Behavioral and Psychological Consequences of Boundary Spanning Burnout for Customer Service Representatives Author(s): Jagdip Singh, Jerry

R. Goolsby, Gary K. Rhoads Source: Journal of Marketing Research, Vol. 31, No. 4 (Nov., 1994), pp. 558-569 Published by: American Marketing Association Stable URL: http://www.jstor.org/stable/3151883 . Accessed: 03/10/2011 10:03
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing Research.

http://www.jstor.org

l~~~~~

and GARYK. RHOADS* JAGDIPSINGH,JERRYR. GOOLSBY,

Marketingboundary spanners cspecially customer service representatives-are notably susceptible to burnout. The authors define the burnout construct and develop hypotheses to examine if burnout acts as a partial mediator between role stressors and key behavioral and psychological job outcomes. Responses from 377 customer service representatives reveal that burnout levels are high relative to other burnout-prone occupations (e.g., police, nursing) and that burnout has consistent, significant, and dysfunctional effects on their behavioral and psychological outcomes. Moreover, burnout mediates the negative effects of role stressors on job outcomes, whereas the positive effects of role stressors are unmediated.

Behavioral
of

and

Boundary

Psychological Consequences Burnout for Spanning

Service

Customer

Representatives

Evidence linking marketing-oriented boundaryspanning positions (e.g., customer service representatives,salespeople) to burnout,the maladyof 1990s, has been accumulating with alarming regularity.In a recent review, Cordes and Dougherty(1993, p. 644) note that customer service representatives are susceptible to "moderateto high" levels of burnout because they "arecaughtin a difficultposition when they perceive that client demandscannot or will not be met In a 1992 surveyof 1299 full time emby the organization." ployees drawnrandomlyfrom U.S. firms, the Northwestern Life Insurancecompanynoted that "sales and service workers are more likely to be candidatesof burnout." The American Instituteof Stress went as far as to categoricallyidentify a customer service workeras one of the ten most stressful jobs in America (Miller et al. 1988). Burnouttendencies1are defined as a psychological dispoIAlthoughburnoutoften refers to an unpleasantstate of being (as in "I am burnedout"),the scientific literature utilizes the termto representa continuumrangingfrom "none"to "high"evidence of psychological symptoms associated with the burnout construct. This construct continuum can be likened to burnouttendencies. The everyday notion of burnoutrepresents only an extreme position at the high end of the construct continuum. Throughoutthe article, we utilize the scientific notion of burnoutconstruct and use the terms burnoutand burnouttendenciesinterchangeably. Weather*JagdipSingh is an Associate Professor,Division of Marketing, head School of Management,Case WesternReserve University.Jerry R. Goolsby is an Assistant Professor, Departmentof Marketing,College of Business Administration, Universityof SouthFlorida.GaryK. Rhoadsis an of Marketing, Associate Professor,Department College of Business Administration,BrighamYoung University.The authorsthankJMR editor Barton Weitz and threeanonymousreviewersfor theirmany helpfulcommentsand the assistanceof Marketing Science suggestions.The authorsalso appreciate Institutefor providingpartialfundingfor this project.Finally,the time and effortdevotedby the participating firmandcustomerservice representatives to make this projectsuccessful is sincerely acknowledged.

sition characterizedby emotional exhaustion, lack of personal accomplishment,and a tendencyto depersonalizeothers. These tendencies are evident in occupations that entail (1) a stronginterpersonal componentand (2) elevatedlevels of role or job stress (Cordes and Dougherty 1993). Marketing boundaryspannerscertainly reflect these occupational characteristics. Often marketing programs require that boundary spanners deliver high levels of service quality through personal interaction, care, and concern for customers.Moreover,due to its innovativenatureand large role set, boundaryspannersare prone to elevated levels of role stress. In addition, burnout plays an importantrole in understanding the impact of role stressors on boundaryspanner performance,satisfaction, commitment, and turnover.Researchersagree that extremerole stress invariablyresults in an erosion of performanceand job-related attitudes (i.e., however, moderatelevels of role stress may in"distress"); crease performanceas stress stimulatesindividualsto excel cf. Seyle 1976). The delineationbetween mod("eustress"; erate and excessive role stress is problematicbecause the notion of moderatestress is subjectto individualvariability. As such, the measurementof role stress is not sufficient to determinewhetherit has a positive or negativeinfluence. In contrast,burnoutoccurs only when stressors overwhelm a person's coping resources. Consequently,burnoutis potentially a more accuratepredictorof criticaljob outcomes than the presence or extent of key role stressors (cf. Bacharach, Bamberger,and Conley 1991). Moreover,burnoutprovides a theoreticallysound approachto differentiate,at the individual level, between positive and negative levels of stress. We aim to evaluate burnout'ssubstantivecontributionto the study of boundaryrole stress processes in marketing. Specifically, our aim is threefold: First, we introduce
Journal of MarketingResearch Vol. XXXI (November 1994), 558-569

558

Burnoutand CustomerService Representatives

burnoutand its related literatureto marketing,focusing on of the burnout the conceptualizationand operationalization construct. In so doing, we draw clear distinction between role stressors and burnout in terms of their influence on boundaryspannerattitudesand behaviors.Second, we posit a set of hypotheses based on the notion thatburnoutpartially mediates the effect of role stressors (e.g., role conflict, ambiguity, overload) on critical behavioral (i.e., performance) and psychological (i.e, satisfaction,turnoverintentions, and commitment)job outcomes. In so doing, we develop the theoretical rationale that inclusion of burnout helps clarify the relationshipbetween role stressorsand job outcomes. Third,we test the partialmediatingthesis by utilizing data from a sample of 377 customerservice representatives in telemarketingpositions and examine how burnout mediatesin role stressprocesses. We close with implications for theory and managerialpractice. AND RESEARCH BACKGROUND HYPOTHESES Burnoutand Role Stressors:Definitional and Discriminant ValidityIssues Burnout. Although several approaches for defining burnoutexist, the particular approachdevelopedby Maslach and Jackson (1981) has received the most attention. Conceptually,Maslach and Jacksondefine burnoutas a psychological condition or syndromethat is characterized by three distinct but interrelatedsymptoms or dimensions: (1) emotional exhaustion (EE), (2) reduced personal accomplishment (RPA), and (3) depersonalization(DP). Notably, EE reflects feelings of being depleted of energy and drainedof sensation ("exhausted") due to excessive psychological ("emotional")demands.Primarily,EE occurs in "intensive" and "people-oriented"occupations in which job tasks involve charged interpersonalinteractions.RPA taps the notion of "learnedhelplessness" (i.e., when efforts repeatedly fail to produceresults)and is characterized of by attributions inefficacy, reduced motivation,and low self-esteem. In extreme burnout,people no longer believe their actions make a difference, and consequently they quit trying. Finally, DP represents the tendency to deindividuateand dehumanize others. In the extreme, DP manifests in a cynical, callous, uncaring,and negative attitudetowardothers and detached referencesto clients as objects. Role stressors. In marketing,role stress researchhas been dominatedby three constructs:(1) Role conflict (RC) is the degree of incompatibility of expectations associated with the role; (2) role ambiguity(RA) is the degree to which clear information is lacking about the expectations associated with a role, methods for fulfilling role expectations,and/or consequences of role performance;(3) role overload (RO) exists when role expectations are far greaterthan the individual's abilities and motivationto performa task. Burnoutversus role stressors.Although embeddedin the same nomological network, researchersdraw clear distinction between burnoutand role stressors,citing two key factors. First, these concepts differ in theirinfluence on various job outcomes. Whereas role stressors are hypothesized to have either a positive (e.g., when stress acts as a stimulus)or negative (e.g., when stress is dysfunctional) influence,

559 burnoutis thought to be consistently dysfunctional. Thus, the relationshipbetween role stressorsand job outcomes is describedby an invertedU curve in accord with the YerkesDodson (1908) law; in contrast,this relationshipfor burnout is linearand negativein slope. In this sense, burnoutandrole stressorshave disparateinfluence on job outcomes. Second, burnoutis not a stressor per se; ratherit arises from the cumulativeimpact of multiple role stressors(Hunsaker 1986). That is, we recognize that in a particularrole, (1) several sources of stress are often present(e.g., conflict, ambiguity,overload),(2) and though individuallythese role stressorsmay not be dysfunctional,(3) the cumulativeeffect of these stressorscan overwhelmthe individual'scoping reThis conceptualization sources,therebyresultingin burnout. of burnoutis consistentwith the notion thateven in an environment of moderate to low role stressors, chronic stress may significantly erode a person's ability to cope and deplete his or her reservoirof support.In this sense, burnout and role stressors differ in their etiology. The preceding should not be taken to imply that role stressorsand burnout are unrelatedconcepts. Rather,we posit that these concepts aredistinctconstructsand relatedthrougha nomological net best representedby a partialmediationthesis. Burnoutin Role StressModels:ThePartialMediationThesis Much previous role stress researchin marketingis based on the proposition of DIRECT EFFECTS, whereby role stressorsare hypothesized as direct antecedentsof job outcomes, includingjob satisfaction(JS),job performance(JP), organizationalcommitment (OC), and turnover intentions (TI). This propositionis rooted in literaturesof role theory, expectancytheory,andjob stress.However,resultsfrom this extensive research have failed to uncover consistent and strongdirecteffects (Brownand Peterson 1993; Churchillet al. 1985). In Churchilland colleague's (1985) meta-analysis, the sharedvariancebetween all role stressorscombined and JP was only 14%. Moreover, Brown and Peterson (1993) revealed wide variabilityin the DIRECTEFFECTS of role stressors.For example, the 95% confidence intervals for the DIRECT relationships between RC and JS was {-.04, -.80} and for RA and JS it was {-.15, -.74}. This wide variabilityis indicativeof a misspecified DIRECTEFFECT.Thus, althoughthereis little doubt that role stressors significantly affect the psychological attitudes and behavioral outcomes of boundary spanners, the evidence lacks consistency and stability. Consistentwith the emerging view of burnoutin occupational psychology (Bacharach, Bamberger, and Conley 1991), we propose that burnoutis (1) an importantpartial mediatorof the effects of role stressorson job outcomes and (2) a morepotentpredictorof variousjob outcomes thanone or more role stressors.This is not to imply that the directeffect of role stressors on job outcomes is always insignificant; rather,the propositionis that the direct effect of role stressors on job outcomes after controlling for burnoutis mostly marginalcomparedwith its "before"effects (i.e., before partiallingthe mediatedeffect of burnout).In this sense, burnoutpartiallymediatesthe role stressorsto job outcomes MEDIATION relationship.We referto this as the PARTIAL thesis (see Figure 1).

560

JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1994 Figure 1

THEESTIMATED PARTIAL BETWEEN MEDIATING THESISFORTHERELATIONSHIP ROLESTRESSORS, ANDJOB OUTCOMES BURNOUT,

E7

El

E8

E9

E2Z

E10

E3

Ell

Theoreticalrationaleand empiricalsupport.The thesis of PARTIALMEDIATIONis based on two key arguments. First,as noted previously,burnoutis thoughtto representthe cumulative effect of several role stressors in the job environment(Hunsaker1986). Thus, when many role stressors simultaneouslyimpinge on the boundaryspanner,it is plausible that though individualstressorsoperate at a moderate to low level, the cumulativeeffect is significantly dysfunctional. In such situations, role stressors are expected to be only moderatelyor weakly relatedto job outcomes, but the pathwaymediatedby burnoutis likely to be strong. Second, researchershave begun to recognize that in and of itself the magnitudeof role stressorsis not as critical as its magnitude vis-A-vis the individual's coping resources (Jackson,Schwab,and Schuler 1986). This distinctionis importantbecause some role stress may help marketerscreatively manage role demandsand possibly enhance productivity (Seyle 1976; Singh 1993). As such, eliminating role stress potentially inhibits creativity and productivity.However, when the coping resources of the individual are exceeded, role stressbecomes unequivocallydysfunctional,resulting in burnout.Thus burnoutclarifies the theoreticalrelationshipbetween role stressorsandjob outcomes by offering an unambiguous approach for assessing the dysfunctional effects of a stressful role environment.

Although the PARTIALMEDIATIONthesis has not been subjected to a direct empirical investigation,recent studies in occupationalpsychology offer indirect support.For example, Leiter and Maslach (1988) show that EE partially mediates the influence of RC and unpleasant supervisor contactson OC. Likewise, Bacharach,Bamberger,and Conley (1991) found that burnoutpartiallymediates the impact of RC and RO on JS. We build on this evolving work by (1) developing the theoreticalrationalefor the partialmediation thesis, (2) proposingspecific hypotheses, and (3) subjecting these hypotheses to empiricalinvestigation. Our frameworkdoes not examine the complex relationships among each of the role stressors,burnoutdimensions, and individualjob outcomes. Rather,we proposehypotheses at the level of higher (second)-orderconstructs-that is, we examine the nomological net of relationshipsthat involves second-order constructs of role stressors (RSTRESS), burnout(BURNOUT) and job outcomes (OUT) (discussed subsequently).We believe that this higher-orderanalysis is more parsimoniousand perhapsmore appropriate given the currentstate of the literature. RelationshipBetweenRole Stressorsand Burnout The second-orderconstructof RSTRESS is composed of three stressors-RC, RA, and RO, each of which increases burnout.RC arises when boundaryspannersare expected to

Burnout and Customer Service Representatives satisfy incompatible demands from multiple role partners. Organizational (e.g., inconsistentpolicies) and interpersonal (e.g., incongruous expectations of boss and customers) sources can contribute to role conflict. When boundary spannersare unable to reconcile incongruentexpectations, role theory suggests that high levels of distress, frustration, and anxiety are generated.This in turnresults in emotionally charged role environmentswherein high levels of effort are requiredto perform tasks. If such an environmentpersists over multiple role episodes, boundary spanners may experienceburnout(cf. Maslach and Jackson 1984). When boundaryspannersface tasks that are unique and require innovative solutions, RA arises. In such complex task environments,it is difficultto provideclear policies and guidelines for all possible situations and contingencies. Thus, some RA is inherentin boundaryspanning roles. In addition, boundaryspannersoftentimes either do not have sufficientresourcesor are not empoweredto implementsatisfactory solutions. This is likely to renderboundaryspanning tasks ambiguous. Finally, because their performance may not be clearly defined or judged (e.g., "quality"service), boundary spanners may lack clarity about performance expectations.Roles withoutclear goals and direction contributeto burnoutbecause performingunderchronicRA conditionsrequiresexcessive levels of energyand mentalresources (Maslach 1982). This drain on energy, combined with a prolonged state of emotional excitement, are conducive to feelings of burnout(Jackson,Schwab, and Schuler 1986). Boundaryspanningpositions in marketingare susceptible to RO because the flow of tasks in boundaryspanningpositions is often uncontrollable(e.g., high traffic periods) and they must interactwith large numbersof customers,as well as many internalemployees and supervisors.Consequently, in many instances, demands exceed the abilities and resources of the individual.Because chronic overload can be emotionally draining,RO can result in burnout. On the basis of the preceding, we propose the following hypothesis for empiricalexamination: becauseof RC, RA, Hi: The higherthe level of role stressors, and/or theburnout. RO,thegreater RelationshipBetween Burnoutand Job Outcomes We conceptualizejob outcomes as composed of two distinct second-orderconstructs:(1) psychological and (2) behavioral outcomes. Psychological outcomes (PSY-OUT) representjob attitudesthat capturethe notion of psychological approachor avoidance towardthe job and include such attitudes as satisfaction, commitment, and turnoverintentions. In contrast,behavioraloutcomes (BEH-OUT) reflect performance outcomes of boundary spanners. These outcomes are not strongly related and differ in their relationships with role stressors(Brown and Peterson 1993). Burnoutand psychological outcomes. We examine three componentsof psychological outcomes relatedto burnouttwo representingattitudesof the "approach" variety,nameorienly JS and OC, and the thirdreflecting an "avoidance" tation, namely TI. The notion thatJS is an important consequenceof burnout is firmly established in the literature (Maslach 1982;

561 Wolpin,Burke,and Greenglass 1991). This view is based on two theoretical arguments. First, because psychological burnoutis the outcome of an appraisalprocess by which an individual evaluates the demands vis-a-vis his or her resources, it is posited that the outcome of this appraisal should affect an individual's psychological well-being on the job, includingJS. Second, because both are affective responses, it is hypothesized that burnoutfeelings should be relatedto JS. Burnout is related negatively to OC because boundary spannerscaught in a burnoutsyndrome generally view the organizationin adversarialterms and tend to withdrawpsychologically from it (Maslach 1982). Thus, emotionally exhausted boundaryspannerswho view others in a detached and callous manner and do not believe their job provides them with a sense of accomplishmentare likely to be less committedto their organization. In addition to its impact on psychological withdrawal from work, chronic burnoutcan also result in withdrawal from the organization(Maslach 1982). Initially, this withdrawalmay takethe form of absenteeism,physical isolation, and extended breaks, as the worker avoids contact with organizationalmembersand customers.Eventually,if burnout persists,the workerwill likely seek permanentavoidanceby leaving the position, the firm, or even the career. On the basis of the preceding, we propose the following hypothesis: the burnout, the lowerthe boundary spanner's H2:The greater lowerJS, OC, outcomes, perceived psychological reflecting and/or TI. higher Burnout and behavioral outcomes. Burnout affects behavioraloutcomes, such as JP,because it reduces the energy at the disposal of the boundaryspannerand leads to reduced efforts at work. Burnoutfeelings also entrapboundaryspanners in a vicious spiral in which they are less prone to seek and obtain supportand, as a result, continue to performineffectively. Finally, burnoutaffects JP directly because the individual perceives little or no control over the job situation, and his or her confidence in tackling work-related problemsdeclines. Thus, we propose: the less effectivea boundary spanH3:Thehigherthe burnout, neris likelyto be in heror his JP,resulting in lowerbehavioraloutcomes. RelationshipBetween Role Stressorsand Job Outcomes In contrast to the prevailing position of DIRECT EFFECTS, we hypothesize a PARTIALMEDIATIONthesis. Specifically, the hypothesis for PARTIALMEDIATIONassertsthatthe direct,nonmediated"after" effect (i.e., afteraccounting for burnout)of role stressorson job outcomes will be small in magnitude compared with the "before"effect (i.e., before controllingfor burnout).This assertionis based on two theoreticalarguments.First,as noted previously,role stressorscan have both positive and negative effects on job outcomes. The burnoutliteraturesuggests that only the negative ("distress") aspect of role stressors contributes to burnout. In contrast, the positive effects of role stressors ("eustress")should not be mediatedby burnout,supporting a direct link between role stressorsand job outcomes. Sec-

562

JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1994 Sample We sought participationfrom service managers of randomly selected medium (10-25 representatives)to large service centers throughout the Unit(26-50 representatives) ed States. In all, 18 service managersagreed to participate. were sent a A total of 770 customer service representatives packet containing their chief executive officer's letter enand a returnenvelope dorsingthe study,a surveyinstrument, to be sent directly to the researchers.After three weeks, managers were sent a second wave of survey-packets.All were promisedanonymity. participants Overall,377 surveys were returned,for a response rate of about 50%. However, 26 received responses were not usable, yielding a usable rate of 45.5%. Fully 69% of the responding sample was female. Most had completed high school or some college (median= 1-3 years of college), are relativelyyoung (median= 31-35 years), and earn less than have $20,000 per year. Over 47% of service representatives been with the firm for two years or less (median = 3-5 years). Service managersindicatedthat our sample was representativeof their employees. Measures Role stressors.Role stressorswere assessed by three separate first-ordermeasures. RC and RA were measuredby three items drawn from House, Rizzo, and Lirtzman's (1970) work.RO involvedfive items from Beehr,Walsh,and Taber's(1976) scale. In addition, an overall item was used to obtaina "global"assessmentfor each role stressor.All responses were obtained on a five-point "strongly disagree"scale. The alphareliabilitiesfor the agree"-"strongly multi-item RC, RA, and RO scales were .68, .73, and .77, respectively. Boundary spanning burnout. The burnout dimensions were measured by 24 items (i.e., 8 items per dimension) drawn from the scale developed by Maslach and Jackson (1981) (see AppendixA).2 Consistentwith Lee andAshforth (1993), the items were modified to include statementsabout customers, boss, coworkers, and top management rather thannonspecific "recipients" (e.g., "I feel indifferenttoward some of my recipients"became "I feel indifferenttoward some of my customers").Separateitems were included for the four role members because in focus groups boundary spannersindicatedthat (1) these were salient role members with whom they interactedfrequently,(2) they present distinct demands and pressures, and (3) burnout tendencies vary with role members. The factor structurefor burnoutitems correspondingto each role member was identical. In each case, a three-di2The specific scale developed by Maslach and Jackson is referredto as the MaslachBurnoutInventory(MBI). The MBI is a copyrightedscale and may be used only with the permissionof its authorsor by contactingConsulting Psychologists Press, 577 College Avenue, Palo Alto, CA 94306, USA. Originally,the MBI scale assessed both the frequencyand intensity of burnoutfeelings. However, this convention has been generally discontinued.Researcherstypically assess the threeburnoutdimensionsusing the MBI items and a response scale providedby Golembiewski, Munzenrider, and Stevenson(1988), which utilizes a Likert-typescale anchoredby "very much unlike me" and "very much like me." The responses from the preceding scale are highly correlatedwith MBI's frequency (p = .89) and intensity (p = .88) scales.

ond, some of the negative impact of role stressors on job outcomes may persist as direct effects when role stressors are not of sufficient magnitudeto foster burnout.That is, as described previously, burnoutoccurs only when role stressors are of sufficient magnitudeto collectively overwhelm the resources of the individual. Role stressors below the thresholdof burnoutwill not follow a mediatedpathway,yet may influencejob outcomes directly. On the basis of the preceding, we propose the following hypotheses for empiricalexamination: effect for burnout, the direct,unmediated H4:Aftercontrolling will be relativeon psychological outcomes of rolestressors ly smallor insignificant. the direct,unmediated effect for burnout, H5:Aftercontrolling will be relatively on behavioral outcomes of role stressors smallor insignificant. RESEARCH DESIGN ResearchSetting and Constraints in We chose a sample of customerservice representatives telemarketingpositions from a large, multinationalFortune areconsideredto 500 firm.Customerservice representatives have one of the most stressful professions in the United States (Miller et al. 1988). We reasoned that if the burnout construct is not relevant in such an environment,it is less likely to prove useful in studying other burnoutprone environments. Furthermore, using a single firm offers the advantage of controllingfor contextualeffects and minimizes research. possible contingenciescommon in cross-industrial The disadvantage, of course, is that the results are less generalizable. Nevertheless, the study sample representsa job environment that is not atypical of many boundaryspanningpositions. Specifically, the customer service representatives were responsible for handling and rectifying billing problems and selling programs.Each representative performeda customer-contactrole by screening and addressing complaints and inquiries. In addition, customer service representativeswere expected to meet daily quotas and were ranCommon to domly monitoredfor service quality standards. much service work, this work setting parallelsmany characteristics of an environmentin which burnoutcould occur. Typically, customer service representativesindicated that they work long hours, lack autonomy, bear responsibility have insufficientresourcesand guidelines withoutauthority, to handle problems, face demandingquotas, and often endure manager apathy. Also, they typically complained of job." being in a "dead-end Interestingly,it became clear that the study of burnout poses additionalconstraintsin designing a field study. Our focus groups revealed that focal boundary spanners were likely to perceive detailed questionnairesas "burdensome" In a sense, a surveythatwas not and "unwanted" paperwork. adapted to the boundary spanners' stressful environment may run the risk of being ignored at best and appearas an additionalrole stressorat worst. Thus, to obtain high-quality data, we balancedthe need for brevitywith thatof multiitem scales by using either a short form of a scale or parsimonious versions of extant measures.

Burnoutand CustomerService Representatives Table 1

UTILIZED ANDCORRELATIONS INTHESTUDY FORTHECONSTRUCTS DEVIATIONS MEANS,STANDARD
Correlations Constructa BurnoutDimensions (BURNOUT)b Emotionalexhaustion(EE) Reduced personalaccomplishment(RPA) Depersonalization(DP) Role Stressors (RSTRESS) Role conflict (RC) Role ambiguity(RA) Role overload (RO) Psychological Outcomes(PSY-OUT) Job satisfaction(JS) Mean 2.71 2.34 2.64 3.24 2.61 2.95 3.45 Standard Deviation .87 .82 .92 .95 .82 1.05 .90 EE RPA DP RC RA RO JS OC TF

563

JPi

.30d .52d .36d .29d .37d -.36d

-.53d

.301 .11 .23d .06 -.22d

-.24d

.28d .20d .25d -.43d

-.55d

.40 .40d -18d

-.38d

.20d -.26d -.30 -.14e -.10 -.09 -.14e -.39" -25d .01 .09

Organizational commitment (OC)

Turnoverintentions(TI)C Behavioral Outcomes(BEH-OUT) Job performance1 Job performance2

3.12
2.86 4.39 4.36

.76
1.20 .64 .59

.55d
49 .10 .11 .54d .08 .07 -.12 .08

-.38d -.13 -.18d

-.10 -. -.35d -.42d

40d -.04 -.04

-.30d .05 .05

.77d

were summedand averagedin orderto obtaina summatedindex. The summarystatisticsare reported aTheitems correspondingto each construct/dimension for this index. bTheBURNOUT dimensions were measuredby utilizing a 6-point scale. All other constructswere measuredby a 5-point scale. The composite reliabilities for the EE, RPA, and DP dimensions of BURNOUT are .80, .81, and .82, respectively. 'The measureof TurnoverIntentionswas reverse scored (i.e., lower numbersmean higher turnover)so that its correlationswith other dimensions of PSYOUT would be in the same direction. dp <.0.
ep < .05.

mensional structurecorrespondingto first-order dimensions of EE, RPA, and DP provided an acceptable fit to the data with fit statistics rangingfrom x2 = 7.4, df = 6, p > .28 (for boss-relateditems) to x2 = 2.6, df= 6, p > .86 (for coworker related items). Furthermore,the three dimensions accounted for more than 67% of the variance in each role member items. Thus, items correspondingto each dimension were summedacross all role membersto obtain a combeposite score for EE, RPA, and DP. The intercorrelations tween composites rangefrom .30 (RPA-DP)to .52 (EE-DP). Job outcomes.Psychological outcomes were measuredby three separatefirst-orderconstructs.JS was operationalized as a 26-item scale adaptedfrom Churchill,Ford, and Walker's (1985) study thathas been frequentlyused in marketing. The alpha reliability of this scale was estimatedas .92. OC was operationalizedby use of a 6-item version of a scale from Mowday,Steers, and Porter(1979). The scale achieved acceptablereliability (a = .79). Finally, TI was assessed by a three-item measure based on Donnelly and Ivancevich (1975). In addition,this measurehas been shown in several studies to be a consistent predictorof actual turnover.The Cronbach's a for this measure was .85. Behavioral outcomes were measuredby a self-rating,6-item measureof JP drawnfrom Dubinsky and Mattson(1979). Boundaryspanners ratedthemselves in comparisonwith theircoworkerson a 5-point scale (rangingfrom "bottom 10%"to "top 10%") on six differentdimensions (quantity,ability,potential,customerrelations,time managementand planning,and knowledge; a = .84). To examine psychological and behavioral second-order dimensions, all outcomes measures were pooled together and analyzedby iterativeprincipalaxis factoranalysis using SPSSx. For this analysis, two composites were utilized for each measureof job outcomes (JS, OC, TI, and JP) by ran-

domly splitting the correspondingitems into two groups. Thus, eight composites were developed (two composites for each of the four measuresof job outcomes). Factoranalysis of these eight composites revealed two eigenvalues larger than 1 (3.62 and 1.86) thatexplained68% of the variance.In line with our expectations,the oblique rotatedfactorpattern indicatedthat measures of JS, OC, and TI loaded together, whereasJP measuresloaded on a separatefactor (factor in= .07). Descriptive statistics and correlations tercorrelation for the study constructsare in Table 1. Methodof Analysis All analyses involved the method of Latent Variable Structural Equationsmodeling using the statisticalsoftware EQS (Bentler 1990). Specifically,we test the hypotheses by using a competing model analysis such that two substantive models are estimated and compared. The first model, referred to as the DIRECT EFFECTS model, involves estimatingthe effects of role stressorson psychological and behavioral outcomes. The second model, the PARTIALMEDIATIONmodel, representsthe posited hypotheses and involves estimatingthe effects of (1) role stressorson burnout, (2) role stressorson psychological and behavioraloutcomes, and (3) burnouton psychological and behavioraloutcomes. The hypothesized mediating effect of burnoutis supported if, comparedwith the DIRECTEFFECTSmodel, the PARTIAL MEDIATIONmodel yields (1) higher explainedvariances (i.e., R2), (2) a significant relationshipbetween role stressors on burnout, (3) substantially diminished or insignificanteffects of role stressorson psychological and behavioral outcomes, and (4) highly significant effects for burnouton psychological and behavioral outcomes. If the PARTIALMEDIATIONis supported,then the model coefficients can be tested.

564 FINDINGS

JOURNAL OF MARKETING NOVEMBER 1994 RESEARCH,

Figure2. In termsof EE, Figure2 reveals that customerservice representatives reportedhigherlevel of burnouttendencies than police, medical residents,and welfare, social, and mentalhealthworkers.However,customerservice representatives'EE levels were lower thanthatof lawyers and childand infant-careworkers. Likewise, for RPA, customer service representativesindicate that their burnout tendencies parallelthose reportedfor the social and federalpublic contact workersand are among the highest across various pro-

Burnout:Magnitudeand DiscriminantValidity The estimated means and standarddeviations of burnout tendencies among our sample of customerservice representatives are in the top panel of Table 1 (underBURNOUT). Figure 2 provides a comparison of means obtained in our study with those reportedin the literaturefor other professions. Only studies that utilized the MBI were included in

Figure2
MEAN VALUES FOR BURNOUT DIMENSIONS FOR CUSTOMER SERVICE REPRESENTATIVES IN COMPARISONTO OTHER PROFESSIONS

6'

5... 4.22 <-- InfantcareWorkersI

fE

45

3.11 <-- Childcare Workersh 2.89 <-- Lawyers d

3
3J

2.71 <-- Customer Service Reps 2.45 <-- Police J 2.34 <-- Medical Residentsc 2.18 <- Public Contact Employees i 2.04 <-- Teachers b 2.00 <-- Nurses ' 1.93 <-- Welfare Workers f 1.87 <-- Mental Health Workerse

2.65 <-- Public Contact Employees i 2.50 <-- Welfare Workers f

2.64 <-- Customer

Service Reps

2.41 <-- Lawyers d 2.34 <-- Customer Service Reps 2.11 <-- InfantcareWorkers I 2.03 <-- Medical Residents c

2.23 <-- Teachers b 2.17 <-- Mental Health Workersc 2.08 <-- Medical Residents c 2.03 <-- Lawyers d

1.70 <-- Nurses a

1.76 <-- Social Workers J

1.20 <-- Childcare Workers h 1.25 <-- Nurses 1.10 <-- Public Contact Employees i 1.03 <-- Mental Health Workerse
1

Emotional Exhaustion
aLeiterand Maslach (1988) bJackson,Schwab, and Schuler (1986). and colleagues (1986). CRafferty Turner,and Brief (1987). dJackson, eLeiter(1990). fLeiter(1991). gLee and Ashforth(1993). hFuquaand Couture(1986). iMaslachand Jackson (1985). JGainesand Jermier(1983).

Reduced Personal
Accomplishment

Depersonalization

Burnoutand CustomerService Representatives

fessions. Finally, in terms of DP, the mean values obtained for our customer service representatives sample are the highest relative to those for other professions. Only lawyers approachthe magnitudeof DP tendencies reportedby customer service representatives. Table 1 also depicts the correlationsbetween the three burnoutdimensions and various role stressors and job outcomes. Withoutexception, the burnoutdimensionscorrelate positively with role stressors and negatively with job outcomes, as expected. Moreover,the correlationsbetween the first-orderdimensions of burnoutand role stressors range from .06 to .37 (average= .24); thus, the sharedvariancebetween these concepts (i.e., redundancy) is no morethan 13% (average<6%). This supportsthe view thatburnoutand role stressorsare empiricallydistinctconstructs.The correlations between the first-orderdimensions of burnoutand psychological outcomes range from -.10 to -.55 (average = -.35) such that the redundancydoes not exceed 31% (average <13%). Also, the average correlationbetween burnoutand psychological outcomes is less than the averagecorrelation among the three psychological outcomes (-.35 versus .53). A similarpatternholds for the relationshipbetween the firstorderdimensions of burnoutand behavioraloutcomes. This suggests that burnoutis a distinct construct,different from other psychological and behavioraljob outcomes. Testof Partial Mediating Thesis Analysis of competingmodels. Table2 summarizesthe results from fitting the DIRECT EFFECTS and PARTIAL MEDIATIONmodels to customer service representatives' data. Several inferencescan be drawnfrom Table2. First, in terms of fit statistics, both models appearto be reasonable fits to the data. These models yields fit indices (i.e., NFI, NNFI, and CFI) that approachor exceed .90 and residuals (i.e., AOSR) that tend to approach .05, the value recommended for acceptable fits. Although the x2 statistic is significantfor both models (at p = .05), problemswith this test statistic for large sample sizes (N > 300) are well documented. Second, in contrast to the first inference, the two competing models differ significantly in terms of the variance they account for in job outcomes. Comparedwith the DIRECT EFFECTS model, the PARTIAL MEDIATION model representsa twofold increase in R2 for psychological outcomes (.72 versus .35) and an over tenfold increasein R2 for behavioraloutcomes (.14 versus .01). Admittedly,both models are less effective in explainingbehavioraloutcomes of customer service representatives.Third, the PARTIAL MEDIATIONmodel shows evidence of a discerniblemediating effect. Specifically, the effect of role stressors on burnoutis highly significant(t = 7.8, p < .01), and so are the effects of burnouton psychological (t = 6.2, p < .01) and behavioraloutcomes (t = 3.9, p < .01). Finally, comparedwith the DIRECTEFFECTSmodel, the PARTIAL MEDIATION model significantlyattenuatesthe effects of role stressorson psychological outcomes (-.59, p < .01 to -.12, p > .10). Surprisingly,the effect of role stressorson behavioraloutcomes shifts from insignificance in the DIRECTEFFECTSmodel (.05, p > .10) to significance in the positive directionunder the PARTIALMEDIATIONmodel (.35, p < .01) (discussed subsequently).Takentogether,the preceding evidence sugMEDIATION model is a significant gests thatthe PARTIAL Table 2

565

THEPARTIAL MEDIATING RESULTS FROM TESTING A THESIS: SECOND-ORDER MODEL OF BURNOUT AS A MEDIATOR BETWEEN ROLESTRESSORSANDJOB OUTCOMES
DIRECTEFFECTS Modela -.59 (7.1) .05 (.8) .78 (11.8) .50 ( 7.9) .56 (8.9)
-

Parameterc

PARTIAL MEDIATION Modelb -.12 (-1.2) .35 (3.0) .64 (7.8) -.77 (-6.2) -46 (-3.9) .75 (11.7) .50 ( 7.7) .58 (9.1) .75e .42 (6.6) 69 (10.5) .65e .86 (11.0) .65f ( 9.4) .77e .99 (21.1) 146.0 38 .91 .90 .93 .05 .72 .14

StructuralCoefficients RSTRESS- PSY-OUT RSTRESS - BEH-OUT RSTRESS -e BURNOUT BURNOUT - PSY-OUT BURNOUT -e BEH-OUT MeasurementCoefficients RSTRESS - RC RSTRESS - RA RSTRESS - RO BURNOUT - EE BURNOUT -- RPA BURNOUT - DP PSY-OUT-- JS PSY-OUT- OC PSY-OUT- TI BEH-OUT - JPi BEH-OUT - JP2 Goodness of Fit Statisticsd X2 df NFI NNFI CFI AOSR R2 for PSY-OUT R2 for BEH-OUT

.66e .85 (10.1) .65s (9.5) .78e .99 (21.6) 71.0 17 .91 .89 .93 .06 .35 .01

aThe DIRECT EFFECTS model includes only the direct effects role stressors -- psychological outcomes and role stressors -+ behavioraloutcomes. tThe PARTIALMEDIATIONmodel includes two direct effects, role stressors -e psychological outcomes and role stressors -+ behavioraloutcomes, and two mediatedeffects, role stressors-- burnout-+ psychological outcomes and role stressors-+ burnout-e behavioraloutcomes. CThe estimates are standardized coefficients obtainedfrom itparameter eratively reweightedgeneralized least squares solution using the software EQS. The t-statisticfor each estimate is in parenthesis. dNFI= Normed-Fit-Index, NNFI = Non-Normed-Fit-Index, CFI is Comand AOSR = the Average-Off-Diagonal-Standardizedparative-Fit-Index, Residual (cf. Bentler 1990). eThe unstandardized coefficient correspondingto this parameterwas set to equal 1.00 to fix the scale of the latent variable. fThe turnoverintentionsmeasure(TI) was reversescored so thatits loading on PSY-OUTis in the same directionas JS and OC.

improvementover a DIRECT EFFECTS model and substantiatesthe partialmediatingrole of burnout. Testof hypotheses.Before testing model coefficients, it is useful to inspect the measurementcoefficients of the PARTIAL MEDIATIONmodel in Table 2. Note that the factor loadings of RC, RA, and RO are significant and range from .50 (t = 9.1, p < .01) to .75 (t = 11.7, p < .01). Consistent with this, the composite reliability of the second-orderrole stressors construct is .65. Likewise, the factor loadings of burnout dimensions are significant (ranging from .42 for RPA to .75 for EE, p < .01) and composite reliabilityof the second-orderburnout construct is .66. The outcome constructs,psychological and behavioraloutcomes, also depict

566

JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1994 velopment of the burnoutconstruct. Nevertheless, our research offers insights heretoforeunavailablein marketing's role stress research by yielding demonstrableevidence of burnout's(1) relevancefor customerservice representatives, (2) consistently dysfunctional consequences for customer service representatives'behavioral and psychological outcomes, and (3) partialmediatingeffect in role stress models. Burnout'sRelevance Our comparisonof burnoutmeans across variousprofessions underscores burnout's relevance for marketing. In some cases, burnoutmean scores are comparableor higher than samples from ocfor customer service representatives cupationsconsideredprototypicalof burnoutand where its consequencesare well documentedandresearched(e.g., police, medical residents, social/mentalhealth workers).Customer service representativesapparently experience substantialamountof EE and a sense of RPA on theirjob. Notendenciesto depertably,customerservice representatives' sonalize interpersonalrelationshipsare among the highest across professions, suggesting that these boundaryspanners likely detach themselves emotionally from the people (i.e., customers)thatthey are expected to serve. This sense of detachment may underlie customer service representatives' feelings of RPA and may contributeto service quality problems. In addition, this lends credence to the popular press claims that some marketingboundary spanners are prime candidatesfor burnout(Miller et al. 1988). Takentogether, it appearsthat the magnitudeof burnouttendencies among are high enough to warrant customerservice representatives andconcerned researchers the seriousattentionof marketing managers. Burnout'sConsequences Our results offer clear and compelling evidence of bumout's consequences for customer service representatives. Specifically,burnouthas a (1) significant,negativeimpact on behavioral outcomes or performanceof customer service representatives (P = -.46), (2) large, negativeimpact on the psychological outcomes of customer service representatives(3 = -.77), and (3) significantlystrongereffect on behavioraland psychological outcomes relativeto the direct effects of role stressors(-.46 versus .05; -.77 versus -.59). Thus, the inclusion of burnoutin a role stress model yields nontrivial increases in explained variance for behavioral (AR2= .13) and psychological outcomes (AR2= .37). These resultsparallelfindings from occupationalpsychology literaturewith samples of "helping" professionals(Lee andAshforth 1990; Leiter and Maslach 1988). Burnout'sMediatingRole Our study offers evidence that substantiatesthe hypothesized theoretical argumentsregardingburnout'smediation in role stress processes. In accord with HI, role stressors have a significant, positive relationshipwith burnout(p3= .64). In addition,burnouthas significant dysfunctionalimpact on psychological and behavioraljob outcomes, supportingH2 and H3, respectively.Takentogether,it is apparent that role stressorsaffect burnouttendencies of customer service representatives and, in turn,burnoutaffects job out-

factorloadings thatexceed .65 (p < .01) and composite reliabilities that are .77 and .87 respectively. Consistent with Hi, Table 2 reveals that role stressors have a significant, positive effect on BURNOUT with a standardized coefficient of .64 (t = 7.8, p < .001). Thus, as boundaryspannersperceive greaterrole stressors,they appearto reflect more burnouttendencies.In turn,BURNOUT results in a substantivelysignificant but negative effect on psychological outcomes as the path coefficient is estimated at -.77 (t = -6.2, p < .01). This offers strong supportfor H2. Moreover,in accordwith H3, burnouthas a significant,negative influence on behavioraloutcomes with a standardized coefficient of -.46 (t = -3.9, p < .01). This suggests that. with increasingburnouttendencies, boundaryspannersdepict greaterdysfunctionaleffects in both theirpsychological attitudestowardand behavioralresponses on the job. However, the impact of burnouton psychological outcomes is aboutone and a half times as large as its effect on behavioral outcomes. The partial,unmediatedeffects of RSTRESS on job outcomes, after controlling BURNOUT, are also provided in Table 2. Note that the unmediatedeffect of role stressorson psychological outcomes is insignificant because the standardizedcoefficient is only estimated at -.12 (t = 1.2, p > .10). This offers supportfor H4. However, the unmediated effects of role stressors on behavioral outcomes is significant and positive because the pathcoefficient is estimatedas .35 (t = 3.0, p < .001). Because of its significanceand the direction of effect, this result is contraryto H5. Note that it is only the partial,unmediatedeffect of role stressors on behavioral outcomes that is positive; the total effect of role stressors on behavioral outcomes is indeed insignificant.3 Evidently,the overall effect of role stressorson behavioral outcomes appearsto be cancelled by two countervailingeffects-a positive, eustress-type effect (i.e., RSTRESS -BEH-OUT), and a negative, distress-type effect (i.e., RSTRESS -e BURNOUT -+ BEH-OUT).More importantly, utilizing burnoutas a mediatorhelps uncover the positive, eustress-typeeffects of role stressors. DISCUSSION We aimed to explore burnout'srelevance for marketingoriented boundary spanners and examine its substantive contributionto the study of role stress processes. As an initial study of boundary spanner burnout, however, certain limitationsshouldbe noted.The findings are based on a specific sample of boundary spanners (i.e., customer service representatives)working in a single firm. Furtherstudies with different samples and across firms and industriesare needed for generalizableevidence. Moreover,the relevance of the burnout construct for various marketing-oriented boundaryspannerswarrantsfurtherexamination.This may require psychometric assessment and, perhaps, furtherde3The total effect of RSTRESS on BEH-OUT is a sum of two components: (1) an unmediated,direct effect thatis estimatedat .35 and (2) a mediated, indirect effect as RSTRESS effects BURNOUT (coeff. = .64) and BURNOUT effects BEH-OUT (coeff. = -.46). The indirect effect is the productof the two involved coefficients and equals -.29. Consequently,the total effect of RSTRESS on BEH-OUT is .06. Note that this total effect equals the correspondingeffect in the DIRECT EFFECTSmodel, within roundingerror.

and Customer ServiceRepresentatives Burnout

comes. Moreover,in accordwith H4, the directeffect of role stressorson psychological outcomes reduces from substan= -.59) to insignificant(13 = -.12) when the mediattial (13 ing effects of burnoutare included.We appearto have sufficient evidence to conclude thatburnoutis a criticalmediator in role stress processes, and as such it clarifies the link between role stressors and job outcomes. Thus, its omission from furtherrole stress researchis not justifiable. The effects involving behavioraloutcomes yield the most interestingresults. Controllingfor burnout,the effect of role stressors on behavioral outcomes shifts from insignificant = .35). We inter(P = .05) to a significant,positive effect (1P pret this shift in light of the Yerkes-Dodsonlaw that posits that role stressorsand performanceare relatedby an inverted-U relationship.We reason that, in a given randomsample, customer service representativesare probably located randomly at different points of this U-shaped curve. As such, at the aggregatelevel, the empiricalestimation of the linear relationshipbetween role stressors and performance yields a flat line with an insignificant slope coefficient. However, after controlling for burnout, customer service representativesare likely located only along the upward sloping part of the U-shaped curve, because customer service representativeswho are located along the downward sloping curve likely evidence burnout tendencies and, hence, are partialledout. For this reason, the empiricalestimation of the linear relationshipbetween role stressorsand behavioraloutcomes, after controllingfor burnout,yields a positive coefficient. In other words, the effect of role stressors on behavioraloutcomes for one partof the range (positive slope) is inconsistentwith second part(negativeslope). This inconsistency is likely a major contributingfactor in wide variability of previous role stress findings. Because burnouthas a consistent, negative effect on behavioraloutcomes, controllingfor burnouthelps uncoverthe eustressinherentin role stressors. Overall, the precedingclear and consistent evidence substantiatesthe theoreticalargumentshypothesizedto account for burnout'simpact in role stress processes. Specifically, these argumentsinvolve three propositions.First,burnoutis thoughtto be a strongerpredictorof job outcomes than role stressors because of its "accumulative"capacity; that is, burnout presents the combined influence of several role stressors.As such, the combined effect is likely to be more significant than its individual components. Second, burnout'seffect is thoughtto be more consistentthanthatof role stressors because it considers the coping resources of the individual boundaryspanner.Because burnouttendencies appearwhen role stressorsexceed the individual'scoping resources,burnoutis thoughtto be consistentlydysfunctional; in contrast,role stressorscan be either dysfunctional (distress) or functional (eustress) (cf. Singh 1993). Third, burnoutis more closely associated with job outcomes because it is a psychological condition attainedas an outcome of a stress appraisalprocess (Lazarusand Folkman 1984). Because job outcomes, especially psychological outcomes, are also consequences of other appraisalprocesses, it is expected that these outcome states be related. MANAGERIAL IMPLICATIONS

567

In describing the burnout'sconsequences, Pines, Aronson, and Kafry (1981, p. 3-4) observe:
Tragically,burnoutimpacts precisely those individuals who had once been among the most idealistic and enthusiastic.We have found, over and over again, that in orderto burnouta person needs to have been on fire at one time. It follows, then, that one of the great costs of burnoutis the diminutionof the effective service of the best people in a given profession. Accordingly, everyone is the poorerfor the existence of this phenomenon.

Indeed, the costs of burnoutare high for the individualand organizationalike. The results of this study imply that managers' focus on identifying and possibly eliminatingall role stressors in boundary spanning positions is probably misguided. Instead, they should view burnout as the critical variableof interestthat unequivocallycapturesthe dysfunctional aspects of a stressful role environment.This strategy is likely to preservethe motivationalelements in a role that come from "solving" challenging problems (and may be perceived as stressful) while ensuring that boundaryspanners do not burnout as chronicproblemsoverwhelmthe resources of the individual.Moreover,because it capturesthe individual'snet affective reactionto the job arising from all chronic and negative aspects of the job (Maslach and Jackson 1981), burnoutserves as a uniquequality-of-life-on-thejob indicator.Managerscan initially utilize benchmarkstudies to determinetolerable limits of burnouttendencies that are specific to their personnel, industry, and job environment. CONCLUSION For over two decades, researchershave studied various role stressors and their direct impact on psychological and behavioraloutcomes of boundaryspanners.This direct, aggregate analysis using global measures has invariablyrevealed that role stressors have dysfunctionaleffects resulting in a conventionalwisdom that boundaryspanningroles ideally must be free of role stressors to maintain optimal performanceand psychological well-being. Although at the aggregate,global level, this conventionalwisdom is probably undisputed,a contraryview is emerging at the margins as researchersbegin to recognize the energizing and functional aspects of certainrole stressors(cf. Singh 1993). By developing and testing the partial mediation thesis, our study contributes to this growing literature.Our findings offer the following new directions into understandingrole stress processes:
*It clarifies the conceptual and etiological distinction between burnoutand role stressors.Thus, to the extent that researchersuse burnout and role stressors interchangeably, such practice is theoretically and empirically not justified. *Itindicates thatbumoutcan account for variabilityinherent in individualvulnerabilityto role stressorsby controlling for excessive role stress and, consequently,uncovering the eustress inherentin role stressors.As such, it appearsuseful to include burnoutin furtherresearchof role stress processes.

568 *It opens new areas of research that provide answers to why and how role stressorsaffect job outcomes. Explicit in the partial mediating frameworkis the view that role stressors result in burnout,and in turn negative job outcomes, only when employee's coping mechanisms are overwhelmed. Thus a research programcan be initiated that examines the role of social support, organizational factors (e.g., autonomy),and coping behaviorsin burnout processes. *It offers insights for burnout management. Evidently, eliminatingrole stress not only is a formidabletask, but it because it purgesjobs of may also be counterproductive motivational elements that come from accomplishing

JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1994

challenging tasks. The burnout literature shifts focus away from role stressorsand emphasizesthe significance of coping resources. The more we know about the etiology of burnout, and the processes by which role stressors positively and negatively affect job outcomes, the greater the likelihood of creating boundary roles that balance the need for retaining the eustress inherent in role stressors and reducing burnout among employees. As Maslach (1982, p. 40) explains, "the promise inherent in understanding burnout is the possibility of doing something about it."

Appendix A
THEOPERATIONAL MEASURE BURNOUT UTILIZED TO MEASURE INCSRS
Dimension Depersonalization Role Member Customer Boss Coworkers TopManagement Reduced Personal Accomplishment Customer Boss Coworkers TopManagement Emotional Exhaustion Customer Boss Coworkers Item Descriptiona I feel I treatsome customersas if they were impersonal"objects." I feel indifferenttowardsome of my customers. I feel a lack of personalconcern for my supervisor. I feel I'm becoming more hardenedtowardmy supervisor. I feel I have become callous towardmy coworkers. I feel insensitivetowardmy coworkers. I feel I am becoming less sympathetictowardtop management. I feel alienatedfrom top management. I feel I performeffectively to meet the needs of my customers.b I feel effective in solving the problemsof my customers.b I feel I am an importantasset to my supervisor.b I feel my supervisorvalues my contributionto the firm.b I feel my coworkerstruly value my assistance.b I feel I am a positive influence on my coworkers.b I feel I satisfy many of the demandsset by top management.b I feel I make a positive contribution towardtop managementgoals.b Workingwith customersis really a strainfor me. I feel I am workingtoo hardfor my customers. Workingwith my boss directly puts too much stress on me. I feel emotionally drainedby the pressuremy boss puts on me. I feel frustrated because of workingdirectly with coworkers. I feel I work too hardtrying to satisfy coworkers.

Top Management

I feel dismayed of topmanagement. by theactions

I feel burnedout from tryingto meet top management's expectations.

aForall items, the respondentswere asked to evaluatethe extent to which each statementaccuratelydescribeshow they feel about theirjob. The responses were obtained by utilizing a 6-point Likert-typescale with response categories as follows: (1) Is very much UNLIKE me, (2) Is UNLIKE me, (3) Is somewhat UNLIKE me, (4) Is somewhatLIKE me, (5) Is LIKE me, (6) Is very much LIKE me. bThese items were reverse scored.

REFERENCES "Work-home ConandS. Conley, S. B., P.Bamberger, Bacharach, the Impactof flict AmongNursesand Engineers: Mediating atWork," Journal andSatisfaction on Burnout RoleStress of Organizational Behavior, 12, 39-53.

(1976), A., JeffreyT. Walsh,andThomasD. Taber Beehr,Terry and Organizationally of Stress to Individually "Relationship Journal of Order Needsas a Moderator," Valued States:Higher
Applied Psychology, 6 (1), 41--47.

in Structural FitIndexes P.(1990),"Comparative Models," Bentler,

Psychological Bulletin, 107, 238-46. Churchill, Gilbert A., Jr., N. Ford, S. Hartley, and 0. Walker,Jr.

A Performance: of Salesperson (1985), "The Determinants

Meta-Analysis," Journal of Marketing Research, 22 (May),

103-18.
, and 0. Walker,Jr. (1974), "Measuringthe Journal of Marketing of Industrial Job Satisfaction Salesmen," 254-60. 11 (August), Research,

Cordes, C.L. and T.W.Dougherty(1993), "A Review and an InteAcademyof Management grationof Researchon Job Burnout," Review, 18, 621-56. Donnelly, James H. and John M. Ivancevich(1975), "Role Clarity 71-74. and the Salesman,"Journal of Marketing,39 (January), Dubinsky,A. and B. Mattson(1979), "Consequencesof Role Conflict and Ambiguity Experiencedby Retail Salespeople,"Journal of Retailing, 55 (Winter),70-86. and Locus of Fuqua,Robertand KristinCouture(1986), "Burnout Controlin Child Day Care Staff,"Child Care Quarterly,15 (2), 98-109. Gaines, Jeannieand John M. Jermier(1983), "EmotionalExhaustion in a High Stress Organization," Academy of Management Journal, 26 (4), 567-86. Golembiewski, Robert, Robert F. Munzenrider, and Jerry G. Stevenson (1988), Stress in Organizations: Toward a Phase Model of Burnout.New York:PraegerPublishers. Hunsaker,Johanna (1986), "Burnout:The Culminationof Long Term Stress," Industrial Management,28 (November/Decem-

Burnout and Customer Service Representatives

ber), 24-26. Jackson, Susan E., Richard L. Schwab, and Randall S. Schuler (1986), "Toward an Understanding of the Burnout Phenomenon,"Journal of Applied Psychology, 71 (4), 630-40. , Jon A. Turner,and ArthurP. Brief (1987), "Correlates of BurnoutAmong Public Service Lawyers,"Journal of Occupational Behavior,8, 339-49. Lazarus,R. S. and S. Folkman(1984), Stress,Appraisal,and Coping. New York:SpringerPublishing. Lee, Raymond and B. Ashforth (1990), "On the Meaning of Maslach's Three Dimensions of Burnout,"Journal of Applied Psychology, 75 (6), 743-47. and (1993), "A Longitudinal Study of Burnout Among Supervisors and Managers: Comparison Between Leiter and Maslach (1988) and Golembiewski et al. (1986) Models," OrganizationalBehavior and HumanDecision Process, 54, 369-98. Leiter,Michael P. (1990), "The Impactof Family Resources, Control Coping, and Skill Utilization on the Development of Burnout: A Longitudinal Study," Human Relations, 43, 1067-83. (1991), "Coping Patternsas Predictorsof Burnout:The Function of Control and Escapist Coping Patterns," Journal of OrganizationalBehavior, 12, 123-44. and ChristinaMaslach (1988), "The Impactof Interpersonal Environment on Burnout and OrganizationalCommitment,"Journal of OrganizationalBehavior, 9, 297-308. Burnout:Definitional Maslach, Christina(1982), "Understanding in Job Stress and Issues in Analyzing a Complex Phenomenon,"

569
Burnout: Research, Theory, and Intervention Perspectives, Whiton StewardPine, ed. Beverly Hills, CA: Sage Publications, 29-40. of Exand Susan E. Jackson(1981), "TheMeasurement perienced Burnout," Journal of Occupational Behavior, 2, 99-113. Setand (1984), "Burnoutin Organizational tings,"Applied Social PsychologyAnnual, 5, 133-53. and (1985), "The Role of Sex and Family Variablesin Burnout," Sex Roles, 12, 837-51. Miller, Annetta, Karen Springen, Jeanne Gordon, Andrew Murr, Bob Cohen, and Lisa Drew (1988), "Stress on the Job," Newsweek(April 25), 40-45. Pines, Ayala M. and Elliot Aronson(1988), CareerBurnout:Causes and Cures.New York:The Free Press. , and Ditsa Kafry (1981), Burnout:FromTe), dium to Personal Growth.New York:The Free Press. Rafferty,J., J.P. Lemkau, R.P. Purdy,and J. Rudisill (1986), "Validity of Maslach BurnoutInventoryfor Family Practice Physicians,"Journal of Clinical Psychology,42 (3), 488-92. Seyle, Hans (1976), The Stress of Life. New York:McGrawHill. Role Ambiguity:Facets,Determinants Singh, J. (1993), "Boundary and Impacts," Journal of Marketing,57 (April), 11-31. Wolpin, J., R. Burke, and E. Greenglass (1991), "Is Job Satisfaction an Antecedent or a Consequence of Psychological Burnout?" HumanRelations, 44 (2), 193-208. Yerkes, Robert M. and John Dodson (1908), "The Relation of Journalof Strengthof Stimulusto Rapidityof HabitFormation," Neurologyand Psychology, 10, 459-82. Comparative