Are Anxious Workers Less Productive Workers? It Depends on the Quality of Social Exchange

14 Pages·2015·0.29 MB·English

Journal of Applied Psychology
Are Anxious Workers Less Productive Workers? It
Depends on the Quality of Social Exchange
Julie M. McCarthy, John P. Trougakos, and Bonnie Hayden Cheng
Online First Publication, August 10, 2015. http://dx.doi.org/10.1037/apl0000044
CITATION
McCarthy, J. M., Trougakos, J. P., & Cheng, B. H. (2015, August 10). Are Anxious Workers
Less Productive Workers? It Depends on the Quality of Social Exchange. Journal of Applied
Psychology. Advance online publication. http://dx.doi.org/10.1037/apl0000044
JournalofAppliedPsychology ©2015AmericanPsychologicalAssociation
2015,Vol.100,No.5,000 0021-9010/15/$12.00 http://dx.doi.org/10.1037/apl0000044
Are Anxious Workers Less Productive Workers?
It Depends on the Quality of Social Exchange
Julie M. McCarthy and John P. Trougakos Bonnie Hayden Cheng
UniversityofToronto–Scarborough HongKongPolytechnicUniversity
Inthisarticle,wedrawfromConservationofResourcesTheorytoadvanceandtestaframeworkwhich
predictsthatemotionalexhaustionplaysanexplanatoryroleunderlyingtherelationbetweenworkplace
anxiety and job performance. Further, we draw from social exchange theories to predict that leader–
y.
shers.broadl mmmeaemndcibaete.erdFeitnxhdcehinlaignnskgeabceartnowdsesecanoww3-oowrrkkaepvrleaecsxetcuahdnayxniogefetypwaoinllildcemjooibtfifpgieacrteefrostrhmseuaphnpacoremr,toefdvueloreuafrfnedmctaosbdooevfl.eaEtnhmxeioeettfiyfoenocantlojefoxcbhoagpunesrittfiioovrne-
dpubliminated ienxthearfuesrteionnce,.wFhuirlethleera,dceorw–morekmerbeexrcehxacnhgaengmeitmigiatitgeadtethdethpeosniteigvaetirveelartieolantiboentwbeeetwneaennxieemtyoatinodnaelmeoxthioanuas-l
salliedisse tviioaneamnodtijoonbalpeexrfhoarumsatinocne.anTdhihsigshtulidgyhtesluthceidvaatelusethoefderfafewcitnsgoofnwsoorckipallarceesoaunrxcieestytooonffrseestouthrecepodteepnlteitailolny
ofitobe harmfuleffectsofworkplaceanxietyonjobperformance.
et
onot Keywords: workplace anxiety, emotional exhaustion, job performance, social exchange, cognitive
n
ors interference
i
ationand
cier
os
su
s
Aal WorkplaceanxietyhasbeenestimatedtocosttheU.S.economy (Boydetal.,2009),andeconomicsuccess(Forsyth,Kelly,Fusé,&
ologicalindividu osevreirou$s4c0obniclelironn,baencnauuaslelyre(cKeenstssleurrv&eyGsrseuegngbeesrtgt,ha2t04012%).Tofhiwsoirska- Ktaalrceoknlas,eq2u0e0n4c).esImfoproretmanptllyo,yaenesxiaentydaolsrgoahnaiszaptoiotennstiianlltyhedeftorirmmeonf-
he ersreportelevatedlevelsofworkplacetension(AmericanPsycho- reduced job performance (Ford, Cerasoli, Higgins, & Decesare,
ch
yt
Psof logical Association, 2012), and anxiety-related absences are on 2011). Given that job performance is one of the outcomes that
ne averagefourtimeslongerthanothernonfatalillnessesorinjuries organizationsandemployeescaremostabout(Motowidlo,2003),
as
cu
erial (Bureau of Labor Statistics, 2001). Although employees vary in considerationoftheroleofworkplaceanxietyinthecontextofjob
Amson their propensity to experience workplace anxiety, there is also performance is crucial and serves as the first step in developing
eer evidencethatitisontherise(D’Mello,2012;Twenge,2000). strategiestomitigateitspotentiallynegativeeffects.
hp
ythe Thisistroublesomefromtheperspectiveofemployees,because Considering the significant role that anxiety can play in em-
bt
yrightedolelyfor iflteavmmeilalsyyolcfifojeonbt(rSidbciuhsstieeamtitsoafnat,chtMeioicnnBc(rBrieeoarys,ed&d,LsveeawpnainrGa,ut&inodnSyab,ge2et0wr,0e23e0)n0a9wn)d.oTrhkhigiashneidsr pqaluoleaynrecgeeesatnatsuitsmuodcbeieasrteaodnfdwebimtehhpaairvniixcoiarelst,ya,artsiitcwilseesnllooatnssutahrnpexriidseeitntyrgimathneadnttatshltrecerosesnasierne-
ps
coed also troublesome from the organization’s perspective, because occupationalhealthjournals(e.g.,Fay&Sonnentag,2002;Fordet
isnd high levels of anxiety have negative implications for ethical be- al., 2011; Gomes, Faria, & Gonçalves, 2013). Despite this work,
entnte havior (Kouchaki & Desai, 2015), organizational effectiveness researchexploringthelinkbetweenworkplaceanxietyandoverall
mi
docucleis jloarbgepleyrfeoxrammainnceedhthaesrbeelaetniolnimsbiteetdw.eRenesgeeanrcehralinantxhiiestydolemvaeilnsahnads
hisarti performance on specific tasks, such as performance on selection
Ts
hi tests(e.g.,Proost,Derous,Schreurs,Hagtvet,&DeWitte,2008),
T
JulieM.McCarthyandJohnP.Trougakos,DepartmentofManagement, job interviews (McCarthy & Goffin, 2004), and school-based
RotmanSchoolofManagement,UniversityofToronto–Scarborough;Bon- examinations(e.g.,Seipp,1991).Thisworkdrawsfromcognitive
nie Hayden Cheng, Department of Management and Marketing, Hong interferencetheories,providingevidencethathighlevelsofanxi-
KongPolytechnicUniversity. ety interfere with the ability to process task-related information,
Thisresearchwassupported,inpart,byresearchgrantsfromtheSocial which results in reduced levels of performance (Eysenck, Derak-
Sciences and Humanities Research Council of Canada awarded to Julie shan, Santos, & Calvo, 2007). While cognitive interference is a
McCarthy(#410-2011-0313,#435-2015-0220)andJohnTrougakos(#410-
viablemechanismforexplainingtheanxiety–performancerelation
2008-0505, #435-2014-0693). We thank Sunjeev Prakash for his invalu-
on specific tasks, we suggest that it is not the only mechanism
ablesupportwiththisprojectandKathleenBoiesforherhelpfulcomments
underlyingtherelationbetweenworkplaceanxietyandjobperfor-
onanearlierversionofthisarticle.
mance.
CorrespondenceconcerningthisarticleshouldbeaddressedtoJulieM.
McCarthy,DepartmentofManagement,UniversityofToronto,1265Mil- Drawing from Conservation of Resources Theory (COR; Hob-
itaryTrail,Scarborough,Ontario,CanadaM1C1A4.E-mail:julie.mccarthy foll,1989),weproposethatakeyexplanatorymechanismforthe
@utoronto.ca relation between workplace anxiety and job performance is emo-
1
2 MCCARTHY,TROUGAKOS,ANDCHENG
tionalexhaustion.Thisisbecausetypicallevelsofjobperformance 1986;Spielberger,1972).Assuch,itrepresentsastressresponse
require the execution of multiple tasks over sustained periods of intheformofastrainsymptom(Beehr,1995;Jex,1998).
time.Assuch,highlevelsofjobperformancearedependentonthe Workplaceanxietyissubsumedunderthebroaderconstructof
protection and facilitation of cognitive and personal resources. performance anxiety, which reflects feelings of apprehension
When employees experience high levels of workplace anxiety, about the execution of specific tasks, including job interviews
these resources are depleted, resulting in emotional exhaustion, (McCarthy & Goffin, 2004), selection tests (Proost et al., 2008),
andultimatelyreducedlevelsofjobperformance. school-basedexaminations(Cohen,Ben-Zur,&Rosenfeld,2008),
Given the consequences of workplace anxiety, an examination artistic performances (van Kemenade, van Son, & van Heesch,
of factors that may serve to buffer its negative effects is also 1995),andsportscompetitions(Kleine,1990).However,whilethe
crucial.Inthecurrentstudy,weconsidersocialexchangerelation- generalconstructofperformanceanxietyiswellgroundedinpast
ships as a critical moderator of the relation between workplace research,theconsequencesofdealingwithworkplaceanxietyhave
anxiety and job performance. The nature of people’s workplace notbeenfullyexplored.
exchange relationships has been shown to provide resources that Workplace anxiety is conceptually and empirically related to
y. canaidinimprovingemployees’workoutcomes(Cohen&Wills, other types of anxiety and related affective constructs, but is not
shers.broadl 1w9o8r5k;plNacge&excShoarnegneserne,la2ti0o0n8s)h.ipBsywbitrhintghiantgotnogweothrkeprlarecseeaanrcxhietoyn, raendxuientdyanbtecwauitsheitnhecsoentcroanststtroucatstr.anItsiiesntdsisittuinactitonfr-ospmecsiftaictet-rbaaits,eidt
publinated weprovideemployeesandorganizationswithinsightintopotential reflectsgeneralfeelingsofwork-relatedanxietythatmanifestover
dmi buffers of the negative consequences of anxiety on job perfor- time (Spielberger, 1972). It differs from general trait anxiety
salliedisse maOncuer.study makes a contribution to the existing research in a btheactauissewowrokrpklapclaecsepeacnixfiiecty(Zreeidflneecrts&anMeavttahleuwatsi,ve2-0b0a5s)e.dOaunrxdieetfy-
ite
ofob numberofways.Tobegin,wedrawfrompasttheoryandresearch inition also distinguishes workplace anxiety from neuroticism,
oneott todevelopaconceptualframeworkthatdelineatesthemechanisms because anxiety is only one of the six components underlying
orsn by which workplace anxiety has an effect on performance. In neuroticism (McCrae & Costa, 1987), and neuroticism is not
i
ationand dthoeinegxtseon,twtoeiwnhteigchrateeminostiiognhatsl ferxohmaussetivoenraalnlditecroatgunrietsivaenidntaesrsfeesrs- workplacespecific.
cier
sous ence mediate the relation between workplace anxiety and work- Emotional Exhaustion and the Workplace Anxiety–Job
s
Aal place effectiveness. Positioning emotional exhaustion as a mech-
ologicalindividu athnaistmhaivsefubnedeanmuesnetdalltyodeixffpelraeinntafnroxmietyp–apsterifnotremrfaenrecnecerelmatoiodnelss. ThefewempiricaPlienrvfeosrtimgaatinocnestRhaetlahtaivoenexaminedanxietyin
he Providingconceptualclarificationofthisissueisvaluable,because the workplace have generally found a negative association be-
ch
yt
Psof it has theoretical relevance for models of workplace anxiety, as tween anxiety and job performance (Ford et al., 2011). These
ne well as practical relevance for treating the implications of work- studies are informative, but have yet to consider the mechanism
as
cu
erial placeanxiety. underlyingobservedrelations.Thereare,however,severalstudies
Amson Wefurthercontributetotheliteraturebyexaminingmoderators examiningthemechanismbetweenanxietyandspecifictask-based
eer oftheworkplaceanxiety–jobperformanceassociation,specifically performance, such as testing situations and sports competitions.
hp
ythe examining the exchange relationship employees have with their Meta-analyticreviewsofthesestudiessuggestthat,consistentwith
bt
yrightedolelyfor stthhueepeeforxvrcmihsoaonrsfg,ecionrwetloharteikofenorsremhxipcohefamlnepgaledoe(yCre–eWmsXehma).vbePearwseittxwhchothareknigrhecaos(LlfloeMcaugXsu)ee,dsa,oninnd tbmheeatwnrceeeelanitsiponenrefgboaerttmiwvaeenec(nee-.agbn.a,xseiBedotyuarnahxniisdet&yjobaAnldpleesrnuf,obr1sm9eq9au2ne;cneKt,ltteahisenker,pe1lear9tf9ioo0rn-;
ps
coed organization or policy-based solutions (e.g., telecommuting, flex Seipp, 1991). Importantly, studies of anxiety and performance
isnd time) to help individuals minimize work anxiety and stress (Ko- focus on cognitive interference as the primary explanatory factor
entnte ssek,Baltes,&Matthews,2011).However,policy-basedsolutions for the observed negative relations. A number of theories are
mi
docucleis oWfteeninftaeiglrtaoteacChiOevRed(Hesoibrefdolrl,es1u9lt8s9()Ntioelsfeonc,uTsaorins,p&erCsoonxa,l-2b0a1s0e)d. sfoulblsouwminegd:upnrdoecrestshiengruebfrfiiccioefnccoyg(nEityisveenicnkte&rfeCreanlvcoe,,i1n9c9lu2d),inagttethne-
hisarti resources that employees can leverage to reduce the potentially tional control (Eysenck et al., 2007), and interference (Wine,
Ts
hi harmfuleffectsofworkplaceanxiety.Indoingso,werespondto 1980). Common to each of these theories is the proposition that
T
calls for theory-based predictions about the differential effects of anxiety interferes with people’s ability to process immediate
supervisorandcoworkersupport(e.g.,Ng&Sorensen,2008).In events,resultinginlowerperformance.Thereisconsiderablesup-
the following section, we detail the theoretical rationale for our portfortheroleofcognitiveinterferenceasitrelatestoanumber
modelandoutlineourhypotheses. of specific tasks, such as performance anxiety and selection test
scores(McCarthy,Hrabluik,&Jelley,2009),examgrades(Sara-
son, 1984), sports outcomes (Wilson & Smith, 2007), and music
Workplace Anxiety
success(Wan&Huon,2005).
Workplaceanxietyisconceptualizedasfeelingsofnervousness Itisnotsurprisingthatinspecificsituations,cognitiveinterfer-
and apprehension about the accomplishment of job tasks (cf., encemediatestherelationsbetweenanxietyandthecorresponding
Eysenck et al., 2007; Muschalla & Linden, 2012; Zeidner & performance outcome. These situations are characterized by high
Matthews,2005).Itisadomain-specificconstructthatisaffected levelsofstate-basedanxietyandhighlevelsofmaximumperfor-
by both individual differences and workplace characteristics (cf., mance because they represent evaluative contexts that require
Ganster&Schaubroeck,1991;Motowidlo,Packard,&Manning, intense effort over relatively short periods of time (Campbell,
AREANXIOUSWORKERSLESSPRODUCTIVE? 3
1990).Further,consistentwiththedefinitionofmaximumperfor- & Grandey, 2007), and general negative affectivity (Thoresen,
mance derived by Sackett, Zedeck, and Fogli (1988), in these Kaplan, Barsky, Warren, & de Chermont, 2003). In turn, high
situationsindividualsareawaretheyarebeingevaluated,theyare levels of emotional exhaustion prompt employees to distance
trying to maximize their performance, and their performance is themselves,orwithdrawfromtheirwork,resultinginlowerlevels
measuredafterarelativelyshortdurationoftime.Giventhathigh of performance (Cropanzano, Rupp, & Byrne, 2003). This is
levelsofconcentrationandfocusarerequired,itisnotunexpected primarily because of the considerable toll that emotional exhaus-
that cognitive interference has been found to draw individuals’ tion has on individuals’ motivation to perform effectively (Hal-
attentionawayfromthetaskathand,resultinginlowerlevelsof besleben & Bowler, 2007). Similarly, emotional labor research
taskperformance(Eysencketal.,2007). linksemotionalexhaustiontomoreworkplaceerrors(Goldberg&
In contrast, typical performance (i.e., overall job performance) Grandey,2007)andreducedservicequality(Grandey,2003).
ischaracterizedbyemployeebehaviorundermoreroutinecondi- The current study moves beyond past models of cognitive
tions and is elicited in contexts requiring sustained effort over interference as the underlying process linking workplace anxiety
extended periods of time, such as the day-to-day performance of and performance. Specifically, workplace anxiety’s impact on
y. jobduties(Campbell,1990;Sackettetal.,1988).Infact,inmost employees’ job performance is likely to be influenced by emo-
shers.broadl oenccguapgaetiionnsli,kaelycobnesicdoemraebsleropuotirntieonovoefrthtiemdea.iOlyvweroarlkl jeombpploeryfeoers- tfieornenalcee.xhWauesptiroonpoasbeovtheeafnodllboewyionngdhtyhpeoethffeescitss:ofcognitiveinter-
dpubliminated mvaarnycinegtdheugsreeenscoomfcpoagssneistivneurmeseorouurcsesty,paesswoeflltaasskashtohsattorfeoqtuhierer Hypothesis1:Emotionalexhaustionwillmediatetherelation
salliedisse pWeresiossn,aBlraersroosu,rc&esM,inacclDuedrimngide,n2e0rg0y5,;eTffroorutg,aaknodsp&ersHisitdeengc,e2(0B0e9a)l., bcoetuwnetienngwfoorrckopglancietivaenxinietetyrfearnendcejo.b performance when ac-
ite
ofob Overall job performance is an aggregate of the expected organi-
oneott zational value of employees’ behavior over time (Motowidlo, The Role of Social Exchange Resources
orsn 2003), and the day-to-day performance of most jobs requires the as Moderators
i
ationand e(Bxeecaulteiotnal.o,f20m0u5l)t.ipTlehuws,owrkhitlaeskcsogonviteirveaninteexrtfeenrednecdetmimaeypbeeritohde CORtheoryalsoprovidesinsightsintofactorsthatmightbuffer
cier
sous primary mechanism driving performance in highly specific situa- thenegativerelationbetweenworkplaceanxietyandperformance.
s
Aal tions that trigger momentary anxiety, experiencing continually AccordingtoCORtheory,inordertomitigatetheeffectsassoci-
ologicalindividu etilmeveaitsedlewssolrikkpellaycteoabnexeiextpylaiinnerdelbatyiocnogtonitpiverefoinrmteirnfegreancjoebaloovneer. afrtoedmwthitehernevsoiruorncmeelonst”s,(pHeoobpfloel“l,c1a9ll8o9n,pre.s5o1u7r)c.eOsnaveasiulacbhleretsootuhrecme,
he InlinewithAjzenandFishbein’s(1977)principleofcompatibil- social support, may be especially critical (Hobfoll, 1989, 2001).
ch
yt
Psof ity, which highlights the importance of matching predictors to Socialsupportcanacttooffsetresourcedrainanditscorrespond-
ne relevantoutcomes,wesuggestthattheunderlyingprocessdriving ing negative consequences in many ways, such as broadening
as
cu
erial the relation between workplace anxiety and job performance is one’spoolofavailableresources(Hobfoll,1989),promotingpos-
Amson emotionalexhaustion. itivecopingskills(Dunahoo,Hobfoll,Monnier,Hulsizer,&John-
eer COR theory provides an excellent framework for detailing the son,1998),andreducingworktaskdemands(Ray&Miller,1994).
hp
ythe processleadingto,andtheconsequencesof,emotionalexhaustion. Combined, these social support functions serve to replenish the
bt
yrightedolelyfor CrtheOseosReurrctehesesoousruryccehhsoailssdsitmimthpeaotratinannddt,ievnbideercugaayulss(eHstrroeibsvofeoulrtlco,e1p9dr8roa9tie)n.ctcCaaonnndsleearbvduiinltdgo rpeex.sh4oa7uu0rs)ct.eiIonpno,tuoprlr,no,rmepsooustliitntiignvgebesinpttie“rraplopssemirtfiaovyremogafafnsicneetst(phHieraaeklfsaf”nece(tHns,ooPbfefeormhllo,ont1ii9eom8n9ail,,
ps
coed increased levels of emotional exhaustion (Maslach, Schaufeli, & &Toppinen-Tanner,2008).
isnd Leiter,2001).Importantly,CORalsoalignswiththeextended,or Socialsupportisparticularlyimportantinworkcontexts.Indi-
entnte aggregate,effectsthatworkplaceanxietymayhaveonemployees. viduals with whom employees interact can provide both material
mi
docucleis depSlpeeticoifnicoafllyre,sCoOurRcehsaosvaer“ltoinmge-te(Hrmo”bfeomll,ph1a9s8is9,).foIcnusfaincgt,oanktehye awnodrksoaccitoiveimtioestio(Nnagl&resSoourrecnessenth,a2t0c0a8n).aHidoewmepvleory,efeewsisntuthdeieirsdhaaivlye
hisarti premise of COR theory, as well as other research specifically examined whether social support can mitigate the potentially
Ts
hi focusingonemotionalexhaustion,isthatthecontinualdepletionof harmfuleffectsofanxietyonworkplaceperformance(cf.,Kamdar
T
personalresources,suchasenergyandfocus,willresultinburnout & Van Dyne, 2007). We extend COR theory to consider two
symptoms such as emotional exhaustion (Richardsen, Burke, & fundamentalwork-relatedsourcesofsocialexchangesupport—the
Leiter, 1992). Moreover, COR theory holds that “ongoing work support employees receive from their fellow employees (CWX;
demands consistently deplete resources at a faster rate than re- Sherony&Green,2002),andthesupportemployeesreceivefrom
sources can be replenished” (Freedy & Hobfoll, 1994, p. 312). theirsupervisors(LMX;Graen&Uhl-Bien,1995).
Anxious workers, by their very nature, are at a greater disadvan- These relations are critical sources of social support because
tage in this regard given their tendency to use up resources wor- coworkersandsupervisorsareideallyplacedtoprovideemployees
rying about task-related issues and to engage in self-critical socioemotionalandmaterialresources(Graen&Uhl-Bien,1995;
thoughtsregardingtheirabilities(Sarason,1984). Sherony & Green, 2002). Indeed, high-quality social exchange
Inlinewiththesepropositions,priororganizationalresearchhas relations are associated with high levels of trust, honesty and
foundemotionalexhaustiontobecausedbynegativeantecedents encouragement(e.g.,Graen&Scandura,1987;Sherony&Green,
such as stress (Garrosa, Moreno-Jiménez, Rodríguez-Muñoz, & 2002).Thus,wepredictthatemployeeswithhighCWXandLMX
Rodríguez-Carvajal,2011),effortfulemotionregulation(Goldberg drawuponthesocioemotionalandmaterialsupportreceivedfrom
4 MCCARTHY,TROUGAKOS,ANDCHENG
peers and supervisors to replenish their resources and aid their thenumberofemotionalandbehavioralresourcesthatcoworkers
performance when experiencing anxiety-induced emotional ex- drawfromtoprovidesocialsupport(Chiaburu&Harrison,2008).
haustion. Pastresearchsupportsthesepropositionsbydemonstratingthat
Further, drawing on Ajzen and Fishbein’s (1977) principle of coworkersplayacriticalroleinimpactingemployees’well-being
compatibility,wesuggestthatCWXandLMXwillplaydifferen- (Halbesleben,2006;Viswesvaran,Sanchez,&Fisher,1999).This
tial roles in buffering the anxiety–exhaustion–job performance is not surprising in light of the frequency with which coworkers
link.AjzenandFishbein(1977)detailtheimportanceofmatching sharetheiremotionalexperiences(Hadley,2014).Whenpeoplein
predictorstorelevantoutcomesandindoingsohighlightthevalue social exchange relationships share their internal feelings with
of considering the fundamental differences between CWX and trustedothers,theyaremorelikelytoreceivesocialsupporttohelp
LMX.Inparticular,CWXrelationsaremorelikelytoberelational dealwithproblems,andaremoreeffectivelyabletocopewiththe
(interpersonal) in nature (Karasek, Triantis, & Chaudhry, 1982; potential negative consequences of work strains (Snow, Swan,
Wu & Hu, 2009), whereas LMX relations are more likely to be Raghavan, Connell, & Klein, 2003). In turn, these factors
subject to power differentials and thus more transactional (eco- strengthenone’sabilitytocopewithstrain(Thoits,2011).Itthus
hers.broadly. nsuocmhi,cw)einsnuagtguerset(tNhagt&theSomroendseerant,in2g00e8f;feScatskuorfatih&eseJerxe,la2t0io1n2s)h.iApss sigtaantedsthteoermeaostoinontahlalyt eelxehvaautesdtincgowefofrekcetsr soufpwpoorrtkpshlaocueldanhxeileptym.it-
blised arelikelytobemostimpactfulatdifferentstagesintheprocess. Although research has yet to examine the potential buffering
punat The moderating role of CWX. The relationship between effect of CWX in the workplace anxiety–emotional exhaustion
itsalliededissemi caoonftwdenotrhpkeoeisrrssesisusspeqsriugvanilsiioftaircstiavn(eCtllyyhiladebissustripunoct&wferrHomwarirttihhsoarntes,bpe2et0cw0te8teo)n.sEuepmmepprvllooisyyoeereesss, rcsetorlwaaitoniorskn.esFrhoipsru,eptxphaoemrrtepliiens,tschooemwreoerlekavetiirodnseunpbcpeetowfroterehnathsgebemeneeonrdaleforsauttnriendsgstoorrosmleainnoid-f
ofob but relatively equal levels of power with respect to coworkers mize the negative effects of strain on affective (e.g., depressed
et
onot (Diefendorff & Greguras, 2009). In addition, unlike roles formal mood; Karasek et al., 1982) and health-related (e.g., depression;
n
oris leadersoccupy,therolescoworkersoccupyrarelyinvolveperfor- LaRocco, House, & French, 1980) outcomes. Coworker support
ciationerand meeasncaendmcoonwitoorriknegr.sAcsenatecronmseoqreueonnce,sorceilaaltiornecsipbreotwcietyenaenmdptlrouys-t hvaissioanlsaonbdeeemnofotiuonndaltoexbhuafufsetriothne(Wreluat&ionHbue,t2w0e0e9n),aabsuswiveellsausptehre-
ssous (Cole, Schaninger, & Harris, 2002), while relations between su- effects of an unsafe work climate on employees’ affective com-
alAdual pervisors and employees center more on economic transactions mitment(vanEmmerik,Euwema,&Bakker,2007).Finally,meta-
ologicindivi asunpdpoaurttehdorbityytr(aKdairtiaosneaklemtoadle.,ls1o9f82th).eTehmisploliynmeeonftrreealastoinoinnsghipis, abnuafflyerticodfatthaeinsdtirceastseosr-tshtaraticnowreolraktieornsotchiaanl ssuupppeorrvtisiosrya ssturopnpgoerrt
ychthe which are generally grounded on economic or transactional ex- (Viswesvaranetal.,1999).Thesefindingsareconsistentwiththe
Psof changes between employee and employer (LMX) and the exis- bufferinghypothesisadvancedbyCohenandWills(1985),which
canuse tence of social or relational exchanges between employee and suggeststhatsocialsupportintervenesintheexperienceofstress-
merinal coworker(CWX).ItisalsoconsistentwithSocialExchangeThe- ful experiences. We predict CWX will play a pivotal role in
ytheAheperso oexryist(Bilnaut,h1e96ex4c),hwanhgiechrfeolactuiosensshoinpthbeetwpeerecneivtwedoopbalirgtiaetsio.nTshtehsaet bhuauffsetrioinng. the link between workplace anxiety and emotional ex-
bt obligations can create two types of relations—an economic ex-
yrightedolelyfor cchhaannggeetthhaattiissbbaasseeddoonntrmusattearnidalirsetciciprroecsiotuyrc(Besl,auo,r1a96s4o)c.ial ex- Hprelylaapctoeiotahnnexwsiiseiltl2y:baeCnWdweeXmakwoetirilolwnmahloeendxeChraaWtuesXtthioeisnr,hesliuagtchiho.nthbaetttwhiesepnowsiotirvke-
scopdeds canWhhailveethseosceiaelxecxhcahnagnegreeslawtioitnhsscuapnecrvroissos,rssuacnhdtheactoenmompliocyeexes- The moderating role of LMX. Whereas coworker support
in
entnte changes with coworkers, authentic social exchanges of a more bufferstheexhaustionassociatedwithanxiety,wesuggestsuper-
documcleisi icnotwimoarkteersa.nTdhpiseroscocnuarlsfnoartutwreoamreainmroeraesolnikse.lFyirtsot,eamrispelobyeetewseaerne vexishoarussutipopnoratnidsmjoobrepelirkfeolrymtaoncime.paAcsttdheesclirnibkebde,twineteernacetmioontsiobnea-l
hisarti morelikelytouseimpressionmanagementtacticsandmaskneg- tween employees and supervisors are less likely to involve emo-
This ative emotions when interacting with individuals who have high tional sharing and displays of emotions by employees than are
T
relative power (Diefendorff, Morehart, & Gabriel, 2010). Thus, interactions between coworkers. This is because employees are
when interacting with supervisors, employees have a stronger morelikelytocontroltheiremotionalexpressionswheninteraction
tendency to suppress negative emotions, such as anxiety, while partnershavehighrelativepower,comparedwithwhenthepower
when interacting with coworkers, employees only partially sup- differential is equal or lower (Diefendorff et al., 2010). Further,
pressnegativeemotion(Diefendorff&Greguras,2009).Similarly, compared with coworker interactions, those between employees
employees have been found to share emotional experiences with andsupervisorsaremorelikelytobebasedoneconomicexchange,
coworkersinalmost80%ofemotionalworkplaceevents(Hadley, whereemployeesprovidehighlevelsofperformanceinexchange
2014). Thus, employees are more likely to be relaxed and share for tangible rewards (Hüffmeier & Hertel, 2011). According to
internalaffectivestateswheninteractingwithcoworkersthanwith social exchange theory, this occurs because supervisory support
supervisors(Ferris&Mitchell,1987).Second,coworkersinteract promotes employees’ sense of obligation and increases their mo-
morefrequentlywitheachotherthanwithsupervisors,providing tivation to perform (Sakurai & Jex, 2012). This elevated motiva-
a more accurate daily picture of each other’s well-being tion should allow employees with higher LMX to overcome the
(Hüffmeier & Hertel, 2011). These frequent interactions increase effects of exhaustion on performance. This is consistent with
AREANXIOUSWORKERSLESSPRODUCTIVE? 5
research on personal resources (e.g., Muraven & Slessareva, population (Hartley, Burchfiel, Fekedulegn, Andrew, & Violanti,
2003), which shows that people who have higher levels of moti- 2011).Thus,workplaceanxietyishighlyrelevantandisacommon
vation are able to maintain performance effectiveness even when occurrence among policing organizations (U.S. Department of
resourcesaredepleted(i.e.,statesofexhaustion).Underconditions Labor,2011).
ofhighersupervisorexchangerelationshipquality,theassociated
elevated motivation will help offset the detrimental effects of Participants and Procedure
emotional exhaustion, allowing employees to perform effectively
Participantsincludedpoliceofficers,theirsupervisors,andtheir
despite feeling internally depleted. Thus, “supervisor support
peers.1 They were recruited using a weekly newsletter that is
shouldleadtomoretechnicallypreciseworkoutcomesthanshould
e-mailedtoallofficers.Thenewsletterinformedofficersthatthe
coworkersupport”(Ng&Sorensen,2008,p.246).
purpose of the study was to obtain feedback on police officer
InadditiontothepotentialmotivationaleffectsofhigherLMX
well-being and job behaviors and directed them to the survey
quality, we suggest that higher LMX relationships provide em-
website.Thewebsitecontainedthestudyquestionnairesandasked
ployees with greater external resources that can help to offset
y. depleted internal personal resources. Supervisors are often gate- participantstoprovidecontactinformationfortheirsupervisorand
shers.broadl kSeceapnedrusrat,o1t9a8n7g)i.blTehereresofourrec,esstrwonitghidnyaandicorrgealantiizoantsiohnips(Gwriatehnsu&- ownoerkcpolawcoeraknexr.ieAtyt.ATtimTeim1e,2o(f3fimceorsnthcsomlaptelre)t,eodfftihceersmceoamsuprleeteodf
publinated pervisorscanhelpemployeesobtainmaterialresourcesnecessary mTiemaseu3re(s6owfeeemksotfioolnlaolweinxghatuhsetiTonimaend2scuorgvneiyti)v,esuinpteerrvfiesroernscec.omA-t
dmi toperformathigherlevels(Ng&Sorensen,2008;Scott&Bruce,
salliedisse 1ef9f9e4ct)i,vaelllyowbiyngletahneimngtomoovreerhceoamveilyinotenrneaxlterernsoaulrrceesoduerfciceist.sTmhourse, pplleetteeddtmheeamsueraessuorefoLfMCXWaXn.d job performance, while peers com-
oroneofitisnottobe sbueptCweorevneisnsiosetrexnhstuapuwpsitotihortnthiasenspdeaprpterircofupoloramsriltayinocnreesl.,evemanptiriincabluffifnedriinnggstshuegglienskt iPannaddAritviactinoidptuaaaavlnleotssrfare7gws7epe0rooefnind4pde1ridevyditdeooaumtrahsilensoasrfneeatcslgpoyeon.ndmMdsaeulodersvtt(oer8yet0hs(%preeo)fsn,iprdCsoetnanssutuescrarvhaseatiydea,n(cid:2)eaan(r78dn78e5%%d9))5a.,
ationand tpheartf,oormvearnaclle,h(Gigehresrtnqeural&ityDLaMy,X19is97re)l,aoterdgatnoizhaigtihoenrallecvietliszeonfshjoipb collegediplomaoruniversitydegree(80%).Onaverage,theyhad
cier been a member of the police force for 15 years and had been
sous behaviors(Ilies,Nahrgang,&Morgeson,2007)andorganizational
Asal attitudes(e.g.,jobsatisfaction;Ng&Sorensen,2008).Supervisor workingundertheircurrentrankfor6years.Avarietyofjobtitles
ologicalindividu sthuapnpocrtowisoarlkseorlisnukpepdortto(sKtroamngdearrre&latiVoannswDiythneta,sk20p0e7r)f,orfmuratnhceer wtthoeeerrexeashmpeolidnnedbewynthtsheatehmeppralrestaiacmnipdpaltinhntegs.gbeCinaoesmrapploapsroeisdpounalsatthiworeenaroetftmpooatldhieceebeoexfttfweicreneearnsl
he bolstering the notion that LMX should be especially relevant in
ycth validity of the study. No significant differences were found be-
Psof facilitating performance compared with CWX. Further, Sakurai tween respondents and the general population of officers with
ne andJex(2012)foundthatsupervisorsupportmoderatedtherela-
caus respecttoage,sex,ethnicity,oreducationlevel.
eAmeriersonal tseiumopnpplbooeryttewweealeesnvleeolmwsoo,tfniowengosartakinvedeffpeoemrrt.fooHtriomonwasnehvceaedrs,uwachhdeetnhtraismtuwpeenhrtevanilssiomurppseuarpcvtpisooornrt proAvitdoetaelvoaflu3a2ti6onosf.fiAceprpsr(o4x1im%a)tealgyreseidxtwoeheakvseatfhteerirosfufipceerrvsishoards
ythhep was high, no significant relations were obtained. We predict the completedthesecondsurvey,supervisorswereinvitedtoratetheir
bt employee’s performance and LMX using a Web based question-
yrightedolelyfor folloHwyinpgo:thesis3:LMXwillmoderatetherelationbetweenemo- nsuapireer.viAsortsotwaleroefm2a6l7e(s8u6p%er)vCisaourscapsiaarntisci(p8a8t%ed).wThhoehmadajeoarrintyedoaf
scopdeds trieolantailoenxwhailulsbtieonwaenadkejrobwpheernfoLrmMaXncies,hsiugchh.thatthisnegative csoorlliengethdeispalommpaleowrausn4iv8eyrseiatyrsdoelgdr,ehea(d8b1e%en).aTphoeliacveeorafgfiecesrufpoerrv2i4-
in
entnte years, and had been working under their current rank for four
mi years. Supervisors held a variety of job titles. Officers were also
docucleis Method asked to identify a peer with whom they had regular contact at
hisarti To test our predictions, we conducted a field study with the work.Likesupervisors,peersweresentaninvitationtorateCWX
Ts approximately 6 weeks after officers had completed the second
hi RoyalCanadianMountedPolice.Apolicecontextishighlyrele-
T survey. A total of 164 peers responded. The majority were male
vant to our scholarly research interests and pragmatic concerns
(77%) Caucasians (88%) who had earned a college diploma or
about workplace anxiety. Police officers have a broad set of
universitydegree(76%).
responsibilitiesthatemphasizetheprovisionof“...publicsafety
Ourfinalsampleincluded267officerswhohadcompletedTime
bymaintainingorder,respondingtoemergencies,protectingpeo-
1 and Time 2 surveys, and had their performance rated by their
pleandproperty,enforcingmotorvehicleandcriminallaws,and
supervisor. Peer ratings were obtained for 154 of these officers.
promotinggoodcommunityrelations”(O’Net,2014).Theirwork
Themajorityofofficersweremale(78%)Caucasians(88%)who
environmentischaracterizedashighinstress,becausetheymust
had earned a college diploma or university degree (80%). On
dealwithviolentoffenders,crimescenes,accidentvictims,victims
average,theyhadbeenamemberofthepoliceforcefor18years
ofabuse,death,andpublicsuspicion,amongotherdemands(An-
andhadbeenworkingundertheircurrentrankfor5.5years.
dersen&Papazoglou,2014).Further,over80%ofpoliceofficers
report that they interact with angry and/or unpleasant individuals
on a daily basis (O’Net, 2014), placing them at a significantly 1Ourdefinitionofpeerincludesanycoworkerthatanofficerhasintheir
higherriskofphysicalandmentalhealtheffectsthanthegeneral network.
6 MCCARTHY,TROUGAKOS,ANDCHENG
Measures played each competency (e.g., “This person independently pro-
ducesconsistent,highqualityresults”and“Thispersonidentifies,
Withtheexceptionofthedemographicvariables,allscaleswere creates, and implements effective solutions to problems”; (cid:3) (cid:2)
assessed using a Likert-type rating that ranged from 1 (strongly
.97). Performance ratings were collected for research rather than
disagree) to 5 (strongly agree). Psychometric properties of the
for administrative purposes because operational ratings are prone
selectedscaleshavebeenfoundtobestronginpastresearch,and
toleniencyerror(Curtis,Darvey,&Ravden,2005).
each scale exhibited adequate internal consistency reliabilities in
thecurrentstudy.
Results
Workplace anxiety. Workplace anxiety was assessed with
eight items modified from the performance anxiety scale devel- Table 1 presentsthedescriptivestatisticsandcorrelationsamong
oped by McCarthy and Goffin (2004). A sample item is: “I am study variables. We tested the hypotheses in our conceptual model
overwhelmedbythoughtsofdoingpoorlyatwork”((cid:3)(cid:2).94).See with structural equation modeling (SEM) using Amos 19 software
Appendixforfullitemlist. (Arbuckle, 2010). Maximum likelihood estimation procedures were
y. Emotional exhaustion. Emotional exhaustion was assessed usedandthreeindiceswereemployedtoassessthefitofourmodels:
s.adl with the five-item subscale of the Maslach Burnout Inventory theChi-Squareindex,theRootMeanSquareErrorofApproximation
herbro GeneralSurvey(Schaufeli,Leiter,Maslach,&Jackson,1996).A (RMSEA),andtheComparativeFitIndex(CFI).Thiscombinationof
s
blied sampleitemis:“Ifeelusedupattheendoftheworkday”((cid:3)(cid:2) fitindicesensuredtheinclusionofanindexthatconsidershowmuch
punat .95). variance is explained in light of how many degrees of freedom are
dmi
salliedisse witChosginxititievmesinatdearpfeterdenfcroe.mtCheogCnoitgivneitiivneteIrnfeterrefnecreenwceasQausessetsiosend- uhsoewdm(i.ue.c,hRvMarSiaEnAce),iassewxeplllaiansedanbiyndtehxetmhaotdiesla(id.ei.r,ecCtFfIu;nBcteionntleorf,
eofittobe nIaaimrea(Stawraosrokn,,IPtohtitnekr,a&boSuatransoonn-,w1o9r8k6a).cAtivsiatimesp”le((cid:3)ite(cid:2)m.i9s0:)“.When 1ca9s9e0o).fFthoerRthMeCSEFAI,,vvaalulueessaapppprrooaacchhiinngg10iinnddiiccaatteeaaggooooddffiitt..GInivtehne
onnot Supervisor-ratedLMX. SupervisorsassessedLMXbycom- thatwealsoconductedcomparisonsofnon-nestedmodels,wealso
oris pletingsevenitemsfromGraenandUhl-Bien’s(1995)LMXscale. includetheAkaikeInformationCriterion(AIC:Akaike,1987)andthe
ationand We asked supervisors to evaluate LMX to mitigate common Bayesian Information Criterion (BIC; Raftery, 1995). For each of
cier sourcebiasesandtomorefullycapturethesupportfocalemploy- these,smallervaluesindicateastrongerfit.Thepowerofouranalyses
os
Assalu eesreceivefromtheirinteractionpartners.Pastresearchsuggests wasfoundtobe1.0foratestofclosefit(RMSEA(cid:2).05;Preacher&
aldu that people are often not aware of the full extent of support they Coffman,2006).
ologicindivi mpriogvhitderesceviavleuafbrolemrtehseoiurricnetseraacntdionbepnaerftintsersb,eaynodndthawthtahtistshueppemor-t ingWietefmirsstwexiathminineedacthhesmcaelaesuirnetmoetnhtremeodpealrcbeylsratnhdaotmsleyrvgerdoupas-
he
ch
yt ployeeisawaretheyarereceiving(Howland&Simpson,2010).A indicators of each latent construct (Williams & Anderson, 1994).
Psof sample item is: “I use my power to help this employee solve Covariances among all latent factors were free to vary. Results
ne
caus problemsatwork”((cid:3)(cid:2).80). indicated that our 6-factor measurement model achieved a strong
merinal Peer-rated CWX. Peers assessed CWX by completing the fittothedata,(cid:4)2(120)(cid:2)180.2,p(cid:5).01;RMSEA(cid:2).04;CFI(cid:2)
eAerso samesevenitemsfromGraenandUhl-Bien’s(1995)LMXscale. .98;AIC(cid:2)318.16;BIC(cid:2)565.68.Moreover,itemparcelloadings
ythhep Consistent with LMX, we asked peers to evaluate CWX to miti- were significant (.72 to .98, p (cid:5) .01), suggesting that the items
bt gatecommonsourcebiasesandtomorefullycapturethesupport representedtheirintendedconstructs.Wecomparedourmodelto
pyrightedsolelyfor fiwtoeocmrakl”iesm:((cid:3)“pIl(cid:2)ouys.ee7e2ms)y.repceoiwveerfrtoomhetlhpeitrhiisnteemrapcltoioyneepsaortlnveersp.roAblseammspalet AageIftCihvee(cid:2)r-,fa(cid:4)4c20t(o91r.265m1);o(cid:2)BdeICl38th(cid:2)4a.7t63g3,r9op.u2(cid:5)p0e;d.a0L1fi;MvReX-MfaaScntEodArjmo(cid:2)bodp.0ee9lr;ftohCramFtIagnr(cid:2)oceu.p9teo3d-;
scoded Supervisor-rated job performance. Supervisors assessed CWXandLMXtogether,(cid:4)2(125)(cid:2)419.22,p(cid:5).01;RMSEA(cid:2)
in
entnte job performance by completing a 20-item scale of police officer .09; CFI (cid:2) .92; AIC (cid:2) 547.22; BIC (cid:2) 776.81; and a one-factor
documcleisi jSoubpecrovmisoprestenevciaelsuatthedatthweereextiednetnttiofiewdhitchhrouthgehiraemjopbloayneaelydsiiss-. 2m7o4d9e.l29th,aptg(cid:5)ro.u0p1e;dRaMllScEonAstr(cid:2)uc.t2s7o;nCtoFIa(cid:2)sin.g3l0e;fAacItCor,(cid:2)(cid:4)22(815375.)2(cid:2)9;
hisarti
Ts
hi
T
Table1
DescriptiveStatisticsandIntercorrelationsAmongScaleCompositesandLatentVariables
Variables M SD 1 2 3 4 5 6
1.Workplaceanxiety 2.34 1.16 — .62(cid:2)(cid:2)(cid:2) .24(cid:2)(cid:2) (cid:6).03 (cid:6).05 (cid:6).16(cid:2)
2.Emotionalexhaustion 2.79 1.22 .50(cid:2)(cid:2)(cid:2) — .29(cid:2)(cid:2) .01 .01 (cid:6).15(cid:2)
3.Cognitiveinterference 2.54 .94 .33(cid:2)(cid:2)(cid:2) .09 — (cid:6).07 (cid:6).05 (cid:6).18(cid:2)
4.Leader–memberexchange(LMX) 4.20 .53 (cid:6).02 (cid:6).03 (cid:6).02 — .05 .78(cid:2)(cid:2)(cid:2)
5.Coworkerexchange(CWX) 4.20 .46 (cid:6).39(cid:2)(cid:2)(cid:2) (cid:6).06 (cid:6).01 .03 — .23(cid:2)(cid:2)
6.Jobperformance 4.19 .64 (cid:6).16(cid:2) (cid:6).14(cid:2) (cid:6).03 .55(cid:2)(cid:2)(cid:2) .16(cid:2) —
Note. N (cid:2) 267 for all variables except CWX, where N (cid:2) 154. Correlations below the diagonal represent
zero-orderrelationsamongtheobservedscales.Correlationsabovethediagonalrepresentrelationsamongthe
latentscalesassessedinourstructuralequationanalyses.
(cid:2)p(cid:5).05. (cid:2)(cid:2)p(cid:5).01. (cid:2)(cid:2)(cid:2)p(cid:5).001.
AREANXIOUSWORKERSLESSPRODUCTIVE? 7
BIC(cid:2)3084.64.Ineachcase,oursix-factormodelwasastronger
fittothedata.
Next, we tested our hypothesized model by following proce-
duresrecommendedfortestingmoderationeffectswithcontinuous
variables in SEM (Moulder & Algina, 2002). Adopting Jaccard
andWan’s(1995)procedure,wemeancenteredmaineffectvari-
ablesbeforecomputingproducttermindicatorsinordertoreduce
collinearity between the main effect and interaction terms. Find-
ings indicated that the hypothesized model yielded an acceptable
fittothedata,(cid:4)2(244)(cid:2)358.21,p(cid:5).01;RMSEA(cid:2).04;CFI(cid:2)
.97;AIC(cid:2)539.25;BIC(cid:2)826.23,andispresentedinFigure1.2
Asillustrated,thefindingssupportedourhypotheses.
Hypothesis 1 predicted that emotional exhaustion will mediate
y. therelationbetweenworkplaceanxietyandjobperformancewhile
shers.broadl aacncxoieutnytidnegmofonrstrcaotgendiativseigniniftiecrafnetrepnocsei.tivAesreplaretidoinctwedit,hwemoroktpiolancael Figure 2. Coworker exchange (CWX) as a moderator of workplace
publinated exhaustion((cid:7)(cid:2).62,b(cid:2).58,SE(cid:2).05,p(cid:5).001),whichinturn anxietyandemotionalexhaustion.
dmi demonstrated a significant negative effect with job performance
salliedisse (d(cid:7)ard(cid:2)ize(cid:6)d.1c3o,efbfic(cid:2)ien(cid:6)t.f0o7r,thSeEin(cid:2)dir.e0c3t,epffe(cid:5)ct.o0f0w1)o.rFkuprlathceera,nthxeietsytaonn- between workplace anxiety and emotional exhaustion, providing
oneofitottobe penercfeowrmaasn(cid:6)ce.1t0hr(obu(cid:2)gh(cid:6)em.0o5t,ioSnEal(cid:2)ex.0h1a)u,satinodntahned9c5o%gnbiitaivs-ecoinrtreercfteerd- spu(cid:5)ppo.0rt1f)owrHasypfoouthnedsitso2m,wohdielreaLteMthXe(r(cid:7)el(cid:2)ati.o1n0,bbet(cid:2)we.1e2n,eSmEo(cid:2)tio.0n6al,
orsn confidence interval ranged from (cid:6).15 to (cid:6).04 (5,000 bootstrap exhaustionandjobperformance,providingsupportforHypothesis
ationandi raensxaimetpyleosn).jIonbcpoenrtfroarsmt,athnecestwanadsa.r0d0i,zendondsirigecntifeifcfaenctt(onfsw;bor(cid:2)kpl.a0c0e, 3an.aFliygsuerser2eviellaulsetdratthesattheemCotWioXnalinetexrhaacutisotinoneffweacst.mSiomreplestrsolonpgelys
sociuser SE (cid:2) .00). These findings indicate that the relationship between relatedtoworkplaceanxietyatlowlevelsofCWX(b(cid:2).78,SE(cid:2)
Asal workplaceanxietyandjobperformanceisfullymediatedbyemo- .11),t(266)(cid:2)6.84,p(cid:5).01,thanathighlevelsofCWX(b(cid:2).39,
hologicaleindividu teaixnohdnaaublseetyixoohnnadeuxscptioloagninnieatidnvdaedcdionigttienorinfteiavrleevnacirneitaen(r(cid:8)cfeeRrie2nnc(cid:2)joeb..pF0e4ur)rf.tohrAemrc,acneomcrdeoiantibgoonvtaoel ieSnxEthea(cid:2)raucs.tt1iioo2nn),ewtf(f2aes6c6tm.)oS(cid:2)riem3ps.t3lre3o,nslpgolpy(cid:5)esr.e0ala1nt.aelFdyisgteousrjeroeb3vepialelluerfsdotrrtamhtaeatsnectmheeoatLtioMlonXwal
nPsyceofth cAugrureinnits,beSnincghhm,aFrikesl,d,th&isPeifefreccet,i2s0m15e)d.iOumurifninmdianggnsitauldsoei(nBdoicscaote, tlhevanelsatohfigLhMleXve(lbso(cid:2)fL(cid:6)M.2X3,(SbE(cid:2)(cid:2)(cid:6)..0043),,StE(2(cid:2)66).0(cid:2)4),(cid:6)t(52.6164),(cid:2)p (cid:6)(cid:5)..6071,,
as
ericalu thatanxietyexplains38%ofthevarianceinemotionalexhaustion, ns.Intermsofthemagnitudeofeffects,ourfindingsindicatethat
Amson 6%ofthevarianceincognitiveinterference,andthatthecombined anxiety and CWX explain 41% of the variance in emotional
eer effectsofanxiety,emotionalexhaustionandcognitiveinterference exhaustion,andthatemotionalexhaustionandLMXexplain59%
hp
ythe explain 63% of the variance in job performance. Overall, our of the variance in job performance. Overall, our results support
yrightedbolelyfort rCeWsOuXlutsrassenucdpopnLodMrtsXeHt.yopAfohstyhipellosuitsshter1sa.etesdfoicnusFeidguorneth1e, mCWodXera(t(cid:7)ing(cid:2)ro(cid:6)le.s1o7f, HelysO.pouFtrihrfseitsneawsle2setatenosdtfed3an.aaly“sreesveinrsveo-lcvaeudsatetisotinn”gmtwoodeall,teirnnawtihviechmojodb-
copeds b(cid:2)(cid:6).20,SE(cid:2).08,p(cid:5).01)wasfoundtomoderatetherelation performancewassetasanantecedenttoemotionalexhaustionand
sd cognitive interference, which in turn were set as antecedents to
in
entnte workplace anxiety. The reverse causation model exhibited a
documcleisi 5w7e2a.k7e4r,fpit(cid:5)to.t0h1e;dRaMtaStEhaAn(cid:2)our.0h7y;pCotFhIes(cid:2)ize.d93m;oAdIeCl, (cid:4)(cid:2)2(274342).7(cid:2)4;
hisarti BIC (cid:2) 1019.72. Second, we tested a “full” model in which both
This LMX and CWX were positioned as moderators of the relation
T between anxiety and emotional exhaustion, and both LMX and
CWX were positioned as moderators of the relation between
emotional exhaustion and job performance. This model retained
cognitiveinterferenceasamediatingmechanism.The“full”model
exhibited a weaker fit to the data than our hypothesized model,
(cid:4)2(392)(cid:2)1037.61,p(cid:5).01;RMSEA(cid:2).08;CFI(cid:2).88;AIC(cid:2)
1179.54; BIC (cid:2) 1,549.03. Nevertheless, the overall pattern of
regression weights replicates that found in our predicted model.
Specifically, CWX (but not LMX) significantly moderates the
Figure1. Structuralequationmodeling(SEM)finalmodelresults.Note.
Standardizedcoefficientsarepresentedhere—pleaseseeresultssectionfor
unstandardized coefficients and standard errors. Fit statistics: (cid:4)2(244) (cid:2) 2Theoverallpatternoffindingsremainsthesamewhenorganizational
358.21,p(cid:5).01;RootMeanSquareErrorofApproximation(RMSEA)(cid:2) tenure,neuroticism(measuredviaJohn&Srivastava,1999),andjobstress
.04;ComparativeFitIndex(CFI)(cid:2).97. (measuredviaParker&DeCotiis,1983),arecontrolled.
8 MCCARTHY,TROUGAKOS,ANDCHENG
betweenemployeewell-beingandjobperformance(e.g.,positive
andnegativeaffect;Kaplan,Bradley,Luchman,&Haynes,2009),
byilluminatingaviablemechanism—emotionalexhaustion—that
mayunderlietherelations.Ultimately,ourarticleisoneofthefirst
to simultaneously examine the process of resource loss with ex-
haustion and cognitive interference as it relates to worker effec-
tiveness,providinganotherimportantcontributiontoCORtheory.
Athirdcontributionofourstudyisthatweconsidermoderators
oftherelationsbetweenworkplaceanxiety,emotionalexhaustion
andjobperformance.DrawingfromCORtheory,ourstudyhigh-
lightstheroleofsocialsupportinbufferingthenegativeeffectsof
workplace anxiety and emotional exhaustion. Our study finds
supportfortheroleofinterpersonalresources(i.e.,socialsupport)
y.
s.adl as an antidote for the harmful effects of anxiety on job perfor-
herbro mance. Thus, our paper integrates the COR and workplace ex-
s
blied Figure3. Leader–memberexchange(LMX)asamoderatorofemotional changeliteratureswhich,untilnow,havelargelyexistedinparal-
punat exhaustionandjobperformance. lel.Wealsoaddtothesocialexchangeliteraturebydemonstrating
dmi
salliedisse thhieghilmigphotrdtainffceereonfcecsoninsidcoerwinogrkreerlaatniodnssuhpipersvoisuorrcer,elaastioonusrhriepssu.lts
eofittobe rwehlaitlieonLMbeXtw(ebeunt nwootrCkpWlaXce) saingxniieftiycanatnldy memodoetiroanteasl tehxehareulsattiioonn,
onnot between emotional exhaustion and job performance. Combined, Practical Implications
ationorandis omuorderelsoufltwsoprrkopvliadceeastnrxoinegtysaunpdpojortbfpoerrfoourmratnhceeo.retically driven froTmhihsigshtuldeyvehlsasofnwotoarbkleplaimcepalincxaiteiotyn,sfoforrininddivivididuualaslswwhohowosurkffienr
cier
sous demandingenvironments(e.g.,roleconflict,highworkload),and
Asal Discussion for individuals who work in high-pressure industries (e.g., police
ologicalindividu exhOauurstrieosnearischaandviamnpcoerstathnetfmieeldchbayndisemmounnsdtreartliynigngthathteemroeltaiotinoanl oanffxicieetrys,cofimnaenscaitaalhtriagdhecros)s.t,Faisrsatn,xiitoduesminodnivstirdauteaslstahraetmwoorerklpiklaeclye
chhe between workplace anxiety and job performance. We also show toexperienceemotionalexhaustion,andinturn,havelowerlevels
yt
Psof thatsocialexchangecanmitigatetheharmfuleffectsofworkplace ofjobperformance.Asaresult,itiscrucialfortheseemployeesto
anse anxiety, such that CWX moderates the relation between anxiety haveaccesstoresourcesthatwillallowthemtorecoverfromthe
cu
erial and emotional exhaustion, and LMX moderates the relation be- resourcedrainthatworkplaceanxietycaninduce.
mn
Aso tween emotional exhaustion and job performance. Thus, while Ourfindingsalsohighlighttheimportantroleofsocialexchange
heper workplace anxiety may come at a high cost, social support can resources,asemployeeswhowereabletodrawonsupervisorsand
bytthe mitigateitsnegativeeffects. coworkers for support were less likely to experience the harmful
pyrightedsolelyfor Contributions to Research ecsufofpneetcirtnvsuisoooufrsslwyisowrpkoaprrlkaaimcnegouatnontx.iimRetepysreoaavnrecdhreeemlvatoiidtoienonsncaewlsietuhxghgcaeouswststoiorthknae.trsTohpaunesnd,
coed There are a number of theoretical contributions of this study. communication is the key to developing these strong relations
isnd First,ourstudyextendstheuseofCORtheorybyproposingthat
entnte individualswhoexperiencehighlevelsofworkplaceanxietymay (Miles,Patrick,&King,1996).Fromtheperspectiveoforganiza-
mi tions,ourfindingshighlighttheimportanceoftrainingallemploy-
docucleis preemrcaeiinveintahehigwhosrtkaptelaocfealaesrtatotahvreoaidterneisnogurecnevliorossn.mGeinvtenatnhdattthhuiss eestodeveloppositiveworkrelationsandtoengageinsupportive
hisarti stateofalertrequireshighlevelsofvigilance,itactuallyservesto behaviors.Wealsoacknowledgethepotentialforsocialsupportto
Ts buffertheeffectsofworkplaceanxietyoncognitiveinterference,
hi deplete resources and is related to emotional exhaustion. We
T andweencouragefutureresearchtoexplorethispossibility.
contribute to COR theory by exploring the role of workplace
Additional strategies that may reduce the harmful effects of
anxietyonresourcedepletion,aswellastherolethatresourceloss
workplaceanxietyincludegivingemployeesbreaks(Trougakos&
playsinemployeejobeffectiveness.
Hideg, 2009), providing them with autonomy over how to use
Second,ourresearchmovesbeyondcognitiveinterferenceand
considersemotionalexhaustionasthekeyprocessunderlyingthe breaktimes(Trougakos,Hideg,Cheng,&Beal,2014),andreduc-
relation between anxiety and performance. Emotional exhaustion ing emotional demands on employees (Goldberg & Grandey,
was found to mediate above and beyond cognitive interference. 2007).Individual-basedstrategies,suchashighqualitysleep(e.g.,
Thesefindingsrepresentanimportantcontributiontothefield,as Sonnentag,Binnewies,&Mojza,2008)andlearningnewhobbies
theysuggestthatmeaningfulmodelsofworkplaceanxietyshould (e.g., Fritz & Sonnentag, 2006), have also been found to be
considerboththeeffectsofworkplaceanxietyonresourcedeple- effectiveresourcerecoverytools.Importantly,futureworkshould
tionviaemotionalexhaustion,aswellastheeffectsofworkplace consider the mechanisms underlying each technique, such as the
anxietyoncognitiveinterferenceviaoff-taskprocessing.Further, extent to which it reduces emotional exhaustion, cognitive inter-
thesefindingsextendpastresearchthathasexaminedtherelations ference,orothermediatingfactors.
AREANXIOUSWORKERSLESSPRODUCTIVE? 9
Strengths, Limitations, and Directions for We would also like to note that the slope for the effect of
Future Research emotional exhaustion on job performance for those with high
levels of LMX (Figure 3) was not statistically significant. While
Acorestrengthofourworkisthatwedrawfrompasttheoryand this might have been a result of supervisors rating subordinates
researchtodevelopaconceptualframeworkdelineatingmediating with better quality LMX consistently higher on job performance,
and moderating mechanisms of the relation between workplace regardless of exhaustion levels, we suggest that it is more likely
anxiety and job performance. We provide a robust test of our that this result was a product of the LMX relationship. To be
predictionsbyadoptingathree-wavemethodologicaldesign,cap- specific, the ability to maintain a consistent level of excellent
turingdatafrommultiplesources,andutilizingadvancedstatistical performance is a critical element of high quality LMX (Graen &
procedurestotestourcompletemodel.Nevertheless,itwouldbe Scandura,1987).HighLMXindividualsarerelieduponbysuper-
advantageous to assess our model over longer periods of time. visorstoperformthemorecriticaltasksoftheworkgroup(Dien-
Futureresearchthatincludesmorethanthreewaveswouldhelpto esch & Liden, 1986). These employees are likely to do what is
fullyelucidatetheserelations. necessarytomeetperformancerequirementsataconsistentlyhigh
y. Anadditionalstrengthisthatourresearchwasbasedonalarge levelregardlessoftheirlevelsofexhaustion,inordertomaintain
shers.broadl sfoamrmpalnecoefbeymspulpoeyreveissoarsn,dasinwcelulldaesdrraatitninggssooffCLWMXXbyancdowjoobrkpeerrs-. trheleastiuocncsehsisp.ofAtthtehweosarkmgertoiumpe,,assuwpeerlvliassortshemqauyalbietymoofrteheliirkeLlMytXo
dpubliminated Terhailsiztaybpielitoyfodfatoauirsfcinhdailnlegnsg.inItgatlosoobhtealipnsatnodbirnidcgreeassecsietnhceegaennd- rpercoovgidneiztehewmhewnithhigghreLatMerXreesmouprlcoeyseteoshaerlepetxhheamusmteadi,ntaanidnpinertfuorrn-
salliedisse pwroarcktipcleacbeyaandxviaentyc,inwghoiluerptrhoevoirdeitnicgalvaulnudaebrlsetainndfoinrmgaotfiothneornolheoowf mlesasncoef.eTmhpeloenydeereesxuhlatuisstaiocnolnesviestles.ntWleeveenlcoofupreargfeorfmutaunrecererseegaarrcdh-
ite
ofob tomanageitspotentiallyharmfuleffects.Acorrespondinglimita- tomorefullyexplorethenatureandmechanismsunderlyingthese
oneott tionisthatoursamplewascomprisedofpoliceofficerswhowere relations.4
orsn predominantly Caucasian males. While these demographics are
ationandi raedpvraensteangteaotiuvsefoorffuNtourrtehreAsemarecrhictaonepxoamlicieneowffhiceethrse,rtihtewreosuulldtsboef Conclusion
ociser thepresentstudyholdforminoritygroupmembers,aswellasfor The current study tested and found support for emotional ex-
ssu haustionasanewtheoreticallinkbetweenworkplaceanxietyand
Aal otherjobincumbentpopulations.Webelievethatourmodelwould
ologicalindividu rpeacteioivnesssuucphpoarstmwiiltihtaorythperergsroonunpesl,,fpiarertfiicguhltaerrlsy,hanigdh-asitralikneespoicloctus-. jetoasnbiztepbdeercmfaouermcsehaainntcisesum,geg,vecesontgswnthihtaiivtleewcionortnketprrfloaelcrleiennagcnefx.oiTrethayispshrfeoinvudilodinubgsleyismhioymdpepolotehrd--
he Thebasisforthisassertionisthatthetheoreticallinksbetweenthe
Psycofth variablesinourmodelarethesame,regardlessofthedemographic iinncoarpmooraretinhgoelimstoictiomnaanlneexrhathuastniohnaasnbdeceongndiotinveeiinntethrfeerpeansctebays
ne backgroundand/ortypeofjobthattheemployeeholds.However,
ericaalus weacknowledgethatthestrengthoftheseeffectsmaydifferacross esoxcpilaalneaxtocrhyanmgeecahsaannismanst.idOotuertosttuhdeyefaflescotsporfovwidoerkdpliancseigahntxiinettoy
Amson jobs. For example, police officers often work in shifts which on job performance. Overall, these findings provide further in-
eer providesperiodsof“down-time.”Thistypeofbreakmayserveto
hp sights into the consequences of workplace anxiety, and demon-
yrightedbytolelyforthe raoaenbusxlrteri.emictyto—dtehjleoibnefplfoeewrcft-ossrtmroefasnsccooegccrneuiltpaivatietoionin.n3steFwrfuoeturuelrdnecbeweoiprnkarttthihceautlwaeroxlyrakmpvilanalceues- smtraanteagtehda.twhileworkplaceanxietycomesatahighcost,itcanbe
ps
coed Wealsoelectedtohavecoworkersandsupervisorsprovidethe 3Wethankananonymousreviewerforthissuggestion.
entisntend rsautpinpgorotftheaxtcehmanpgloeyreeelastrieocnesihviepdqfuraolmityt.hoWseeiwnetrheeiirntweroerskteednvinirothne- 4Wethankananonymousreviewerforthissuggestion.
mi
docucleis mwietnhtoaunrdgwoaelsw.aHnatevdintgoceonwsuorreketrhsatanodursumpeearsvuisroemrsepnrtowviadsearlaitginnegds References
Thissarti of CWX and LMX is also consistent with research on invisible Ajzaenna,lyIs.,is&aFnidshrbeeviine,wMo.f(1e9m7p7i)r.icAatltirteusdeea-rbcehh.aPvisoyrchreollaotgioicnasl:AButhlleeotirne,ti8c4al,
Thi support, which demonstrates that targets of support may not be 888–918.http://dx.doi.org/10.1037/0033-2909.84.5.888
completelyawareoftheexactnatureofsupporttheyreceive,and
Akaike,H.(1987).FactoranalysisandAIC.Psychometrika,52,317–332.
that this invisible support can influence their emotional and per- http://dx.doi.org/10.1007/BF02294359
formanceeffectiveness(e.g.,Bolger&Amarel,2007).Therefore, AmericanPsychologicalAssociation.(2012).Workplacesurvey:Psycho-
we obtained other ratings of exchange quality to capture the logicallyhealthyworkplaceprogram.Retrievedfromhttps://www.apa
supportthatemployeesmaynotrecognize.Italsoprovideduswith .org/news/press/releases/phwa/workplace-survey.pdf
a second and third source of data, thus reducing concerns of Andersen, J. P., & Papazoglou, K. (2014). Friends under fire: Cross-
common source variance and strengthening our methodology. At culturalrelationshipsandtraumaexposureamongpoliceofficers.Trau-
matology,20,182–190.http://dx.doi.org/10.1037/h0099403
the same time, we acknowledge that our measures captured the
Arbuckle,J.(2010).Amos19.Chicago,IL:AmosDevelopmentCorpora-
level of support perceived by the giver of the support, and may
tion.
thusincludebiases.Itwouldbeadvantageousforfutureresearch
Beal, D. J., Weiss, H. M., Barros, E., & MacDermid, S. M. (2005). An
to adopt a multiple-rater approach and compare the ratings of episodicprocessmodelofaffectiveinfluencesonperformance.Journal
socialsupportmadebythetargetemployeetotheratingsmadeby ofAppliedPsychology,90,1054–1068.http://dx.doi.org/10.1037/0021-
supervisorsandpeers. 9010.90.6.1054