Perceptions of intergroup tipping differences, discriminatory service, and tip earnings among restaurant servers

International Journal of Hospitality Management 46 (2015) 15–25

Contents lists available at ScienceDirect

International Journal of Hospitality Management journal homepage: www.elsevier.com/locate/ijhosman

Perceptions of intergroup tipping differences, discriminatory service, and tip earnings among restaurant servers Zachary W. Brewster ∗ Wayne State University, 2272 Faculty Administration Building, Detroit, MI 48202, United States

a r t i c l e

i n f o

Keywords: Tipping Restaurant service Discrimination

a b s t r a c t In addition to encouraging good service, there are concerns that the custom of tipping may also motivate restaurant servers to discriminate in their service delivery by giving relatively less attention to members of groups thought to be poor tippers. Surprisingly, however, there is a notable scarcity of studies that have directly assessed the relationship between the custom of tipping and discriminatory restaurant service. Further, to the degree that servers discriminate in response to their predictions of customers’ tipping intentions they are presumed do so in order to maximize their tip earnings and yet there have been limited attempts to directly assess the efficacy of discrimination as a profit maximizing strategy. In response, this study analyzes data (n = 954) from a large online survey of current and former restaurant servers to explore the relationships between perceptions of intergroup tipping differences, discriminatory service, and tip earnings. Results indicate that servers’ who harbor negative attitudes about customer types stereotypically thought to be poor tippers are also more likely to report that they discriminate in their service delivery. Harboring stereotypic attitudes towards customer aggregates thought to be especially good tippers, however, was not found to be predictive of service discrimination. Further, in contrast to popular beliefs this study’s results suggest that discriminatory service in response to servers’ a priori predictions of customers tipping intentions may not be an effective way for servers to enhance their tip earnings. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction The custom of leaving gratuities, or tips, following many service encounters is quite pervasive in the United States and beyond (e.g., bellmen, taxi drivers, barbers, hairstylist, food delivery drivers, etc. see Lynn et al., 1993). Tipping is a particularly salient characteristic of the full-service restaurant industry. In fact, in the US alone there are nearly 3 million waiters, waitresses, and bartenders who are economically dependent on the estimated $40 billion in tips that consumers voluntarily relinquish to food service workers each year (Azar, 2009). By delegating control responsibilities to consumers who leave tips in accordance with the quality of service they received, tipping has been shown in this literature to be a cost effective way for restaurant operators to motivate servers to provide quality service (see Lynn and Withiam, 2008 for a review of these advantages; also see Azar, 2003, 2004, 2009, 2011; Kwortnik et al., 2009; Lynn and Sturman, 2010; Ogbonna and Harris, 2002).

∗ Tel.: +1 313 577 2930; fax: +1 313 577 2735. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ijhm.2015.01.001 0278-4319/© 2015 Elsevier Ltd. All rights reserved.

However, according to conventional wisdom dating back to at least the 19th century, in addition to encouraging good service, tipping may also function to motivate service providers to discriminate in their service delivery by giving relatively less attention to customers thought to be poor tippers (cf. Brewster, 2013; Margalioth, 2006; Wang, 2014). In fact, concerns about discriminatory service stemming from the custom of tipping continue to be expressed and cited as a primary reason why the custom should be abolished and replaced with inclusive pricing or automatic service charges (e.g., Palmer, 2013; Peterson, 2013; Wang, 2014; Wells, 2013). Despite such widespread and enduring concerns there is a notable scarcity of studies that have directly assessed the relationship between servers’ perceptions tipping differences across social aggregates and their tendencies to discriminate in their service delivery. Further, those existing studies that have assessed this relationship (see Brewster, 2012a, 2013) have been plagued with methodological shortcomings. Additionally, to the degree that servers do extend differential service according to their predictions of customers’ tipping intentions they are presumed do so in order to maximize their tip earnings. Yet, attempts to assess the nature of the alleged relationship between discriminatory service and tip earnings have been

16

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

limited. Thus, it is unknown if servers do in fact garner greater tips as a result of providing service that is predicated on their a priori predictions of customers’ tipping intentions. If this is the case, discriminatory service delivery could rightfully be considered, at least in part, an economically rational server behavior and could logically be curtailed by replacing voluntary tipping with inclusive pricing or by adding automatic service charges to customers’ bills. Given the uncertainties inherent in predicting the tipping behaviors of individual customers, however, it is equally plausible that discriminatory service is either economically ineffective or even counterproductive (cf., Maynard and Mupandawana, 2009). If either of these latter scenarios is true, restaurant proprietors who decide to implement an alternative remuneration system in lieu of voluntary tipping as a way to encourage servers’ to provide equitable service may effectively be throwing the baby out with the bath water. That is, by abolishing tipping in their establishments they may be forfeiting all of the documented advantages of tipping (see Lynn and Withiam, 2008) in order to reduce an undesirable employee behavior that could otherwise be curtailed via designing and implementing training initiatives to teach servers that discriminatory service is not economically effective. This study advances the literature on the causes and consequences of service discrimination in two ways. First, I attempt to replicate and extend extant findings linking servers’ perceptions of tipping differences across social aggregates and discriminatory restaurant service (cf. Brewster, 2013; Brewster et al., in press). Second, this study aims to explore the relationship between service discrimination and tip earnings in an attempt to assess the validity of servers’ claims that discriminatory service is an economically effective response to perceived variability in customers’ tipping behaviors. 2. Background and predictions 2.1. Perceptions of intergroup tipping practices and service discrimination Server anecdotes (and from other tip dependent service employees), past and present, portraying some groups of customers to be more generous in their tipping practices than others are common. Only recently, however, have scholarly efforts been taken to empirically identify those customer characteristics that are most often associated by restaurant servers with good or bad tipping behaviors. One of the most robust findings to emerge from these studies is that African American, Hispanic, and Asian customers tend to be perceived by servers to be poor tippers relative to their white counterparts (Brewster et al., in press; Brewster and Rusche, 2012; McCall and Lynn, 2009; Noll and Arnold, 2004). Among other customer types that have been shown to be perceived by servers to be comparatively inadequate tippers are foreigners, women, teenagers, elderly adults, tables with small children, Christians/religious people, and anyone bearing coupons (Harris, 1995; Lynn, in press; Lynn and Katz, 2013; Maynard and Mupandawana, 2009; McCall and Lynn, 2009)1 .

1 Studies assessing the validity of some of these perceived intergroup differences in tipping (e.g., religious customers, women, coupon users) have produced mixed results (Boyes et al., 2004; Grossman and Parrett, 2011; Lynn and Thomas-Haysbert, 2003; Lynn, 2013; Lynn and Katz, 2013; Lynn and Brewster, 2015; Maynard and Mupandawana, 2009; Parrett, 2006). Nevertheless, the totality of existing evidence indicates that servers’ perceptions of intergroup tipping differences are not groundless (Maynard and Mupandawana, 2009). Studies have found, for instance, that some of the customer types that servers perceive to be inadequate tippers (e.g., Blacks, Hispanics, Asians, lower income) may on average tip their servers statistically less than their counterparts, in part, as a result of being comparatively less familiar with U.S. tipping norms (cf., Lynn, 2004, 2006a, 2006b, 2011, 2013; Lynn and Brewster, 2015).

Thus, according to conventional logic any customer who is a member of these social aggregates are at risk of receiving less of their servers’ attention vis-à-vis their counterparts who are members of aggregates perceived to be better tippers. While direct tests of this relationship are rare, findings from three recent studies do support the idea that service quality varies systematically by servers’ perceptions of intergroup tipping differences. First, in an analysis of data derived from a small community (n = 175) survey of restaurant servers Brewster (2012a) found that positivity toward the tipping practices of Blacks was, as predicted, inversely related to servers’ self-reported tendencies to vary their service according to their customers’ race. The link between servers’ perceptions of interracial tipping differences and race-based discriminatory service was subsequently replicated and extended by Brewster et al. (in press). In an analysis of survey data collected from a large geo-demographically diverse sample of U.S. restaurant servers (n = 872), the authors found that servers who harbor negative attitudes towards the tipping practices of customers of color (i.e., Blacks or Hispanics) or positive attitudes towards Whites’ tipping behaviors were also more likely to report withholding effort from their Black and Hispanic patrons. In a third study, Brewster (2013) analyzed survey items that asked respondents to consider eighteen scenarios, each representing realistic inter and intra-table variation in customer attributes (e.g., three white college age men, three Hispanic teenagers, two elderly couples, etc.), and report on a five-point scale how each table is typically thought to tip (very good = 1−very bad = 5). Servers who were more sensitive to tipping differences across these scenarios were also found to be more likely to report that they discriminate in their service delivery by differentially extending excellent service in response the predicted likelihood that they would be rewarded with a commensurately good tip. While these studies document a link between servers’ perceptions of intergroup differences in tipping behaviors and discriminatory service delivery many questions surrounding this relationship remain. First, given Brewster’s (2012a, 2013) reliance on a small community sample of servers, the generalizability of the aforementioned results has not been adequately established. Second, while Brewster (2012a), Brewster et al. (in press) demonstrate that discriminatory service is, in part, the result of server attitudes towards the tipping practices of African Americans and Hispanics that are congruent with stereotypes casting these customers as poor tippers it is unclear whether server attitudes that are congruent with stereotypes depicting other customer types as poor tippers (e.g., elderly, teens, etc.) also encourage discriminatory service as suggested by conventional logic. Third, it is unclear if or how server attitudes that are congruent with stereotypes depicting certain customer types as relatively better tippers (e.g., whites, middle aged, men, etc.) affect their propensities to discriminate in their service delivery. To the degree that servers’ discriminate against customer types perceived be relatively poor tippers by withholding effort from these clientele they should also be motivated to discriminate in their service delivery by allocating extra effort to customer types perceived to be above average tippers. Alternatively, given that servers’ negativity towards some customers’ tipping practices is likely to surface only in relation to their perceptions of other customer types as being especially good tippers, and vice-a-versa, it seems equally plausible that discriminatory service is primarily the outcome of server attitudes that are congruent with both positive and negative tipping stereotypes. Finally, while the results of Brewster, 2013 study suggests that discriminatory service may stem from servers’ perceptions of tipping differences across customer aggregates the measure of subjects’ sensitivity to such differences did not differentiate servers’ who harbor consistently negative attitudes towards customers’ tipping practices from those who harbor consistently positive

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

attitudes. Under both conditions servers were considered to be equally insensitive to intergroup tipping differences and thus predicted to be less likely to discriminate in their service delivery. As a result of the author’s operationalization of this sensitivity construct the nature of the observed effect of servers’ perceptions of tipping differences on discriminatory service is ambiguous. It is not clear, for instance, whether the effect observed in this study is driven primarily by servers’ sensitivity to tipping differences within or between groups conventionally thought to be below average or above average tippers, respectively. Further, Brewster (2013) finding that those who harbor consistently negative perceptions towards customers’ tipping practices (e.g. insensitive to tipping differences) discriminate less than their counterparts is inconsistent with the idea that discriminatory service stems from server attitudes that are congruent with stereotypes casting certain customer types as poor tippers (e.g., African Americans and Hispanics, cf. Brewster, 2012a; Brewster et al., in press). The current study addresses the limitations evident in these extant studies by assessing the effects of servers’ adherence to tipping stereotypes that cast certain customer aggregates as below average tippers and other aggregates as above average tippers on their reports of providing service that is informed by their predictions of customers’ tipping intentions. I posit that harboring attitudes that are congruent with either positive or negative tipping stereotypes will motivate servers to discriminate in their service delivery. Further, I anticipate that servers’ propensities to discriminate in their service delivery will be most pronounced when their attitudes are congruent with both negative and positive tipping stereotypes. That is, the effects of harboring attitudes that are congruent with negative (or positive) aggregate tipping stereotypes on discriminatory service will be stronger as a function of servers’ adherence to positive (or negative) aggregate tipping stereotypes. H1 Harboring attitudes that are congruent with negative (or positive) aggregate tipping stereotypes will be positively associated with servers’ self-reported proclivities to provide discriminatory service. H2 The effects of harboring attitudes that are congruent with negative (or positive) aggregate tipping stereotypes on discriminatory service will be moderated by servers’ adherence to positive (or negative) aggregate tipping stereotypes.

2.2. Service discrimination and tip earnings To the degree that servers do extend differential service according to their predictions of customers’ tipping intentions, as the previously reviewed studies suggest, they are presumed to do so in order to maximize their tip earnings. Service discrimination, in other words, is thought to be an outcome of servers’ instrumentally rational deliberations concerning the predicted odds that a customer will leave an acceptable tip (15–20% of the bill) after services have been rendered. In the event that the predicted odds of a table leaving an acceptable tip are low, many servers believe that they can enhance their overall earnings by withholding effort from these guests and redistributing their efforts to tables that are perceived to be predictably more generous in their tipping behaviors (cf. Brewster, 2012a, 2013; Brewster et al., in press). The instrumental motivation and presumed economic effectiveness underling service discrimination is exemplified by a respondent in Erickson’s (2004, p. 557) study when he explains, “. . .The people who don’t tip right, they just don’t get the service, it’s not that we don’t serve them, but we know. I don’t work for free. I mean I’m not there to have fun. I’m here to make money.” This response echoes that given by Dirks and Rice (2004, p. 43) respondent, who states: “I. . .I hate to admit, but. . .I try to give everyone, um, same service, but I try to concentrate myself on tables who I know are going to tip well.”

17

Despite the lack of research assessing the economic effectiveness of discriminatory service, there are empirical reasons to doubt that restaurant servers can typically increase their tips by differentially extending service in response to their perceptions of customers’ tipping intentions. First, customers’ evaluations of service quality in full-service restaurants have consistently been shown to be positively associated with tips (Lynn and McCall, 2000; Lynn and Sturman, 2010). Thus, all else being equal, any incremental decreases in perceived service quality that might result from discriminatory service delivery would logically result in a cumulative net decrease in servers’ tipped income2 . Second, there a host of server behaviors that are not necessarily associated with customers’ rating of service quality but that have nevertheless been shown to increase tip earnings (e.g., writing “Thank You” on the check, forecasting good weather, telling a joke, calling customers by name, etc., cf., Lynn, 2003; Lynn and McCall, 2000). If servers utilize such tip-enhancing techniques indiscriminately, then they would logically be able to increase their overall tipped income. Alternatively, if such techniques are differentially utilized by servers in response to their a priori predictions of receiving a good/bad tip, their overall tipped income would likely be restricted in a self-fulfilling fashion. That is, by failing to utilize tipenhancing techniques or other value-adding gestures when waiting on customers who are predicted to leave an inadequate tip, servers may effectively be creating the conditions to ensure that a relatively smaller tip is realized from these customers (see Barkan and Israeli, 2004; Brewster, 2012b; Brewster and Mallinson, 2009; Dirks and Rice, 2004). Third, given the limitations inherent in predicting individual’s actions more generally, it is unlikely that servers would be capable of predicting customer’s tipping practices with the precision that would be necessary to make discriminatory service an economically rational response to variable tipping practices. Existing studies have shown, for instance, that many of the customer characteristics/attitudes that are strongly associated with tipping behaviors cannot be readily observed by restaurant servers (e.g., motives underlying tipping practices and familiarity with tipping norms, cf., Lynn, 2006b; Lynn and Brewster, 2015; Lynn, 2009). The customer/table characteristics that servers would be able to readily observe and thus draw from to predict customers’ tipping intentions (e.g., age, gender, race, dress, party size, etc.) have, however, been found to have relatively small effects on actual tipping behaviors and are collectively able to explain a very modest amount of the variation in bill adjusted tip rates. In short, and as Maynard and Mupandawana (2009, p. 602) point out, “there is so much unexplained variation in tipping rates that it is not productive to base staffing or service quality decisions on diner-specific tipping expectations.” Finally, some research on the service-tipping relationship has found that the tipping practices of some customers are more

2 The correlation between customers’ rating of service quality and size of the gratuity that they leave their server is on average about 0.2 (cf., Azar, 2009; Lynn and McCall, 2000; Lynn, 2001). Intuitively this might suggest that any effects of discriminatory service through service quality would likewise be small. However, in the absence of further research readers should reserve drawing such a conclusion. First, studies that have found a small relationship between tips and service quality have primarily used between-subject correlational designs. A within-subject study by Lynn and Sturman (2010), however, found that the strength the relationship between service quality and tips varies considerably across consumers and that for some customers the strength of this relationship is quite strong. Further, even if the strength of the between-subject service-tipping relationship is on average quite modest this would not preclude there from being a substantively meaningful effect of variable service delivery on servers’ tip earnings over time. In other words, the potentially trivial between-customer differences in tip earnings that stem from servers’ propensities to discriminate in their service delivery could lead to nontrivial earning losses over time.

18

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

sensitive to service quality than others (Lynn and Sturman, 2010). Such differences in the strength of the tipping-service relationship across customers could further undermine the economic effectiveness of discriminatory service in response to severs’ predictions of customers’ tipping intentions3 . For example, consider that servers tend to perceive customers who are older, female, Hispanic, and Asian to be below-average tippers (McCall and Lynn, 2009) and, as such, might be more likely to discriminate against clients from these groups in an attempt to maximize their tipped income (cf., Ayres et al., 2005; Brewster, 2012a; Brewster et al., in press). However, some research suggests that the tipping practices of these customers may be more strongly associated with service quality compared to their younger, male, White counterparts (Lynn et al., 2012; Lynn and Katz, 2013; Lynn and Thomas-Haysbert, 2003). As such, rather than maximizing tipped incomes, differences in customers’ sensitivities or reactions to service quality instead may result in inefficient predictions of customers’ tipping intentions and possibly negative returns on servers’ discriminatory service strategies. In sum, rather than discrimination being economically rational, both logic and existing empirical evidence call into question servers’ abilities to predict customers’ tipping intentions with enough precision to make discriminatory service an economically advantageous response to variable tipping practices. In fact, if there is an association between discriminatory service delivery and tip earnings, then the evidence suggests that the nature of the relationship is most likely to be negative. H3 Servers’ self-reported proclivities to provide discriminatory service will be negatively associated with tip earnings. 3. Data and methods 3.1. Data collection and Sample The predicted relationships between perceptions of intergroup tipping differences, discriminatory service, and tip earnings are tested in a large (n = 954), and geo-demographically diverse data set derived from a larger survey that was anonymously completed by individuals who were either currently employed as a restaurant server or had been so at some point in the past year. The survey asked a variety of questions intended to solicit information 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 various websites that are known to attract restaurant servers4 . Of the 1,786 individuals who clicked on the survey link, 221 did not reside in the United States, had not worked as a restaurant server in the past year, or had worked in a restaurant that prohibited tipping, leaving a potential sample of 1565 respondents who are current or recent servers in tipping restaurants in the U.S. Of these, 59% (n = 929) clicked “submit” to complete the questionnaire. However, 25 respondents who completed a substantial portion of the survey, including

3 While research on the service-tipping relationship has found that the tipping practices of some customers are more sensitive to service quality than others (Lynn and Sturman, 2010) the specific factors underlying such differences are not clear. Studies that have tested for moderators of the service-tipping relationship have either found the relationship to be quite robust across customer and server characteristics or have found inconsistent or isolated moderation effects (cf. Lynn et al., 2012). Analyzing two distinct data sources Lynn et al. (2012) did, however, find that tips increased with service more strongly the larger the size of the bill. Thus, servers might be able to realize a greater marginal return on their service efforts by focusing disproportionately on tables that are going to have large bills. 4 The blogs posting links to the survey were: www.stuckserving.com, waiterextraordinaire.blogspot.com, and theseamericanservers.blogspot.com, www.facebook.com/pages/Bitchy–Waiter.

providing responses to each of the dependent variables in this study (i.e., service discrimination, relative tip earning, average percent tip) were retained. While less than 6% of the remaining 954 cases had missing values on any one of the independent or control variables in this study, multivariate listwise deletion across these variables would result in the loss of an additional 23% of cases (n = 219). Thus, to retain these cases, a multiple imputation procedure was used to estimate values for observations with missing data on each of the independent and control variables included in this analysis. Specifically, a Markov Chain Monte Carlo (MCMC) method in SPSS was used to substitute each missing value with a list of ten different simulated values, resulting in the construction of ten different imputed datasets. Regression models are then estimated using each of the ten datasets, and results are combined “to obtain overall estimates and standard errors that reflect missing-data uncertainty as well as finite-sample variation” (Schafer and Graham, 2002:165). These procedures yield a geographically diverse (48 states and the District of Columbia) final analytic sample of 954 restaurant servers. 3.2. Measures 3.2.1. Dependent variables Service discrimination. Service discrimination in this study is conceptualized as the deliberate varying of service in response to servers’ predictions of customers’ tipping intentions. Respondents were asked to indicate on a 7-point scale (1 = strong disagree; 7 = strongly agree) how much they disagree or agree with the following statements: “I give better service to customers I expect to be good tippers than to those I expect to be bad tippers,” “I always give my best effort when serving regardless of who or what my customers are (reverse coded).” On a 5-point scale (1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = all of the time) respondents were also asked to indicate how frequently they “give substandard service to customers expected to be poor tippers.” Responses to these items were standardized and averaged to create single index measuring service discrimination where higher values indicate subjects’ greater propensities to self-report that they discriminate in their service delivery (Cronbach’s ˛ = 0.71)5 . Five cases had information on two of the items but were missing values on the remaining item used to create this index. In these cases index scores are based on the information the respondents provided (Table 1). Tip earnings. This study includes two measures of servers’ tip earnings. First, servers’ average tip earnings was ascertained by asking respondents to approximate on an 8-point scale the average tip percentage that they receive from customers at their restaurant (1 = 0%, tips are uncommon, 2 = 1–5%, 3 = 6–10%, 4 = 11–15%, 5 = 16–20%, 6 = 21–25%, 7 = 26–30%, and 8 = over 30%). A limited number of respondents reported their average tip percentage to be 5% or less (n = 3) or greater than 26% (n = 8). As such, these attributes were combined with the 6–10% attribute or the 21–25% attribute, respectively (1 = 0–10%; 2 = 11–15%; 3 = 16–20%; 4 = >20%). While tips are conventionally measured as a percentage of food sales, this operational definition of tip earnings does not take into consideration the full range of unmeasured establishment-level factors that could logically affect servers’ average tip percent (e.g., price, location, consumer traffic, etc.). Thus, as a test of robustness, I also include in the analysis a relative measure of tip earnings ascertained with a single question asking respondents to indicate on a 7-point semantic differential scale (1 = much smaller than most others’ tips and 7 = much larger

5 A rotated (Promax) principle components factor analysis of these items produced a single factor with an eigenvalue greater than one (1.90), which accounted for 63% of the total item variance.

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

19

Table 1 Positive and negative socio-demographic tipping stereotypes. Negative stereotypes

Positive stereotypes

Customer/table characteristic

N

Mean

% Congruent

Customer/table characteristic

n

Mean

% Congruent

Teenagers Elderly Christians Jews Coupon users Foreigners Asians Blacks Hispanics Tables with kids Women All Female dining parties Women dining alone

953 952 938 932 942 944 931 940 935 937 947 944 936

1.70 2.43 2.39 2.71 2.05 1.76 2.58 2.04 2.42 2.85 3.01 2.92 3.17

87.4 55.3 49.2 28.0 72.9 83.1 39.4 67.2 49.6 34.1 19.5 31.3 18.8

Young adults Middle aged adults Gay men Gay women Smokers Whites Couples on a date Men All male dining parties Mixed sex dining parties Men dining alone

948 949 947 947 935 929 937 950 939 938 940

3.03 3.67 3.98 3.36 3.37 3.26 3.61 3.84 4.12 3.52 3.78

28.4 54.6 65.6 37.0 31.4 23.8 53.6 65.3 76.7 43.2 60.7

Notes: Mean values represent the average server’s perception towards the tipping practices of each respective social groups on a 5-point scale (1 = very bad tippers, 3 = average tippers, 5 = very good tippers). Percent (%) congruent represents the percentage of servers who perceive each respective socio-demographic group to be below average or very bad tippers (negative stereotypes) or above average or very good tippers (positive stereotypes).

than most others’ tips) how their tips compare with those earned by other servers in their restaurant. Most of the servers in this study viewed their tip earnings quite favorably relative to their peers. In fact, only 17 cases had values of 1–3 on this scale. These attributes were thus combined and responses recoded to range from 1 to 5 resulting in a variable that is clearly measuring perceived differences among servers who harbor favorable perceptions of their tip earnings relative to their peers6 . 3.2.2. Primary independent variables of interest Adherence to aggregate tipping stereotypes. Subjects’ were asked to consider 24 different aggregates, or groups, of customers and to indicate on a 5-point scale whether they have found members of each group to be very bad tippers (=1) or very good tippers (=5). Some respondents indicated that they did not know how specific customer types tipped. These responses were recoded to indicate that the subject perceived the group in question to be average tippers (=3). Next, variables measuring subjects’ perceptions of tipping practices among teenagers, elderly, Christians, Jews, coupon users, foreigners, Asians, Blacks, Hispanics, tables with kids, women, all female dining parties, and women dining alone were recoded to reflect subjects’ agreement with stereotypes casting these customer aggregates as poor tippers (very bad tippers = 2, below average tippers = 1, and average or above average tippers = 0). These items were then summed and averaged to create an index wherein higher values indicate greater agreement with these negative tipping stereotypes. Fifty-three cases had information on at least 8 of the items but were missing values on one or more of the remaining five items used to create this index. In these cases index scores are based on the information the respondents provided. Variables measuring subjects’ perceptions of tipping practices among young adults, middle aged adults, gay men, gay women, smokers, Whites, couples on a date, men, all male dining parties, mixed sex dining parties, and men eating alone were recoded to reflect subjects’ agreement with stereotypes casting these customer aggregates as above average tippers (very good tippers = 2, above average tippers = 1, and average or below average tippers = 0). These items were then summed and averaged to create an index wherein higher values indicate more agreement with positive tipping stereotypes. Fifty-six cases had information on at least 7 of the items but were missing values on one or more of the

6 The bivariate association between these alternative measures of servers’ tip earnings is 0.17 (p < .001).

remaining four items used to create this index. In these cases index scores are based on the information the respondents provided7,8 .

3.2.3. Control variables The analysis includes controls for numerous exogenous factors that might account for spurious links between servers’ perceptions of tipping differences, service discrimination, and relative tip earnings. First, I control for the effects of subjects’ disposition towards customer service (e.g., service orientation, cf. Brown et al., 2002; Donavan and Hocutt, 2001; Kim et al., 2003; Kim, 2011). In accordance with Gwinner et al. (2005) subjects were asked to indicate on a 7 point scale how much they agreed with the following statements (1 = strongly disagree to 7 = strongly agree): “I enjoy helping others,” “I can get along with most anyone,” “I pride myself in providing courteous service,” “It is natural for me to be considerate of others’ needs,” and “The best job I can imagine would involve assisting others in solving their problems.” Responses to these four questions were averaged to create an index measuring servers’ customer-orientation (Cronbach’s ˛ = 0.84). Ten cases had information on at least 3 of the items but were missing values on one or more of the remaining two items used to create this index. In these cases index scores are based on the information the respondents provided.

7 Given the lack of extant information on servers’ perceptions of the tipping practices of gay men/women and mixed sex dining parties I was precluded from including these variables in the positive stereotype congruency index on an a priori basis. It was only after discovering that the average server in this study perceives gay mean (mean = 3.98), gay women (mean = 3.36), and mixed sex dining parties (mean = 3.52) to be above average in their tipping practices that these variables were included in the index measuring servers’ agreement with positive tipping stereotypes. The variables measuring servers’ perceptions of the remaining 21 aggregates that were used to create the negative and positive tipping stereotype indices were directly or indirectly derived from extant studies on servers’ perceptions of customers’ tipping behaviors (cf., Lynn, forthcoming; Lynn and Katz, 2013; Maynard and Mupandawana, 2009; McCall and Lynn, 2009). Further, item means confirmed that each of the aggregates of customers included in the negative and positive stereotype measures are perceived by the typical server in this study to be below average or above average tippers, respectively, with the exception of the variables “women” and “women dining alone” who were perceived to be average (mean = 3.01) or slightly above average (mean = 3.17) in their tipping practices. However, relative to ¨ = 3.84) and “men dining alone” (mean = 3.78), these their counterparts, “men(mean customers were still considered to be poor tippers and thus included in the negative stereotype congruency index. 8 The bivariate association between measures of servers’ adherence to positive and negative aggregate tipping stereotypes is small and unreliable (r = 0.058, p = 0.08) thus suggesting that these measures are capturing unique and independent aspects of servers’ attitudes towards customers’ tipping behaviors.

20

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

Potential regional differences in perceptions customers’ tipping intentions, server discrimination, and tip earnings are controlled by including dummy variables for Northeast (=1), Midwest (=1), and West (=1) with those residing in the South (=0) serving as the reference group. I also control for the effects of subjects’ sex, race, and age. Respondents’ sex is dummy coded as female (=1) and male (=0) and respondents’ race is dummy coded as white (=1) and nonwhite (=0) (i.e., black, Hispanic/Latino, Asian, Native American, and other). Subjects’ age is measured in years and was calculated from a question asking them to indicate the year they were born9 . Additionally, respondents were asked to indicate the number of years that they have worked as a restaurant server (ranging from 0 = less than 1 year to 10 = 10 or more years), whether they are currently employed as a server (1 = yes), work in a restaurant where service charges (e.g., tips) are automatically added to all checks (1 = yes), or work in a restaurant that is commonly patronized by a customers who are ethnic minorities (1 = yes), foreign (1 = yes), or elderly (1 = yes). Also included in the analysis are two nested dummy variables measuring reported expensiveness of the restaurant wherein the respondent is employed (expensive = 1; moderately expensive = 1; inexpensive = 0). Response patterns for these variables were not mutually exclusive; that is, some subjects reported to be employed in a combination of restaurants that differed by expensiveness (n = 119). Since this response pattern might reflect respondents’ employment in two different restaurants, a dummy variable is included in the analysis to adjust for this overlap (multiple responses perhaps indicating two or more jobs = 1; mutually exclusive responses = 0). Finally, at the end of the survey, respondents were asked to indicate on a five point scale (from 1 = strongly disagree to 5 = strongly agree) how much they agreed with each of the following four statements: “I took the survey seriously”; I was completely honest when answering the questions”; “I read each question carefully”; I tried to make my answers as accurate as I could.” Subjects’ responses to these questions were averaged to create an index measuring engagement/honesty when completing the questionnaire (Cronbach’s ˛ = 0.95). 3.3. Sample summary statistics Table 2 provides summary statistics for the analytic sample used in this analysis. The average subject in this study reported to receive an average tip percentage between 16 and 20%. Subjects harbored quite favorable perceptions of their tip earnings relative to their coworkers. The typical server in this study admittedly sometimes provides discriminate service. Further, the average respondent harbors attitudes, which are congruent with tipping stereotypes casting some aggregates of customers as below average tippers and other aggregates as above average in their tipping practices. The typical subject has approximately 8 years of experience as a server, has a strong orientation to customer service, and is currently employed as a server (90%) in a restaurant that does not automatically add gratuities to customers’ checks (98%). A considerable number of the servers in this sample reported to work in a restaurant that is commonly frequented by ethnic minority (47%), foreign (38%), and/or elderly customers (66%). In addition, while respondents are geographically dispersed throughout the U.S. they are likely to be White (91%) and female (84%). The average respondent is 33 years of age. Roughly thirty-percent are employed in inexpensive restaurants, 74% in moderately expensive restaurants, and 8% in expensive restaurants. Finally, the vast majority of the

9 Values were imputed for three cases wherein the original values on the age variable were nonsensical (e.g., 1070).

subjects in this analysis reported to be engaged with the survey and to answer each question honestly.

4. Results To test the hypothesized relationships between measures of servers’ adherence to positive/negative tipping stereotypes, discriminatory service delivery, and tip earnings, nested OLS regression models were estimated. Table 3 presents the results from models predicting self-reported discriminatory service (H1–2) and Table 4 presents results from models predicting server reported tip earnings (H3).

4.1. Tipping stereotypes and discriminatory service In Model 1, a baseline model predicting server reported discriminatory service is estimated that includes only the covariates in this analysis. The covariates included in this baseline model are able to account for 12% of the variation in server reported discriminatory service. Model 2 provides a test of the predicted main effects of server attitudes that are congruent with positive and/or negative tipping stereotypes on reported service discrimination (H1). The results of this model provide partial support for the first hypothesis tested in this analysis. Harboring attitudes that are congruent with positive tipping stereotypes is not found to be associated with server reported discriminatory service (B = 0.040, p = 55). However, harboring attitudes that are congruent with stereotypes casting certain aggregates of customers as below average tippers is shown to be reliably associated with increased server reported discriminatory service (B = 1.30, b = 0.41, p < 001). Further, the statistically significant negative second order effect (B = −0.508, b = −0.24, p < 01) observed in this model indicates that discriminatory service increases with servers’ agreement with negative tipping stereotypes at a decreasing rate and then plateaus10 . The nonlinear effect of servers’ adherence to negative aggregate tipping stereotypes that is observed in model 2 accounts for roughly 6% of the variation in self-reported discriminatory service in this sample11 . In contrast to the H2, and as shown in model 3, the effects of servers’ adherence to negative tipping stereotypes are not found to be moderated by their adherence to positive tipping stereotypes (B = −0.087, p = 669).

10 In the process of developing operational definitions for measures of servers’ perceptions of aggregate tipping differences, I began by taking an exploratory and empirically driven approach in an effort to first clarify the ambiguity inherent in Brewster (2013) measure of servers’ “sensitivity to tipping differences.” It was during this process that the nonlinear effect of servers’ adherence to negative tipping stereotypes was discovered. 11 Existing research has shown customers’ of color to be particularly vulnerable to discriminatory service as a result of servers’ perceptions of these customers as poor tippers (cf. Brewster, 2012a, 2012b, 2013; Dirks and Rice, 2004; Noll and Arnold, 2004). Thus, it is possible that the effects of servers’ adherence to negative tipping stereotypes that are observed in model 2 are driven primary by respondents’ negativity towards racial/ethnic minorities. To explore this possibility, I estimated a supplemental model after creating a separate index measuring servers’ negativity towards the tipping practices of Jews, foreigners, Asians, Blacks, and Hispanics. Results show that servers’ negativity towards racial/ethnic minorities (B = 0.278, p < 0.001) and other customer aggregates perceived to be poor tippers (B = 0.266, p < 0.01) have independent effects on their propensities to report discriminating in their service delivery in response to their predictions of customers’ tipping intentions. However, the nonlinear effect of servers’ negativity towards customer types stereotypically thought to be poor tippers was found to be limited to nonracial/ethnic customer attributes (e.g., teens, elderly, Christians, coupon users, tables with kids, women, all women, and woman alone).

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

21

Table 2 Descriptive statistics for variables in analysis (n = 954). Variable

Min.–Max.

Primary variables Relative tip earnings Average percent tip Discriminatory service Positive tipping stereotypes (0 = non-congruent attitudes) Negative tipping stereotypes (0 = non-congruent attitudes) Covariates Age Race White (=1) Non-White (=0) Gender Male (=0) Female (=1) Region South (=0) Northeast (=1) Midwest (=1) West (=1) Current server (yes = 1) Customer-orientation Experience (years) Automatic tip (yes = 1) Lots of minority clientele (yes = 1) Lots of foreign clientele (yes = 1) Lots of elderly clientele (yes = 1) Inexpensive restaurant (=0) Moderately expensive restaurant (yes = 1) Expensive restaurant (yes = 1) Employed in two restaurants (yes = 1) Survey engagement/honesty

%

1–5 1–4 −1.22–2.63 0.00–1.82 0.00–2.00

Mean

SD

3.00 2.88 0.000 0.663 0.659

0.90 0.67 0.80 0.36 0.32

19–72

–

32.76

9.10

0–1 –

90.9 9.1

–

–

0–1

16.1 83.9

–

–

–

–

– 5.79 7.58 – –

– 0.95 2.88 – –

4.73

.50

0–1 – 0–1 1–7 0–10 0–1 0–1

0–1 0–1 0–1 0–1 1–5

4.2. Discriminatory service and tip earnings Table 4 presents results from OLS regression models testing the hypothesized (H3) negative effect of discriminatory service on relative tip earnings (models 4–5) and average tip percentage

38.2 22.1 24.6 15.1 90

2.0 47.4 37.7 66.5 30.8 73.8 8.1 12.5 –

(models 6–7). As shown in model 5, net of the effects of the covariates and servers’ stereotypic attitudes towards aggregate level tipping practices, discriminatory service delivery is reliably associated with a decrease in relative tip earnings (B = −0.077, p < 05). Nevertheless, the effect size of discriminatory service is

Table 3 Metric and standardized coefficients from OLS regression analyses predicting Server Reported Service Discrimination (n = 954). Model 1 Independent variables Positive tipping Stereotypes Negative tipping Stereotypes Negative tipping Stereotypes squared Negative × positive stereotypes Covariates Age (years) Subject White (yes = 1) Subject Female (yes = 1) South (yes = 1) Midwest (yes = 1) West (yes = 1) Current server (yes = 1) Customer orientation Serving experience Automatic tip (yes = 1) Minority clientele (yes = 1) Foreign clientele (yes = 1) Elderly clientele (yes = 1) Moderately expensive Restaurant (yes = 1) Expensive restaurant (yes = 1) Employed in two restaurants (yes = 1) Survey engagement/honesty Constant R2 Note: Standardized coefficients are in parentheses. * p < 0.05. ** p < 0.01. *** p < 0.001.

−0.017*** (−0.154) −0.031 (−0.031) −0.160* (−0.160) −0.023 (−0.023) −0.059 (−0.059) −0.002 (−0.002) 0.102 (0.102) −0.177*** (−0.168) 0.007 (0.020) −0.010 (−0.010) 0.134* (0.134) −0.063 (−0.063) −0.120* (−0.120) −0.134* (−0.134) −0.143 (−0.143) 0.022 (0.022) −0.122* (−0.061) 2.35 (0.242) 0.12

Model 2

Model 3

0.040 (0.015) 1.30*** (0.408) −0.508** (−0.242)

0.093 (0.013) 1.35*** (0.408) −0.508** (−0.242) −0.087 (−0.010)

−0.014*** (−0.127) −0.042 (−0.042) −0.148* (−0.148) −0.045 (−0.045) −0.072 (−0.072) −0.004 (−0.004) 0.088 (0.088) −0.146*** (−0.138) 0.003 (0.009) −0.071 (−0.071) 0.108* (0.108) −0.103 (−0.103) −0.112* (−0.112) −0.183** (−0.183) −0.157 (−0.157) 0.033 (0.033) −0.156* (0.078) 1.73 (0.323) 0.18

−0.014*** (−0.127) −0.041 (−0.041) −0.145* (−0.045) −0.045 (−0.045) −0.071 (−0.071) −0.003 (−0.003) 0.087 (0.087) −0.146*** (−0.138) 0.003 (0.008) −0.072 (−0.072) 0.107* (0.107) −0.102 (−0.102) −0.112* (−0.112) −0.182** (−0.182) −0.156 (−0.156) 0.034 (0.034) −0.155* (−0.077) 0.277 (0.321) 0.18

22

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

Table 4 Metric coefficients from OLS regression analyses predicting relative tip earnings and average percent tip (n = 954). Relative tips

Independent variables Positive tipping stereotypes Negative tipping stereotypes Discriminatory service Covariates Age (years) Subject White (yes = 1) Subject female (yes = 1) South (yes = 1) Midwest (yes = 1) West (yes = 1) Current server (yes = 1) Customer orientation Serving experience Automatic tip (yes = 1) Minority clientele (yes = 1) Foreign clientele (yes = 1) Elderly clientele (yes = 1) Moderately expensive restaurant (yes = 1) Expensive restaurant (yes = 1) Employed in two restaurants (yes = 1) Survey engagement/honesty Constant R2

Average tip percent

Model 4

Model 5

Model 6

Model 7

0.451*** (0.180) −0.066 (−0.023)

0.455*** (0.181) −0.023 (−0.008) −0.077* (−0.085)

0.223*** (0.121) −0.224** (−0.106)

0.224*** (0.121) −0.221** (−0.104) −0.006 (−0.009)

0.008 (0.079) −0.051 (−0.057) 0.062 (0.068) 0.063 (0.069) 0.050 (0.055) 0.057 (0.063) 0.031 (0.034) 0.060 (0.063) 0.039** (0.124) −0.080 (−0.089) 0.094 (0.104) 0.106 (0.117) −0.048 (−0.053) −0.027 (−0.030) −0.256* (−0.283) 0.159 (0.176) 0.007 (0.004) 1.70*** (−0.115) 0.09

0.007 (0.068) −0.055 (−0.061) 0.051 (0.056) 0.059 (0.065) 0.045 (0.049) 0.057 (0.063) 0.038 (0.042) 0.049 (0.051) 0.039** (0.125) −0.087 (−0.096) 0.102 (0.113) 0.099 (0.109) −0.057 (−0.063) −0.041 (−0.045) −0.269* (−0.298) 0.161 (0.178) −0.005 (−0.003) 1.86*** (−0.088) 0.09

−0.006* (−0.082) −0.040 (−0.061) −0.147* (−0.221) 0.041 (0.062) 0.040 (0.060) −0.085 (−0.127) 0.075 (0.113) 0.009 (0.013) 0.038*** (0.163) −0.116 (−0.174) −0.177*** (−0.266) 0.063 (0.095) −0.106 (−0.159) 0.223*** (0.335) 0.341*** (0.512) −0.091 (−0.136) −0.067 (−0.050) 3.09*** (0.051) 0.13

−0.006* (−0.084) −0.041 (−0.061) −0.148* (−0.223) 0.041 (0.062) 0.040 (0.060) −0.085 (−0.127) 0.076 (0.114) 0.008 (0.012) 0.038*** (0.163) −0.117 (−0.175) −0.177*** (−0.265) 0.063 (0.094) −0.106 (−0.160) 0.222*** (0.333) 0.340*** (0.510) −0.090 (−0.136) −0.067 (−0.050) 3.10*** (0.054) 0.13

Note: Standardized coefficients are in parentheses. * p < 0.05. ** p < 0.01. *** p < 0.001.

quite small (b = −0.085) and explains less than a half of a percentage point of the variation in relative tip earnings. Further, and as shown in model 7, discriminatory service is not found to be associated with servers’ reported average tip percentage (B = −0.006, p = 0.83). Taken as a whole these results provide no evidence that discriminatory service is an economically effective strategy to enhance servers’ tip earnings. To the contrary, the results from Models 5 and 7 suggest that there is either no effect of discriminatory service on tip earnings or the effect is negative but quite modest in size. Notably, harboring attitudes that are congruent with positive tipping stereotypes is shown in Table 4 (models 5 and 7) to be predictive of an increase in both relative tip earnings (B = 0.455, b = 0.18, p < 001) and average tip percentage (B = 0.224, b = 0.12, p < 001). Further, harboring attitudes congruent with negative tipping stereotypes is shown in model 7 to be predictive of a decrease in average tip percent (B = −0.221, b = −0.104, p < 01)12 . It is possible that the effects of servers’ adherence to aggregate-level tipping stereotypes and discriminatory service delivery interact to predict tip earnings such that the effects of adhering to tipping stereotypes on tip earnings are observed or are stronger when coupled with discriminatory service delivery. To explore this possibility, I also estimated models predicting relative tip earnings and average tip percent that included product terms between reported discriminatory behaviors and servers adherence to negative/positive tipping stereotypes. Results of these models did not support this possibility.

12 Baseline models predicting relative and absolute tip earnings were estimated that included only the covariates. The explanatory power of these models were then juxtaposed with models 4 (relative tip earnings) and 6 (tip percentage) in order to ascertain the unique variance in tip earnings explained by servers’ stereotypical attitudes towards customers’ tipping practices. Relative to the baseline models tipping stereotypes were found to account for 3% of the variance in both measures of tip earnings.

5. Discussion and conclusions 5.1. Discussion of contributions This study offers several contributions to the literature on service discrimination in full-service restaurants. First, in a large geo-demographically diverse sample of current and former restaurant servers results indicate that servers’ who harbor attitudes towards customer aggregates stereotypically thought to be below average tippers are also more likely to profess that their delivery of service is predicated on their predictions of customers’ tipping intentions. Further, this effect was found to be nonlinear thus suggesting that servers who harbor some negative attitudes towards the tipping practices of customer aggregates may discriminate in their service delivery as frequently as their counterparts who strongly embrace such stereotypes. Third, results suggest that servers’ adherence to positive tipping stereotypes is inconsequential towards understanding servers’ instrumental motivations to discriminate in their service delivery. That is, perceptions of some customer aggregates as being above average tippers does not appear to be a sufficient condition to motivate discriminatory service. These results replicate, extend, and further clarify recent findings reported by Brewster (2013). Taken as a whole, results substantiate what has conventionally been thought to be true—many restaurant servers deliver inequitable service in response to their predictions of customers’ tipping intentions and that such predictions are informed, in part, by their negative perceptions of tipping practices among certain aggregates of customers. Further, this study’s findings suggest that this process operates net of the effects of a relatively large set of potential confounds. It is particularly noteworthy that the association between negative tipping stereotypes and discriminatory service was observed to be reliable net of the effects of servers’ orientation to customer service, thus suggesting that the association cannot be understood as simply a spurious manifestation of variability in servers’ dispositional proclivities to meet customers’

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

needs (cf. Brown et al., 2002; Donavan and Hocutt, 2001; Kim et al., 2003). Nevertheless, it remains possible that the effects of servers’ adherence to negative tipping stereotypes on their proclivities to discriminate in their service delivery are moderated by their orientation to customer service. That is, servers with personalities that are conducive to customer service may refrain from discriminating in response to their a priori predictions of customers tipping intentions, or do so less frequently, despite harboring attitudes that are congruent with stereotypes casting some customer aggregates as poor tippers. To explore this possibility, model 2 (Table 3) was re-estimated after including a product term between subjects’ adherence to negative tipping stereotypes and their orientation to customer service. The interaction term in this supplementary model was not statistically reliable (B = −0.081, p = 0.358). Thus, hiring restaurant servers with the right personality for customer service (e.g., a disposition to be caring, courteous, and friendly; cf. Kim, 2011) is alone unlikely to be an adequate business strategy to curtail the threat to optimal and equitable service quality that is posed by servers’ adherence to negative tipping stereotypes. Fourth, the current study constitutes the first attempt to assess the association between tip earnings and discriminatory service delivery. Contrasting with the beliefs of some servers, discriminatory service in response to variable tipping practices was not found to be associated their average percent tip. This finding constitutes good news for restaurant operators, as it suggest that servers are not monetarily rewarded for discriminatory service practices. However, these results also suggest that servers are not financially penalized for discriminating in their service delivery. As such, all else being equal there may be no economic advantage associated with providing optimal or equitable service to all restaurant patrons. In fact, in the event that discriminating servers, vis-à-vis non-discriminators, withhold more effort from tables predicted to tip below average than is otherwise redistributed to tables predicted to tip above average they could theoretically realize a net gain in their tip earnings relative to the amount of effort they expend. To the degree that this is the case, non-discriminating servers may have to work harder to secure the same average percent tip as their discriminating counterparts. However, results also suggest that discriminating servers may in fact earn incrementally less in tips relative to their nondiscriminating counterparts. The inconsistency in the associations between discriminatory service and the two measures of tip earnings included in this study may suggest that the service-tipping relationship is stronger at the server level of aggregation than it is at the customer-level of aggregation. If this is the case, compared with average percent tip, the relative measure of tip earnings may be comparatively more sensitive to variability in service quality that would logically result from discriminatory service delivery. That is, if non-discriminating servers compare tips with those of other servers, namely discriminating servers, they may notice that they tend to earn more tips each shift even if they are not able to notice the relationship between service quality (e.g., discrimination) and tips at the customer level (e.g., average tip percent; see Lynn and McCall, 2000, p. 212). Future studies should assess this possibility. Finally, I find in this study that harboring attitudes that are congruent with positive and negative tipping stereotypes have divergent effects of similar magnitude on servers’ reported average percent tip. Although these effects were not found to be mediated by discriminatory service, in the absence of further research it would be premature to reject variable service delivery as the primary mechanism by which harboring negative and/or positive tipping stereotypes affects tip earnings. Rather than variable service more generally the measure of discriminatory service used in this study may have only tapped servers’ conscious proclivities

23

to deviate, albeit incrementally, from the standard service requirements imposed by management (e.g., friendliness, attentiveness, promptness, etc.). If this is the case, this measure would be insensitive to the vast array of subtle gestures and behaviors that may not be conceived by servers as requisites of standard service but which could nevertheless be consciously or unconsciously differentially extended to customers in response to positive/negative perceptions of aggregate tipping differences (cf. Brewster and Mallinson, 2009; Lynn, 2003; Lynn and McCall, 2000). Further, if the measure of discriminatory service used in this study is only tapping servers’ proclivities to deviate from the standard service requirements imposed by management it might likewise be insensitive to systematic differences in how service is delivered (cf., Brewster et al., 2014). For instance, the literature on emotional labor has identified deep acting and surface acting as strategies (cf., Hochschild, 1983) utilized by service providers to regulate their emotions in order to conform to organizational display rules. While both deep and surface acting are effectively used by frontline employees, studies have found deep acting to be a stronger predictor of favorable customer evaluations of their service providers (e.g., Grandey et al., 2005; Groth et al., 2009; Hennig-Thurau et al., 2006). Additionally, in a study of restaurant server-customer dyads, Chi et al., 2011 found that servers who utilized deep acting were more likely to exceed their customers’ service expectations relative to those who used a surface acting strategy and as a result were rewarded with greater tips. Thus, if differential utilization of these emotional labor strategies stem from or are associated with servers’ adherence to positive/negative aggregate tipping stereotypes it might explain the divergent effects of such stereotypes on servers’ tip earnings that were observed in this study. That is, servers who adhere to negative/positive tipping stereotypes may also earn less/more in tips as a result of their tendencies to utilize surface/deep acting as a method of regulating their workplace emotions. 5.2. Implications for restaurant operations Taken as a whole, this study’s findings suggest that implementing an inclusive pricing or automatic service charge system (or a combination of the two) in lieu of voluntary tipping would reduce servers’ motivations to allocate differential levels of service. However, given the various advantages associated with the custom of tipping (cf., Azar, 2003; Lynn and Withiam, 2008; Lynn et al., 2011; Kwortnik et al., 2009), restaurant proprietors in the United States, with few exceptions (cf. Wells, 2013), have been reluctant to experiment with alternative ways to compensate their wait staff. This study’s findings suggest, however, that restaurant operators may be able diminish the discriminatory service that stems from this system of compensation without abolishing the custom in their establishments. Some servers would logically be less inclined to embrace negative tipping stereotypes and to discriminate in their service delivery if they knew that doing so may not be economically productive. To the degree that such educational efforts are successful, service discrimination that directly results from both servers’ perceptions of intergroup tipping differences and their predictions of customers’ tipping intentions, more generally, should abate. Admittedly, convincing many servers that what they believe to be true (e.g., I can increase my tips by devoting “extra” service to good tippers at the expense of poor tippers) may be a fallacy will not be an easy undertaking and will likely require a multifaceted approach. To thwart the threat of this information being dismissed by some servers, who will likely find it to be incongruent with their beliefs, it is recommend that it be conveyed regularly and through multiple mediums (e.g., training workshops, training manuals/videos, shift meetings, staff meeting, posters in the back of the house, etc.).

24

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25

Rather than presenting such information matter-of-factly restaurant operators should consider using the analogy of gambling to teach servers that embracing negative tipping stereotypes and discriminating in their delivery of service may not be in their economic interest. By using the analog of gambling servers could be encouraged to recognize that their negative attitudes and discriminatory behaviors while in some cases might be rewarded, over the course of a shift, week, or year such attitudes and behaviors may leave them with a net earnings loss. In short, by making service discrimination analogous to gambling servers might be more inclined to concede that “the house eventually always wins” and thereby be discouraged from playing the game. While these practical ideas for curtailing service discrimination in U.S. restaurants flow logically from this study’s results there are some notable limitations that operators should be cognizant of before altering their existing approaches to staff training. 5.3. Limitations While the sample size, geographic diversity, and uniqueness of these data constitute important strengths of this study, the generalizable limits of this study’s findings are unknown because subjects were not randomly selected from the population of all tipped servers in the United States. For instance, respondents in this analytic sample might differ in meaningful ways from those servers who refrain from visiting the types of websites that were used to passively recruit participants, or from servers who visit these websites yet failed to provide a completed survey. One potential limitation imposed by this type of selection bias is that the processes observed in this sample may not operate similarly among servers that are unrepresented in these data. Thus, these results should be replicated using nationally representative samples or, at the very least, using data from respondents who diverge in meaningful ways from the sample used in this study (i.e., nonwhite servers and males). Additionally, the use of cross-sectional data precludes conclusive causal inferences about the relationships between the key variables in this analysis. As I previously conjectured, for instance, it is quite possible that harboring attitudes that are congruent with positive and/or negative tipping stereotypes lead servers’ to systematically vary their service delivery in some salient way thereby increasing/decreasing their tip earnings. The relationship between servers’ positive/negative perceptions of aggregate tipping practices and tip earnings, on the other hand, may reflect high tip earners’ tendencies to attribute their higher earnings to the better tipping practices of those customers’ stereotypically thought to be good tippers. In contrast, low tip earners might be inclined to attribute their lower earnings to those customer characteristics stereotypically associated with lower tips. Alternatively these relationships may not be causal but rather spurious as a result of exogenous factors that were not included as controls in the current study. Readers should reserve drawing strong conclusions until these findings can be replicated using data and modeling techniques that are better suited to approximate the proper temporal ordering among variables. Relatedly, given the self-reported nature of these data the associations observed in this study may have been artificially attenuated and/or inflated as the result of a variety of common method biases (e.g., satisficing, social desirability, acquiescence, negative/positive affect, item context, etc., see Podsakoff et al., 2012). Future assessments of the effects of servers’ attitudes (e.g., perceptions of customers’ tipping behaviors) and behaviors (e.g., discriminatory service) on tip earnings, in particular, would benefit from a multi-method multi-trait design in a real service context. In this vein, I recommend these associations be tested using data derived from within-restaurant surveys of customer-server dyads

over an extended period of time (cf., Chi et al., 2011; Groth et al., 2009; Medler-Liraz, 2014). Such a design would not only diminish validity threats posed by common method biases be would also be well suited to test for server, customer, and environmental characteristics that may function to moderate the discriminatory service–tip earnings relationship. It is likely that future studies will find that discriminatory service as a strategy to maximize tip earnings is more/less effective under certain conditions. The results of this study also underscore the need for further research on the antecedent factors implicated in servers’ predictions of customers’ tipping intentions. In this study, servers’ agreement with stereotypes portraying aggregates of customers as poor tippers accounted for only 6% of the variability in their reports of providing service that is informed by their predictions of customers’ tipping intentions. Thus, perceived intergroup tipping differences appear to constitute only a small part of what is likely to be a complex process by which servers’ predictions of customers’ tipping intentions materialize and become manifest in service delivery. Finally, whereas the results from this study adds to a growing body of literature indicating that restaurant servers deliver service that is predicated on their predictions of customers tipping intentions the nature of this variable service remains unclear. As a result of this lack of clarity, and as previously discussed, the measure of discriminatory service used in this study was quite crude and thus may have been insensitive to the multitude of subtle ways in which servers might systematically discriminate in their service delivery. Steps should thus be taken to replicate this study’s results using more sensitive measures of service discrimination. This research could reveal a much stronger relationship (positive or negative) between discriminatory service and tip earnings than was observed in the current study. I hope that the current study inspires such research. Acknowledgements I would like to thank Jon Brauer, David Merolla, and the anonymous referees for their assistance at various stages of stages of this project. I would also like to thank Michael Lynn for providing me with the data analyzed in this study. References Ayres, I., Vars, F.E., Zakariya, N., 2005. To insure prejudice: racial disparities in taxicab tipping. Yale Law J. 114 (7), 1613–1674. Azar, O.H., 2003. The implications of tipping for economics and management. Int. J. Soc. Econ. 30 (10), 1084–1094. Azar, O.H., 2004. Optimal monitoring with external incentives: the case of tipping. Southern Econ. J. 71 (1), 170–181. Azar, O.H., 2009. Incentives and service quality in the restaurant industry: the tipping—service puzzle. Appl. Econ. 41 (15), 1917–1927. Azar, O.H., 2011. Business strategy and the social norm of tipping. J. Econ. Psychol. 32, 515–525. Barkan, R., Israeli, A., 2004. Testing servers’ roles as experts and managers of tipping behaviour. Serv. Ind. J.V 24 (6), 91–108. Boyes, W.J., Mounts Jr., W.S., Sowell, C., 2004. Restaurant tipping: free-riding, social acceptance, and gender differences. J. Appl. Soc. Psychol.V 34 (12), 2616–2628. Brewster, Z.W., Mallinson, C., 2009. Racial differences in restaurant tipping: a labour process perspective. Serv. Ind. J. 29 (8), 1053–1075. Brewster, Z.W., 2012a. Racialized customer service in restaurants: a quantitative assessment of the statistical discrimination framework. Soc. Inquiry 82 (1), 3–28. Brewster, Z.W., 2012b. Racially discriminate service in full-service restaurants: the problem, cause, and potential solutions. Cornell Hospitality Q. 53 (4), 274–285. Brewster, Z.W., Rusche, S.N., 2012. Quantitative evidence of the continuing significance of race: tableside racism in full-service restaurants. J. Black Stud. 43 (4), 359–384. Brewster, Z.W., 2013. The effects of restaurant servers’ perceptions of customers’ tipping behaviors on service discrimination. Int. J. Hospitality Manage. 32 (1), 228–236. Brewster, Z.W., Lynn, M., Cocroft, S., 2014. Consumer racial profiling in U.S. restaurants: exploring subtle forms of service discrimination against black diners. Soc. Forum 29 (2), 476–495.

Z.W. Brewster / International Journal of Hospitality Management 46 (2015) 15–25 Brewster, Z.W., Brauer, J. R., Lynn, M., in press. Economic motivations and moral controls regulating discrimination against Black and Hispanic diners. The Sociological Quarterly. Brown, T.J., Mowen, J.C., Donavan, D.T., Licata, J.W., 2002. The customer orientation of service workers: personality trait effects on self- and supervisor performance ratings. J. Marketing Res. 39 (1), 110–119. Chi, N., Grandey, A.A., Diamond, J.A., Krimmel, K.R., 2011. Want a tip? Service performance as a function of emotional regulation and extraversion. J. Appl. Psychol. 96 (6), 1337–1346. Dirks, D., Rice, S.K., 2004. Dining while black: tipping as social artifact. Cornell Hotel Restaurant Admin. Q. 45 (1), 30–47. Donavan, D.T., Hocutt, M.A., 2001. Customer evaluation of service employee’s customer orientation: extension and application. J. Qual. Manage. 6 (2), 293–306. Erickson, K., 2004. To invest or detach? Coping strategies and workplace culture in service work. Symbolic Interact. 27 (4), 549–572. Grandey, A., Fisk, G.M., Mattila, A.S., Jansen, K.J., Sideman, L.A., 2005. Is ‘service with a smile’ enough? Authenticity of positive displays during service encounters. Organiz. Behav. Hum. Decis. Process. 96 (1), 38–55. Grossman, P.J., Parrett, M.B., 2011. Religion and prosocial behavior: a field test. Appl. Econ. Lett. 18 (6), 523–526. Groth, M., Hennig-Thurau, T., Walsh, G., 2009. Customer reactions to emotional labor: the roles of employee acting strategies and customer detection accuracy. Acad. Manage. J. 52 (5), 958–974. Gwinner, K.P., Bitner, M.J., Brown, S.W., Kumar, A., 2005. Service customization through employee adaptiveness. J. Serv. Res. 8, 131–148. Harris, M.B., 1995. Waiters, customers, and service: some tips about tipping. J. Appl. Soc. Psychol. 25 (8), 725–744. Hennig-Thurau, T., Groth, M., Paul, M., Gremier, D.D., 2006. Are all smiles rreated equal? How emotional contagion and emotional labor affect service relationships. J. Mark. 70 (3), 58–73. Hochschild, A.R., 1983. The Managed Heart: Commercialization of Human Feeling. University of California Press, Los Angeles, CA. Kim, H.J., McCahon, C., Miller, J., 2003. Service orientation for contact employees in Korean casual-dining restaurants. Int. J. Hospitality Manage. 22 (1), 67–83. Kim, H.J., 2011. Service orientation, service quality, customer satisfaction, and customer loyalty: testing a structural model. J. Hospitality Mark. Manage. 20 (6), 619–637. Kwortnik Jr., R.J., Lynn, M.W., Ross Jr., W.T., 2009. Buyer monitoring: a means to insure personalized service. J. Marketing Res. 46 (5), 573–583. Lynn, M., 2001. Restaurant tipping and service quality: a tenuous relationship. Cornell Hotel Restaurant Admin. Q. 42 (1), 14–20. Lynn, M., 2003. Restaurant tipping and service: an update, extension, and reconciliation. Cornell Hotel Restaurant Admin. Q. 44 (5/6), 139–148. Lynn, M. Forthcoming Negative Perceptions of Christian Tippers: How Widespread are They? Journal of Foodservice Business Research; in press. Lynn, M., 2004. Black-white differences in tipping of various service providers. J. Appl. Soc. Psychol. 34 (11), 2261–2271. Lynn, M., 2006a. Race differences in restaurant tipping: a literature review and discussion of practical implications. J. Foodservice Bus. Res. 9 (4), 99–113. Lynn, M., 2006b. Geodemographic differences in knowledge about the restaurant tipping norm. J. Appl. Soc. Psychol. 36 (3), 740–750. Lynn, M., 2009. Individual differences in self-attributed motives for tipping: antecedents, consequences, and implications. Int. J. Hospitality Manage. 28 (3), 432–438. Lynn, M., 2011. Race differences in tipping: testing the role of norm familiarity. Cornell Hospitality Q. 52 (1), 73–80.

25

Lynn, M., 2013. A comparison of Asians’, Hispanics’ and Whites’ restaurant tipping. J. Appl. Soc. Psychol. 43 (4), 834–839. Lynn, M., Zinkhan, G.M., Harris, J., 1993. Consumer tipping: a cross-country study. J. Consum. Res. 20 (3), 478–488. Lynn, M., McCall, M., 2000. Gratitude and gratuity: a meta-analysis of research on the service-tipping relationship. J. Socio-Econ. 29 (2), 203–214. Lynn, M., Thomas-Haysbert, C., 2003. Ethnic differences in tipping: evidence, explanations, and implications. J. Appl. Soc. Psychol. 33 (8), 1747–1772. Lynn, M., Withiam, G., 2008. Tipping and its alternatives: business considerations and directions for research. J. Serv. Marketing 22 (4), 328–336. Lynn, M., Sturman, M.J., 2010. Tipping and service quality: a within-subjects analysis. J. Tourism Hospitality Res. 34 (2), 269–275. Lynn, M., Kwortnik Jr., R.J., Sturman, M.J., 2011. Voluntary tipping and selective attraction and retention of service workers in the United States: an Application of the ASA model. Int. J. Hum. Res. Manage. 22 (9), 1887–1901. Lynn, M., Jabbour, P., Kim, W.G., 2012. Who uses tips as a reward for service and when? An examination of potential moderators of the service-tipping relationship. J. Econ. Psychol. 33 (1), 90–103. Lynn, M., Katz, B., 2013. Are Christian/religious people poor tippers? J. Appl. Soc. Psychol. 43 (5), 928–935. Lynn, M., Brewster, Z.W., 2015. Racial and ethnic differences in tipping: the role of perceived descriptive and injunctive tipping norms. Cornell Hospitality Q. 56 (1), 68–79. Margalioth, Y., 2006. The case against tipping. J. Labor Employment Law 117 (9), 1–30. Maynard, L.J., Mupandawana, M., 2009. Tipping behavior in Canadian restaurants. Int. J. Hospitality Manage. 28 (4), 597–603. McCall, M., Lynn, A., 2009. Restaurant servers’ perceptions of customer tipping intentions. Int. J. Hospitality Manage. 28 (4), 594–596. Medler-Liraz, H., 2014. Negative affectivity and tipping: the moderating role of emotional labor strategies and leader-member exchange. Int. J. Hospitality Manage. 36, 63–72. Noll, E.D., Arnold, S., 2004. Racial differences in restaurant tipping: evidence from the field. Cornell Hotel Restaurant Admin. Q. 45 (1), 23–29. Ogbonna, E., Harris, L.C., 2002. Institutionalization of tipping as a source of managerial control. Br. J. Ind. Relat. 40 (4), 725–752. Palmer, B., 2013. Tipping is an Abomination, (accessed November 21, 2013). Parrett, M., 2006. An analysis of the determinants of tipping behavior: a laboratory experiment and evidence from restaurant tipping. Southern Econ. J. 73 (2), 489–514. Peterson, K., 2013. 5 Reasons We Should Ban Tipping, http://money.msn.com/ now/post.aspx?post=54a75d58-55e8-44ed-9593-385604bfe455 (accessed November 21, 2013). Podsakoff, P.M., Mackenzie, S.B., Podsakoff, N.P., 2012. Sources of method bias in social science research and recommendations on how to control it. Ann. Rev. Psychol. 63, 539–569. Schafer, J.L., Graham, J.W., 2002. Missing data: our view of the state of the art. Psychol. Methods 7 (2), 147–177. Wang, L., 2014. At the tipping point: race and gender discrimination in a common economic transaction. Virginia J. Soc. Policy Law 21, 101–166. Wells, P., 2013. Leaving a tip: a custom in need of changing? N.Y. Times, September 4, A1.