / m = .05 / 90 = .0006; p > .0006 for all 90 interactions). Overall, these supplementary models show that the strong morality associations observed across a diffuse set of behavioral outcomes are largely insensitive to individual and contextual variations reflected in respondent demographic characteristics, personality domains, serving experience, service aptitude, instrumental attitudes, and restaurant settings.
Discussion and Conclusion
In this study, we contribute to the sociology of work literature by identifying employees’ MCSE as a neglected but salient motivation influencing various everyday workplace interactions. In an analysis of a large, geographically diverse occupational dataset comprised of current and recently employed restaurant servers, we find that MCSE is associated with a diffuse set of instrumental, strategic, and reactive workplace behaviors. Overall, the findings intersect with much sociological research, and ethnographic accounts of workplaces in particular (e.g., Kolb 2014; Sallaz 2015), in challenging unidimensional depictions of employees as overly calculative, utility-maximizing agents, while underscoring a pervasive (if oft-implicit) theme in this literature—that value-rational, normative, or moral motivations tend to supersede economic concerns in governing human action.
In particular, three outcome measures—discrimination, service sweethearting, and preferential treatment to regulars—explicitly tap economic motives by asking respondents whether they engage in these forms of differential treatment to increase tip earnings. Consistent with prior work showing that morality suppresses economically motivated racial/ethnic service discrimination (Brewster et al. 2015), we find that servers with strong MCSE are less likely to engage in these three distinct workplace behaviors despite the economic rewards often perceived to be attached to these types of unequal service.
In addition, our findings indicate that servers’ MCSE are inversely associated with flirting, workplace venting, and open disrespect toward clientele. Each of these behavioral outcomes frequently reflect servers’ economic concerns about garnering tips. However, unlike the discrimination, service sweethearting, and employee theft outcomes, the measures of flirting, venting, and disrespect did not explicitly specify these as economically motivated workplace behaviors. While workplace venting and open disrespect toward customers have been linked to server dissatisfaction with customers’ tipping practices, there are other factors that could precipitate such behaviors that may not conflict with servers’ MCSE. For instance, many servers, including those with strong MCSE, might respond to customer aggression or abuse by venting and/or blatantly disrespecting the transgressor (Grandey et al. 2004; Sosteric 1996). In fact, under some conditions, servers and other tipped employees may feel morally compelled to punish abusive customers who violate occupational codes that prescribe how people deserve to be treated (e.g., “street justice”; cf. Hoffmann 2008; also see Gatta 2002:67). Thus, in light of our theoretical expectation that MCSE would curb economically motivated differential service, the observed associations between servers’ MCSE, flirting, workplace venting, and open disrespect toward customers might have been even stronger if we had isolated economic motives from alternative noneconomic motivational confounds.
Taken together, this study’s findings suggest that employees’ moral commitments may overshadow economic motives and other factors to inhibit unequal service. It is particularly notable that these patterns emerge among employees working in an occupation in which the institutionalized remuneration system heightens economic insecurities, formally structures employee-client interactions as market exchanges, and encourages unequal interactions with clients. Essentially, a central finding of our study is that workers often act in ways that are incongruent with their economic self-interests due to competing moral commitments—even in these economically hyperrational contexts. Furthermore, that this finding is consistent across numerous behavioral outcomes, is largely insensitive to individual and contextual variations, and is observed among tipped restaurant employees lends credence to our conclusion that moral commitments often dominate decision making irrespective of specific behaviors or contexts.
Generally, these findings highlight the potential for sociological concepts like values, norms, and moral commitments to contribute to a rapidly emerging area of interdisciplinary social science research that aims to document and explain systematic deviations from economically rational action (cf. Ariely 2008; Thaler 2015). Certainly, economic or rational choice models offer broad, powerful explanations of human behavior. Likewise, it is relatively easy to see how each of the six workplace actions examined in this study (i.e., discrimination, preferential treatment, service sweethearting, flirting, venting, and disrespect) might reflect reward and cost considerations (e.g., concern about tips). However, our finding that moral motives govern a diffuse set of instrumental, strategic, and reactive workplace behaviors illustrates both the utility and the necessity of developing and testing coherent theoretical alternatives to economic models. Furthermore, given the relative size of the morality associations documented here compared with other plausible mechanisms (e.g., personality, attitudes or stereotypes, worker orientation, workplace context), we encourage efforts aimed at developing general theories that place value-rational or moral motives at front and center in explaining human action in context (cf. Etzioni 1988; Wikström et al. 2012).
Limitations and Cautions
Reliance on cross-sectional data and self-reports of morality, which may be confounded by behavioral projections or retrospective rationalizations, prohibits us from drawing strong causal inferences. Nonetheless, the variety of behavioral correlates of servers’ MCSE makes reverse causality interpretations and third-variable confounds unlikely. First, the various relationships we report are more parsimoniously explained as “morality effects” on servers’ workplace behaviors than as separate behavioral dispositions that all affect servers’ expressed commitment to service equality. Second, the variety of behavioral outcomes we find related to servers’ MCSE also makes it difficult to identify any one third-variable confound that could explain all those relationships. In fact, by drawing from research in business and psychology, we were able to identify and control for several factors that might logically produce spurious relationships between MCSE and employee behavioral outcomes, including servers’ orientations to customer service, propensities to deliver personalized service to clientele, perceptions of customers’ tipping behaviors, and personality traits like extraversion, conscientiousness, and agreeableness. Nevertheless, future research should explore alternative measures of morality and longitudinal designs that would allow for temporal ordering of constructs to be more clearly established.
In addition, our nonprobability sampling method limits the generalizability of our findings and specifically threatens the validity of our inferential conclusions about the statistical significance of observed associations. In other words, it remains unknown whether the correlations we observe in this sample between MCSE and workplace behaviors are likely to exist in the larger theoretical population of U.S. restaurant servers. Nonetheless, prior research indicates that the correlational structures between theoretical constructs tend to be quite similar even when Internet survey respondents differ demographically from larger populations (Alvarez, Sherman, and VanBeselaere 2003). Likewise, many previously documented empirical patterns are replicated in our sample. For example, existing studies have used a variety of methodologies to identify a host of customer types that restaurant servers tend to perceive as inadequate tippers, including customers of color (e.g., African Americans, Hispanics, and Asians), international visitors, females, teenagers, elderly adults, tables with small children, Christians/religious people, and anyone bearing coupons (see McCall and Lynn 2009). Severs’ negativity toward the tipping practices of each of these customer aggregates is replicated in our sample (see Brewster 2015). In short, we have no reason to believe that the associations documented in this sample would differ substantially in a nationally representative probability sample of servers.
Moreover, reliance on nonprobability sampling does not threaten the validity of our descriptive conclusions about the presence and strength of associations observed in this sample. Put differently, our sampling method does not pose limits to our conclusion that servers’ MCSE are associated—and quite strongly—with a diffuse set of workplace behaviors among the large, national, demographically diverse sample of restaurant servers surveyed in this study. Thus, given the relative strength of these observed morality associations (which are typically much larger than those found among other common correlates), and considering the size and geographic diversity of our sample (which constitute important strengths over much prior occupation-specific research examining related outcomes), our results certainly warrant replication efforts using nationally representative probability samples, or using data from respondents who diverge in other meaningful ways from those who participated in this study (e.g., servers of color).
Toward a More Nuanced View of Morality at Work
While the findings of this study offer support for the notion that moral motivations affect server-customer interactions and regulate specific work-related behaviors, we note that our measure of MCSE likely offers only a limited window into the vast and complex ways that moral motives influence interactions, attitudes, and behaviors in the workplace. For example, we find that servers with weak MCSE are more likely to give preferential service to regular clientele who are known to tip well. We interpret this as reflecting these servers’ greater tendencies to act on economic motivations in the workplace. In other words, we argue that some servers are less morally motivated to provide equally optimal service to all of their customers and, as such, are more likely to devote their best efforts to customers known to be good tippers. However, the story may be more complex, as moral rather than monetary motives may be governing some of these servers’ behaviors as well.
In other words, servers who report weak moral commitments may not be acting solely, or even primarily, on motivations to maximize tips when giving preferential service to regulars. Rather, they may feel these clients deserve better treatment precisely because they are regulars—clients who devote more resources to the restaurant and who have cultivated a relationship with the waitstaff (see Sosteric 1996). Therefore, we are not suggesting that servers with weak moral convictions to service equality are necessarily “less moral” than their strong-MCSE counterparts. Rather, these servers’ behaviors may simply be governed more strongly by alternative moral motivations. As Haidt and Graham (2007) noted, concerns about fairness and proportionality represent but one of (at least) five moral foundations (i.e., harm/care, fairness/reciprocity, in-group/loyalty, authority/respect, and purity/sanctity), each of which is “akin to a kind of taste bud, producing affective reactions of liking or disliking when certain patterns are perceived in the social world” (p. 104).
Thus, rather than merely reflecting economic motives, servers’ preferential treatment to regulars may also reflect strong moral commitments to favor in-group members, or a relatively greater internalized commitment to “loyalty” compared with “fairness.” Furthermore, interpersonal variability in the notion of “fairness” itself may result in both types of servers acting “morally” in diametrically opposite ways. For instance, most servers with strong MCSE may give the same level of service to their regulars as they give to all of their clientele because it is the “fair” thing to do—they believe strongly that all customers are deserving of the same great dining experience. Others, however, may give preferential service to regular clientele who are known to tip well because it is only “fair” that these customers be given the “extra” services that they (will) pay for.
Additional studies are thus clearly needed to elucidate the complex and nuanced ways that moral motivations influence customer-employee relationships within the restaurant context and across other tipped and nontipped professions. These studies should include alternative, improved, and more diverse measures of moral motivations (e.g., see Haidt and Graham 2007). Studies that sufficiently take into account diverse moral motivations that may be relevant to tipped employees’ behaviors and interactions may find that economic considerations are even less salient than commonly assumed in existing research. Moreover, research assessing the genesis, sustenance, and downstream effects of employees’ MCSE is needed. As a point of departure, we encourage studies that take into account the workplace culture within which employees’ moral convictions become (or fail to become) manifest in their attitudes and interactions with customers, coworkers, and managers. Interpersonal variability in the emergence, strength, and effectiveness of servers’ MCSE might reflect, in part, structural and cultural differences in the relative emphasis placed on profit maximization and broader concerns about justice across workplaces that we were not able to account for in the current study (cf. Crowley and Hodson 2014; Erickson 2004; Sosteric 1996).
We also encourage research assessing the robustness of our findings across professions. The most obvious extensions of this work include examinations of whether the general patterns observed in this study are found among employees in other tipped professions (e.g., taxi drivers, hair stylists, etc.). However, MCSE are also likely to influence the attitudes and actions of nontipped service employees who have an instrumental stake in the outcomes of their interactions with clientele. For instance, according to Pierce (1995:7; see also Leidner 1993:154), male lawyers engage in “strategic friendliness” by using charm and/or flattery to manipulate clients, judges, and juries in an attempt to enhance their chances of successful litigations. Furthermore, it is possible that equality convictions and other moral motives (e.g., Haidt and Graham 2007) are even more salient in caring professions such as nursing, social work, counseling, victim advocacy, and homecare where economic motives are less paramount, where interaction-based economic incentives are less obvious, and where the nature of work itself is thought to be altruistic and morally motivated (Kalleberg et al. 2006; Kolb 2014). Overall, then, we encourage future sociological research broadly focused on identifying how moral motivations operate in tandem with economic and other instrumental considerations to influence the attitudes, behaviors, and interactions of employees, employers, and clients across a variety of workplace domains.
Footnotes
Acknowledgements
The authors would like to thank three anonymous referees for the extremely helpful and insightful feedback that they provided on earlier versions of this article.
Authors’ Note
A version of this article was presented at the 2015 Annual Meeting of the Midwest Sociological Society in Kansas City, Missouri.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The writing of this article was supported by the Humanities Center at Wayne State University.