When Are Fakers Also Drinkers? A Self-Control View of Emotional Labor and Alcohol Consumption Among U.S. Service Workers

Abstract

Some employees tend to drink more alcohol than other employees, with costs to personal and organizational well-being. Based on a self-control framework, we propose that emotional labor with customers—effortfully amplifying, faking and suppressing emotional expressions (i.e., surface acting)—predicts alcohol consumption, and that this relationship varies depending on job expectations for self-control (i.e., autonomy) and personal self-control traits (i.e., impulsivity). We test these predictions with data drawn from a national probability sample of U.S. workers, focusing on employees with daily contact with outsiders (N = 1,592). The alcohol outcomes included heavy drinking and drinking after work. Overall, surface acting was robustly related to heavy drinking, even after controlling for demographics, job demands and negative affectivity, consistent with an explanation of impaired self-control. Surface acting predicted drinking after-work only for employees with low self-control jobs or traits; this effect was exacerbated for those with service encounters (i.e., customers, the public) and buffered for those with service relationships (i.e., patients, students, clients). We discuss what these results mean for emotional labor and propose directions for helping the large segment of U.S. employees in public facing occupations.

Over a quarter of the U.S. workforce reports having an alcoholic drink within two hours of leaving work during the prior month and close to a third admit to binge drinking (Frone, 2019). When employees drink frequently or heavily their health is at risk and their sleep quality is poor (Rehm, Taylor, & Room, 2006; Reynolds et al., 2003; Roehrs & Roth, 2001). There are also potential business costs such as accidents and injuries, absenteeism, and lost productivity (Frone, 2019; Normand, Lempert, & O’Brien, 1994; Rehm et al., 2009), and excessive alcohol use cost U.S. society $249 billion in 2010 (Sacks, Gonzales, Bouchery, Tomedi, & Brewer, 2015). Therefore, identifying the factors that determine such problematic drinking behavior has serious implications for employee and organizational well-being, as well as society in general.

Employees in occupations that require frequently interacting with the public (i.e., customers, patients, students) with positive expressions (Hochschild, 1983) are some of the heaviest drinkers (e.g., food servers, sales, Frone, 2013; Mandell, Eaton, Anthony, & Garrison, 1992). One common explanation for excessive drinking in these occupations is that these jobs involve emotionally demanding and stressful interactions (Dormann & Zapf, 2004; Grandey, Kern, & Frone, 2007), and drinking is a maladaptive response to stress. Although emotional demands and interpersonal stressors are related to drinking, the relationships are indirect through affective states and only under certain conditions (Frone, 2008; Frone, 2019; Hight & Park, 2018; Liu, Wang, Zhan, & Shi, 2009; Shepherd, Fritz, Hammer, Guros, & Meier, 2018).

A second plausible explanation for heavier drinking among employees in service occupations has yet to be empirically tested: reduced self-control due to emotional labor (Grandey & Gabriel, 2015). Specifically, the more that service employees suppress annoyance and act concerned or enthusiastic with customers, also called surface acting, the more they are exerting self-control over their natural feelings. Across studies, exerting self-control tends to impair later self-control, though there is some debate over whether this is due to reduced capacity for self-control or shifting motivations towards more enjoyable behavior (Baumeister, Tice, & Vohs, 2018; Inzlicht, Schmeichel, & Macrae, 2014; Lian, Yam, Ferris, & Brown, 2017). Our purpose is not to test specific mechanistic views about why self-control impairment occurs, but instead to use a self-control lens to predict that (a) employees who chronically surface act at work are less likely to regulate their alcohol consumption, and (b) personal and work-related self-control factors determine when this is more or less likely to occur.

Overall, our paper offers conceptual, empirical, and practical contributions. First, we answer the call to expand the study of emotional labor beyond work attitudes and burnout to include health and home outcomes (Grandey & Gabriel, 2015; Hülsheger & Schewe, 2011). We assess whether and when emotional labor predicts a behavior with serious health, societal and organizational implications—alcohol consumption. We adopt a self-control perspective to propose that surface acting with customers is positively associated with alcohol consumption, while also identifying mitigating factors. Specifically, a self-control lens leads us to propose that surface acting is not as strongly linked to alcohol consumption when the work context permits employees to decide how and when to regulate (work autonomy), and when employees have natural self-control tendencies (low trait impulsivity) (Moller, Deci, & Ryan, 2006; Muraven & Slessareva, 2007; Tangney, Baumeister, & Boone, 2004).

Second, we empirically take into account other factors that might create a spurious relationship between surface acting and drinking, such as job demands (Hight & Park, 2018; Liu et al., 2009; Shepherd et al., 2018) and trait negative affect (Semmer, Messerli, & Tschan, 2016). For example, Shepherd et al. (2018) reasoned that expended self-control explained why emotional job demands were associated with drinking in their study, but they did not direct test this nor did they control for employee trait negativity. By demonstrating whether surface acting contributes to drinking behavior beyond stress-related factors, we are able to more confidently conclude that the relationship is due to self-control impairment. Furthermore, we test our question with a large probability sample of over 1500 U.S. employees who interact with customers, clients, patients, students, and the general public. This approach provides the statistical power to test our interaction predictions and is more generalizable than emotional labor research that is often limited to a single occupation (e.g., bus drivers, hospital workers, prison officers, Grandey, Foo, Groth, & Goodwin, 2012; Shepherd et al., 2018; Wagner, Barnes, & Scott, 2014). Overall, our empirical approach permits us to be more confident about inferences that can be made to the population of public-facing employees.

Finally, our inquiry has practical value for the occupational health of employees in service jobs, who comprise a large segment of today’s workers (Bureau of Labor Statistics, 2014). Excessive drinking is linked to absenteeism, lost earnings, reduced health, and increased health care costs (Sacks et al., 2015; Frone, 2019). Identifying whether and when emotional labor results in excessive drinking illuminates avenues for practices and interventions that are distinct from the prior focus on job stressors (Gabriel, Cheshin, Moran, & Van Kleef, 2016; Grandey & Diamond, 2010). Thus, our research findings will allow managers to identify (and minimize) factors that exacerbate serious costs for service workers and organizations.

Emotional Labor and Alcohol Use: A Self Control Lens

According to self-control theories and evidence, people who effortfully exert self-control at one time are more likely to show self-control failure later (Baumeister, Vohs, & Tice, 2007; Evans, Boggero, & Segerstrom, 2016; Hagger, Wood, Stiff, & Chatzisarantis, 2009; Inzlicht et al., 2014). There has been debate over the precise explanation for this self-control impairment, specifically whether impairment is due to “cutting back of exertion to conserve its remaining energy” (i.e., limited resource model, Baumeister et al., 2018, p. 142), or that “initial control leads to shifts in motivation away from ‘have-to’ goals and toward ‘want-to’ goals” (i.e., motivational shift model, Inzlicht et al., 2014, p. 131). Comparing these mechanisms effectively requires lab-based or momentary data with successive self-control tasks and measurement of intervening mechanisms—although self-control theorizing is also used to explain person-level differences in unhealthy or deviant behavior (e.g., Yam, Fehr, Keng-Highberger, Klotz, & Reynolds, 2016).

For our purposes, we draw on the self-control framework to propose employees who expend more self-control effort at work (i.e., surface acting) are less likely to exert self-control over behavior in another domain (i.e., alcohol use), while controlling for other personal and work factors that could explain the relationship. To develop our predictions (see Figure 1), we describe both the limited resource (Baumeister et al., 2007) and motivation shift (Inzlicht et al., 2014) models of self-control impairment, which lead to similar propositions at the employee level. We do not test the process or try to tease apart the specific mechanisms, but instead seek to learn whether surface acting differentially relates to drinking behavior depending on self-control conditions. Both of these models point to similar factors likely to reduce the self-control impairment: work autonomy (i.e., the extent that one’s work behaviors feel self-controlled) and trait impulsivity (i.e., tendency to act without thinking).

Figure 1:

Proposed self-control model of alcohol consumption by service workers

Surface Acting as Self-Control Effort

Emotional labor is frequently studied through the lens of self-control. A recent review has identified emotional labor as one of the most frequently studied predictors of regulatory depletion (Lian et al., 2017), and ego depletion theory is one of the most commonly studied explanations for emotional labor’s detrimental effects (Grandey & Gabriel, 2015). A vast majority of this work focuses on surface acting, or amplifying, faking or suppressing felt emotions in order to appear positively and evoke customer satisfaction and return business (Grandey, 2000; Grandey & Gabriel, 2015; Hochschild, 1983).

Consistent with the definition of self-control, surface acting involves inhibiting behavioral impulses to conform to rules and meet long-term goals (Tice & Bratslavsky, 2000), which in this case is the emotional expectations of service work (i.e., service with a smile) (Diefendorff & Gosserand, 2003). The extent of surface acting an individual must engage in, however, varies due to both job and personal factors. Some occupations have higher display expectations and lower familiarity with customers, such that more regulatory effort is required (Brotheridge & Grandey, 2002; Gabriel, Diamond, & Grandey, 2015). Surface acting is also more frequent in jobs with more rude or distressing interactions (Sliter, Jex, Wolford, & McInnerney, 2010), and by employees who tend to feel more negatively in general (Kammeyer-Mueller et al., 2013), in order to override the impulse to aggress toward or avoid customers (Lian et al., 2014; Wang, Liao, Zhan, & Shi, 2011). Consistent with the definition of self-control, surface acting involves inhibiting behavioral impulses to conform to rules and meet long-term goals (Tice & Bratslavsky, 2000), which in this case is the emotional expectations of service work (i.e., service with a smile) (Diefendorff & Gosserand, 2003).

Surface Acting and after-work Alcohol Use

Given the inhibition inherent to surface acting, employees who perform more surface acting are exerting more self-control effort at work. Self-control is also needed to resist or limit one’s alcohol consumption (Muraven & Shmueli, 2006), especially given the many opportunities for alcohol consumption available to employees (i.e., happy hours, passing bars on the commute, liquor cabinet at home), or to drink in moderation due to the pleasurable initial experiences of alcohol (Baumeister, Heatherton, & Tice, 1994). Both lab and field studies have shown that exerting more self-control results in consuming more alcohol in a “taste test” (Muraven, Collins, & Nienhaus, 2002) or violating one’s self-imposed drinking limits later that day (Muraven, Collins, Shiffman, & Paty, 2005). These effects can be explained by reduced self-control capacity such that willpower is weak (Baumeister et al., 2007) and reduced motivation to suppress impulses and readiness to engage in more pleasurable activities (Inzlicht et al., 2014).

More specific to our purposes, exerting self-control over emotional behavior and impulses is linked to later self-control impairment. In lab studies, suppressing or amplifying an emotional display impairs self-control on a variety of later tasks (Demaree, Schmeichel, Robinson, & Everhart, 2004; Hagger, Wood, Stiff, & Chatzisarantis, 2010; Richards & Gross, 2000; Schmeichel, Vohs, & Baumeister, 2003; Zyphur, Warren, Landis, & Thoresen, 2007). Organizational field evidence supports that surface acting is linked to more deviant behavior toward others, suggesting less self-control while at work (Chi & Grandey, 2016; Deng, Walter, Lam, & Zhao, 2016; Yam et al., 2016). We extend these ideas to propose that employees who more frequently surface act while at work engage in less constrained drinking behavior, i.e., heavy or excessive drinking, and increased drinking after work. Both behaviors are consistent with the rationale that they lack the regulatory capacity to control drinking away from work (Baumeister et al., 2007) and that they shift motivations towards these more desirable behaviors (Inzlicht et al., 2014).

To more clearly conclude that the surface acting-drinking relationship is due to self-control, we test whether this effect exists beyond emotional job demands. Employees with more emotional job demands—i.e., difficult or distressing events at work—will need to engage in more surface acting to maintain appropriate emotions in the wake of emotionally demanding interactions, and may also consume more alcohol to cope with the job stress (Cooper, Frone, Russell, & Mudar, 1995; Liu et al., 2009; Shepherd et al., 2018; Sliter et al., 2010). Therefore, it is important to minimize the possibility that any association between surface acting and drinking is spuriously due to a shared relationship with felt stress (Semmer et al., 2016). Thus, we predict:

Hypothesis 1: Beyond emotional job demands and other known covariates, employees who more frequently perform surface acting with customers consume more alcohol than those who surface act less.

Job Self-Control: Work Autonomy

Not all employees in service jobs engage in excessive drinking after work; self-control impairment can be mitigated by the extent of autonomy around controlling those impulses. Both the limited resource (Baumeister et al., 2007) and motivation shift (Inzlicht et al., 2014) models lead us to predict that autonomy at work can mitigate the likelihood of drinking afterward. Employees in service jobs differ in the extent that their behaviors feel autonomous, or freely chosen, initiated, and sustained (Deci & Ryan, 1985). In low autonomy jobs (e.g., fast food cashier), service behaviors may be explicitly required, scripted, monitored and evaluated, whereas in high autonomy jobs (e.g., sales, teachers), employees have discretion with how they act, based on interpersonal or professional norms (Bolton & Boyd, 2003; Goldberg & Grandey, 2007; Grandey, Fisk, & Steiner, 2005a).

From a limited resource perspective, “autonomously motivated self-control is less depleting because it is more energizing and vitalizing than self-control that feels forced upon the person” (Muraven, Gagne, & Rosman, 2008, p. 769). A series of lab studies found that feeling like one has autonomy to perform self-control in a first task improves one’s self-control in the second task, compared to impaired self-control when lacking autonomy (Moller et al., 2006). Autonomy also permits one to take breaks as needed, allowing for resource replenishment (Grandey et al., 2012; Trougakos, Hideg, Cheng, & Beal, 2014). Thus, surface acting in a high autonomy job should be less likely to reduce regulatory capacity compared to a low autonomy job. From a motivation shift perspective (Inzlicht et al., 2014), exerting self-control over behavior in autonomous conditions is more satisfying compared to the same behaviors in coercive conditions (Gagné & Deci, 2005; Trougakos, et al., 2014). Thus, surface acting when feeling autonomous is less likely to motivate a shift to a more gratifying, yet potentially costly, behavior after work compared to when that surface acting feels coerced.

Consistent with these arguments, prior research shows that work autonomy buffers the effect of surface acting on exhaustion, job satisfaction, and performance (Christoforou & Ashforth, 2015; Goldberg & Grandey, 2007; Grandey, Chi, & Diamond, 2013; Grandey, Fisk, & Steiner, 2005b; Johnson & Spector, 2007; Wharton & Erickson, 1993). We extend this evidence to see whether autonomy moderates the association of surface acting with drinking. We expect:

Hypothesis 2: The positive relationship between surface acting and alcohol consumption is weaker for employees who have more work autonomy compared to those with less autonomy.

Employee Self-Control: Trait Impulsivity

Another mitigating factor that may explain why all service workers do not drink heavily is the employees’ personal tendency to control their behavior for long-term benefits. Trait impulsivity represents the extent to which people either act on impulses without thinking about consequences and costs or control impulses and think things through (Baumeister et al., 1994; Muraven & Baumeister, 2000; Tangney et al., 2004). We expect that the relationship between surface acting and drinking will be stronger for highly impulsive employees than less impulsive employees.

From a limited resources perspective, employees with high trait impulsivity have less capacity for self-control than those with low trait impulsivity, who act with self-control more automatically. As such, the self-control required from surface acting at work is more likely to deplete resources for highly impulsive employees, leaving them less able to resist drinking after work than less impulsive employees for whom such control is more automatic (Baumeister et al., 2018). From a motivational shift perspective, less impulsive employees find self-control more personally congruent and rewarding than impulsive people, such that after work they are less motivated to seek out other forms of gratification such as alcohol (Hofmann, Friese, & Wiers, 2008; Inzlicht et al., 2014). Overall, more impulsive employees who tend to surface act are likely to stop exerting self-control and “give in” to the temptation of more gratifying – though costly – behavior (e.g., staying late at the happy hour or cracking another beer at home) compared to less impulsive employees.

Evidence supports the moderating role of trait self-control on surface acting impairing later work behavior. A field study found the relationship of surface acting with deviant behavior (i.e., aggression) was weaker among those with more trait self-control (Yam et al., 2016). In a service context, the link between surface acting and deviant work behaviors was buffered for conscientious individuals, conceptualized as a trait measure of self-control (Chi & Grandey, 2016). Extending this to employees’ health-related behavior after work, we expect that:

Hypothesis 3: The positive relationship between surface acting and alcohol consumption is stronger for employees with higher trait impulsivity compared to those who are less impulsive.

Method

Sample and Study Design

Our focal sample was 1,592 participants that came from a larger study of 2,975 U.S. workers who took part in a random telephone survey, called the National Survey of Work Stress and Health (NSWSH). The population from which the study participants were sampled was all non-institutionalized adults, ages 18–65 years, who were employed in the civilian labor force, and who resided in households in the 48 U.S. contiguous states and the District of Columbia. A group of 29 extensively trained interviewers collected data from December 2008 to April 2011 using computer-assisted telephone interviewing (CATI). In households with more than one eligible individual, the next birthday method was used to select at random one individual for participation in the study (Potthoff, 1994). Of all eligible individuals, 47% participated in the NSWSH. On average, the interview lasted 55 minutes and participants were paid $25.00 (USD).

Sampling weights.

For all analyses, the participants were weighted, following standard procedures for sample survey data, in order to improve generalizability to the target population (Korn & Graubard, 1999; Levy & Lemeshow, 1999). The sampling weights account for differences in the initial selection probabilities for the reached telephone number, the number of different telephone lines through which the household could be reached, and the number of eligible adults in the household. The weights further adjust for differential nonresponse and were post-stratified to population totals (joint distribution of gender, race/ethnicity, age, and region of the country) obtained from the Current Population Survey (Bowler & Morisi, 2006) for the months during which the study was in the field (December 2008 to April 2011). Post-stratification adjusts for known differences between the sample and population on key variables that may result from sampling error, undercoverage, or nonresponse.

Study inclusion criteria.

Our focal sample represented respondents who were wage and salary workers (owner/operators were dropped) and met the criteria of working in an emotional labor occupation, in order to hold constant performance goals for all participants (Diefendorff, Richard, & Croyle, 2006; Hochschild, 1983). Emotional labor occupations have frequent contact with the public (i.e., outsiders to the organization) and are expected to manage emotions to conform to certain displays with the public. We identified respondents who reported daily contact with targets outside the organization (i.e., customers, clients, patients, students, or the general public), which was 54% of the nationally representative sample¹. The daily contact sample (N = 1, 592) reported greater emotional job demands and more surface acting than employees with limited contact (“weekly”, “monthly”, or “less than monthly”; N = 399) or no contact (“never”, N = 541; ps < .01).² See Table 1 (left side) for descriptive statistics.

Table 1.

Means and Standard Deviations by Customer Contact for Emotional Labor and Drinking

Frequency of Service Interactions N	No Contact 541	Limited Contact 399	Daily Contact 1,592	Daily: Encounters (customers, general public) 749 (52.6%)	Daily: Relationships (clients, patients, and students) 843 (47.4%)
Job Title Examples	Computer analyst, Software developers, Laboratory technician, Cooks, Janitors, Housecleaners, Accountants, Assembly line workers	Managers, HR specialists, Software developers, Lawyers, Legal support, Grounds maintenance, Truck drivers	(see columns to the right)	Barista, Cashier, Customer service representative, Sales Associate, Pharmacist, Restaurant server; Bus driver, Custodian, Police officer, Security guard	Banker, Massage therapist, Paralegal, Social worker, Nurse, Dental hygienist, Physician, Child care worker, Coach, Teacher, School nurse, Yoga instructor
Emotional Job Demands	1.42 (1.07)a	1.55 (.69)a	2.02 (1.09) b	1.85 (1.10)a	2.19 (1.04) b
Surface Acting	--	1.64 (.94)a	1.97 (.98)b	2.09 (1.03)a	1.83 (.90)b
After Work drinking	.77a (1.19)a	1.05 (1.19)b	1.00 (1.28) b	1.10 (1.40)a	.89 (1.12)b
Heavy drinking	.45 (.81)	.43 (.65)	.47 (.79)	.56 (.85)a	.37 (.70)b

Note: Job examples are titles that were represented multiple times within each category. Unweighted sample sizes are presented. Percentages, means, and standard deviations are based on weighted data. Responses were on a 0 (never) to 4 (always) scale for emotion regulation and a 0 (never) to 5 (6 to 7 days a week) scale for the alcohol use variables.

¹Means within a row with different superscripts for columns 2 – 4 are significantly different (p < .05).

²Means within a row with different superscripts for columns 5 and 6 are significantly different (p < .05). Surface acting was not gathered for those with no customer contact.

Sample characteristics.

We describe our respondent (i.e., population) characteristics with weighted means and percentages. Our focal sample represented a wide range of front-line employees (see Table 1) based on the types of outsiders they reported interacting with: customers (i.e., customer service; 42.9%), clients (i.e., professional service; 19.3%), students (i.e., education; 17.4%), the general public (i.e., public service; 9.7%), and patients (i.e., health care; 10.7%). The sample was slightly more likely to be female (52%) than in the U.S. civilian workforce (47%, www.bls.gov). In terms of racial/ethnic background, 72.3% were White, 12.4% were Black, 7.4% were Hispanic, and 7.9% were of other racial/ethnic makeup. The average age of the participants was 41 years (SD = 12.87); they worked an average of 41 hours per week (SD = 12.37), in the same job for an average of 6 years (SD = 6.94), with a median personal income of $37,865 (SD = $116,546.72). The distribution of highest level of education was: no high school diploma (4.8%); high school diploma or GED (18.2%); trade, technical, or vocational training (2.8%); some college (18.7%); Associate’s degree (9.4%); Bachelor’s degree (23.7%); some graduate school (2.9%); Master’s degree (13.9%); and a doctoral level degree (5.6%).

Measures

Table 2 shows means, standard deviations, reliabilities, and correlations for the entire sample. The time frame of the past 12 months was used for the job measures and drinking to ensure we were capturing current working conditions rather than general trait-like tendencies.

Table 2.

Means, Standard Deviations, and Correlations Between Study Variables Among Daily Contact Employees

Variables	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Gender	.48	.50	—
2. Race	.28	.45	−.01	—
3. Age	40.75	12.87	−.04	−.06	—
4. Education	5.95	2.29	−.01	−.10	.25	—
5. Income	50675	116547	.12	−.04	.08	.18	—
6. Negative affectivity	1.67	.59	.00	−.08	−.09	−.03	.02	(.84)
7. Trait impulsivity	1.82	.69	.04	−.10	−.11	−.17	−.05	.49	(.86)
8. Emotional job demands	2.02	1.09	−.02	−.02	.12	.20	.10	.27	.13	(.78)
9. Type of service interactions	.47	.50	−.25	−.06	.17	.40	.08	−.03	−.08	.16	—
10. Work autonomy	3.53	.61	.13	−.02	.10	.07	.09	−.14	−.06	−.03	.03	(.85)
11. Surface acting	1.97	.98	−.03	.06	−.18	−.12	−.08	.16	.17	.15	−.14	−.10	(.74)
12. After work drinking	1.00	1.28	.21	−.14	−.09	.07	.08	.11	.13	.03	−.08	.03	.06	(.77)
13. Heavy drinking	.47	.79	.19	−.07	−.33	−.15	.01	.13	.19	−.02	−.13	.00	.16	.58	(.80)

Note. N = 1,592. Values are based on weighted data. Coefficient alpha is presented on the diagonal. Correlations with absolute values greater than .05 are significant at p < .05. Gender: Female = 0, Male = 1; Race: White = 0, Minority = 1; Type of service interactions = Customers or general public = 0; Clients, patients, or students = 1.

Surface acting.

We created a brief five-item measure that captured the full range of surface acting strategies: amplifying, faking, and suppressing emotions while interacting with others (Grandey, 2000)³. After reporting the target with whom they have the most direct contact (e.g., patient, customer), respondents were asked “While interacting with [that target], how often do you…” [0 (never) to 4 (always)] with two items about amplifying positive and caring emotions, two for faking positive and caring emotions, and one for suppressing negative emotions (α = .74; see Appendix).

Alcohol consumption.

We measured after work drinking to capture drinking that happens directly after one is done interacting with customers. We used an established two-item measure (Frone, 2008) that asks for: (1) frequency during the past 12 months [0 (never) to 5 (6 to 7 days a week)] of commencing drinking within two hours of leaving work and (2) the typical number of drinks consumed when drinking after work. These items were then combined to represent the extent of after work drinking (α = .77; M = 1.00, SD = 1.28). Our other measure, heavy drinking, indicates impaired regulation of drinking and likely indicates drinking done after work (Frone, 2015). Respondents rated the frequency in the past 12 months of [0 = never to 5 = 6 to 7 days a week] of two drinking behaviors: (1) drinking four (if female)/five (if male) or more drinks in two hours and (2) drinking to intoxication (α = .80; M = .47, SD = .79).

Work autonomy.

Four items adapted from the Work Design Questionnaire (Morgeson & Humphrey, 2006) assessed autonomy over work behavior ranging from (1) strongly disagree to (4) strongly agree. Example items include “I can use personal initiative or judgment in carrying out my work” and “I can decide on my own how to go about doing my work” (α = .85).

Trait impulsivity.

To assess trait self-control tendencies, respondents were asked to rate their agreement [1(strongly disagree) to 4 (strongly agree)] with six items (Magid & Colder, 2007). Example items are “I have trouble controlling my impulses” and “When I’m upset I often act without thinking” (α = .86).

Job covariates.

With our inclusion criteria, we objectively constrained the included jobs to have high frequency of customer contact and positive display occupational expectations. We statistically control for emotional job demands, which might spuriously create a relationship between surface acting and drinking by increasing felt negative affect (Semmer et al., 2016). We used a three item composite (Frone, 2015) that is similar to an existing scale (Kristensen, Hannerz, Høgh, & Borg, 2005): In the past year, how often “was your work emotionally demanding?”, “did your job put you in emotionally unpleasant or disturbing situations?”; and “was the work you do emotionally unpleasant or disturbing” (α = .78). Response anchors ranged from 0 (never) to 4 (everyday). Furthermore, emotional labor occupations can be distinguished into two main types – those with daily interactions with clients, patients or students tend to have ongoing service relationships and those who interact with customers or the public tend to have one-time service encounters (Gutek, Bhappu, Liao-Troth, & Cherry, 1999). We control for the type of service interactions, which is linked to surface acting and drinking via differences in job stress and satisfaction (Bakker & Demerouti, 2007; Erickson & Stacey, 2013; Wang & Groth, 2014). Employees who worked with clients, students, and patients (typically repeated) were coded 1 (N = 749), whereas those who interact with customers or the general public (typically anonymous, one-time) were coded 0 (N = 843).

Personal covariates.

We also controlled for individual factors linked to both surface acting and excessive drinking, to ensure any found relationship is beyond those effects. Trait negative affectivity represents the extent that one tends to feel stressed, and is found to increase likelihood of alcohol use (Frone, 2013; Kuntsche, Knibbe, Gmel, & Engels, 2006) as well as surface acting (Kammeyer-Mueller et al., 2013). This construct was assessed with seven items (Denollet, 2005) on response anchors ranging from 1 (strongly disagree) to 4 (strongly agree) (e.g., “I take a gloomy view of things” and “I am often in a bad mood” (α = .84). Based on prior evidence for demographic relationships with surface acting (Johnson & Spector, 2007; Maneotis, Grandey, & Krauss, 2014) and/or alcohol use (Bacharach, Bamberger, & Sonnenstuhl, 2002; Frone, 2013), we control for gender (0 = women, 1 = men), race (0 = White, 1 = minority), age (in years), education (1 = less than high school to 10 = doctoral level degree) and income (in U.S. dollars).

Results

Descriptives by Group

Our data offers a unique description of the emotional labor and drinking behaviors of a representative sample of U.S. workers (see Table 1, N = 2,532). Similar to an earlier occupational comparison (Mandell et al., 1992), we found that after work drinking was more frequent for those with daily customer contact (M = 1.00, SD = 1.28) compared to those with none (M = .77, SD = 1.19; p = .01), yet not different from those with limited contact (M = 1.05, SD = 1.19, p < .48)⁴. More specifically, after work drinking (M_relationship = .89, SD = 1.12; M_encoumter = 1.10, SD = 1.40, p < .001) and heavy drinking (M_relationship = .37, SD = .70; M_encoumter = .56, SD = .85, p < .001) is greater for those who interact in service encounters (i.e., customers and general public). Furthermore, though emotional job demands are higher in service relationships (N = 843; M = 2.19, SD = 1.04) compared to service encounters (N = 749; M = 1.85, SD = 1.10, p < .001); surface acting – like drinking – is higher in encounters than relationships (M_relationship = 1.83, SD = .90; M_encoumter = 2.09, SD = 1.03; p < .001). Overall, this replicates and extends prior occupational differences in emotional labor (Brotheridge & Grandey, 2002), and is consistent with the idea that drinking is linked to surface acting rather than the occupational demands⁵. We turn to regression analyses to further test our predictions.

Hypothesis Testing

Following conventions for analyzing sample survey data (Lehtonen & Pahkinen, 2004), all linear regression analyses described below employed the sampling weights and robust standard errors based on Taylor linearization. Tests of overall model fit and increments in fit for the regression analyses were based on adjusted Wald F-tests. Hierarchical moderator regression analyses (Aiken & West, 1991) were conducted with mean-centered variables. First, the covariates were entered on Step 1, followed by surface acting on Step 2 (Hypothesis1), followed by the surface acting x impulsivity and surface acting x work autonomy interactions on Step 3 (Hypotheses 2 and 3). For significant interactions, the conditional relations of surface acting to alcohol consumption were estimated and plotted for low (1 SD below the mean) and high (1 SD above the mean) levels of the moderator variable. Results of the hierarchical regression analyses are reported in Table 3.

Table 3.

Results from Moderated Hierarchical Regression Predicting Alcohol Consumption (weighted)

	After Work Drinking		Heavy Drinking
	b	se	b	se
Step 1
Gender	.44***	.091	.26***	.061
Race	−.33**	.118	−.13	.075
Age	−.01**	.004	−.02***	.003
Education	.07***	.021	−.02	.013
Personal income	.005	.003	.001	.001
Trait negative affectivity	.10	.098	.059	.069
Type of service interactions	−.19*	.085	−.02	.062
Trait impulsivity (IMP)	.17	.096	.13*	.058
Work autonomy (AUT)	.04	.079	.03	.049
Emotional job demands	.00	.045	.01	.026
R2	.100***		.175***
Step 2
Surface Acting	.06	.049	.07*	.034
ΔR2	.002		.007*
Step 3
Surface Acting x IMP	.21**	.071	.08	.049
Surface Acting x AUT	−.16**	.062	−.08	.043
ΔR2	.019***		.009**

Note: N = 1,592. Values are unstandardized regression coefficients. To avoid small coefficients, income was rescaled so that a unit increase equals $10,000. Gender: female = 0, male = 1; race: White = 0, minority = 1; Type of service interactions: customers or general public = 0; clients, patients, or students = 1.

^*p < .05;

^**p < .01;

^***p < .001.

Surface acting and alcohol consumption.

In the first step of our regression analyses, the personal (i.e., gender, race, age, income, education, negative affectivity) and job (i.e., emotional job demands, type of service interactions) covariates, along with trait impulsivity and work autonomy, explain a significant amount of variance in after work drinking (ΔR² = .10, p < .001) and heavy drinking (ΔR² = .18, p < .001). Hypothesis 1 proposed that surface acting predicts after work alcohol consumption beyond emotional job demands and these covariates. As shown in Table 3, surface acting did not significantly predict after work drinking (b = .06, p = .21, ΔR² = .002), but did predict heavy drinking (b = .07, p < .05, ΔR² = .007), which partially supports Hypothesis 1.

Moderators of surface acting and alcohol consumption.

Adding the interaction terms of surface acting with both autonomy and impulsivity increased the variance explained in after work drinking (ΔR²= .019, p < .001) and heavy drinking (ΔR²= .009, p < .01). Hypothesis 2 states that high work autonomy weakens a positive relationship between surface acting and alcohol use compared to low work autonomy. The results in Table 3 show that the surface acting x work autonomy interaction is significant for after work alcohol use (b = −.16, p = .01). Figure 2 shows that the conditional relation of surface acting and after work alcohol use was significant and positive at low levels of work autonomy (b = .17, p < .001), and non-significant when work autonomy was high (b = −.02, n.s.), consistent with predictions. The interaction term did not significantly predict heavy alcohol use, but the effect was in the same direction (b = −.08, p = .056). These results partially support Hypothesis 2.

Figure 2:

Graph of the Interaction of Surface Acting and Work Autonomy on After Work Drinking

Hypothesis 3 states that trait impulsivity moderates the relationship between surface acting and alcohol consumption such that the relationship is stronger for employees who tend to be more impulsive than less so. The results in Table 3 show that the surface acting x trait impulsivity interaction was significant for after work alcohol use (b = .21, p < .01). Figure 3 shows that the conditional relationship of surface acting to after work alcohol consumption at high levels of trait impulsivity was significant and positive (b = .22, p < .01), and was nonsignificant at low levels of trait impulsivity (b = −.07, n.s.), as expected. The same steps were taken for heavy alcohol use; the interaction term did not have a significant coefficient (b = .08, p = .12). Overall, these results partially support Hypothesis 3.

Figure 3:

Graph of the Interaction of Surface Acting and Trait Impulsivity on After Work Drinking

Post-Hoc Analyses

We note that there may be a concern that our including a large set of covariates might distort the observed relationships or make things appear significant due to collinearity (Spector & Brannick, 2011). We re-ran analyses with and without the personal and work covariates, and our findings and conclusions remain the same. We report results with the covariates included to be more conservative in our conclusions about the contribution of surface acting on drinking.

We conducted additional exploratory tests to assess the robustness of our findings, and to further understand the results as evidence for self-control as the explanation.

Emotional job demands and job stress explanation.

Our reasoning for the two proposed moderators is that they mitigate whether self-control exertion (i.e., surface acting) is linked to drinking, but we do not directly test that mechanism. To more confidently conclude that autonomy and impulsivity inform us about when a self-control explanation, we also test them as moderating the relationship of emotional job demands and drinking. Based on job stress models, emotional job demands would predict more drinking when there is low job autonomy than high (Goldberg & Grandey, 2007; Bakker & Demerouti, 2006; Karasek, 1979; Shepherd, et al., 2018), and a tendency to react to stress with impulsive behavior rather than more self-control (Tangney et al., 2004).

We reran our regression analyses adding the two-way interactions of emotional job demands with autonomy and trait impulsivity along with our predicted two-way interactions. Neither of the two-way interactions involving emotional job demands predicted after work drinking or heavy drinking beyond the predicted interactions, nor were the associations involving the predicted interactions changed. Thus, trait impulsivity and work autonomy uniquely moderate the association of surface acting, but not emotional job demands, to after work drinking, consistent with self-control theorizing.

At-work drinking versus after-work drinking.

In our theoretical reasoning and analysis, we focus on emotional labor at work predicting drinking occurring after work. However, given surface acting is linked to deviant behavior while at work (Yam et al., 2016), it is possible we would find similar results for at-work drinking. We tested whether surface acting predicted a 2-item composite of the frequency and the amount of drinking alcohol during the workday (α = .92). Consistent with prior work, employee gender (males) significantly predicted drinking while working, but surface acting was unrelated (b = .01, se = .02, p =.46), and neither interaction term with impulsivity (b = .03, se = .03, p =.29) nor autonomy (b = −.01, se = .03, p =.72) was significant.

We note that the proportion of individuals reporting any form of workday alcohol use was substantially smaller (5.9%) than those reporting after work alcohol use (48.5%) and heavy alcohol use (39.2%). Given range restriction in the behavior, we have low statistical power to find interaction effects. Moreover, this evidence fits with a self-control explanation: most employees have less opportunity to drink and stronger motivation to resist (i.e., due to professional consequences) at work, compared to after work, so they override self-control depletion to resist drinking until later (Muraven et al., 2006.

Three-way interaction with job and trait self-control.

It is possible that our two-way interactions are conditioned by a three-way interaction that produces a “perfect storm”: frequent surface acting in low autonomy jobs (i.e., feel controlled) might be more effortful and unpleasant for those who are more impulsive (i.e., lack self-control) compared to other combinations. However, after adding the lower-order two-way interaction (impulsivity x autonomy) to the equation, the three-way interaction was not significant for after work drinking (b = −.02, p = .82) or heavy drinking (b = −.05, p = .36). Thus, the moderating effects of impulsivity and autonomy do not depend on each other, such that either trait or job self-control alone conditions the effect of surface acting on drinking.

Occupational service encounter/relationship as moderator.

In our main analyses, we controlled for service relationships (versus encounters); however, service relationships may also moderate whether surface acting at work impairs self-control over drinking behavior.⁶ An ongoing service relationship is more satisfying to both parties compared to one-time encounters with strangers (Gutek et al., 1999; Wang & Groth, 2014), such that self-control is more rewarding and later impairment might be less likely (Muraven, Shmueli, & Burkley, 2006).

We explored this possibility by first adding the two-way interaction between surface acting and service relationship to the regression, which was unrelated to both heavy drinking (b = −.10, p = .09) and after work drinking (b = −.04, p = .63). Then, we explored whether service relationships conditioned the hypothesized two-way interactions of surface acting with autonomy and impulsivity. After adding the lower-order two-way interactions (service relationships with autonomy and impulsivity), the two three-way interactions contributed significant variance explained in after work drinking (R² = .12, ΔR²= .01, p < .01) and heavy drinking (R² = .20, ΔR²= .01, p < .05). This seemed to be driven by the interaction with impulsivity: the three way interaction term of surface acting with autonomy and service relationships was not significant for either outcome (after work drinking; b = .15, SE = .11, p = .18; heavy drinking; b = .04, SE = .08, p = .59) but the three-way interaction with impulsivity was significant for both outcomes (after work drinking; b = −.31, SE = .12, p = .01; heavy drinking; b = −.18, SE = .08, p = .03). Simple slopes shown in Figures 4a and 4b reveal that among those with both high impulsivity and working in service encounters, surface acting is associated with after work drinking (b = .31, p = .002) and heavy drinking (b = .21, p = .005), but not for employees. We discuss this below.

Figure 4. Graph of Exploratory Three-way interaction of Surface Acting, Trait Impulsivity and Service Relationship on Both Drinking Behaviors.

Discussion

Our primary aim was to understand when employees in service occupations – those who have daily contact with customers and offer “service with a smile”– tend to drink more than others, and whether self-control over emotions contributes to this harmful outcome. Consistent with alcohol research that focuses attention on higher risk drinkers (Bacharach et al., 2002; Muraven et al., 2005), we focused on service workers, who tend to have higher average rates of drinking, to understand the personal and work factors that explain the variability in drinking in this group of workers. Using a large representative sample of the employed U.S. population, we found that employees with daily contact with customers (including patients, students, clients, and the general public) comprised over half the workforce. This suggests that our results are relevant to a large segment of the U.S. working population.

We adopted a self-control lens to argue that the more regulatory exertion over emotions in the work domain, the less self-control over drinking after work (Baumeister et al., 2007). We recognize that there are competing explanations for why self-control impairment occurs, with the limited resource and motivational shift models being the dominant approaches (Baumeister et al., 2018; Inzlicht et al., 2014). Our cross-sectional method also cannot truly test the processes proposed by these theories, though we controlled for confounds and conducted post hoc analyses to rule out other possible explanations. Our focus was not to tease these approaches apart, but to use self-control to identify likely explanations and theoretically relevant moderators.

Research Findings and Implications

First, we compared our focal sample of employees with daily customer contact to employees with limited or no customer contact, and employees whose daily contact was in service encounters (i.e., customers and general public) versus service relationships (i.e., students, patients, clients). Consistent with prior studies (Brotheridge & Grandey, 2002; Mandell et al., 1992), our nationally-representative sample supported that prototypical emotional labor occupations (i.e., daily contact in low autonomy jobs or service encounters) performed the most surface acting and drank more after work than other occupations.

We then conducted additional analyses among employees with daily customer contact, controlling for demographic variables known to predict drinking (e.g., gender, age), the job’s emotional demands (e.g., negative or stressful work events) and employee’s trait negativity (Semmer et al., 2016). Beyond these factors, the frequency of surface acting related to the extent of drinking after work depending on job self-control (work autonomy) and trait self-control (trait impulsivity), whereas surface acting predicted heavy drinking directly.

Autonomy as moderator.

Autonomous work conditions neutralize the link between surface acting and after-work drinking. Thus, we extend the existing body of evidence showing perceived work autonomy mitigates the costs of surface acting to health and performance (Christoforou & Ashforth, 2015; Grandey et al., 2005a; Johnson & Spector, 2007), to include the mitigating effect on health-related behaviors after work. These findings are consistent with research showing that when self-regulating behavior is done by choice or in one’s preferred way, it is less likely to have self-regulatory costs and more likely to offer benefits (Moller et al., 2006; Muraven, 2008; Trougakos et al., 2014). If drinking was a response to stress, we might expect work autonomy to also moderate the relationship of emotional job demands to alcohol (Bakker & Demerouti, 2007; Karasek, 1979). However, the moderating effect of work autonomy only affected the association of surface acting to drinking behavior. Our findings do not deny that job demands may predict drinking via job stress or negative mood, under certain conditions (Frone, 2016b, 2019; Shepherd et al., 2018); rather, they suggest an additional explanation for drinking that depends on the extent of self-control exertion and the autonomy of that exertion.

Impulsivity as moderator.

In general, more unhealthy behavior is expected when one has chronic low self-control (i.e., high trait impulsivity) (Tangney et al., 2004). Our study shows that highly impulsive employees are more susceptible to acting in unhealthy ways (i.e., by drinking more) when they are frequently performing surface acting at work. This moderating effect is consistent with the limited capacity view of self-control (Muraven, Shumeli, & Burkley, 2006), in that employees high in trait impulsivity do not habitually engage in self-control, thus controlling emotions is effortful. It is also consistent with the motivational shift perspective, in that employees high in trait impulsivity would find controlling their behavioral impulses around emotions more fatiguing and unpleasant and thus be ready to turn to more rewarding behavior after work (Inzlicht et al., 2014). We also tested whether distressing work experiences in general (emotional job demands) predict more drinking among impulsive employees as a maladaptive coping strategy. Trait impulsivity did not moderate the effect of emotional job demands on drinking, meaning the moderating role of trait self-control was specific to surface acting on drinking after work—consistent with a self-control lens.

Direct effect on heavy drinking.

It is notable that employees who tend to surface act with their customers are also more likely to be heavy drinkers, i.e., more than 4–5 drinks at a time and to intoxication. One might argue that this effect is simply due shared variance from job stressors, or employees who tend to feel negatively or impulsively; yet the effect emerged with those variables as covariates. Our exploratory analysis incorporating service relationships in tandem with our other self-control variables provided some insight into what mitigates the link of surface acting and heavy drinking. We found that for those with service encounters (i.e., customers, general public), surface acting was strongly linked to heavy drinking for highly impulsive employees but not less impulsive employees, whereas employees in service relationships (i.e., clients, patients, students) did not show the link between surface acting and drinking regardless of trait impulsivity. Employees with service relationships may find self-control of expressions less effortful and more rewarding (Gagné & Deci, 2005), consistent with prior evidence showing surface acting to be less problematic when employees identify with their work (Humphrey, Ashforth, & Diefendorff, 2015; Schaubroeck & Jones, 2000). However, given that this three-way interaction was post-hoc it should be interpreted with caution.

It is possible that the direct effect of surface acting and heavy drinking could be buffered by recovery behaviors after work that replenish resources (Shepherd et al., 2018) or by a strong motivation to stop drinking, such as when taking care of young children (Muraven, Shumeli, et al., 2006). Overall, a self-control impairment explanation seems to fit: the more employees exert control over their expressions for customers, the more likely they drink to intoxication.

Research Limitations and Future Directions

Our study has a unique strength in that we have a large representative sample of US workers who interact with organizational outsiders, as opposed to a focus on one occupation (e.g., Grandey, 2003; Wagner et al., 2014) or a convenience sample of students (Brotheridge & Grandey, 2002). This study provides information about surface acting and drinking behavior that can be generalized to the broader population of U.S. workers in public-facing occupations. However, this approach also comes with conceptual and methodological limitations.

Conceptual model limitations.

Based on the self-control perspectives, we demonstrated surface acting at work predicts alcohol consumption, but we did not directly test the mechanism of impaired self-control. Given our cross-sectional data, we could not adequately test mediation, so we relied on moderation tests to infer a process of self-control (Spencer, Zanna, & Fong, 2005). Future research should try to explicitly compare mechanisms of self-control with job stress or negative mood. However, identifying a distinct measure of self-control capacity is a challenge (Lian et al., 2017) because negative mood sometimes is used to indicate self-control depletion (Liu et al., 2017). We propose that this could be teased apart by examining an interesting corollary across the alcohol and self-control literature: the importance of beliefs. Specifically, negative affect is more likely to predict drinking when people believe that alcohol has tension-reduction properties [i.e., tension-reduction expectancies; (Frone, 2013)] and regulatory demands are more likely to diminish later regulatory performance when one believes that self-control is a limited resource (Job, Dweck, & Walton, 2010). Comparing tension-reduction expectancies and limited resource beliefs as moderators would help to understand the mechanisms predicting drinking behavior.

More specifically, we acknowledge there are differing theoretical views about why self-control impairment occurs from one domain to another, but we are limited in our ability to compare these views. The idea that self-control impairment occurs because initial self-control (a) depletes a regulatory resource pool (i.e., limited resource model) or (b) is unpleasant and motivates one to stop regulating, both are able to explain why employees who do more surface acting are likely to drink more on average and why our moderators work. To effectively tease apart these mechanisms, within-person changes at the momentary level are needed—but given the infrequency of some drinking behaviors, this requires longer study periods and narrower samples of individuals who drink at a certain level at least regularly. Comparing the role of breaks (i.e., replenishing energy resources) and rewards (i.e., increasing self-control gratification) for surface acting would help to untangle the mechanism as well. In addition, examining more than one indicator of self-control impairment would help to test the self-control paradigm. For example, future research could test whether and when surface acting influences verbal aggression with a partner when faced with disrespect, or healthy eating when faced with junk food (Liu et al., 2017).

Our study is also limited in how we conceptualized and measured emotional labor. We used a different measure than is typically used (i.e., Brotheridge & Lee, 2003), to more fully capture the full range of surface acting strategies of amplifying, faking, or suppressing. Furthermore, some might argue that these individual strategies may function differently (Cote, 2005). In fact, in an exploratory test we found that faking positive is linked to heavy drinking beyond the covariates (b = .07, se = .028, p = .008), but amplifying (b = −.01, se = .021, p = .615) and hiding (b = .010, se = .021, p = .652) were unrelated. Perhaps faking requires exerting more self-control than the others, or perhaps acting phony at work motivates one to feel disinhibited by large amounts of alcohol after work. More attention to the specific strategies and processes is warranted to determine why fakers are drinkers (as per our title).

We do not take into account deep acting, or strategies to modify feelings such as reappraisal or perspective taking (Grandey, 2000; Hochschild, 1983). These strategies improve mood and customer reactions, and thus might exhibit differential relationships with alcohol consumption. Furthermore, emotional labor can also be performed with organizational members (Kim, Bhave, & Glomb, 2013). Given that our post-hoc moderating results show that surface acting is not related to alcohol consumption when in the context of ongoing relationships, we propose that emotional labor is more problematic when performed with the public than with coworkers and supervisors. Future research should compare deep and surface acting and contrast outcomes when performing emotional labor with customers or with coworkers.

Methodological design limitations.

We controlled for variables that might create a spurious relationship between surface acting and drinking (e.g., impulsivity, negative affect), but of course we could not rule out all possible third variables that might make employees have certain emotional labor jobs and drink more (e.g., conscientiousness, cognitive ability). Furthermore, our focus on employees with daily customer contact and high emotional labor expectations may be overly limiting. Given the similarity of job demands for those with limited customer contact, we reanalyzed our hypotheses to include those with weekly contact as well as daily. The results remain the same, meaning they are robust across employees with regular customer contact. Importantly, these results may only generalize to U.S. workers; in other contexts, there may be stronger norms and values for surface acting that reduce its costs (Allen, Diefendorff, & Ma, 2014). In addition, we cannot assess or tease apart the effects of work group norms (i.e., happy hours) or occupational norms for drinking, which likely impact after work drinking (Frone & Brown, 2010; Liu, Wang, Bamberger, Shi, & Bacharach, 2015).

Because all of the data were self-reported, shared method variance may have caused some level of inflation in our reported coefficients. We attempted to address this issue in several ways (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003), using different response scales to reduce effects of consistency biases, and controlling for negative affective tendencies and other likely confounds (Semmer et al., 2016). Moreover, our predicted interaction effects are unlikely to be artifacts of common method variance; CMV does not inflate estimates of interaction effects, and can even make them harder to detect statistically (Siemsen, Roth, & Oliveira, 2010).

Finally, our tested model assumes that surface acting predicts alcohol consumption, but we cannot be sure of the causal order of our variables given our cross-sectional data. It is possible that the positive relationship is because employees who tend to drink heavily then feel more negatively at work (McKinney, 2010; Roehrs & Roth, 2001), and thus have to surface act more. Yet, a reversed causal flow might reveal a negative relationship: heavy drinking harms sleep quality, which impairs self-control and makes one more likely to act impulsively rather than regulate emotions (Diestel, Rivkin, & Schmidt, 2015). To test such questions effectively, future research needs daily diary studies with lagged and reciprocal effects (Liu et al., 2009).

Practical Implications

Our evidence supports a self-control approach to understanding surface acting and alcohol use with practical implications. First, our specific findings correspond to the “dark side” of emotional labor (Grandey, Rupp, & Brice, 2015) , and contribute to growing research on the costs to the employee’s life away from work (e.g., Krannitz, Grandey, Liu, & Almeida, 2015; Wagner et al., 2014). The cost is especially steep for low autonomy jobs and employees who have anonymous, one-time service encounters, and we urge managers to consider whether benefits of emotional labor are worth the costs.

Second, managers can also take steps to mitigate the impact of emotional labor on drinking through changes to the workplace. Our evidence further supports the value of an autonomy-supportive workplace (Gagné & Deci, 2005), which buffers the costs of surface acting in terms of job attitudes and burnout (Grandey et al., 2005), as well as after work drinking. Managers could protect employees by permitting recovery breaks to avoid the feeling of diminished self-control upon leaving work (Shepherd et al., 2018) and offer more support and rewards for positive displays – including the opportunity to build ongoing relationships with customers as suggested by our post-hoc findings –may reduce the need to turn to drinking after work (Grandey et al., 2013). Finally, training for self-control might be effective for service employees (Muraven, Baumeister, & Tice, 1999). Oaten and Cheng (2006) found that an intervention to practice academic self-control improved self-control for alcohol use among college students. Thus, an intervention that helps employees learn self-control for work behaviors (e.g., attendance, Frayne & Latham, 1987) could help with drinking as well.

Summary

Overall, the evidence presented here suggests the value of considering self-control and expanding the potential outcomes of emotional labor (see Grandey & Gabriel, 2015) to understand the issue of excessive drinking by service employees. Our findings suggest a cost to the “service with a smile” requirement: unless service employees have strong self-control tendencies or they are permitted to be self-governed at work, “fakers” tend to also be “drinkers”.

Appendix

Instructions: “While Interacting With [Target], How Often do You . . .” (0 [Never] to 4 [Always])

Surface Acting Measure (α = .72)

1. Exaggerate or amplify displays of positive emotions, such as friendliness, happiness, enthusiasm, or gratitude?

2. Pretend to feel positive emotions, such as friendliness, happiness, enthusiasm, or gratitude?

3. Exaggerate or amplify displays of caring, such as concern, sympathy, compassion, or empathy?

4. Pretend to feel a sense of caring, such as concern, sympathy, compassion, or empathy?

5. Hide negative emotions, such as sadness, disappointment, frustration, or anger?