8755263

Michael K. Brady, Clay M. Voorhees, & Michael J. Brusco

Service Sweethearting: ItsAntecedents and Customer

ConsequencesThis research is the first to examine service sweethearting, an illicit behavior that costs firms billions of dollarsannually in lost revenues. Sweethearting occurs when frontline workers give unauthorized free or discounted goodsand services to customer conspirators. The authors gather dyadic data from 171 service employees and 610 oftheir customers. The results from the employee data reveal that a variety of job, social, and remuneration factorsmotivate sweethearting behavior and several measurable employee traits suppress its frequency. The results fromthe customer data indicate that although sweethearting inflates a firm’s satisfaction, loyalty, and positive word-of-mouth scores by as much as 9%, satisfaction with the confederate employee fully mediates these effects. Thus,any benefits for customer satisfaction or loyalty initiatives are tied to a frontline worker that the firm would rathernot employ. Marketing managers can use this study to recognize job applicants or company settings that areparticularly prone to sweethearting and as the basis for mitigating a positive bias in key customer metrics.

Keywords: sweethearting, customer satisfaction, frontline employees, loyalty, marketing metrics

Michael K. Brady is Carl DeSantis Professor of Business Administration(e-mail: [email protected]), and Michael J. Brusco is Synovus Professor ofMarketing (e-mail: [email protected]), Department of Marketing, Collegeof Business, Florida State University. Clay M. Voorhees is Assistant Pro-fessor, Department of Marketing, Eli Broad College of Business, MichiganState University (e-mail: [email protected]). The authors thankSterling Bone, Bart Bronnenberg, Wayne Hochwater, Kay Lemon, SandraRobinson, Stacey Robinson, Brent Scott, and three anonymous JMreviewers for their helpful comments on previous drafts of this article. Thisarticle was accepted under the editorship of Ajay Kohli.

© 2012, American Marketing AssociationISSN: 0022-2429 (print), 1547-7185 (electronic)

Journal of MarketingVolume 76 (March 2012), 81–9881

Employee theft and fraud cost U.S. firms up to $600billion annually (Mazar, Amir, and Ariely 2008). Inthe retailing sector, approximately two-thirds of these

losses and 35% of annual profit losses (Amato-McCoy2009; Tarnowski 2008; Terris and Jones 1982) can be attrib-uted to an act of employee deviance in which frontlineemployees give unauthorized free or discounted goods orservices to a friend or acquaintance. This behavior, whichwe term “service sweethearting,” is common in hospitalityindustries, in which staff members may provide food andbeverages that never appear on the bill (Hawkins 1984).However, potential exists for this behavior in virtually anyindustry in which employees interact with customers at thepoint of sale. For example, retail cashiers slide productsaround the bar-code scanner to give the false impressionthat a friend is paying for an item. Repair service employ-ees provide repairs without notifying the billing depart-ment. Installation technicians initiate service without sub-mitting a work order. The problem is now so pervasive thatcompanies such as IBM and Stoplift have developedsophisticated algorithms to detect sweethearting in videosurveillance recordings (Dannen 2009). To the best of ourknowledge, no prior research has examined either the

employee or the customer side of sweethearting dyads. Thelack of research on sweethearting is surprising, as it wouldbe difficult to identify a topic that has such a considerableimpact on profitability yet has received so little attention.

From a scholarly perspective, sweethearting presents auseful sidebar to research that espouses close relationshipsbetween customers and frontline service employees. Thedevelopment of commercial friendships (Price and Arnould1999), social bonding (Bendapudi and Berry 1997), rapport(Gremler and Gwinner 2000), and long-term relationships(Reynolds and Beatty 1999) between customers andemployees are known to have several positive outcomes forfirms, including positive word of mouth (WOM) andenhanced customer satisfaction and loyalty scores (Gremlerand Gwinner 2008; Reynolds and Beatty 1999). Ourresearch indicates that there may be an insidious downsideto close relationships between customers and frontlineemployees, and we show that a symptom of the problem iselevated satisfaction, loyalty, and positive WOM scores.

The current research draws on literature in marketing,organizational behavior, psychology, and criminology tooffer insights into three broad questions about the sweet-hearting phenomenon: (1) What are the antecedents tosweethearting behavior for customer contact employees?(2) Are there certain identifiable employee traits that sup-press the sweethearting phenomenon? and (3) What are theeffects of sweethearting on prominent customer outcomemeasures? With respect to employees, researchers know lit-tle about the contextual, individual, or preventive factorsthat may influence the onset of sweethearting behavior. Anexploratory study and insights from various literaturesinform our understanding of sweethearting antecedents.Further analysis shows that specific employee traits, such as

risk aversion and low need for social approval, buffersweethearting frequency. With respect to customers, weexamine how sweethearting influences key customer met-rics, such as satisfaction measures. Because customers whoparticipate in sweetheart deals are apt to rate the firm and itsemployees more favorably than nonsweethearts who payfull price, this otherwise detrimental behavior may manifestitself in the form of inflated customer feedback scores. Weinvestigate the nature of this ironic effect on firm-level out-come measures, with a particular interest in whether it isanchored to the employee or the firm.

Research BackgroundA Cautionary Caveat to Strong Customer–Employee RelationshipsA central theme of the relationship marketing literature isthat firms benefit from developing strong, long-term rela-tionships with customers. Material gains for the firm occurin the form of cost efficiencies (Reichheld 1996), increasedcustomer commitment (Mavondo and Rodrigo 2001),higher satisfaction scores (Reynolds and Beatty 1999),increased customer loyalty (Beatty et al. 1996), and adampening effect on the negative implications of servicefailures (DeWitt and Brady 2003; Matilla 2001). However,some scholars suggest caution in assuming that close rela-tionships between customers and employees are always inthe best interest of the firm (Bove and Johnson 2006).Employees with close ties to customers may leave for acompetitor and take lucrative customers with them (Beattyet al. 1996; Bendapudi and Leone 2002). As an exemplar,American Express estimates that approximately 30% ofcustomers would follow their financial advisor to a differentfirm (Tax and Brown 1998).

Our research offers a similar cautionary tale withrespect to strong customer–employee relationships. In thecase of sweethearting, employees purposefully underchargecustomers—or do not charge them at all—for services ren-dered. In effect, they are literally “giving away the store” tocustomers with whom they have a close relationship.Unlike the case of defecting employees and customers,sweethearting is difficult for firms to detect; thus, firmsoften sustain revenue losses over time (Tarnowski 2008).Even worse, firms may compound their losses by rewardingdeviant employees for their inflated customer satisfactionscores.Comparison with Other Forms of Employee andCustomer Deviance Several deviant employee behaviors exist in organizations,including antisocial behaviors (Giacalone and Greenberg1997), complaint concealment (Harris and Ogbonna 2009),workplace aggression (Baron and Neuman 1996), sabotage(Harris and Ogbonna 2002, 2006), organizational retaliation(Skarlicki and Folger 1997), and theft (Greenberg 1997).Likewise, customer deviance takes several forms, such asfraud (Cole 1989), cheating (Wirtz and Kum 2004), abu-siveness (McColl-Kennedy et al. 2009), and theft (Fullertonand Punj 2004). All these behaviors are deviant because

82 / Journal of Marketing, March 2012

they depart from established rules and expectations in waysthat have negative implications for a firm.

Although sweethearting also contradicts establishedrules and expectations, it differs from the preceding formsof deviance in at least three ways. First, whereas mostforms of employee and customer theft behaviors areintended to be carried out without the knowledge of theother party, both parties are aware of sweethearting. Thecollaborative aspect of sweethearting is important becausecustomer involvement sets sweethearting apart from otherforms of workplace deviance and emphasizes the relevanceof the behavior to marketers.1 Second, research in organiza-tional behavior suggests that employees typically engage indeviant acts because of some combination of individualtraits and workplace states (Greenberg 2002; Hollinger andClark 1983). Sweethearting reinforces the point that deviantacts also have roots in social and economic exchange (e.g.,Greenberg and Scott 1996). Third, unlike other forms ofdeviant behaviors, in which either the customer or theemployee may benefit from the act (but not both), sweet-hearting simultaneously offers benefits to the employee andthe customer. Therefore, sweethearting is different fromother forms of deviance in ways that should be of interest tomarketers.Fertile Ground for Sweethearting DyadsService purchases, especially those that require interactionwith frontline personnel, represent an opportune context inwhich to study sweethearting. Services are characterized byinseparability (Keh and Pang 2010), credence properties,and heightened risk (Czepial 1990), all of which increasethe likelihood that customer–employee interactions willplay a prominent role in the purchase process. Likewise, wenote that employee deviance is common in the services sec-tor. Estimates suggest that 35% of retail employees, 33% ofhospital employees, 43% of grocery store employees, and62% of fast-food employees engage in theft behaviors(Greenberg and Barling 1996). Therefore, we focus atten-tion on sweethearting behavior in frontline service encoun-ters. To do so, we begin with an exploratory investigation ofthe employee side of sweethearting dyads and then turn ourattention to modeling the customer side.

The Employee Side ofSweethearting

In general, research on customer–employee interactionstakes two broad perspectives. The first and most commonperspective is that employees are working for the firm andwith customers to provide excellent service (Berry 1995).

1We note that sweethearting collaboration does not need toinvolve a prior arrangement known in advance but that customersmust be complicit in the act. For example, a frontline employeewho comes across a friend unexpectedly and then offers free bene-fits is engaging in sweethearting. In contrast, an employee whooffers free benefits to an unwitting customer, perhaps in retaliationto the firm, is engaging in a form of organizational retaliation(Skarlicki and Folger 1997) or workplace aggression (Baron andNeuman 1996).

Various intrinsic and extrinsic rewards serve as motivationsfor providing such exemplary service. The second, morerecent perspective is that service employees sometimeswork against customers in such a way that they purpose-fully provide poor service (Warren 2003). The latter per-spective is consistent with service sabotage, in which front-line employees act as saboteurs who intentionally providepoor service to customers (Harris and Ogbonna 2002). Sub-par working conditions and various personality types arecommon antecedents to sabotage behavior (Harris andOgbonna 2006). Sweethearting behavior belongs to neitherof these two perspectives. Employees in sweetheartingdyads are neither working for the firm nor against the cus-tomer; rather, they work with complicit customers in a waythat is against the best interests of the firm. Therefore,antecedents to sweethearting are unknown. We examinesweethearting antecedents next, with a particular interest inthose that are unique to sweethearting behavior.

We used a two-stage process to identify sweetheartingantecedents. We began with an open-ended survey of front-line service employees. Then, we reviewed prior researchon employee deviance to determine whether other knownforms of deviant behaviors shared the emergent factors.Although the deviance literature does not address sweet-hearting per se, there may be common factors that influencea variety of deviant behaviors (Robinson and Greenberg1998). Next, we discuss the exploratory study and developour research model.The Exploratory StudyAn open-ended survey was administered to 40 people whowere either currently employed or had previously worked inservices industries within the past two months. The surveyrequired respondents to answer three questions and severaldemographic and classification items. The first questionprovided respondents with a clear definition of sweetheart-ing behavior and asked them if they had participated in asweethearting dyad. Then, respondents described the mostrecent sweethearting incident in which they had participatedduring the past two months. If the employee had not partici-pated in a sweethearting dyad, he or she was instructed tomove on to the final question, in which respondents listedand elaborated on the factors that influenced their decisionto either engage in or abstain from sweethearting.

We obtained responses from upper-level business stu-dents at a large university. Respondents were given extracredit for participating in the study and were screened toensure that they were currently employed by a service com-pany or had been employed in the past two months. Thesample consisted of service employees working at bars andrestaurants (52%), retail stores (32%), and other miscella-neous service firms, such as golf courses and tanning salons(16%). In accordance with prior studies on similar behav-iors (Greenberg and Barling 1996), 67% of the respondentsindicated participation in sweethearting within the past twomonths.

Following established protocol for qualitative dataanalysis (Glaser and Strauss 1967; Spiggle 1994), weallowed categories to emerge from the data and placed no

Service Sweethearting / 83

restrictions on the length of text required for inclusion.Coders identified and categorized the factors that influ-enced sweethearting behavior and indicated whether theyhad a positive or a negative effect. The coder reliability esti-mate was 94%. In cases of disagreement, coders discussedtheir responses until they reached a consensus. This processidentified 11 antecedents.Sweethearting AntecedentsWe organized the 11 sweethearting antecedents thatemerged from the data into themes (Spiggle 1994). Weorganized antecedents with prior support in a particular lit-erature under themes named for the literature from whichthey emerged. For antecedents unique to sweetheartingbehavior, we attempted to organize them under existingthemes; when this was not possible, we created a newtheme. This process resulted in the establishment of threeexisting themes (job related, deterrence, and traitantecedents) and the creation of a new theme that we term“remuneration antecedents.”

Remuneration antecedents. Material gain is the mostconspicuous reason to engage in theft behaviors. Priorresearch has attributed this link to factors such as financialpressures employees face (Merton 1938) and greed (Astor1972). However, remuneration in sweethearting dyads isnot straightforward because the obvious beneficiary is thecustomer, not the employee. Our qualitative data nonethe-less revealed two ways that employees are reimbursed, sug-gesting a link between sweethearting and remuneration.

The first remuneration factor, reciprocity, underscoresthe collaborative and interactive aspect of sweetheartingbehavior. We found that there is often a social obligation forsweetheart customers to provide employees with reciprocalgifts at their respective service jobs (Belk 1976). For exam-ple, as one employee stated, “All of my friends are in theservice industry, so it is standard to ‘hook up’ your friendswhen they come in so that they’ll do the same for you inreturn.” Thus, sweethearting often involves a tit-for-tatexpectation in which frontline service employees buildinformal credits at other service firms. As one employeeexpressed, “It’s just one of those things where if you scratchmy back I’ll scratch yours.” A significant implication asso-ciated with reciprocity is that sweethearting behavior maybe contagious in the sense that an incident at one firm mayinvoke another at a completely unrelated firm. We areaware of no other deviant behavior that has a similar impacton the well-being of other firms.

The second remuneration method is through financialgain in the form of informal payments. Many frontlineemployees mentioned an expectation that sweethearts willreturn the favor through better tips. As one employee stated,“When I give them free drinks, they save money and sothey tip better.” Although this behavior was most commonin restaurants and bars, it was not limited to these settingsor even to industries in which tipping is the norm. As oneretail employee stated, “I worked at an office supply storeand when my friends came in I’d switch the UPC codes onitems and they would get charged for pencils.… I did this to

make more money.” Thus, we propose that remunerationfactors directly affect sweethearting:

H1: Sweethearting frequency is greater when there is opportu-nity for (a) reciprocal sweethearting and (b) financial gain.

Trait antecedents. Implicit in the idea that employeetraits influence sweethearting behavior is the notion thatsome people are naturally prone to deviant behaviors(Osgood et al. 1988). The existence of a deviant personalityis consistent with Freud’s theory of personality and withstudies advanced in the organizational psychology literature(e.g., Lykken 1995; Miner and Capps 1996). This belief isevident in the widespread use of personality and integrity testsin preemployment screening (Murphy 1993; Sackett 1994).

The first trait factor, need for social approval, is rarelydiscussed in the deviance literature. Nonetheless, the socialmotive for sweethearting is particularly strong because thebehavior is a conspicuous source of social capital inemployee peer groups. Consider the following example:“My friends love to talk about how cool it is for them tocome in to eat and drink at the restaurant I work at becausethey only pay a fraction of what they should. Hearing themtell the stories about the fun they had and how much theylike me and my restaurant makes me feel good.” Greenbergand Barling (1996) refer to this effect as “social needtheory” and suggest that it is particularly relevant to sweet-hearting behavior.

The second employee trait that may influence sweet-hearting is risk-seeking propensity. Research in the crimi-nology (Szockyj and Geis 2002) and psychology (Lernerand Keltner 2001) literatures indicates that criminals tend tohave an inherent propensity for high-risk activities. Thisview is consistent with Gottfredson and Hirschi’s (1990)theory of crime and with Zuckerman’s (1980) research onsensation seekers. Consider the following statement from aretail employee: “Why wouldn’t I do it? It’s fun and excit-ing to give your friends free stuff.”

The third trait antecedent that we uncovered is personalethics, which refers to the individual moral beliefs employ-ees hold (Mazar, Amir, and Ariely 2008). For example, anemployee remarked, “It’s immoral, so I don’t do it.”Another employee said, “I don’t like the idea of stealingand would not intentionally do anything that would makethe people I work for lose profits.” On the basis of insightsfrom the exploratory study and corroboration from severalliteratures, we propose the following:

H2: Sweethearting frequency is greater when employees (a)have higher levels of need for social approval, (b) havehigher levels of risk-seeking propensity, and (c) haveweaker ethical values.

Job-related antecedents. Job characteristics are knownto influence the onset of workplace deviance (Lau, Au, andHo 2003; Robinson and Greenberg 1998). Positive jobcharacteristics and organizational climates suppress deviantbehaviors, whereas negative characteristics and climatespromote them. The first job-related antecedent we identi-fied, work group norms, refers to whether employeedeviance is a normal or accepted behavior among employ-ees in an organization (Greenberg 1997; Robinson and


O’Leary-Kelly 1998). Consider the following two respon-dents: “I do it because all the other servers do it,” and “Otheremployees had told me how to do it and did it quite often.”

Previous research has established a link between jobsatisfaction and numerous deviant workplace behaviors,including employee theft (Terris and Jones 1982), absen-teeism (Bolin and Heatherly 2001), and sabotage (Jermier1988). The following verbatim quotation indicates that jobsatisfaction also plays a role in sweethearting: “I have neverdone this at previous jobs where I was actually satisfiedwith my work environment.”

Likewise, employees with a lack of organizational com-mitment are known to engage in deviant acts at a high rate(Hollinger, Slora, and Terris 1992; Randall 1987). The ratio-nale for this relationship is that employees with future stakesin the organization have more to lose than more transientemployees if they were caught (Hollinger 1986). Considerthe following comment: “To my employers, I was just a namein a time slot, so I had no loyalty to the store and hooked upcustomers in return.” Thus, we propose the following:

H3: Sweethearting frequency is greater when (a) it is consis-tent with established work group norms, (b) employee jobsatisfaction is lower, and (c) organizational commitment islower.

Deterrence antecedents. Although sweethearting seemsto be a pervasive activity, we identified several ways topotentially deter the behavior. The first two deterrenceantecedents involve the perceived threat of sanctions if anemployee were to be caught sweethearting. These factorsaddress two logical questions that an employee might con-sider: (1) “Will I get caught?” and (2) “If I do get caught,how severely will I be punished?” The first question refersto punishment certainty. Employees who perceive a lowlikelihood of detection are more likely to engage in deviantacts (Kantor 1983). For example, as one employee remarked,“My managers are completely oblivious,” and as anothermentioned, “Nobody ever seems to notice.” The secondquestion addresses punishment severity (Tittle 1980). Asone employee stated, “Some people have gotten fired oversimilar situations, so I don’t hook up customers anymore.”

The third deterrence antecedent, job control, is relatedto the suppression of deviant behaviors through strategiesthat limit job autonomy. Prior studies have shown thatemployee theft is directly related to opportunities to steal(Hollinger and Clark 1983). Thus, unscrupulous employeeswho are given autonomy in their jobs are especially likelyto engage in deviant behaviors (Murphy 1993). The follow-ing comment made by a restaurant employee exemplifiesthe effect of job control on sweethearting: “I had completecontrol over seating and serving customers, so it was easyto do it.” On the basis of the preceding evidence, we pro-pose the following:

H4: Sweethearting frequency is greater when (a) detection isless certain, (b) punishment is less severe, and (c) employ-ees have more control over their jobs.

Interactions. There is considerable evidence that, inaddition to direct links, employee traits moderate the effectsof situational factors on deviant behaviors (Cullen and

Sackett 2003; Henle 2005). In other words, a specificemployee trait may be associated with theft directly andalso heighten the effect of a negative work environment ontheft behaviors (Strutton and Lumpkin 1992). Of particularinterest to this research is the opposite scenario, wherebycertain employee traits suppress the onset of sweetheartingbehavior (Colbert et al. 2004). We consider several suchmoderated relationships in an effort to highlight the breadthof trait effects on sweethearting and to identify importantboundary conditions. The establishment of boundary condi-tions is particularly important for this research becauseantecedents such as financial gain and reciprocity are diffi-cult for frontline managers to directly control. For otherantecedents, the very act of controlling them may conflictwith best practice. For example, efforts to limit employeejob control contradict research that indicates that employeeswith some control over their jobs are less likely to becomestressed and burned out (Singh 2000). Identification ofmeasurable trait factors that suppress the allure of financialgain, reciprocal deals, and opportunistic behavior offers ameans to manage sweethearting behavior without the nega-tive repercussions associated with oppressive tactics such aseliminating job control.

In making our predictions, we draw on literature thatindicates that deviant employee behavior is influenced bysituational factors “only if such behavior is consistent withtheir personality traits” (Colbert et al. 2004, p. 599). We pro-pose that need for social approval and risk-seeking tenden-cies are consistent with deviant behavior, whereas personalethics is inconsistent. With regard to personal ethics, employ-ees with weak personal ethics may be more likely to takeadvantage of job control and opportunities for financial gainbecause they lack the protection from misdeeds afforded bya strong ethical code (Lau, Au, and Ho 2003). Alternatively,employees with strong personal ethics would be less likelyto take advantage of having control over their jobs andopportunities for financial gain because doing so violates theirpersonal codes of ethics. Thus, we propose the following:

H5: Personal ethics moderates the direct effects of (a) job con-trol and (b) financial gain on sweethearting frequency suchthat the effects are weaker when personal ethics is strong.

As the employees in our exploratory study noted, recip-rocal exchange carries a social obligation of compliance(Gouldner 1960). Receivers of sweethearting gifts oftenhave a future obligation to reciprocate at their own frontlineservice jobs. Employees who violate this norm risk socialsanctions, such as unpopularity and exclusion from peergroups (Cialdini 2007). We expect that such informal sanc-tions would be particularly troublesome to people with a highneed for social approval. In contrast, employees with a lowneed for social approval should feel less compelled to reci-procate, and therefore a low need for social approval shouldsuppress the effect of reciprocity on sweethearting. Thus:

H6: Need for social approval moderates the direct effect ofreciprocity on sweethearting frequency such that the effectis weaker when need for social approval is low.

Like other forms of deviance, sweethearting occurswhen favorable conditions exist and the perpetrators are


willing to accept the risk of detection (Greenberg 2002).Therefore, opportunity and risk acceptance are importantbuilding blocks for sweethearting, but both must be presentfor the seed to germinate. For example, risk-averse employ-ees may have the opportunity to engage in sweethearting byhaving a certain amount of control over their jobs, yet theymay be unwilling to accept even a minimal risk of detec-tion. Thus, we expect that a low propensity for risk sup-presses the effect of job control on sweethearting.

H7: Risk-seeking propensity moderates the direct effect of jobcontrol on sweethearting frequency such that the effect isweaker when risk-seeking propensity is low.

Sweethearting ModelWe developed the model presented in Figure 1 to assess theeffects of the 11 antecedents on sweethearting frequency.Next, we discuss our methods and the results.MeasuresWe adopted established measures for 9 of the 12 variablesin the model and developed sweethearting frequency, finan-cial gain, and reciprocity scales specifically for this study.We also included a series of control variables in the analy-ses. The Appendix provides details about the measures forthe employee constructs.Data CollectionWe drew the employee and customer samples from variousservice companies in a medium-sized metropolitan area.Because of the sensitive nature of the issues under investiga-tion, we adopted a snowball sampling design according torecommendations for snowball and chain referral sampling(e.g., Biernacki and Waldorf 1981; Zinkhan, Burton, andWallendorf 1983) as well as precedents in the marketing andorganizational behavior literatures. Our recruitment of par-ticipants, initiation of referral chains (Bennett and Robinson2000; Groth, Hennig-Thurau, and Walsh 2009), surveydesign (Harris and Ogbonna 2002, 2006), and data collectionprocedures (Liao, Joshi, and Chuang 2004) were all basedon previous research designs for the study of customer–employee dyads or deviant behaviors.

A team of students from an upper-level marketingresearch class screened and recruited respondents. Studentsare often used in sampling for deviant behaviors (e.g.,Henle 2005; Trevino and Victor 1992) because they tend tohave “privileged access” to desired respondents (Griffiths etal. 1993). In the case of sweethearting behavior, studentsare relatively young, which fits the profile of employeeswho are especially prone to deviant behaviors (Lau, Au, andHo 2003), and they tend to be involved in active social net-works with close access to frontline service employees.2

2We acknowledge that our sampling method is likely to empha-size relatively young frontline service workers who are involvedin active social networks. Although this sample profile may notmatch all frontline service workers, it matches the profile of front-line employees who are likely to engage in this particular form ofdeviant behavior (Hollinger and Clark 1983; Lau, Au, and Ho2003), thus providing a valid sample frame for our study.

Moreover, the use of confederates to initiate referral chainsmay provide more valid data than if researchers attemptedto directly access respondents involved in deviant activities(Zinkhan, Burton, and Wallendorf 1983).

To gather the data, student recruiters identified onemember of a sweethearting dyad and asked this person tocomplete a survey and then distribute a survey packet to theother member. The packet included a postage-paidemployee survey and four postage-paid customer surveys.Anonymity was guaranteed in writing. Respondents wereinformed that the study was being conducted by academicsnot affiliated with, or paid by, any particular company. Toensure respondent validity, we used a callback verificationprocess whereby student recruiters provided an independentlisting of their key informants’ contact information, and wesubsequently contacted 20% of the informants to validatethat the surveys were properly distributed. Other investiga-tions involving customer–employee relationships (e.g.,Groth, Hennig-Thurau, and Walsh 2009) have used similarsampling and verification approaches.

The data collection process resulted in 185 serviceemployee responses, of which we removed 14 question-naires (7.6%) because of missing values. The demographiccharacteristics of the employees were typical for these ser-vice industries, as reported in Brown, Cowles, and Tuten(1996) and by the U.S. Bureau of Labor Statistics. Themean age of the employees was 22.4 years, and they had anaverage tenure of 1.6 years with their firms. Of the respon-dents, 49% were employed in restaurants, 19% wereemployed in retail, 16% were employed in other hospitalitysettings, 7% were employed in pure service settings (e.g.,


cable television installation and repair, car washes, tanningsalons), and 9% were employed in other service industries. Assessment of the MeasuresThe results of a confirmatory factor analysis indicated thatthe measurement model provided a good fit to the data (2 =1904.27, d.f. = 968; comparative fit index = .91; standardizedroot mean square residual = .06; and root mean square errorof approximation = .06). Moreover, follow-up analyses basedon the recommendations of Fornell and Larcker (1981) pro-vided support for the reliability, convergent validity, anddiscriminant validity of the constructs (for a completereporting of the measurement statistics, see Table 1).Tests for Common Method BiasTo assess common method bias, we applied the markervariable assessment technique that Lindell and Whitney(2001) and Malhotra, Kim, and Patil (2006) developed.This method involves the identification of the two lowestcorrelations (r = .0003, .004) among the manifest variablesin the data set. Using these correlations as estimates ofmethod bias in the data, we calculated a discounted correla-tion matrix using the more conservative method bias esti-mate (r = .004) and compared it with the unadjusted matrix.Neither the significance nor the signs of any of the correla-tions changed across matrices, indicating that method biasdoes not pose a risk to the interpretation of the data.Partial Least Squares AnalysisWe employed partial least squares analysis (PLS) to testH1–H7. The results of the PLS analysis revealed that 11 of

Antecedents: Employee Data (N = 171)

FIGURE 1Antecedents and Customers Consequences of Service Sweethearting

Trait Factors•Risk-seeking propensity (+)•Personal ethics (–)•Need for social approval (+)

Remuneration Factors•Financial gain (+)•Reciprocity (+)

Service Sweethearting

Perceptions ofInequity

Satisfaction withthe Employee

Job-Related Factors•Deviant work group norms (+)•Personal ethics (–)•Need for social approval (+)

Deterrence Factors•Punishment severity (–)•Punishment certainty (–)•Job control (+)

Customer Outcomes•Satisfaction with the firm•Loyalty to the firm•Positive word of mouth

Consequences: Customer Data (N = 610)

+

+

+

+


TABLE 1

Scale Statistics: Means, Standard Deviations, M

easure Reliabilities, Average Variances Extracted, and Correlations

A: Employee Sam

ple

Variable

MSD

MR

AVE

12

34

56

78

910

111.

Sweethearting frequency

5.22

1.83

.86

.56

2.Risk-seeking propensity

6.43

1.51

.79

.57

.34

3.Personal ethics

6.04

1.51

.85

.60

–.27

–.21

4.Need for social approval

5.16

1.54

.79

.50

.16

.00

.28

5.Financial gain

3.93

2.95

.98

.93

.44

.22

–.30

–.04

6.Reciprocity

5.74

2.25

.90

.70

.52

.38

–.31

.04

.48

7.Job satisfaction

5.94

1.91

.86

.67

–.03

–.08

.33

.18

–.19

–.14

8.Organizational commitment

3.48

1.88

.86

.61

–.12

–.20

–.03

–.24

–.06

–.07

.29

9.Deviant work group norms

4.96

1.83

.85

.59

.53

.31

–.21

.09

.33

.45

–.06

.01

10.Job control

5.39

1.95

.76

.51

.20

.15

.19

.16

.28

.08

.15

.18

–.06

11.Punishment severity

5.13

2.27

.94

.77

–.04

.02

–.04

–.06

.08

.00

–.17

.24

–.10

–.07

12.Punishment certainty

4.03

1.82

.89

.61

–.13

–.04

.22

.04

–.04

–.01

.04

.07

–.15

.06

.35.

B: Customer Sam

ple

Variable

MSD

MR

AVE

12

34

51.

Sweethearting involvement

.50

.50

NANA

2.Perceptions of inequity

6.24

1.98

.96

.86

.46

3.Satisfaction with employee

7.56

1.64

.94

.86

.27

.36

4.Satisfaction with the firm

7.29

1.66

.96

.89

.15

.18

.67

5.Loyalty to the firm

6.31

2.08

.95

.86

.14

.15

.47

.66

6.Positive WOM

7.05

1.81

.97

.91

.11.13

.54

.71

.75

Notes: MR = measure reliability, and AVE = average variance extracted. In the employee sample, all correlations greater than .14 are significant (p

< .05); in the customer

sample, all correlations are significant (p

< .01).

the 15 hypothesized effects were significant and the modelexplained 57% of the variance in sweethearting frequency.Specifically, both remuneration factors (p < .05), all threetrait variables (p < .05), two job-related factors (deviant workgroup norms and organizational commitment, p < .01), andone deterrence factor (job control, p < .05) had significant,direct effects on sweethearting frequency. Moreover, weidentified significant interactions for the effects of financialgain (personal ethics ¥ financial gain, p < .05), reciprocity(need for social approval ¥ reciprocity, p < .01), and jobcontrol (risk-seeking ¥ job control, p < .05) on sweetheart-ing frequency. However, the interaction between personalethics and job control (H5a) was not significant. Table 2 pre-sents complete results of the hypothesis tests.Probing Interaction EffectsTo probe the three significant interactions, we calculatedsimple slopes at high and low levels of the moderators(Aiken and West 1991) and regions of significance for eachinteraction effect (Bauer and Curran 2005). We first calcu-lated the simple slopes for the personal ethics ¥ financialgain interaction (H5b). As we expected, the results indicatethat financial gain had weaker effects on sweethearting fre-quency at high levels of personal ethics ( = .03, p > .05)than low levels of personal ethics ( = .15, p < .01). Indeed,at high levels of personal ethics, financial gain was not asignificant driver of sweethearting frequency. Next, we esti-mated the regions of significance in an effort to assess theexact level of personal ethics at which the direct effect offinancial gain on sweethearting was buffered. Figure 2 pre-


TABLE 2Partial Least Squares Results for theSweethearting Antecedents Model

Predictor CoefficientRemuneration FactorsH1a: Reciprocitya .26**H1b: Financial gaina .13*

Trait FactorsH2a: Need for social approval .13*H2b: Risk seeking .12*H2c: Personal ethics –.15*

Job-Related FactorsH3a: Deviant work group norms .27**H3b: Job satisfaction .07n.s.H3c: Organizational commitment –.14**

Deterrence FactorsH4a: Punishment certainty –.04n.s.H4b: Severity of punishment –.02n.s.H4c: Job controla .14*

Interaction EffectsH5a: Personal ethics ¥ job control .01n.s.H5b: Personal ethics ¥ financial gain –.11*H6: Need for social approval ¥ reciprocity .18**H7: Risk seeking ¥ job control .15*

R2 .57*p < .05.**p < .01.aThe effect is qualified by an interaction.Notes: n.s. = nonsignificant.

!!!!

1.00

p < .05 p > .05.60.50.40.30.20.100

Employees’ Personal Ethics6.20 9.00

Simple Slopes of Financia

l Gain

!!!

!!!

!!!

!!!

1.00

p > .05 p < .05.60.50.40.30.20.100

Employees’ Need for Social Approval4.26 9.00

Simple Slopes of R

eciprocity

!

!!!

!!!

!!!

!!

1.00

p > .05 p < .05.60.50.40.30.20.100

Employees’ Risk-Seeking Propensity6.47 9.00

Simple Slopes of Job Control

FIGURE 2Plots of the Regions of Significance for theSimple Slopes of Independent Variables on

Sweethearting FrequencyA: Personal Ethics ¥ Financial Gain Interaction

B: Need for Social Approval ¥ Reciprocity Interaction

C: Risk Seeking ¥ Job Control Interaction

Interpretation: The effect of financial gain on sweethearting fre-quency is buffered for employees who score higher than 6.20 on the nine-point scale. The mean for the personal ethics scale was 6.04/9.00 with astandard deviation of 1.51.

Interpretation: The effect of reciprocity on sweethearting frequency isbuffered for employees who score below 4.26 on the nine-point need forsocial approval scale. The mean for the need for social approval scale was5.14/9.00 with a standard deviation of 1.54.

Interpretation: The effect of job control on sweethearting behaviorbecomes nonsignificant for employees who score below 6.47 on the nine-point risk-seeking scale. The mean for the risk seeking scale was 6.43/9.00with a standard deviation of 1.51.Notes: Simple slope estimates in the shaded area are not signifi-

cant (p > .05).

sents the results of the regions of significance testing; itdemonstrates that when personal ethics scores exceeded6.20 on our nine-point scale, financial gain was not a sig-nificant predictor of sweethearting frequency.

Next, we calculated simple slopes and regions of signifi-cance for the need for social approval ¥ reciprocity interac-tion (H6). The results of the simple slopes calculations sup-ported H6 in that reciprocity had a strong and significanteffect on sweethearting frequency at high levels of need forsocial approval ( = .31, p < .01) but was not a significantdriver of sweethearting at low levels of need for socialapproval ( = .08, p > .05). Regions-of-significance calcu-lations revealed that when need for social approval was lessthan 4.26, the effect of reciprocity on sweethearting fre-quency became nonsignificant.

Finally, we probed the risk seeking ¥ job control inter-action (H7), which revealed that job control had a substan-tially weaker effect on sweethearting frequency at low lev-els of risk seeking ( = .06, p > .05) than at high levels ofrisk seeking ( = .23, p < .01). Regions-of-significance test-ing demonstrated that if the level of risk seeking was below6.47, job control no longer significantly influenced sweet-hearting frequency. We now turn to the customer conse-quences of service sweethearting.

The Customer Side ofSweethearting

Perhaps the most interesting and ironic aspect of the sweet-hearting phenomenon involves its impact on customer feed-back about firms and frontline employees. Whereas otherforms of workplace deviance have negative implications forkey customer metrics (Harris and Ogbonna 2006), sweet-hearting seems to have positive effects. Sweetheart cus-tomers have close personal relationships with frontlineworkers (Price and Arnould 1999), and they receive exclu-sive benefits provided at some risk to the employee andsome cost to the firm. We expect these relationships andbenefits to influence customers in much the same way asother contextual biases that inflate customer outcome mea-sures (Peterson and Wilson 1992).

We make two key predictions with respect to the effectsof sweethearting on firm-level outcome measures. The firstprediction addresses why we expect sweethearting to havepositive effects on firm evaluations, and the secondinvolves the locus of that effect. Specifically, we proposethat involvement in sweethearting triggers positive inequityand satisfaction with the employee. Moreover, we contendthat the positive effects of sweethearting on firm outcomesare indirect; that is, they are fully mediated by satisfactionwith the deviant employee. The second half of Figure 1depicts these and our other predictions, and we discussthese hypotheses next.Model Development

Direct effects. Equity theory (Adams 1963) providesclear support for an immediate customer response toinvolvement in a sweethearting exchange. Equity theorysuggests that feelings of positive inequity result when the


ratio of inputs to outcomes decreases relative to the per-ceived ratio of others. Thus, exchanges that require less ofone customer than other customers will result in positiveinequity. In frontline exchanges in which sweetheart cus-tomers receive free or discounted products and other cus-tomers pay full price, the sweethearts should feel as thoughtheir inputs to the exchange are reduced (i.e., they pay lessor nothing) for the same types of outcomes. Thus, we pro-pose the following:

H8: Involvement in a sweethearting dyad directly increasescustomer perceptions of positive inequity.

In addition to enhanced perceptions of the exchange,sweethearting also has direct and positive implications forcustomer perceptions of the employee. Research in psy-chology suggests that people show gratitude for intention-ally rendered benefits that are valuable to the beneficiaryand potentially costly to the benefactor (Emmons 2004). Inturn, the beneficiary (i.e., the customer) reciprocatesthrough actions that contribute to the future well-being ofthe benefactor (i.e., the employee) (McCullough et al.2001). In the case of sweethearting behavior, the employeeplays the role of the benefactor who risks financial and per-haps even legal recourse to provide a benefit to the cus-tomer. As the beneficiary of the exchange, the customerowes the employee a debt of gratitude, which often takesthe form of social goodwill (Belk and Coon 1993). Becausesatisfaction is closely tied to feelings of goodwill (Oliver1997), we expect elevated satisfaction levels for theemployee involved in the sweethearting dyad.

H9: Involvement in a sweethearting incident directly increasescustomer satisfaction with the employee.

Mediation effects. The preceding processes detail howsweethearting triggers positive evaluations of the exchange(H8) and the employee (H9); however, noticeably missingfrom this discussion is the connection between sweetheart-ing and the firm. Our attention here focuses on the nature ofthe positive baseline effect of sweethearting and, in particu-lar, on whether its impact on firm evaluations is mediatedby satisfaction with the employee.

Sweethearting involves an undisclosed interpersonalrelationship that yields a positive imbalance for the cus-tomers involved in the exchange. Because the focalexchange partners in sweethearting transactions are the cus-tomer and the employee, not the firm, it stands to reasonthat the direct benefits derived from the positive inequitywill be awarded to the primary partner (the employee)rather than the secondary partner (the firm). This circum-stance is especially relevant for a sweethearting dyadbecause the firm is unaware of the employee’s actions and,if known, would object to the transaction. Thus, we expectthe effect of positive inequity on firm outcomes to be medi-ated by satisfaction with the employee.

H10: Satisfaction with the employee fully mediates the effectof positive inequity on customer evaluations of the firm(firm satisfaction, loyalty, and positive WOM intentions).

Given that the firm is the stakeholder that incurs thefinancial responsibility, rational decision models suggest

that the firm should be the direct beneficiary of a cus-tomer’s gratitude associated with a sweethearting exchange.However, as sociologists note, gratitude is neither efficientnor rational (Camerer 1988). Gratitude is tied closely toassessments of the sacrifice and effort required to provide agift, whereas monetary value is a secondary considerationin assigning indebtedness.

In the case of sweethearting, although the firm bears thefinancial burden of the exchange, the employee is responsi-ble for the risk and sacrifice associated with the provisionof the free service. Therefore, we expect the employee toabsorb the customer’s goodwill, which is then transferred tothe firm. Similar relationships have been documented in thecustomer relationship management and sales literatures(e.g., Bendapudi and Leone 2002; Reynolds and Beatty1999) whereby satisfaction with the employee is transferredto aggregated evaluations of the firm (Reynolds and Beatty1999).

H11: Satisfaction with the employee fully mediates the effectof sweethearting on customer evaluations of the firm(firm satisfaction, loyalty, and positive WOM intentions).

Methods and ResultsCustomer sampling. We derived the customer data from

the same collection effort used to gather the employee data.Because our primary objectives were to (1) explore differ-ences in employee and firm evaluations across sweetheartsand nonsweethearts and (2) test the process by whichinvolvement in a sweethearting transaction affects firm-level evaluations, we designed the customer data collectionas a quasi-natural experiment (Meyer 1995). Specifically,we employed a field manipulation in which we explicitlytargeted two types of customers: sweethearts and a controlgroup. This approach allowed for a controlled test of sweet-hearting effects by limiting the potential noise associatedwith a population of randomly selected customers.

To implement this design, employees identified in Part1 of the study were instructed to give surveys described ascustomer satisfaction assessments to two customersinvolved in a sweethearting incident and two customerswho were not involved. Customers who were not membersof a sweethearting dyad were control participants used todevelop baseline levels for evaluations of the employeesand firms. All surveys were identical with the exception ofa numeric label that allowed for the identification of dyadmembers but not for the identification of the names of theemployee or customers. All participants were asked to com-plete the survey, seal it in the provided envelope, and returnit to the researchers. This process resulted in 610 usablecustomer responses, of which 302 were sweetheart cus-tomers and 308 were control customers. The sweetheartcustomer group had an average age of 23.7 years, whereasthe control customer group had an average age of 24.9years.

Measures. We operationalized customer involvement ina sweethearting dyad using a binary variable created from acode embedded in the survey identification numbers: Wecoded customers involved in a dyad with a 1 and nonsweet-hearts (the control group) with a 0. We used established


measures from the customer satisfaction and loyalty litera-tures to measure the other constructs in the research model(see the Appendix). We used an inequity scale as a manipu-lation check of the field manipulation. As we expected, thesweetheart group reported significantly higher levels ofpositive inequity (Msweethearts = 7.14) than the control group(Mcontrol = 5.32; p < .01).

Assessments of measures. We conducted a comprehen-sive confirmatory factory analysis to assess the customermeasures. The model provided good fit to the data (2 =219.61, d.f. = 94; comparative fit index = .99; standardizedroot mean square residual = .02; and root mean square errorof approximation = .05). With respect to reliability andvalidity, all scales demonstrated adequate reliability, con-vergent validity, and discriminant validity according to For-nell and Larcker’s (1981) criteria. Table 1 provides com-plete scale statistics.

Tests for common method bias. Because we collectedthe measures of the endogenous variables from a singlesource, we assessed them for method bias using the sametests employed in the employee sample. The results sug-gested that method bias did not pose a risk to the interpreta-tion of the results, because neither the sign nor significanceof the values in the adjusted correlation matrix (second low-est r = .002) differed from the original correlation estimates.

We tested the customer consequences model in twosequential steps. First, we established baseline effects ofsweethearting on firm outcomes using multivariate analysisof variance. In the second phase, we used hierarchical linearmodeling (HLM) to examine the process by which thesepositive effects occurred.

Baseline effects. Our predictions rest on the assumptionthat sweethearting has positive effects on firm-level cus-tomer assessments through its impact on employee evalua-tions. We began by testing the firm-level effects using mul-tivariate analysis of variance, where the binarysweethearting involvement variable was the fixed factorand the employee and firm-level outcomes were the depen-dent variables. The results indicated that sweethearts hadsignificantly (p < .01) more favorable assessments of, andintentions toward, both the employees (Memployee satisfaction =8.00) and the firm (Mfirm satisfaction = 7.51, Mfirm loyalty =6.59, Mpositive WOM = 7.25) than nonsweethearts (Memployeesatisfaction = 7.11, Mfirm satisfaction = 7.04, Mfirm loyalty = 6.01,Mpositive WOM = 6.85). These results suggest that notaccounting for sweethearting in customer experience sur-veys could inflate a firm’s satisfaction, loyalty, and WOMscores by 6.7%, 9.6%, and 5.8%, respectively. Moreover,customer satisfaction with employees could be inflated by12.5%. With the baseline effects established, we turn to therelationships specified in the four research hypotheses.

Analysis of the research model. Given that customerresponses were nested within employees, we adopted HLM(see Raudenbush and Bryk 2002) to test H8–H11. The firstlevel of data consisted of the 610 customers who providedassessments of positive inequity, satisfaction with theemployee, satisfaction with the firm, loyalty to the firm, andpositive WOM intentions. The second level consisted of the

171 employees’ self-reports of sweethearting frequency. Asa result, the Level 1 data may vary within employees, andthe Level 2 results demonstrate differences betweenemployees not accounted for at the customer level.

We began by estimating null models and calculatingintraclass correlation coefficients for each Level 1 variable.This first step established the percentage of variance at eachlevel of analysis. Then, we estimated a series of equationsthat included all Level 1 predictors as fixed effects andsweethearting frequency as a Level 2 predictor of the Level1 intercept. This approach enables us to control for thepotential effects of sweethearting frequency on the meanvalues of all outcome variables, thus providing a more con-servative test of the Level 1 relationships. According to therecommendations of Hoffman, Griffin, and Gavin (2000),we centered Level 1 predictors at people’s (i.e., customers’)means and grand mean-centered the Level 2 variable(sweethearting frequency).

Variance decomposition. We estimated null models foreach endogenous Level 1 variable. The results suggestedthat little variance was present at Level 2 for inequity(.16%) and satisfaction with the employee (5.00%), but asignificant amount of variance existed at the employee levelfor the ultimate dependent variables (satisfaction with thefirm = 12.73%, loyalty to the firm = 14.91%, positiveWOM = 20.77%). Because of the variance at Level 2 forthe three outcome variables, HLM is an appropriate form ofanalysis, as it takes into account the nonindependence in thedata and allows for the controlled testing of varianceexplained by employee- and customer-level variables.

Hypothesis testing. We first tested the direct effects (H8and H9) by estimating two equations. The results, shown inthe top panel of Table 3, provide support for both H8 andH9. Specifically, sweethearting involvement had a signifi-cant and direct effect on perceptions of positive inequity(H8). Because of the binary coding of the sweetheartinginvolvement variable, the unstandardized coefficient of1.92 indicates that perceptions of inequity averaged 1.92scale points higher for sweethearts than for nonsweethearts,and this difference was significant (p < .01), in support ofH8. Sweethearting involvement ( = 0.28, p < .05) also sig-nificantly influenced customer satisfaction with theemployee (H9). Because these results lie at Level 1, theyreveal that, for customers interacting with the sameemployee, sweethearts perceived greater positive inequityand were more satisfied with that employee than nonsweet-hearts, and these results were not confounded by differ-ences between employees.

Next, we formally examined the mediation effects intro-duced in H10 and H11 using Zhao, Lynch, and Chen’s (2010)approach. The results reveal that the indirect effects (throughsatisfaction with the employee) of both sweetheartinginvolvement and perceptions of inequity were significant onthe three outcome variables (p < .05). However, neitherindependent variable had a significant direct effect on theoutcomes (p > .05), which provides evidence of indirect-only mediation (i.e., full mediation) and support for H10 andH11. These results further suggest that omission of an alter-native mediator is unlikely (Zhao, Lynch, and Chen 2010).


The bottom panel of Table 3 presents complete results fromthe mediated model.

DiscussionClose customer–employee relationships are known toincrease customer satisfaction and stimulate positive WOM.However, our study reveals a potential dark side to theserelationships that presents itself in the form of the precisebenefits that close relationships are suggested to produce.Specifically, our results reveal that customers involved insweethearting dyads provide inflated satisfaction, loyalty,and positive WOM scores, any of which could bias bench-marking efforts or disrupt employee rewards programs. In aworst-case scenario, managers might reward the veryemployees responsible for up to 35% of profit losses.

Our employee model provides the first known insightsinto antecedents and moderators of sweethearting behaviorand, in turn, offers guidance to managers who may be strug-gling to address the problem. These findings are importantbecause sweethearting forces managers to maintain a pre-

TABLE 3HLM Results for Customer Consequences ModelA: Effects of Sweethearting Involvement on Perceptions

of Inequity and Satisfaction with the EmployeePredictor CoefficientDependent Variable = Perceptions of InequityIntercept (0) 6.30**Sweethearting frequency (Level 2) (01) .05Sweethearting involvement (1) 1.92**

Dependent Variable = Satisfaction with the EmployeeIntercept (0) 7.59**Sweethearting frequency (Level 2) (01) .04Sweethearting involvement (1) .28*Perceptions of inequity (2) .32**

B: Mediated ModelsPredictor CoefficientDependent Variable = Satisfaction with the FirmIntercept (0) 7.32**Sweethearting frequency (Level 2) (01) .08*Sweethearting involvement (1) –.04Perceptions of inequity (2) –.03Satisfaction with the employee (3) .69**

Dependent Variable = Loyalty to the FirmIntercept (0) 6.33**Sweethearting frequency (Level 2) (01) .07Sweethearting involvement (1) .03Perceptions of inequity (2) .02Satisfaction with the employee (3) .60**

Dependent Variable = Positive WOMIntercept (0) 7.02**Sweethearting frequency (Level 2) (01) .08Sweethearting involvement (1) .13Perceptions of inequity (2) .01Satisfaction with the employee (3) .61**

*p < .05.**p < .01.Notes: Coefficient = unstandardized regression coefficient obtained

in HLM.

carious balance between encouraging strong bonds withcustomers and the reality that some of these close relation-ships will hurt the firm’s bottom line. The results of ourmoderation analysis suggest that at least three measurableemployee traits buffer sweethearting frequency. Takentogether with the direct effects of other measureable or con-trollable factors, our results indicate that it may be possibleto curtail sweethearting before it begins. We discuss theimplications of these findings next.Managerial ImplicationsCustomer and employee deviant behaviors are an increasingproblem for retailers. For example, shrinkage and otherlosses at the point of sale increased 5.9% globally from2008 to 2009, accounting for 1.4% of global retail sales(Bamfield 2009). Retail managers responded to the problemin 2010 by increasing spending on loss prevention byalmost 10% (Bamfield 2010). Our research addresses aninconspicuous and elusive behavior that accounts for a largepercentage of these losses (Tarnowski 2008).

A general conclusion from our findings is that manag-ing sweethearting requires adjustments to established deter-rence strategies. Our results indicate that the threat of pun-ishment, even relatively severe punishment, has littleinfluence on sweethearting frequency. Moreover, becausesweethearting involves collaboration between customersand employees, traditional theft paradigms in whichemployees are pitted against devious customers do notapply. Our recommendations are aimed at managers look-ing for solutions to sweethearting that do not involveoppressive regulations for frontline employees. We alsooffer insights into the magnitude and management of theeffect of sweethearting on key customer metrics.

Profiling frontline employees. Because frontline serviceworkers represent the firm to customers and account for alarge portion of the variance in customer assessments (Rust,Zahorik, and Keiningham 1996), managers are interested inidentifying the trait profile of an ideal frontline worker.Prior research has addressed this topic largely from the per-spective of the “Big Five” personality traits, in which extro-version (Hurley 1998; Liao and Chuang 2004), agreeable-ness (Brown et al. 2002; Frei and McDaniel 1998), andconscientiousness (Brown et al. 2002) are hallmarks of suc-cessful frontline workers. Our research indicates that man-agers attempting to control sweethearting among frontlineworkers should consider a broad range of trait factors. Forexample, firms that use preemployment screening tests canhead off sweethearting if they add measures of personalethics and need for social approval and then target appli-cants who are high and low on these scales, respectively.Our findings show that these specific trait profiles bufferremuneration motivations for sweethearting. Along theselines, managers who want to allow for frontline workerautonomy while minimizing the frequency of sweetheartingactivity should consider avoiding applicants at the veryhigh end of the risk-seeking scale. Although these preven-tive steps might not eliminate sweethearting entirely, ourresearch indicates that they may provide an effective buffer


that circumvents the need to implement oppressive securitymeasures that alienate all frontline workers.

Frontline worker training. In response to the growingthreat from shrinkage losses, more than 90% of retail firmsincreased spending on worker training in 2010, and morethan 75% expect this trend to continue in 2011 (Bamfield2010). Our results for personal ethics indicate that man-agers should amend training programs to include discus-sions on personal integrity and the consequences of deviantbehaviors such as sweethearting. In line with the conclu-sions of Mazar and Ariely (2006), contextual cues can beimplemented at the point of sale that remind workers oftheir ethical obligations. Research has shown such cues toactivate employees’ self-awareness, which can reduce par-ticipation in deviant acts (Mazar and Ariely 2006).

Sweethearting and key customer metrics. The resultsfrom the customer model demonstrate that sweetheartingleads to significantly higher scores on customer metrics forboth the confederate employee and, to a lesser extent, theservice firm. Specifically, our results indicate that, on aver-age, sweethearting inflates firm-level satisfaction scores byapproximately 7%. A gap of this magnitude is greater thanthe relative difference in customer satisfaction between thebest and worst full-service restaurants in the American Cus-tomer Satisfaction Index. Moreover, sweethearting inflatescustomer satisfaction with confederate employees by morethan 12%. Therefore, frontline managers should be wary ofusing raw satisfaction scores as the lone criterion to rewardemployees. Adding internal criteria such as productivitymeasures and rewards for helping other staff members aregood ways to avoid rewarding the wrong behaviors.

Managers should also be aware of how sweetheartingdeterrence efforts can affect customer satisfaction pro-grams. The results from the mediation analyses in the cus-tomer data (H10 and H11) demonstrate that any positiveeffect of sweethearting is filtered through satisfaction withthe collusive employee. In other words, the positive firmbenefit seems to be tied to a frontline worker that the firmwould rather not employ. As a result, firms that terminatedeviant employees or invest in surveillance technology toeradicate sweethearting will likely experience a temporarydecline in frontline service measures. It is important formanagers to resist the temptation to overcorrect for short-term changes in satisfaction scores that may be symptomsof positive change in the organization.Theoretical Implications and Further ResearchThis study investigates a previously unexplored and costlybehavior that has a unique array of motivations and incor-porates concepts drawn from several literatures. Accord-ingly, there are numerous research implications, which wediscuss next.

Customer–employee relationships. Our investigationextends our understanding of the breadth and depth offrontline service exchanges. Whereas most prior researchon service interactions has focused on professional relation-ships that develop between customers and employees (e.g.,Price and Arnould 1999), our study recognizes that personalrelationships may exist before the professional relationship

begins. The potential for these existing relationships tochange the dynamics of the service exchange provides afruitful area for further research. For example, it may be thecase that services provided by friends produce beneficialeffects such as the ones noted for customer participation incoproduction (Bendapudi and Leone 2002).

Social exchange. Our study considers a broader set ofsocial exchange resources than those typically examined incustomer experience research. Like other customer– employeeexchanges, goods, services, and monetary payments aretransferred in sweethearting transactions. However, unlikenormal business transactions, social status, reciprocalexpectations, and supplemental employee compensation arealso transferred. These findings highlight the need to con-sider a wider range of resources when assessing marketingexchanges. Further research should examine these resourcesand their relative value to customers more deeply. In particu-lar, future studies should examine whether the type and levelof resource (e.g., a free physical good, a free service, anupgrade, a discount) influences customer metrics differently.

Boundary conditions. Our results reinforce the need toconsider boundary conditions when modeling frontlineinteractions and deviant behaviors. Employee trait variablesproved to have significant moderating effects on severaldirect relationships in the research model, and the regions-of-significance analyses give an indication of the point atwhich the relationships become nonsignificant. Thus, fail-ure to account for boundary conditions could lead to erro-neous conclusions about sweethearting motivations andtheir relative effects on behavior. Further research couldfocus on extending this line of inquiry to understand howother trait variables, such as desire for decision latitude(Bone and Mowen 2010), may affect sweethearting behav-ior. Further studies could also embrace experience samplingdesigns that focus on tracking sweethearting dyads overtime in an effort to understand how they develop and howcertain situational factors—or combinations thereof—triggera sweethearting incident. Longitudinal designs such as thiswould enable researchers to discern changes in sweetheart-ing frequencies within employees and examine how theymap to changes in customer satisfaction and loyalty scores.

Service theft. Finally, our research indicates that serviceemployees and customers may downplay the moral and ethi-cal ramifications of service theft compared with physicalgoods theft. This is particularly true for services with intan-gible core products. Because there is no obvious cost of goodssold, employees and customers may view service theft as a“no sum lost” scenario for the firm. A study that addressesthese viewpoints and identifies strategies to counteract themwould have significant value for service managers.

ConclusionUltimately, this research illuminates a dark side to closeemployee–customer relationships and suggests that market-ing researchers and managers should adopt a broader viewof social exchange in service transactions. Because of thepotential negative side effects of these commercial friend-ships, marketing managers must more carefully consider


how they recruit, train, and supervise customer contact per-sonnel. Doing so will allow firms to leverage the benefits ofclose relationships while limiting the potential damagingeffects of deviant acts such as sweethearting. We encouragefurther research on sweethearting and other topics thatbridge the divide between marketing and other disciplines.

Appendix: MeasurementEmployee Measures

Overview of measures. All constructs were measuredusing nine-point scales (1 = “strongly disagree,” and 9 =“strongly agree,” unless otherwise noted).

Sweethearting frequency. We measured sweetheartingfrequency using five items developed for this study andbased on measures for similar behaviors, such as altruisticproperty deviance (Hollinger, Slora, and Terris 1992),socially based theft (Hawkins 1984), and property deviance(Hollinger 1986). On the basis of the results of a pretest, weremoved one item, and the resultant five-item scale offereda pretest reliability estimate of .90. The items assessed thefrequency at which employees engaged in sweetheartingbehaviors and were anchored by 1 (“never”) and 9 (“all thetime”).

1. How often do you give away goods and services to friends?2. How often do you intentionally undercharge customers forservices?

3. How often do you hook your friends up with free or dis-counted goods and services?

4. How often do you provide some customers with goods andservices that they didn’t pay for?

5. How often do you hook up people that you like?Risk-seeking propensity. We measured risk-seeking

propensity using three items adapted from a scale devel-oped by Simpson and Joe (1992).

1. I like to take chances.2. I like to do things that are exciting.3. I like friends that are wild.Personal ethics. We measured personal ethics using

Vitell, Rallapalli, and Singhapakdi’s (1993) four-item scale.These items measured the extent to which an employeebelieves that a person should adhere to moral standards andregulations.

1. One should always adhere to all applicable laws and regulations.

2. I try to be ethical in everything that I do.3. One must always be honest in serving consumers.4. One should not knowingly participate in unethical behavior.Need for social approval. We measured need for social

approval using four items from a scale constructed byLennox and Wolfe (1984). The need-for-social-approvalscale measures the extent to which a person is driven by hisor her desire to be accepted by his or her peers.

1. The slightest look of disapproval in the eyes of a personwith whom I am interacting is enough to make me changemy approach.

2. It is important to me to fit into the group that I am with.3. My behavior often depends on how I feel others want me tobehave.

4. If I am the least bit uncertain as to how to act in a social sit-uation, I look to the behavior of others for cues.

Financial gain. We measured financial gain using threeitems developed for this study. We developed these itemsfrom interviews with service employees and managers, andthen we pretested them. We retained all items throughoutpretesting, and the resultant three-item scale offered apretest reliability estimate of .96. The items assessed theextent to which employees feel that they can make addi-tional money by engaging in sweethearting.

1. I can increase my pay by giving away free goods and services.2. When I hook my friends up, I make more money that way.3. By hooking up friends and acquaintances, I can make extramoney.

Reciprocity. We measured reciprocity using four itemsdeveloped for this study. Before conducting the final datacollection, we developed five items using interviews withservice employees and managers, and then we pretestedthem. On the basis of the results of a pretest, we removedone item, and the resultant four-item scale offered a pretestreliability estimate of .94. The items assess whether a recip-rocal exchange relationship may exist with other serviceemployees.

1. If I hook up a customer, then the customer will return thefavor when I visit them at work.

2. I hook up my friends because my friends hook me up inreturn.

3. By giving away free goods and services to friends oracquaintances, I know that they will be more likely to giveme something in return.

4. My friends and I hook each other up whenever we can.Job satisfaction. We measured job satisfaction using

three items from the scale used by Hartline and Ferrell(1996). The items were anchored by 1 (“very dissatisfied”)and 9 (“very satisfied”). These items assessed the degree towhich the employee was satisfied with various aspects oftheir job.

1. How satisfied are you with your overall job?2. How satisfied are you with your supervisor(s)?3. How satisfied are you with your organizational policies?Organizational commitment. We measured organiza-

tional commitment using four items from studies by Hunt,Chonko, and Wood (1985) and Hunt, Wood, and Chonko(1989).

1. I would be willing to change companies if the new joboffered a pay increase. (reverse scored)

2. I would be willing to change companies if the new joboffered more creative freedom. (reverse scored)

3. I would be willing to change companies if the new joboffered more status. (reverse scored)

4. I would be willing to change companies if the new job waswith people who were more friendly. (reverse scored)


Work group norms. We measured work group normsusing four items based on a scale introduced by Beattie,Longabough, and Fava (1992). These items assessed theextent to which sweethearting was a common behavior inthe workplace.

1. Members of my work group typically talk about hooking uptheir friends with free goods and services.

2. The people I work with usually give discounts to peoplethey know.

3. It is normal for employees of my firm to give productsaway to customers that they like.

4. If another employee saw me giving away free products,s/he would react positively.

Job control. We measured job control using three itemsfrom Dwyer and Ganster’s (1991) scale. The items wereanchored by 1 (“none”) and 9 (“very much”).

1. In general, how much overall control do you have overwork and work-related manners?

2. How much control do you have over job tasks?3. How much control do you have over the amount you earn atyour job?

Punishment severity. We measured punishment severityusing five items based on the work of Hollinger and Clark(1983). Employees were provided with a series of activitiesrelated to sweethearting behaviors and were asked to iden-tify how severe the punishment would be for the behaviordescribed in each of the five items. We assessed the itemsusing a nine-point scale that was anchored by 1 (“nothing atall”) and 9 (“fired”).

Which response reflects the most common punishmentyou would receive if you were caught…

1. Giving away goods and services to friends.2. Providing products to certain customers at unauthorizeddiscounts.

3. Intentionally undercharging customers for their service.4. Hooking your friends up with discounted goods and services.5. Providing some customers with goods and services thatthey didn’t pay for.

Punishment certainty. We measured punishment cer-tainty using five items adapted from a scale developed byRobinson and O’Leary-Kelly (1998) that assesses the riskand likelihood of detection for theft behavior. We assessedthe items using a nine-point scale anchored by 1 (“veryunlikely”) and 9 (“very likely”).

How likely is it that you would get punished for intentionally…

1. Giving away goods and services to friends.2. Providing products to certain customers at unauthorizeddiscounts.

3. Intentionally undercharging customers for their service.4. Hooking your friends up with discounted goods and services.5. Providing some customers with goods and services thatthey didn’t pay for.

Control variables. We also included control variables inthe survey instrument. Specifically, we included one ques-tion to assess each of the following control variables.

1. Industry: Please describe the type of service firm where youare employed.

2. Tenure: How long (in months) have you been working atthis firm?

3. Interpersonal Equity: My supervisor does not treat mefairly.

4. Guilt: I feel guilty about hooking up my friends.

Customer MeasuresOverview of measures. We measured all constructs

using nine-point scales anchored by 1 (“strongly disagree”)and 9 (“strongly agree”), unless otherwise noted.

Involvement in sweethearting. We coded involvement insweethearting following the data collection based on identi-fication numbers in consumer surveys. Customers involvedin a sweethearting relationship were coded as 1, and cus-tomers not involved in sweethearting were assigned 0.

Perceptions of inequity. We measured perceptions ofinequity using four semantic differential items based on astudy by Oliver and Swan (1989) that assessed the extent towhich a customer received inequitable treatment.

1. Other customers got a better deal than me. … I got a betterdeal than other customers.

2. Other customer got more than they deserved. … I got morethan I deserved.

3. Other customers got more benefits than me. … I got morebenefits than other customers.


4. Other customers received greater outcomes. … I receivedgreater outcomes.

Customer satisfaction. We measured customer satisfac-tion with employees and the firm with three items fromOliver’s (1980) consumption satisfaction scale.

1. I am happy that I got service from this employee (servicefirm).

2. I am satisfied with my decision to get service from thisemployee (service firm).

3. I think I did the right thing when I got service from thisemployee (service firm).

Customer loyalty. We measured customer loyalty to thefirm using three items based on Zeithaml, Berry, and Para-suraman’s (1996) behavioral intentions battery.

1. I am loyal to this service firm.2. This service firm is my first choice when I purchase thistype of service.

3. I am dedicated to doing business with this service firm.Positive WOM intentions. We measured positive WOM

intentions using three items based on Zeithaml, Berry, andParasuraman’s (1996) behavioral intentions battery.

1. I will recommend visiting this service firm to friends.2. I will say good things about this service firm to others.3. I will encourage friends and relatives to visit this servicefirm.

Experiences,” Journal of Consumer Research, 20 (December),393–417.

Bendapudi, Neeli and Leonard L. Berry (1997), “Customers’Motivations for Maintaining Relationships with ServiceProviders,” Journal of Retailing, 73 (1), 15–37.

——— and Robert P. Leone (2002), “Managing Business-to-BusinessCustomer Relationships Following Key Contact EmployeeTurnover in a Vendor Firm,” Journal of Marketing, 66 (April),83–101.

Bennett, Rebecca J. and Sandra L. Robinson (2000), “Develop-ment of a Measure of Workplace Deviance,” Journal ofApplied Psychology, 85 (3), 349–60.

Berry, Leonard L. (1995), On Great Service. New York: The FreePress

Biernacki, Patrick and Dan Waldorf (1981), “Snowball Sampling:Problem and Techniques of Chain Referral Sampling,” Socio-logical Methods and Research, 10 (2), 141–63.

Bolin, Aaron and Linette Heatherly (2001), “Predictors ofEmployee Deviance: The Relationship Between Bad Attitudesand Bad Behaviors,” Journal of Business and Psychology, 15(3), 405–418.

Bone, Sterling A. and John C. Mowen (2010), “‘By-the-Book’Decision Making: How Service Employee Desire for DecisionLatitude Influences Customer Selection Decisions,” Journal ofService Research, 13 (2), 184–97.

Bove, Liliana L. and Lester W. Johnson (2006), “Customer Loy-alty to One Service Worker: Should It Be Discouraged?” Inter-national Journal of Research in Marketing, 23 (1), 79–91.

Brown, Stephen W., Deborah L. Cowles, and Tracy L. Tuten(1996), “Service Recovery: Its Value and Limitations as aRetail Strategy,” International Journal of Service IndustryManagement, 7 (5), 32–46.

REFERENCESAdams, J. Stacey (1963), “Toward an Understanding of Inequity,”

Journal of Abnormal and Social Psychology, 67 (5), 422–36.Aiken, Leona S. and Stephen G. West (1991), Multiple Regres-

sion: Testing and Interpreting Interactions. Thousand Oaks,CA: Sage Publications.

Amato-McCoy, Deena M. (2009), “The ‘Not-So Sweet’ Side ofRetail Loss,” Chain Store Age, 85 (7), 38–39.

Astor, Saul D. (1972), “Who’s Doing the Stealing?” ManagementReview, 61, 34–35.

Bamfield, Joshua (2009), Global Retail Theft Barometer. Notting-ham, UK: Centre for Retail Research

——— (2010), Global Retail Theft Barometer. Nottingham, UK:Centre for Retail Research.

Baron, Robert A. and Joel H. Neuman (1996), “Workplace Vio-lence and Workplace Aggression: Evidence on Their RelativeFrequency and Potential Causes,” Aggressive Behavior, 22 (3),161–73.

Bauer, Daniel J. and Patrick J. Curran (2005), “Probing Interac-tions in Fixed and Multilevel Regression: Inferential andGraphical Techniques,” Multivariate Behavioral Research, 40(3), 373–400.

Beattie, Martha C., Richard Longabaugh, and Joseph Fava (1992),“Assessment of Alcohol-Related Workplace Activities: Devel-opment and Testing of Your Workplace,” Journal of Studies onAlcohol, 53 (5), 469–75.

Beatty, Sharon E., Morris Mayer, James E. Coleman, Kristy EllisReynolds, and Jungki Lee (1996), “Customer–Sales AssociateRetail Relationships,” Journal of Retailing, 72 (3), 223–47.

Belk, Russell W. (1976), “It’s the Thought that Counts: A SignedDigraph Analysis of Gift-Giving,” Journal of ConsumerResearch, 3 (December), 155–62.

——— and Gregory S. Coon (1993), “Gift Giving as Agapic Love:An Alternative to the Exchange Paradigm Based on Dating

Brown, Tom J., John Mowen, D. Todd Donavan, and Jane W.Licata (2002), “The Customer Orientation of Service Workers:Personality Trait Influences on Self and Supervisor Perfor-mance Ratings,” Journal of Marketing Research, 39 (Febru-ary), 110–20.

Camerer, Colin (1988), “Gifts as Economic Signals and SocialSymbols,” American Journal of Sociology, 94 (Supplement),180–214.

Cialdini, Robert B. (2007), “Descriptive Social Norms as Under-appreciated Sources of Social Control,” Psychometrika, 72 (2),263–68.

Colbert, Amy E., Michael K. Mount, James K. Harter, L.A. Witt,and Murray R. Barrick (2004), “Interactive Effects of Personal-ity and Perceptions of the Work Situation on WorkplaceDeviance,” Journal of Applied Psychology, 89 (4), 599–609.

Cole, Catherine A. (1989), “Deterrence and Consumer Fraud,”Journal of Retailing, 65 (Spring), 107–120.

Cullen, Michael J. and Paul R. Sackett (2003), “Personality andCounterproductive Work Behavior,” in Personality and Work:Reconsidering the Role of Personality in Organizations, Mur-ray R. Barrick and Ann Marie Ryan, eds. San Francisco:Jossey-Bass, 150–82.

Czepiel, John A. (1990), “Service Encounters and Service Rela-tionships: Implications for Research,” Journal of BusinessResearch, 20 (1), 13–21.

Dannen, Chris (2009), “Stoplift’s Grocery Cameras Peg ‘Sweet-hearting’ Cashiers,” Fast Company, (May 11), (accessed August11), [available at http:// www. fastcompany. com/ blog/ chris- dannen/ techwatch/grocery-cameras-identify-sweethearting].

DeWitt, Tom and Michael K. Brady (2003), “Rethinking ServiceRecovery Strategies: The Effect of Rapport on ConsumerResponses to Service Failure,” Journal of Service Research, 6(2), 193–207.

Dwyer, Deborah J. and Daniel C. Ganster (1991), “The Effects ofJob Demands and Control on Employee Attendance and Satis-faction,” Journal of Organizational Behavior, 12 (7), 595–608.

Emmons, Robert A. (2004), “The Psychology of Gratitude,” inPsychology of Gratitude, Robert A. Emmons and Michael E.McCullough, eds. New York: Oxford University Press, 3–18.

Fornell, Claes and David F. Larcker (1981), “Evaluating StructuralEquation Models with Unobservable Variables and Measure-ment Error,” Journal of Marketing Research, 18 (February),39–50.

Frei, Richard L. and Michael A. McDaniel (1998), “Validity ofCustomer Service Measures in Personnel Selection: A Reviewof Criterion and Construct Evidence,” Human Performance, 11(1), 1–27.

Fullerton, Ronald A. and Girish Punj (2004), “Repercussions ofPromoting an Ideology of Consumption: Consumer Misbehav-ior,” Journal of Business Research, 57 (11), 1239–49.

Giacalone, Robert A. and Jerald Greenberg (1997), Antisocial Behav-ior in Organizations. Thousand Oaks, CA: Sage Publications.

Glaser, Barney and Anselm Strauss (1967), The Discovery ofGrounded Theory. Chicago: Aldine.

Gottfredson, Michael R. and Travis Hirschi (1990), A GeneralTheory of Crime. Stanford, CA: Stanford University Press.

Gouldner, Alvin W. (1960), “The Norm of Reciprocity: A Prelimi-nary Statement,” American Sociological Review, 25 (2),161–78.

Greenberg, Jerald (1997), “The STEAL Motive: Managing theSocial Determinants of Employee Theft,” in Antisocial Behav-ior in Organizations, Robert A. Giacalone and Jerald Green-berg, eds. Thousand Oaks, CA: Sage Publications, 85–108.

——— (2002), “Who Stole the Money, and When? Individual andSituational Determinants of Employee Theft,” OrganizationalBehavior and Human Decision Processes, 89 (1), 985–1003.

——— and Kimberly S. Scott (1996), “Why Do Workers Bite theHands That Feed Them? Employee Theft as a Social Exchange


Process,” in Research in Organizational Behavior, Vol. 18,B.L. Staw and L.L. Cummings, eds. Greenwich, CT: JAI Press,111–55.

Greenberg, Liane and Julian Barling (1996), “Employee Theft,” inTrends in Organizational Behavior, C.L. Cooper and D.M.Rousseau, eds. New York: John Wiley & Sons, 49–64.

Gremler, Dwayne D. and Kevin P. Gwinner (2000), “Rapport-Building Behaviors Used by Retail Employees,” Journal ofRetailing, 84 (3), 308–324.

——— and ——— (2008), “Customer–Employee Rapport in Ser-vice Relationships,” Journal of Service Research, 3 (1),82–104.

Griffiths, Paul, Michael Gossop, Beverly Powis, and John Strang(1993), “Reaching Hidden Populations of Drug Users by Privi-leged Access Interviewers: Methodological and PracticalIssues,” Addiction, 88 (12), 1617–26.

Groth, Markus, Thorsten Hennig-Thurau, and Gianfranco Walsh(2009), “Customer Reactions to Emotional Labor: The Rolesof Employee Acting Strategies and Customer Detection Accu-racy,” Academy of Management Journal, 52 (5), 958–74.

Harris, Lloyd C. and Emmanuel Ogbonna (2002), “Exploring Ser-vice Sabotage: The Antecedents, Types, and Consequences ofFrontline, Deviant, Antiservice Behaviors,” Journal of ServiceResearch, 4 (3), 163–83.

——— and ——— (2006), “Service Sabotage: A Study ofAntecedents and Consequences,” Journal of the Academy ofMarketing Science, 34 (4), 543–58.

——— and ——— (2009), “Hiding Customer Complaints: Studyingthe Motivations and Forms of Service Employees’ ComplaintConcealment Behaviors,” British Journal of Management, 21(2), 262–79.

Hartline Michael D. and O.C. Ferrell (1996) “The Management ofCustomer Contact Service Employees: An Empirical Investiga-tion,” Journal of Marketing, 60 (October), 52–70.

Hawkins, Richard (1984), “Employee Theft in the RestaurantTrade: Forms of Ripping Off by Waiters at Work,” DeviantBehavior, 5 (1–4), 47–69.

Henle, Christine A. (2005), “Predicting Workplace Deviance fromthe Interaction Between Organizational Justice and Personal-ity,” Journal of Managerial Issues, 18 (2), 247–63.

Hoffman, David A., Mark A. Griffin, and Mark B. Gavin (2000),“The Application of Hierarchical Linear Modeling to Organi-zational Research,” in Multilevel Theory, Research, and Meth-ods in Organizations: Foundations, Extensions, and NewDirections, K.J. Klein and S.W.J. Kozlowski, eds. San Fran-cisco: Jossey-Bass, 467–511.

Hollinger, Richard C. (1986), “Acts Against the Workplace: SocialBonding and Employee Deviance,” Deviant Behavior, 7 (1),53–75.

——— and John P. Clark (1983), Theft by Employee. Lexington,MA: Lexington Books.

———, Karen B. Slora, and William Terris (1992), “Deviance inthe Fast-Food Restaurant: Correlates of Employee Theft, Altru-ism, and Counterproductivity,” Deviant Behavior, 13 (2),155–84.

Hunt, Shelby D., Lawrence B. Chonko, and Van R. Wood (1985),“Organizational Commitment and Marketing,” Journal of Mar-keting, 49 (Winter), 112–26.

———, Van R. Wood, and Lawrence B. Chonko (1989), “Corpo-rate Ethical Values and Organizational Commitment in Market-ing,” Journal of Marketing, 53 (July), 79–90.

Hurley, Robert F. (1998), “Customer Service Behavior in RetailSettings: A Study of the Effect of Service Provider Personal-ity,” Journal of the Academy of Marketing Science, 26 (2),115–27.

Jermier, John (1988), “Sabotage at Work: The Rational View,” inResearch in the Sociology of Organizations, Vol. 6, Samuel B.Bachrach, ed. Greenwich, CT: JAI Press, 101–134.

Kantor, Seth (1983), “How to Foil Employee Crime,” Nation’sBusiness, 71 (July), 38–39.

Keh, Hean Tat and Jun Pang (2010), “Customer Reactions to Ser-vice Separation,” Journal of Marketing, 74 (March), 55–70.

Lau, Vivian C.S., Wing Tung Au, and Jane M.C. Ho (2003), “AQualitative and Quantitative Review of Antecedents of Coun-terproductive Behavior in Organizations,” Journal of Businessand Psychology, 18 (Fall), 73–98.

Lennox, Richard and Raymond Wolfe (1984), “Revision of theSelf-Monitoring Scale,” Journal of Personality and Social Psy-chology, 46 (6), 1349–64.

Lerner, Jennifer S. and Dacher Keltner (2001), “Fear, Anger, andRisk,” Journal of Personality and Social Psychology, 81 (1),146–59.

Liao, Hui and Aichia Chuang (2004), “A Multi-Level Investiga-tion of Factors Influencing Employee Service Performance andCustomer Outcomes,” Academy of Management Journal, 47(1), 41–58.

———, Aparna Joshi, and Aichia Chuang (2004), “Sticking outLike a Sore Thumb: Employee Dissimilarity and Deviance atWork,” Personnel Psychology, 57 (4), 969–1000.

Lindell, Michael K. and D.J. Whitney (2001), “Accounting forCommon Method Variance in Cross-Sectional ResearchDesigns,” Journal of Applied Psychology, 86 (1), 114–21.

Lykken, David T. (1995), The Antisocial Personalities. Mahwah,NJ: Lawrence Erlbaum Associates.

Malhotra, Naresh, Sung S. Kim, and Ashutosh Patil (2006), “Com-mon Method Variance in IS Research: A Comparison of Alter-native Approaches and Reanalysis of Past Research,” Manage-ment Science, 52 (12), 1865–83.

Matilla, Anna (2001), “The Effectiveness of Service Recovery in aMulti-Industry Setting,” Journal of Services Marketing, 15 (7),583–96.

Mavondo, Felix T. and Elaine M. Rodrigo (2001), “The Effect ofRelationship Dimensions on Interpersonal and Interorganiza-tional Commitment in Organizations Conducting BusinessBetween Australia and China,” Journal of Business Research,52 (2), 111–21.

Mazar, Nina, On Amir, and Dan Ariely (2008), “The Dishonestyof Honest People: A Theory of Self-Concept Maintenance,”Journal of Marketing Research, 45 (December), 633–44.

——— and Dan Ariely (2006), “Dishonesty in Everyday Life andIts Policy Implications,” Journal of Public Policy & Market-ing, 25 (Spring), 117–26.

McColl-Kennedy, Janet, Paul G. Patterson, Amy K. Smith, andMichael K. Brady (2009), “Customer Rage Episodes: Emo-tions, Expressions and Behaviors,” Journal of Retailing, 85(2), 222–37.

McCullough, Michael E., Shelley D. Kilpatrick, Robert A.Emmons, and David B. Larson (2001), “Is Gratitude a MoralAffect?” Psychological Bulletin, 127 (2), 249–66.

Merton, Robert K. (1938), “Social Structure and Anomie,” Ameri-can Sociological Review, 3 (October), 672–82.

Meyer, Bruce D. (1995), “Quasi and Natural Experiments in Eco-nomics,” Journal of Business and Economic Statistics, 13 (2),151–61.

Miner, John B. and Michael H. Capps (1996), How Honesty Test-ing Works. Westport, CT: Quorum Books.

Murphy, Kevin R. (1993), Honesty in the Workplace. PacificGrove, CA: Brooks/Cole.

Oliver, Richard L. (1980), “A Cognitive Model of the Antecedentsand Consequences of Satisfaction Decisions,” Journal of Mar-keting Research, 17 (November), 460–69.

——— (1997), Satisfaction: A Behavioral Perspective on the Con-sumer. New York: McGraw-Hill.

——— and John E. Swan (1989), “Consumer Perceptions of Inter-personal Equity and Satisfaction in Transactions: A Field Sur-vey Approach,” Journal of Marketing, 53 (April), 21–35.


Osgood, Wayne D., Lloyd D. Johnston, Patrick M. O’Malley, andJerald G. Bachman (1988), “The Generality of Deviance inLate Adolescence and Early Adulthood,” American Sociologi-cal Review, 53 (1), 81–93.

Peterson, Robert A. and William R. Wilson (1992), “MeasuringCustomer Satisfaction: Fact and Artifact,” Journal of the Acad-emy of Marketing Science, 20 (1), 61–71.

Price, Linda L. and Eric J. Arnould (1999), “Commercial Friend-ships: Service Provider–Client Relationships in Context,”Journal of Marketing, 63 (October), 38–56.

Randall, Donna M. (1987), “Commitment and the Organization:The Organization Man Revisited,” Academy of ManagementReview, 12 (3), 460–71.

Raudenbush, Stephen W. and Anthony S. Bryk (2002), Hierarchi-cal Linear Models: Applications and Data Analysis Methods.Newbury Park, CA: Sage Publications.

Reichheld, Frederick (1996), “Learning from Customer Defec-tions,” Harvard Business Review, 74 (2), 56–70.

Reynolds, Kristy E. and Sharon E. Beatty (1999), “CustomerBenefits and Company Consequences of Customer–SalespersonRelationships in Retailing,” Journal of Retailing, 75 (1),11–32.

Robinson, Sandra L. and Jerald Greenberg (1998), “EmployeesBehaving Badly: Dimensions, Determinants, and Dilemmas inthe Study of Workplace Deviance,” in Trends in Organiza-tional Behavior, C. Cooper and D. Rousseau, eds. New York:John Wiley & Sons, 1–30.

——— and Anne M. O’Leary-Kelly (1998), “Monkey See, Mon-key Do: The Influence of Work Groups on the AntisocialBehavior of Employees,” Academy of Management Journal, 41(6), 84–85.

Rust, Roland T., Anthony J. Zahorik, and Timothy L. Keiningham(1996), Service Marketing. New York: HarperCollins.

Sackett, Paul R. (1994), “Integrity Testing for Personnel Selec-tion,” Current Directions in Psychological Science, 3 (3),73–76.

Simpson, D. Dwayne and George W. Joe (1992), “Measurement ofSocial Functioning in Addicts: Psychometric and ValidityAnalyses” (IBR Technical Report), Texas Christian University,Institute of Behavioral Research.

Singh, Jagdip (2000), “Performance Productivity and Quality ofFrontline Employees in Service Organizations,” Journal ofMarketing, 64 (April), 15–34.

Skarlicki, Daniel P. and Robert Folger (1997), “Retaliation in theWorkplace: The Roles of Distributive, Procedural, and Interac-tional Justice,” Journal of Applied Psychology, 82 (June),434–43.

Spiggle, Susan (1994), “Analysis and Interpretation of QualitativeData in Consumer Research,” Journal of Consumer Research,21 (3), 491–504.

Strutton, David and James R. Lumpkin (1992), “RelationshipBetween Optimism and Coping Strategies in the Work Envi-ronment,” Psychological Reports, 71 (December), 1179–86.

Szockyj, Elizabeth and Gilbert Geis (2002), “Insider Trading Pat-terns and Analysis,” Journal of Criminal Justice, 30 (4),273–86.

Tarnowski, Joseph (2008), “Good Night, Sweethearting,” Progres-sive Grocer, 87 (6), 126.

Tax, Stephen S. and Stephen W. Brown (1998), “Recovering andLearning From Service Failure,” Sloan Management Review,40 (1), 75–88.

Terris, William and John Jones (1982), “Psychological FactorsRelated to Employee Theft in the Convenience Store Industry,”Psychological Reports, 51 (3), 1219–38.

Tittle, Charles (1980), Sanctions and Social Deviance. Westport,CT: Praeger Publishing.

Trevino, Linda K. and Bart Victor (1992), “Peer Reporting ofUnethical Behavior: A Social Context Perspective,” Academyof Management Journal, 35 (1), 38–64.

Vitell, Scott J., Kumar Rallapalli, and Anusorn Singhapakdi(1993), “Marketing Norms: The Influences of Personal MoralPhilosophies and Organizational Ethical Culture,” Journal ofthe Academy of Marketing Science, 21 (Fall), 331–37.

Warren, Daniel E. (2003), “Constructive and Destructive Deviancein Organizations,” Academy of Management Review, 28 (4),622–32.

Wirtz, Jochen and Doreen Kum (2004), “Consumer Cheating onService Guarantees,” Journal of the Academy of MarketingScience, 32 (2), 159–75.


Zeithaml, Valarie A., Leonard Berry, and A. Parasuraman (1996),“The Behavioral Consequences of Service Quality,” Journal ofMarketing, 60 (April), 31–46.

Zhao, Xinshu, John G. Lynch, and Qimei Chen (2010), “Reconsid-ering Baron and Kenny: Myths and Truths About MediationTests,” Journal of Consumer Research, 37 (August), 197–206.

Zinkhan, George M., Scot Burton, and Melanie Wallendorf(1983), “Marketing Applications for Snowball Sampling:Potential Benefits and Problems,” in Research Methods andCausal Modeling in Marketing, Vol. 1, W.R. Darden, K.B.Monroe, and W.R. Dillon, eds. Chicago: American MarketingAssociation, 5–8.

Zuckerman, Marvin (1980), “Sensation Seeking and Its BiologicalCorrelates,” Psychological Bulletin, 88 (1), 187–214.

Copyright of Journal of Marketing is the property of American Marketing Association and its content may not

be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.

8755263

Documents

Transcript of 8755263