Journal of Business Research 66 (2013) 2163–2170

Contents lists available at SciVerse ScienceDirect

Journal of Business Research

To be engaged or not to be engaged: The antecedents and consequences of serviceemployee engagement

Bulent Menguc a,⁎, Seigyoung Auh b,1, Michelle Fisher a, Abeer Haddad a

a Brock University, Faculty of Business, Department of Marketing, Int'l Business, and Strategy, St. Catharines, Ontario, Canada L2S3A1b Thunderbird School of Global Management, 1 Global Place, Glendale, AZ 85306, USA

⁎ Corresponding author. Tel.: +1 905 688 5550x5074E-mail addresses: bmenguc@brocku.ca (B. Menguc),

seigyoung.auh@thunderbird.edu (S. Auh).1 Tel.: +1 602 978 7296.

0148-2963/$ – see front matter © 2012 Elsevier Inc. Alldoi:10.1016/j.jbusres.2012.01.007

a b s t r a c t

a r t i c l e i n f o

Article history:Received 2 March 2011Accepted 18 January 2012Available online 7 February 2012

Keywords:Employee engagementJob Demand-Resources (JD-R) modelSupervisory supportSupervisory feedbackPerceived autonomyService employee performance

Drawing on the Job Demand-Resource (JD-R) model, this study explores the antecedents and consequencesof service employee engagement. The model examines the main effect of resources (autonomy, feedback, andsupport) on engagement and how the interaction among resources impacts engagement. Further, the modelalso examines the mediating role of engagement in linking resources to customers' perceived level of serviceemployee performance. The study uses multi-level modeling on data from 482 service employees and cus-tomers in 66 retail stores. Results suggest that supervisory feedback is positively related to engagementwhile supervisory support is not. More engagement is related to more positive service employee perfor-mance. Regarding the interactions, supervisory support had a positive effect while supervisory feedbackhad a negative effect on engagement at high levels of perceived autonomy. Also, engagement was a full me-diator between supervisory feedback and service employee performance. Implications for retail service man-agement are discussed.

© 2012 Elsevier Inc. All rights reserved.

1. Introduction

Harter, Schmidt, and Hayes's meta-analysis (2002) reports thatemployee engagement is positively associated with customer satisfac-tion, customer loyalty, productivity, and profitability, and negativelyassociated with employee turnover. Also, according to a global work-force study conducted by Towers Perrin (2008), firms that employedhighly engaged employees enjoyed a spread of more than 5% in operat-ingmargin and 3% in netmargin compared to companies that employedhighly disengagedworkers. Despite its theoretical andmanagerial ram-ifications, however, service employee engagement has received limitedempirical examination in the services marketing literature.

Against this backdrop, we draw on the Job Demand-Resources (JD-R)model (Ashill & Rod, 2011; Babakus, Yavas, & Ashill, 2009; Bakker &Demerouti, 2007; Demerouti, Bakker, Nachreiner, & Schaufeli, 2001) todevelop and test a conceptual model that explicates how employeeswho receive resources from management can provide superior serviceperformance (e.g., providing prompt service, identifying products thatfit customers' needs), as assessed by customers, via employee engage-ment.We also suggest that perceived autonomymoderates the relation-ship between such resources as supervisory support and supervisoryfeedback and engagement.

; fax: +1 905 984 4188.

rights reserved.

Upon careful review of the relevant literature, we identify foursignificant gaps. First, the challenging job demands of satisfyingcustomers' requests while fulfilling management's expectations,coupled with scarce job resources (e.g., autonomy, supervisory feed-back, and supervisor support), make engagement seem very difficultto come by in the services context (Ashill & Rod, 2011). When em-ployees are disengaged, their negative mindset can be contagiousand affect how they treat and serve customers. Consequently, it isvital that we understand what types of organizational, task, and so-cial resources need to be in place to encourage service employeeengagement.

Second, most studies that have examined the consequences of en-gagement have focused on engagement from an internal perspectiveby studying how engagement affects employee attitude and perfor-mance (e.g., Rich, LePine, & Crawford, 2010). In contrast, we exploreemployee engagement from an external perspective to understandhow employee engagement affects customers' views of the service per-formance they receive.

Third, drawing on the previous two points, resources may be toofar removed (i.e., distal) to predict employee performance and weposit that engagement, as a mediator, is a more proximal constructto employee performance that can link resources to customers' eval-uation of service employee performance.

Fourth, while it is important to understand which resources posi-tively affect engagement, in practice, managers typically tend to employmultiple resources simultaneously. As a result, moving from an “addi-tive” model to an “interaction” model can shed light on how differentresources work together or against each other in affecting engagement.

2164 B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

No research to date has examined how the simultaneous deployment ofmultiple resources affects engagement.

In the sections to follow,we beginwith an explanation of our concep-tual model (Fig. 1) and develop our hypotheses, followed by hypothesestesting, which we perform based on matched data collected from twosources (customer service employees and customers) from a nationalchain of specialty retailers. We conclude with implications, limitations,and directions for future research.

2. Conceptual background and hypotheses

Drawing on the JD-R model, we argue that service employees'engagement will be affected by the resources they receive frommanagement. The JD-R model is a conceptual framework used to ex-plain employee engagement in the workplace (Bakker & Demerouti,2007; Demerouti et al., 2001). The model asserts that resources arewhat the organization provides to its employees; those resourcescan include autonomy, supervisory support, or supervisory feed-back. Such resources are expected to (a) lessen the strain of thejob's demands and the psychological and physiological costs thataccompany these arduous demands, (b) be instrumental in theachievement of employee work objectives, and (c) motivate andstimulate personal growth, development, and learning (Demeroutiet al., 2001). Our model attempts to explain how resources can beused to enhance engagement and how the interplay between vari-ous resources affects engagement.

The resources that we examine in this study are supervisory support,supervisory feedback, and perceived autonomy. Our model also assertsthat engaged employees will deliver customer-oriented behaviors andthat such behaviors will be reflected in customers' assessment of serviceemployee performance (Yoon & Suh, 2003; Young, Meterko, Mohr,Shwartz, & Lin, 2009). Further, we position engagement as a mediatorthat links resources and employee performance. Finally, we investigatehowperceived autonomy interactswith supervisory support and supervi-sory feedback in affecting engagement.

2.1. Engagement

A service employee who is engaged can be characterized as enthusi-astic, energetic, motivated, and passionate about his or her work,whereas a disengaged worker is one who is apathetic, robotic, deperso-nalized, estranged, and withdrawn from her or his job (Salanova, Agut,& Peirό, 2005). Engagement has been defined as “a positive, fulfilling,work-related state of mind that is characterized by vigor, dedication,and absorption” (Schaufeli, Salanova, Gonzalez-Roma, & Bakker, 2002,p. 74). Vigor refers to a willingness and determination to exert energyand effort in one's work and to be resilient and persistent when con-frontedwith obstacles. Dedication is analogous to an emotional compo-nent of engagement in that dedication refers to finding meaning and

Work

Supervisory Support

Supervisory Feedback

Employee Level Perceived Autonomy

H1 –H2

H5 –H6

H4

Fig. 1. Concept

purpose in one's work and being enthusiastic, inspired, and proud ofone'swork. Absorption parallels the cognitive component of engagement.Absorption refers to being totally immersed and content with one's worksuch that time passes quickly and to finding it difficult to detach oneselffrom work (Salanova et al., 2005). Consequently, engagement can becharacterized as a “persistent, positive affective-motivational state of ful-fillment” (Maslach, Schaufeli, & Leiter, 2001, p. 417). The literaturehas dif-ferentiated this notion of engagement from job involvement and jobsatisfaction. In fact, Rich et al. (2010) show that engagement affects taskperformance positively even after controlling for job involvement, job sat-isfaction, and intrinsic motivation.

2.1.2. The influence of supervisory support and supervisory feedback onwork engagement

Babin and Boles (1996, p. 60) define supervisory support as “thedegree to which employees perceive that supervisors offer employeessupport, encouragement and concern.” As employees perceive moresupervisory support, they feel more secure and sense that the firmtakes care of their welfare (DeConinck, 2010). According to the JD-Rmodel, supervisory support is a key resource that motivates em-ployees to be engaged in their workplace. Having a supervisor to fallback on under adverse circumstances and who is willing to listencan be a significant motivational boost for employees (DeConinck,2010). Further, supervisory support can alleviate some of the stressand strain imposed by the high demands associated with the job(Babin & Boles, 1996). Consequently, when employees feel that theyare furnished with adequate resources such as supervisory support,high job demands feel less daunting and employees remain engagedin their work (Sand & Miyazaki, 2000). Conversely, when supervisorysupport is lacking, employees question their value and contributionto the organization and feel detached, frustrated, and helpless.

H1. Supervisory support is positively related to engagement.Drawing on Jaworski and Kohli (1991), we define supervisory feed-

back as employees' perception that they are receiving clear informationabout their performance outcome and suggestions for improvement.When employees perceive sufficient developmental feedback, theyhave accurate guidance on how to become more effective (Jaworski &Kohli, 1991). This, in essence, fosters more communication betweenthe two parties and helps the firm [or supervisor] map out ways toimprove performance (Ashford & Cummings, 1983). When employeesperceive that they are receivingmore candid and accurate developmen-tal feedback, they sense that supervisors are interested in their growth,development, and learning (Ashford & Cummings, 1983). Providingcorrective measures to get employees back on track or reinforcingtheir effectiveness motivates employees to be more engaged. In con-trast, a lack of feedback can create ambiguity, conflict, and confusionabout what is expected (Jaworski & Kohli, 1991). The absence of devel-opmental feedback can create a lack of stimulation and fewer

Engagement

Customers’ Evaluation of

Service Employee Performance

H3

ual model.

2165B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

opportunities for change and innovation. Collectively, this can lead toless enthusiasm, energy, passion, and inspiration regarding the job.

H2. Supervisory developmental feedback is related positively toengagement.

2.1.3. Consequences of engagementAs employees become more engaged, they find their work more

meaningful, self-fulfilling, and inspirational and, accordingly, becomemore dedicated, concentrated, and engrossed in their jobs. This posi-tive and motivated state of mind should carry over to how they treatand serve customers. As such, research has shown that engagementinfluences not only in-role behavior but also proactive behavior(Sonnentag, 2003) and extra-role behavior such as organizationalcitizenship behavior (Rich et al., 2010). It seems that engagedemployees have an expanded view of their job role and reach out toa broader set of activities in their jobs. This suggests that, at thevery least, engagement will have a positive effect on how employeeshandle their in-role duties, including providing superior service tocustomers.

Harter et al. (2002) reported that engagement is positively relatedto important business performance metrics such as customer satisfac-tion, loyalty, profitability, and productivity. We expect similar resultsin that engaged employees will work out of happiness, excitement,and the sheer joy of providing excellent service to customers. Weexpect engaged employees to be friendly, attentive to customer prob-lems, prompt in service delivery, and motivated to recommend ap-propriate products based on customer needs (Young et al., 2009).

H3. Work engagement is positively related to customers' evaluation ofservice employee performance.

2.1.4. Mediating role of engagementTogether, H1, H2, and H3 indicate that engagement should mediate

the relationship between (a) supervisor support and employee perfor-mance and (b) supervisor feedback and employee performance. Ourprediction of a mediation hypothesis rests on the argument thatresources such as supervisory support and supervisory feedback are dis-tal concepts from the customer's perspective. That is, customerswill notbe able to feel and assess how service employees treat them based sole-ly on the resources that employees receive from management becausethe resources are too far removed from employee–customer interac-tions to be predictors of service employee performance. Unless theresources provide an impetus for greater levels of employee engage-ment, predicting employee performance based on resources will be in-adequate. As such, we state that engagement is a proximal constructthat reflects how employees feel about the resources they receive, andthis in turn functions as a bridge that connects resources to employeeperformance.

H4. Engagement mediates the relationship between (a) supervisorysupport and customers' evaluation of service employee performanceand (b) supervisory feedback and customers' evaluation of serviceemployee performance.

2.1.5. Moderating role of perceived autonomyWe define perceived autonomy at the employee level as the de-

gree to which employees feel they have independence, flexibility, dis-cretion, and control in performing their jobs (Hackman & Oldham,1976). We posit that there is a complementary relationship betweenperceived autonomy and supervisory support. That is, as employeesperceive greater autonomy, their engagement benefits from highersupervisory support. With higher perceived autonomy, employeesfeel a greater sense of motivation, empowerment, and competence(Marinova, Ye, & Singh, 2008). When employees perceive more au-tonomy, they have a greater tendency to “internalize work rules,

standards, and procedures” (Stone, Deci, & Ryan, 2009, p. 78). Conse-quently, when employees receive supervisory support under condi-tions of high perceived autonomy, they will be able to embrace andintegrate the task and social support received from their supervisors.This suggests that at high levels of perceived autonomy, supervisorysupport will have a positive effect on engagement.

Conversely, when supervisory support is sufficient but employeesperceive little autonomy, they feel less motivation and empowermentto actually put the support into action. That is, employees feel thatthey are not provided with the control and discretion they need tocapitalize on the help that they receive from their supervisors. Em-ployees may receive verbal or psychological comfort and assistancefrom their supervisors but be restricted regarding the latitude, flexi-bility, and freedom to make use of that support. As Velthouse (1990,p. 16) posits, the effect of social support on engagement may be lim-ited. This suggests that when perceived autonomy is low, the effect ofsupervisory support on engagement should be null.

H5. Perceived autonomy positively moderates the relationship betweensupervisory support and engagement such that (a) when perceivedautonomy is high, supervisory support will have a positive effect onengagement while (b) when perceived autonomy is low, supervisorysupport will have no effect on engagement.

Next, we posit that supervisory feedback has a positive effect onengagement under conditions of low perceived autonomy. When em-ployees perceive low autonomy, they have little latitude, discretion,and empowerment to make decisions on their own. As Marinova etal. (2008) suggest, as jobs are designed with less autonomy, thelocus of control shifts from the employee (internal) to management(external). That is, as employees perceive less autonomy, they feelthat there is more control and pressure frommanagement to performtasks in certain ways. Consequently, when employees receive feed-back, it is specific and concrete performance feedback informationon what and how to perform to become more effective (Jaworski &Kohli, 1991). With low perceived autonomy, developmental feedbackgains importance as a guideline and roadmap for how to performbased on supervisor input; therefore, feedback will be a more valu-able and diagnostic resource under low perceived autonomy thanhigh (Ashford & Cummings, 1983).

In contrast, under conditions of high perceived autonomy, super-visory feedback may not contribute as much and may be perceivedas less effective in influencing engagement because more autonomysuggests that employees have more internal control over how toperform tasks. That is, they are less dependent and influenced by su-pervisory feedback than they would be under conditions of low per-ceived autonomy. Therefore, at high levels of perceived autonomy,the benefits of supervisory feedback on engagement are expectedto be limited.

H6. Perceived autonomy negatively moderates the relationship betweensupervisory feedback and engagement such that (a) at low levels ofperceived autonomy, supervisory feedback has a positive effect onengagement while (b) at high levels of perceived autonomy, supervisoryfeedback will have no effect on engagement.

3. Research method

3.1. Sample and data collection

We selected the specialty retail industry for our study. The chief rea-son behind this is that customers of specialty retailers are willing to payhigher prices in order to receive better customer service. Thus, it is im-portant for those retailers to invest in resources thatwill elevate serviceemployees' engagement so that they will deliver customer-orientedbehaviors, whichwill then be reflected in customers' assessment of ser-vice employee performance. Although employee engagement is critical

Table 1CFA results.

Constructs and items Factorloading

t-values

Service employee survey (n=422)Supervisory support (α=.90; CR=.91; AVE=.77)My manager is very concerned about the welfare of thoseunder him/her

.90a –

My manager is willing to listen to work-related problems .88 9.63My manager can be relied upon when things get difficultat work

.85 9.71

Supervisory feedback (α=.86; CR=.87; AVE=.70)My manager gives me sufficient information about workgoals

.84a –

My manager gives me feedback on my performance .86 9.89My managers gives me feedback on how I can improve myperformance

.80 8.27

Perceived autonomy (α=.78; CR=.78; AVE=.55)I can use my own personal judgment on carrying out my job .70a –

I have the freedom to decide what I do on my job .75 7.74I can make my own decisions in carrying out my job .77 7.90Vigor (α=.81; CR=.83; AVE=.50)At work, I feel full of energy .76a –

In my job, I feel strong and vigorous .71 6.90When I get up in the morning, I feel like going to work .65 5.12I can continue working for very long periods at a time .70 6.81In my job, I am mentally very resilient .79 8.14At work, I always persevere, even when things do notgo well

.61 4.83

Dedication (α=.86; CR=.87; AVE=.58)I find the work that I do full of meaning and purpose .76a –

I am enthusiastic about my job .78 8.09My job inspires me .75 7.83I am proud of the work I do .70 6.79I find my job challenging .81 8.43Absorption (α=.78; CR=.80; AVE=.45)Time flies when I am working .65a

When I am working, I forget everything else around me .70 6.77I feel happy when I am working intensely .71 6.94I am immersed in my work .70 6.78I get carried away when I am working .54 3.90It is difficult to detach myself from my job .61 4.80

Customer survey (n=488)Customer evaluation of service employee performance(α=.87; CR=.89; AVE=.55)

Being friendly and helpful to customers .65a –

Approaching customers quickly .79 8.04Asking good questions and listening to find out what acustomer wants

.72 7.21

Being able to help customers when needed .74 7.77Pointing out and relating item features to a customer'sneeds

.71 7.32

Suggesting items customers might like but did not think of .75 7.90Explaining an item's features and benefits to overcome acustomer's objections

.80 8.33

α = Cronbach's alpha; CR = Composite reliability; AVE = Average variance extracted.

2166 B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

across all retail types, we believe that employee engagement is particu-larly relevant in the specialty retail industry where customers expectemployees to bemotivated and passionate about their work. Therefore,we acknowledge that the effectiveness of employee engagement maynot generalize to all retail environments and may be more applicablein a retail context where stores consider customer service an integralpart of their competitive strategy and also where customers expectsuch behaviors.

We collected the data from the service employees and customersof a Canadian retail company. The company specializes in fashionand apparel, footwear, accessories, and sports equipment, and oper-ates 68 stores in predominantly English-speaking provinces. Duringthe period of time we conducted our survey, the retail organizationemployed 1020 service employees across its 68 stores.

We prepared a mailing packet in line with Dillman's (1978) totaldesign method. Accordingly, the packet contained a cover letterpersonally addressed to store managers; service employee surveys,postage-paid return envelopes, and cover letters explaining the pur-pose of the survey; and customer surveys and cover letters explainingthe purpose of the survey. The HR manager informed store managersabout the organization's support and endorsement of the surveys andasked for their support and assistance in distributing the surveys andpostage-paid return envelopes to service employees. To assure theconfidentiality and anonymity of their responses, service employeeswere asked to mail the survey directly to us.

After two reminders, we received 484 employee surveys from 68stores. Since the nested data analysis which we employ requires atleast two responses per store, stores for which only one useable surveywas obtainedwere removed from the sample. As a result, we had a sam-ple of 482 service employees (47.25%) from 66 stores (97%) with an av-erage of 7.5 responses per store. The demographic distribution of theservice employees (on average) is as follows: age (27.5 years), male(56%); store experience (3.5 years), and total work experience in a sim-ilar position (5.4 years).

Store managers conducted the customer survey over two days. Asper the request of the organization, we sent ten customer surveys toeach store. Store managers chose customers randomly. Customerswho approached the checkout counter were asked to respond to thesurvey. Those customers who were willing to do so were then askedto fill out the survey at a desk opposite to the checkout counter. Cus-tomers put the completed surveys in sealed envelopes and returnedthem to store managers. Store managers then mailed the completedsurveys to us. Since wematched the customer surveys with employeesurveys from the same store branch, a total of 488 usable customerresponses from 66 stores were included in this study. Fifty-five per-cent of the surveyed customers were female, the average age was39 years, 77% of customers reported at least one shopping experiencewith the store chain, and 52% of customers reported an annual house-hold income of $50,000 CDN or higher.

3.2. Measures

Table 1 reports the scales we used to measure the constructs ofour model.

Store employees responded to the scales of work engagement, su-pervisor support, supervisor feedback, and perceived autonomy. Wemeasured work engagement by capturing three dimensions: vigor (6items), dedication (5 items), and absorption (6 items) (Salanova et al.,2005). We measured perceived autonomy with a three-item scale bor-rowed from Spreitzer (1995). We measured supervisor support with athree-item scale adapted from House (1981). We measured supervisorfeedback with three items adapted from Zhou and George (2001).

We obtained demographic information from service employees:gender (1-male; 2-female), age (in years), and experience. We formedthe overall index of experience by averaging the z-scores of yearswith the organization and years in the profession.

Customers responded to the scale of service employee perfor-mance. We measured perceived service employee performance witha seven-item, five-point Likert scale (1-completely unsatisfactory;5-extremely good) borrowed from Liao and Chuang (2004). Cus-tomers also provided information about their age (in years), gender(1-male; 2-female), total household income (annual), and previousexperience with the store chain (1-yes, 2-no).

We obtained store size (i.e., the number of full-time employees)from company records.

3.3. Measurement validation

We employed confirmatory factor analysis (CFA) to assess the valid-ity, unidimensionality, and reliability of the measures for both sets ofconstructs (i.e., the data collected from store employees and customers)separately. According to Hair, Black, Babin, Anderson, and Tatham

2167B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

(2006), based on a sample of more than 250 respondents and a modelwith fewer than 30 observed variables, evidence of a good fit would indi-cate a significant χ2 value and aRMSEAof less than .07with a CFI of .92 orhigher. Accordingly, the CFA for the service employee data provided agood fit (χ2

(284)=596.4, GFI=.92, TLI=.91, CFI=.92, RMSEA=.06).Hair et al. (2006) also suggest that based on a sample of more than 250respondents and a model with fewer than 12 observed variables, evi-dence of a good fit would indicate an insignificant χ2 value and aRMSEA of less than .07 with a CFI of .97 or higher. In line with Hair etal. (2006), the CFA for the customers data also indicated a good fit(χ2

(14)=19.8, GFI=.97, TLI=.97, CFI=.98, RMSEA=.04). Table 1shows the reliability estimates for the constructs in both surveys.

The convergent validity of the scales was supported because factorloadings were significant (Anderson & Gerbing, 1988) and the esti-mates of the average variance extracted (AVE) were equal to orhigher than .50 (with the exception of the scale of absorption)(Bagozzi & Yi, 1988). We found the tests for discriminant validity tobe supportive; that is, no confidence intervals of the correlationsbetween any of the constructs included 1.0 (pb .05). The squared in-tercorrelations between two constructs were less than the AVE esti-mates of the respective two constructs for all pairs of constructs(Fornell & Larcker, 1981) (Table 2). In addition, we compared theunconstrained model with the constrained model in which the corre-lation between two constructs was set to 1 (Anderson & Gerbing,1988). For every pair of constructs, the results revealed a significantchi-square difference between the constrained and unconstrainedmodel (Δχ2>3.84), which supports the discriminant validity of theconstructs.

Since the three dimensions of work engagement (i.e., vigor, dedi-cation, absorption) were highly correlated to one another, we createda higher-order construct of work engagement by averaging the totalscore of these three dimensions (e.g., Salanova et al., 2005).

3.4. Hypotheses testing

The data display a hierarchical structure with service employeesand customers nested in stores. Hence, we tested our model and itshypotheses by employing a random-parameter modeling techniquein LIMDEP Version 9.0 (Greene, 2007).

First, we created two interaction effects (perceived autonomy∗su-pervisory support and perceived autonomy∗supervisory feedback)by multiplying the grand mean-centered variables of supervisorysupport, supervisor feedback, and perceived autonomy (Hofmann &Gavin, 1998). Second, we operationalized the construct of customers'perception of employee performance at the store level. We aggregat-ed customers' responses to calculate a single performance score for

Table 2Descriptive statistics and intercorrelations.

Variables 1 2 3 4 5

1. Gender2. Age −.043. Experience .04 .47⁎⁎

4. Supervisory support −.01 −.09⁎ −.12⁎

5. Supervisory feedback −.05 −.08 −.14⁎⁎ .34⁎⁎

6. Autonomy −.03 −.02 .03 .37⁎⁎ .27. Work engagement .01 .08 −.10⁎ .30⁎⁎ .38. Store size .04 −.04 .04 .15⁎⁎ −.19. Customer age −.14 .17 −.04 .02 −.010. Customer gender −.17 .02 −.01 .06 .111. Customer income −.39⁎⁎ .04 .08 .03 .112. Customer experience .02 .09 −.30⁎ −.06 .013. Service employee performance −.03 .27⁎ .01 .11 .0Mean 1.44 27.5 .00 3.71 3.5SD .50 5.1 .82 .77 .8

Employee responses (n=482); Customer responses (n=488).⁎ pb .05.⁎⁎ pb .01 (two-tailed test).

each store. One-way analysis of variance revealed a significant storeeffect (F(65,422)=1.61; pb01). The intraclass correlation coefficientsand median within-group agreement (ICC1=.29, ICC2=.51,rwg=.89) exceeded the recommended values for group-level con-structs (James, Demaree, & Wolf, 1984). Therefore, there was statisti-cal support for this aggregation.

Table 3 (Model 2) indicates that supervisory support (γ=.09; ns)is not significantly related to work engagement (H1 is not supported),whereas supervisory feedback (γ=.18; pb .001) is related positivelyand significantly to work engagement (H2 is supported).

Hypotheses 3 and 4 in combination posit that work engagement isrelated positively to customers' perception of employee performanceand that work engagementmediates the relationship between supervi-sory support/feedback and employee performance. We tested the me-diating role of work engagement as per Baron and Kenny (1986) andmore recent work in the statistical literature (e.g., Schneider, Ehrhart,Mayer, Saltz, & Niles-Jolly, 2005). As Schneider et al. (2005, p. 1023)suggest, “the conditions for full mediation are satisfied when (1) thereis a significant relationship between the predictor and the mediatorand between the mediator and the outcome variable, and (2) the pre-dictor does not explain any additional variance of the outcome (beyondthe mediator).”

In support of Hypotheses 3, the relationship between work en-gagement and employee performance is positive and significant(γ=.11; pb .01). Since supervisory support is not significantly relatedto work engagement, the necessary condition for mediation is notsatisfied. Hence, Hypothesis 4a is not supported. However, superviso-ry feedback is related positively to work engagement and workengagement is also related positively to performance. In addition,when we added work engagement to the model that predictsthe feedback–performance relationship, we observed a significantincrease in model fit (Δχ2

(1)=4.04; pb .05); yet the model fit didnot increase when we added supervisory feedback to the workengagement–performance model (Δχ2

(1)=.14; ns). Hence, Hypothe-sis 4b is supported in that work engagement is a full mediator in thefeedback–performance relationship. We performed the Sobel test,which provides a direct test of the indirect effect of an independentvariable on the dependent variable through the mediator (Sobel,1982). We found that supervisory feedback has an indirect effect onperformance through work engagement (z=2.31; pb .05), whichsuggests that work engagement fully mediates the relationship be-tween supervisory feedback and employee performance. This findingfurther supports Hypothesis 4b.

We found that the interaction of supervisory support and per-ceived autonomy is positive and statistically significant (γ=.22;pb .001), which supports Hypothesis 5. Further, we conducted simple

6 7 8 9 10 11 12 13

8⁎⁎

8⁎⁎ .34⁎⁎

1⁎⁎ −.13⁎⁎ −.12⁎

8 −.16 .01 −.026 −.02 .10 .09 .35⁎

0 −.17 −.07 .08 .42⁎ .18⁎

2 −.04 .14 .03 .02 .15 .096 .07 .31⁎⁎ .11 .10 −.06 .09 .17⁎

2 3.30 3.57 3.75 39.0 1.52 2.25 1.19 4.351 .69 .59 .69 5.28 .22 .70 .24 .28

Table 3Results.

Predictors Work engagement Customers' perceptionof service employeeperformance

Intercept 3.00⁎⁎⁎ 3.82⁎⁎⁎

Supervisor support .09Supervisory feedback .18⁎⁎⁎

Perceived autonomy .25⁎⁎⁎

Work engagement .11⁎⁎

InteractionsSupervisory support×autonomy .22⁎⁎⁎

Supervisory feedback×autonomy −.15⁎⁎

Control variablesStore size −.01 .02Employee gender .04Employee age .03⁎⁎⁎

Employee experience −.09⁎⁎

Customer gender .02Customer age −.01Customer income .02Customer previous experience −.01Deviance −352.84 −390.03ΔDeviance (Δdf) 147.54⁎⁎⁎ (9)a 13.75⁎ (6)a

Total R2 .38 .22

Employee-level variables (n=482); customer-level variables (n=488); store-levelvariables (n=66).

a Based on comparison with null model where there is no a priori predictor.⁎ pb .05.

⁎⁎ pb .01.⁎⁎⁎ pb .001 (two-tailed test).

1

1.5

2

2.5

3

3.5

4

4.5

5

low

Wor

kE

ngag

emen

t

Supervisory Support

low (Autonomy)

high (Autonomy)

1

1.5

2

2.5

3

3.5

4

4.5

5

low

Wor

kE

ngag

emen

t

Developmental Feedback

low (Autonomy)

high (Autonomy)

high

high

B

A

Fig. 2. (Panel A): Moderating effect of perceived autonomy on the supervisorysupport–work engagement relationship. (Panel B): Moderating effect of perceivedautonomy on the developmental feedback–work engagement relationship.

2168 B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

slope analysis and plotted graphs for one standard deviation aboveand below the mean value of perceived autonomy (Aiken & West,1991). At high levels of perceived autonomy, supervisory support isrelated positively and significantly to work engagement (γ=.23;pb .01), which supports H5a, whereas at low levels of perceived au-tonomy, supervisory support is not related to work engagement(γ=−.06; ns), which supports H5b. We found support for Hypothe-sis 6 in that the interaction of supervisory feedback and perceived au-tonomy is negative and statistically significant (γ=−.15; pb .01).Simple slope analysis indicates that at low levels of perceived auton-omy, feedback is related positively and significantly to engagement(γ=.27; pb .001), which supports H6a, whereas at high levels of per-ceived autonomy, feedback is not significantly related to engagement(γ=.03; ns), which supports H6b. Fig. 2 shows the plot for the inter-action effects.

4. Discussion

The goal of this study was to examine (a) the antecedents to ser-vice employee engagement, (b) the consequences of engagement,(c) the mediating role of engagement in linking the antecedents tothe consequences, and (d) the moderating effect of perceived auton-omy on the antecedent–engagement relationship. We tested theabove relationships by drawing on the JD-R model as our theoreticalframework. Our findings can be grouped into the following fourconclusions.

First, supervisory feedback and perceived autonomy promotedservice employee engagement. Clear and objective expectationsand guidelines for performance and suggestions for how to makeimprovements made a difference. A potential reason for this maybe that receiving such feedback can reduce role ambiguity and con-flict (Babin & Boles, 1996). Feedback also implies more communica-tion and dialog between the two parties, which can explain its linkto greater employee motivation and desire to be engaged in theworkplace.

In contrast, supervisory support was not positively and signifi-cantly related to engagement. This is somewhat surprising becausescholars have argued that support is an important resource that alle-viates stress and burnout, strenuous workplace factors that can leadto disengagement (e.g., Ashill & Rod, 2011; Babin & Boles, 1996). Not-withstanding, a possible reason for this null finding may be that su-pervisor feedback is sufficient and supervisor support becomesredundant when both are provided. In fact, what employees may de-sire is coworker support, rather than supervisor support, when super-visor feedback is also provided. Coworker support is the task andsocial support that employees receive from their colleagues. Suchaid may represent a different dimension of support (i.e., horizontalas opposed to vertical support) which could be lacking from supervi-sory support. Receiving support from peers may alleviate some of thepsychological costs and burdens that employees lacking supervisorysupport may feel.

Second, engagement has implications for customer-related out-comes. In this respect, when retail managers invest in improving en-gagement, customers perceive superior employee performance. Thisfinding is important because most studies in the engagement litera-ture have focused primarily on how engagement is associated withemployee attitude and performance. As such, our approach is a depar-ture from the path taken by many and gives credence to the practicaland strategic implications of the engagement construct as it is per-ceived by customers.

Third, engagement operates as a full mediator between superviso-ry feedback and customers' evaluation of employee performance. Webelieve that this is important because it illustrates the process by

2169B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

which some resources such as feedback can be related to customers'opinions of how the organization is performing. These resources aretoo far removed and thus distal to affect customers' opinions aboutemployees' delivery of service performance directly; therefore, it isnecessary to include a more proximal concept to customers' assess-ment of employee performance. Our findings suggest that when em-ployees are engaged, this affects how they behave and interact withcustomers, which is evidenced in higher evaluations of performancefrom customers. This implies that when organizations desire tochange customers' evaluations of employee performance by providingmore resources such as feedback, managers need to ensure that suchfeedback translates into engagement if the investment is to have achance of altering customers' perceptions of service performance.

Fourth, although prior studies in the literature have explored the di-rect and main effects of resources on engagement, little was knownregarding the interactive effect between resources. Our study revealsthat there is a positive interaction between supervisory support andperceived autonomy, and a negative interaction between supervisoryfeedback and perceived autonomy. This implies that implementing su-pervisory support and feedback in an organizational setting that hasample perceived autonomy can affect engagement in completely differ-entways.What these results suggest is that employees appreciate feed-backmore in an environment where there is little perceived autonomy.When employees are not given the freedom and discretion to maketheir own judgments regarding how to carry out tasks, receiving feed-back from supervisors that clearly and objectively demonstrates whatthey need to do to improve their performance is beneficial and im-proves engagement. Also, the lack of a direct main effect of supervisorysupport on engagement indicates that support alone may not beenough and that such support needs to be complemented with per-ceived autonomy. Employees may feel that simply being listened toand being taken care of in terms of work-related issues and employeewelfare is not sufficient but that they need to be reassured that theycan control and implement decisions on their own without necessarilyseeking supervisor approval and consent.

Our study also has practical implications. Retail managers need tobe wary of providing support and feedback simultaneously. Our re-sults suggest that supervisory support has no impact on engagementwhen feedback is also provided to employees. Also, if retail managersdesire to offer either support or feedback, perceived autonomy playsan important contextual role. That is, supervisory support shouldhave the greatest impact on engagement when employees perceivehigh autonomy. In contrast, for supervisory feedback to make an im-pact on engagement, such feedback is most desirable under low per-ceived autonomy. Overall, how much autonomy employees perceivein the retail environment can inform managers' decisions as towhether they should use support or feedback to affect engagement.If perceived autonomy is high, supervisory support may be more ef-fective in enhancing engagement. Conversely, if perceived autonomyis low, supervisory feedback may be a useful tool for improving en-gagement. Another interpretation could be that if there are twostores, A and B, and store A has in place supervisory support whilestore B implements supervisory feedback, store A should considerproviding more autonomy to its employees while store B should notif the goal for both is to improve employee engagement.

5. Limitations and directions for future studies

Our study is not without limitations, which we see as opportunitiesfor future research. First, we focused on three resources (supervisorysupport, supervisory feedback, and perceived autonomy). Includingother resources such as coworker support could enrich our model.Therefore, future studies should attempt to include a wider range of re-sources. Second, our final dependent variable was customers' evalua-tion of service employee performance. Future studies could includefinancial performancemetrics such as sales and profits to further bolster

the validity of our model and address any criticisms that may exist re-garding accountability and return on engagement. Finally, it would beworthwhile to assess whether engagement has a positive effect on cus-tomers' evaluation of employee performance even after controlling forconstructs such as involvement, satisfaction, and commitment (Rich etal., 2010).

References

Aiken, L. S., & West, S. G. (1991).Multiple regression testing and interpreting interactions.Newbury Park, CA: Sage.

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A re-view and recommended two-step approach. Psychological Bulletin, 103, 411–423.

Ashford, S. J., & Cummings, L. L. (1983). Feedback as an individual resource: Personalstrategies of creating information. Organizational Behavior and Human Performance,32(3), 370–399.

Ashill, N. J., & Rod, M. (2011). Burnout processes in non-clinical health service encoun-ters. Journal of Business Research, 64(10), 1116–1127.

Babakus, E., Yavas, E., & Ashill, N. J. (2009). The role of customer orientation as a mod-erator of the job demand–burnout–performance relationship: A surface-level traitperspective. Journal of Retailing, 85(4), 480–492.

Babin, B. J., & Boles, J. S. (1996). The effects of perceived coworker involvement and su-pervisor support on service provider role stress, performance and job satisfaction.Journal of Retailing, 72, 57–75.

Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journalof the Academy of Marketing Science, 16(1), 74–94.

Bakker, A. B., & Demerouti, A. (2007). The job demands-resources model: State of theart. Journal of Managerial Psychology, 22(3), 309–328.

Baron, R. M., & Kenny, D. (1986). The moderator–mediator variable distinction in socialpsychological research: Conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychology, 51(6), 1173–1182.

DeConinck, J. B. (2010). The effect of organizational justice, perceived organizationalsupport, and perceived supervisor support on marketing employees' level oftrust. Journal of Business Research, 63(12), 1349–1355.

Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001). The jobdemands-resources model of burnout. Journal of Applied Psychology, 86(3),499–512.

Dillman, D. A. (1978). Mail and telephone surveys: The total design method. New York:John Wiley & Sons, Inc..

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobser-vable variables and measurement error. Journal of Marketing Research, 18(1),39–50.

Greene, W. H. (2007). Econometric analysis. Englewood-Cliffs, NJ: Prentice Hall.Hackman, R. J., & Oldham, G. R. (1976). Motivation through the design of work: The test

of a theory. Organizational Behavior and Human Performance, 16(August), 250–259.Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006). Multivariate

data analysis (Sixth edition). Upper Saddle River, New Jersey: Pearson PrenticeHall.

Harter, J. K., Schmidt, F. L., & Hayes, T. L. (2002). Business-unit-level relationshipbetween employee satisfaction, employee engagement, and business outcomes:A meta-analysis. Journal of Applied Psychology, 87(April), 268–279.

Hofmann, D. A., & Gavin, M. B. (1998). Centering decisions in hierarchical linearmodels: Implications for research in organizations. Journal of Management, 24,623–641.

House, J. S. (1981). Work stress and social support. Reading, MA: Addison-Wesley.James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating within-group interrater

reliability with and without response bias. Journal of Applied Psychology, 69, 85–98.Jaworski, B. J., & Kohli, A. K. (1991). Supervisory feedback: Alternative types and their

impact on salespeople's performance and satisfaction. Journal of Marketing Re-search, 28(2), 190–201.

Liao, H., & Chuang, A. (2004). A multilevel investigation of factors influencing employeeservice performance and customer outcomes. Academy of Management Journal,47(February), 41–58.

Marinova, D., Ye, J., & Singh, J. (2008). Do frontline mechanisms matter? Impact ofquality and productivity orientations on unit revenue, efficiency, and customersatisfaction. Journal of Marketing, 72(March), 28–45.

Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). In S. T. Fiske, D. L. Schacter, & C.Zahn-Waxler (Eds.), Job burnout. Annual Review of Psychology, 52. (pp. 397–422).

Rich, B. L., LePine, J. A., & Crawford, E. R. (2010). Job engagement: Antecedents andeffects on job performance. Academy of Management Journal, 53(3), 617–635.

Salanova, M., Agut, S., & Peirό, J. M. (2005). Linking organizational resources and workengagement to employee performance and customer loyalty: The mediation ofservice climate. Journal of Applied Psychology, 90(6), 1217–1227.

Sand, G., & Miyazaki, A. D. (2000). The impact of social support on salesperson burnoutcomponents. Psychology and Marketing, 17(1), 13–26.

Schaufeli, W., Salanova, M., Gonzalez-Roma, V., & Bakker, A. B. (2002). The measure-ment of engagement and burnout: A two sample confirmatory factor analyticapproach. Journal of Happiness Studies, 3, 71–92.

Schneider, B., Ehrhart, M. G., Mayer, D. M., Saltz, J. L., & Niles-Jolly, K. (2005). Under-standing organization-customer links in service settings. Academy of ManagementJournal, 48(6), 1017–1032.

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structuralequation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290–312).Washington, DC: American Sociological Association.

2170 B. Menguc et al. / Journal of Business Research 66 (2013) 2163–2170

Sonnentag, S. (2003). Recovery, work engagement, and proactive behavior: A new lookat the interface between nonwork and work. Journal of Applied Psychology, 88(3),518–528.

Spreitzer, G.M. (1995). Psychological empowerment in the workplace: Dimensions, mea-surement, and validation. Academy of Management Journal, 38(October), 1442–1465.

Stone, D. N., Deci, E. L., & Ryan, R. M. (2009). Beyond talk: Creating autonomousmotivation through self-determination theory. Journal of General Management,34(3), 75–91.

Towers Perrin (2008). Global workforce study. http://www.towersperrin.com/tp/showhtml.jsp?url=global/publications/gws/index.htm&country=global

Velthouse, B. (1990). Creativity and empowerment: A complementary relationship. Re-view of Business, 12(2), 13–20.

Yoon, M. H., & Suh, J. (2003). Organizational citizenship behaviors and service qualityas external effectiveness of contact employees. Journal of Business Research,56(8), 597–611.

Young, G., Meterko, M. M., Mohr, D., Shwartz, M., & Lin, H. (2009). Congruence in theassessment of service quality between employees and customers: A study of a pub-lic health care delivery system. Journal of Business Research, 62(11), 1127–1135.

Zhou, J., & George, J. M. (2001). When job dissatisfaction leads to creativity: Encourag-ing the expression of voice. Academy of Management Journal, 44, 682–696 (August).