JournalofAppliedPsychology Copyright2006bytheAmericanPsychologicalAssociation 2006,Vol.91,No.4,762–776 0021-9010/06/$12.00 DOI:10.1037/0021-9010.91.4.762 Loving Yourself Abundantly: Relationship of the Narcissistic Personality to Self- and Other Perceptions of Workplace Deviance, Leadership, and Task and Contextual Performance Timothy A. Judge, Jeffery A. LePine, and Bruce L. Rich UniversityofFlorida Theauthorsreportresultsfrom2studiesassessingtheextenttowhichnarcissismisrelatedtoself-and otherratingsofleadership,workplacedeviance,andtaskandcontextualperformance.Study1results revealedthatnarcissismwasrelatedtoenhancedself-ratingsofleadership,evenwhencontrollingforthe Big Five traits. Study 2 results also revealed that narcissism was related to enhanced leadership self-perceptions;indeed,whereasnarcissismwassignificantlypositivelycorrelatedwithself-ratingsof leadership,itwassignificantlynegativelyrelatedtootherratingsofleadership.Study2alsorevealedthat narcissismwasrelatedtomorefavorableself-ratingsofworkplacedevianceandcontextualperformance comparedtoother(supervisor)ratings.Finally,ashypothesized,narcissismwasmorestronglynegatively relatedtocontextualperformancethantotaskperformance. Allthatislovelyinhimselfheloves, arearrogantandhaughty.Withintherealmofnormalpsychol- andinhiswitlesswayhewantshimself: ogy,narcissismcameunderseriousstudyinthelate1970sand hewhoapprovesisequallyapproved; has since intensified. Some writers have argued that society as heseeks,issought,heburnsandheisburnt. a whole has become more narcissistic (Lasch, 1979), while Butwhy,Ofoolishboy, others have argued that individuals differ in their narcissist sovainlycatchingatthisflittingform? tendencies and such differences could be measured in the nor- Thecheatthatyouareseekinghasnoplace. Avertyourgazeandyouwillloseyourlove, mal population (Raskin & Hall, 1981). forthisthatholdsyoureyesisnothingsave Inpersonality/socialpsychology,researchhasfocusedontopics theimageofyourselfreflectedbacktoyou. such as whether narcissism predicts aggression (e.g., Bushman, Itcomesandwaitswithyou;ithasnolife Bonacci,vanDijk,&Baumeister,2003),hownarcissistsviewand itwilldepartifyouwillonlygo. are viewed by others (e.g., Taylor, Lerner, Sherman, Sage, & –PubliusOvidiusNaso(“Ovid”)(B.More[trans.]) McDowell,2003),andtheroleofnarcissisminself-enhancement (Wallace & Baumeister, 2002) and in interpersonal relationships TheGreekmythofNarcissustellsthestoryofamansovain and proud that he fell in love with his own image. In the field (e.g.,Campbell,Foster,&Finkel,2002).Researchershavedistin- of psychology, Freud (1914/1991) used the term narcissism to guished narcissism from related concepts such as extraversion, describe the relationship between libido and the ego. Freud hostility, and self-esteem. Of these, perhaps most focal is self- considered narcissism to result from a perturbed childhood esteem. Research has indicated that narcissism is positively but transition from subject-directed to other-directed libido, and moderatelyrelatedtoself-esteem,withthecorrelationvaryingby wasmanifestedintendenciestopreferfantasytoreality.Since self-esteemmeasures(r¯(cid:1).35acrossmeasures;Brown&Zeigler- Freud’s libidinal-based treatment of the concept, psychologists Hill, 2004). Campbell, Rudich, and Sedikides (2002) conclude, considered narcissism to lie within the domain of clinical psy- “Narcissism does not appear simply to reflect exceptionally high chology, though in a manner somewhat different from Freud’s self-esteem”(p.365). treatment. According to the Diagnostic and Statistical Manual Numerousstudieshavesoughttodeveloporvalidatemeasures of Mental Disorders (DSM–IV; American Psychiatric Associa- ofnarcissism.Oftheextantmeasures,themostfrequentlyusedis tion, 2000), narcissism is, broadly, a grandiose sense of self- theNarcissisticPersonalityInventory(NPI;Raskin&Hall,1981). importance.AccordingtotheDSM–IV,narcissistsarepreoccu- Emmons(1984)investigatedtheconstructvalidityoftheNPIand pied with fantasies of unlimited success, believe they are found four factors, which he labeled: (1) Exploitiveness/Entitle- special and unique, require excessive admiration, have a sense ment(thebeliefthatoneisadeptatmanipulatingpeople,andalso ofentitlement,areinterpersonallyexploitive,lackempathy,and a sense of entitlement to do so); (2) Leadership/Authority (the belief that one possesses an extraordinary ability to influence others, and the preference for leadership and authority roles in general); (3) Superiority/Arrogance (the belief that one is just TimothyA.Judge,JefferyA.LePine,andBruceL.Rich,Departmentof “better”thanothersandisabornleader);(4)Self-absorption/Self- Management,WarringtonCollegeofBusiness,UniversityofFlorida. admiration(anelevatedsenseofvanityandthebeliefthatoneis CorrespondenceconcerningthisarticleshouldbeaddressedtoTimothy A. Judge, Department of Management, Warrington College of Business, special). Emmons (1987) and Watson and Biderman (1993) pro- UniversityofFlorida,Gainesville,FL32611-7165.E-mail:[email protected] videdadditionalconstructvalidityevidenceonahierarchicalrep- 762 NARCISSISM 763 resentationofnarcissism,wherebythefourdimensionsalsoindi- textual performance, and task performance (Borman & Brush, cateahigher-ordernarcissismfactor. 1993; Campbell, 1990; Rotundo & Sackett, 2002). Traditionally, Given the intense interest in narcissism in personality/social leadership has constituted an aspect of performance in jobs that psychology, one might be quite surprised to find that narcissism includedexplicitaccountabilityfortheperformanceofacollective hasbeenstudiedverylittleinindustrial-organizational(I-O)psy- (group,team,organization)orunitofwork.However,thechang- chology. In fact, of the 4,010 studies identified in the PsycINFO ing nature of the workforce (e.g., increased diversity) and of the databasewithnarcissisminitskeywordsorabstract,therewereno natureofworkitself(e.g.,increaseduseofteams)haveincreased studies published in the two flagship I-O journals, Journal of the prevalence of leadership as an aspect of job performance Applied Psychology or Personnel Psychology. However, drawing (Lord&Smith,1999).Elementsofworkplacedeviance(orcoun- from the extant research in personality/social psychology, we terproductive performance), defined as voluntary behavior that suggest that narcissism may play an important role in predicting harms the well-being of the organization (Rotundo & Sackett, ratingsofcriteriathatconstitutealargeportionofthejobperfor- 2002), have been included as aspects of job performance by mance domain. In fact, the relationship between narcissism and severalscholars(e.g.,Murphy,1989;Robinson&Bennett,1995). ratings to performance criteria is likely to be relevant to I-O Infact,recentresearchsuggeststhatmanagersweighdeviance scholarsandpractitionersfortwofundamentalreasons. as much or more than performance of specific job duties when First,becausenarcissismbroadlyreflectsstrongself-admiration completing overall performance evaluations (Rotundo & Sackett, andbehavioraltendencieswhichmaynotbeviewedpositivelyby 2002).Contextualperformanceincludesthosebehaviorsthatcon- others (Penney & Spector, 2002), it is possible that narcissism tribute to the organization by fostering a positive social and psy- influencesself-andotherperceptionsdifferently,andinsightinto chological climate (Borman & Motowidlo, 1997). Examples of thispossibilitymaybeimportantgiventhatdifferencesinpercep- contextual performance include assisting coworkers when they tionsarethefoundationforcertaintypesofperformancemanage- needit,beingcourteousandrespectfultopeers,makingconstruc- ment and development practices (Brett & Atwater, 2001). In tivesuggestions,beingagoodsport,andputtingforthextraeffort essence,narcissismmayprovideanexplanationfordifferencesin onthejob.Finally,taskperformanceincludesthosebehaviorsthat thewayweperceiveourbehaviorrelativetothewayothersseeour aregenerallyrecognizedaspartofthejobanddirectlycontribute behavior,andthisexplanationisdifferentthanwhatisassumedby totheorganization’stechnicalcore(Borman&Motowidlo,1997). scholars who may attribute differences to a number of factors Although these four criteria do not exhaust the domain of job includingtheamountofinformationavailabletotheraterregard- performance criteria, they likely represent a large portion of this ingthebehaviorbeingrated,ageneralself-servingbias,attribution domain across a broad set of jobs, and as we explain in the next processes, or differences in understanding regarding the value of section,weexpecteachtoberelatedtonarcissism. certainbehaviorstotheorganization(Murphy&Cleveland,1995). To date, however, there has been no research published (in any Relationship of Narcissism to Self Versus Other Ratings journalinthePsycINFOdatabase)thatcomparestheinfluenceof narcissism with both self- and other ratings of job performance Given the obvious link between narcissism and self- criteria. Thus, one purpose of our research is to investigate rela- enhancement(Robins&Beer,2001),aninterestingandimportant tionships between narcissism and self- and other ratings of job questioniswhethernarcissistshaveenhancedviewsoftheirabil- performancecriteria. itiesandcompetence,relativetothosewhoscorelowonthetrait. Second, we suggest that narcissism may have value as a pre- Althoughthisquestionhasbeenstudiedinpersonalitypsychology dictor of job performance criteria over and above other well- (Campbell,Rudich,&Sedikides,2002;John&Robins,1994),the known personality traits. Although relationships between the Big self-enhancing tendencies of narcissists are less studied in the Fivetraitsandjobcriteriasuchasleadership(Judge,Bono,Ilies, workdomain.Giventhesocialdesirabilityofleadership(Meindl, & Gerhardt, 2002), contextual performance or citizenship behav- 1985) and performance, one reasonable application of narcissism iors (Organ & Ryan, 1995), and job performance (Barrick & isintheareaofself-andotherviewsofthesefocalcriteriainthe Mount, 1991) are well established, narcissism is an aspect of workplace. personality that is not reflected well in the Big Five (Paulhus & Research by Campbell, Rudich, & Sedikides (2002) revealed Williams, 2002). Moreover, narcissism reflects a set of attitudes, that narcissists (high overall scorers on the NPI) tended to have beliefs,andbehavioraltendenciesthatlikelyimpacteachofthese inflated views on traits reflecting an agentic orientation (e.g., organizational criteria in unique ways. To date, however, only a intelligent), but not on traits reflecting a communal orientation few studies have even linked narcissism to these criteria. Excep- (e.g., caring). Narcissists tend to see themselves as superior to tions include Deluga (1997) who linked narcissism to the per- others,perhapsasameansofconstructingforthemselvesaposi- ceivedgreatnessofU.S.presidentsandPenneyandSpector(2002) tive self-image (Morf & Rhodewalt, 2001). Indeed, the DSM–IV who linked narcissism to self-reported counterproductive behav- defining characteristics of narcissists include grandiosity and an iors.Asanotherexample,Soyer,Rovenpor,andKopelman(1999) exaggerationoftheirtalentsandaccomplishments(AmericanPsy- found a nonsignificant relationship between narcissism and self- chiatric Association, 2000). John and Robins (1994) found that reported sales performance. Thus, a second purpose of our re- narcissiststendtoexhibitaself-enhancementbias,suchthatthey searchistoinvestigatethedegreetowhichnarcissismpredictsjob judged themselves more favorably. These authors speculate that performancecriteriaoverandabovetheBigFivetraits. theinflationaryeffectsofnarcissismarelikelytobemostapparent Before continuing, we note that our choice of performance when the judgment is in an ego-involving context. Because, in criteriawasguidedbyresearchonmultidimensionalmodelsofjob general, few individuals would be indifferent to being labeled a performance,whichincludeleadership,workplacedeviance,con- poorleader,ordeviant,orapoorperformer,weassumethatsuch 764 JUDGE,LEPINE,ANDRICH judgmentsareego-involving.Thus,thoughnoresearchisdirectly is threatened (Stucke & Sporer, 2002). Bushman and Baumeister on-point,webelievethatnarcissistswillhaveenhancedviewsof (1998) found that narcissists were more likely to engage in ag- their leadership, workplace deviance, and contextual and task gressive behavior because they are hypervigilant to perceived performance compared to people who are low in narcissism. Be- threats. Narcissists may be predisposed to engage in aggressive cause, as John and Robins (1991) note, “All clinical accounts of and other deviant behavior because they are predisposed to see narcissismconcurthatnarcissisticindividualsholdunrealistically their work environment in negative, threatening ways. Finally, exaggerated beliefs about their abilities and achievements” (p. Soyer et al. (1999) found that narcissists were more comfortable 209), those scoring high on narcissism should be especially sus- withethicallyquestionablesalesbehaviors,suggestingthatnarcis- ceptibletoself-enhancementwithrespecttothesecriteria. sists are less bound to organizational rules of propriety. Putting Thisself-enhancementhypothesis,however,onlyaddressesthe theseperspectivestogether,narcissismmaybelinkedtodeviance effect of narcissism in predicting self-perceptions (i.e., how nar- throughbothaperceptualandbehavioralprocess:narcissistsmay cissismrelatestoself-ratings).Thus,whentryingtounderstandthe bepredisposedtoperceivethreatsintheworkplace,andtheymay relative effects of narcissism on self- and other ratings, it is be more likely to respond aggressively to those threats that are necessarytocomparetheeffectsofnarcissismonself-reportswith perceived. Thus, we expect narcissism to be positively related to the effects on other reports of performance criteria. Given the otherratingsofindividuals’workplacedeviance. self-enhancingnatureofnarcissism,weexpectnarcissiststoeval- uate themselves more positively irrespective of the criterion. An equallyinterestingandimportantissue,then,istherelationshipof Contextual and Task Performance narcissism to other reports of the criteria. Given the nature of narcissismanditsoriginsinclinicalpsychology,itappearsthatit Given their agentic focus (Campbell, Rudich, & Sedikides, generallywouldbeanundesirablecharacteristicatworkfromthe 2002),thattheirself-esteemislikelytobeespeciallycontingenton perspective of those who work with narcissists. However, the the hallmarks of success (Roberts & Robins, 2000), and that reason for this undesirability (at least as perceived by others) is projecting an image of competence is especially importance to likely to vary as a function of the criterion under consideration. narcissists(Elliot&Thrash,2001),onewouldexpectnarcissiststo Thus,inthenextsectionwediscusslinkagesbetweennarcissism haveunrealisticviewsoftheirperformancecapabilities,suchthat andeachofthecriteria. theybelievethemselvesmorecapableorcompetentthantheyare inreality(Wallace&Baumeister,2002).Thus,onewouldexpect narcissists to see themselves as effective performers. However, Leadership suchself-enhancingtendenciesmaybedetrimentaltoperformance Narcissists are motivated to gain the admiration of others and because they are built on a fragile base of self-esteem (which is receive affirmation of their superiority. Morf and Rhodewalt easilyshatteredinthewakeoffailure)andthetendencytoexcuse (2001) argue that the process of gaining admiration and affirma- away failures (Robins & Beer, 2001). The narcissistic pursuit of tion is self-defeating in the long-term because the tactics so used self-esteem also may hinder performance through diminished (e.g., aggressing at and derogating others, self-aggrandizement, learningandpoorerself-regulation(Crocker&Park,2004).Over- lowintimacystrivings)undermineinterpersonalrelationships.In- all, this research leads us to expect that narcissism will be nega- deed, the pursuit of self-esteem, something narcissists are partic- tivelyrelatedtobothcontextualandtaskperformance.Soyeretal. ularlypredisposedtoward,isarguedtobecostlyintermsofothers’ (1999)foundthatnarcissismwasunrelatedtosalespersonperfor- perceptions (Crocker & Park, 2004). Moreover, the excessive mance,althoughthemeasureofnarcissismwassomewhatunusual agentic focus of narcissists, which causes them to focus on their (aclinicalmeasurebasedontheDSM–III)andthemeasureofsales goals at the expense of others’ goals (Morf & Rhodewalt, 2001), performancewasself-reportedbyindividuals. also should serve to undermine narcissists in a social context. Asaresultoftheforegoingsupport,nextwepresenthypotheses Finally,narcissistscharacteristicallylackempathy(Brown&Bos- foreachofthecriteriaincludedinthisstudy: son,2001).Giventhecentralityofempathytoleadership(Hogan &Hogan,2002),thisfurthersuggeststhatwhereasnarcissistsmay H-1a: Narcissism will be positively related to self-ratings of viewthemselvesassuperiorleaders,otherswillformtheopposite leadership. conclusion.Thus,weexpectnarcissismtobenegativelyrelatedto H-1b:Narcissismwillbenegativelyrelatedtootherratingsof others’perceptionsofleadership. individuals’leadership. Workplace Deviance H-2a:Narcissismwillbenegativelyrelatedtoself-ratingsof workplacedeviance. PenneyandSpector(2002)foundthatnarcissismwaspositively relatedtodeviantorcounterproductiveworkbehaviors(r(cid:1).27,p H-2bNarcissismwillbepositivelyrelatedtootherratingsof (cid:2).05).Whymightnarcissismbelinkedtodevianceinthework- individuals’workplacedeviance. place?Becausenarcissistsarecoercive(Baumeister,Catanese,& Wallace,2002),andmaybemotivatedtoderogateothers(Morf& H-3a: Narcissism will be positively related to self-ratings of Rhodewalt, 2001), one would expect narcissists to be more pre- contextualperformance. disposed to engage in behaviors that ultimately harm the organi- zation. Moreover, research suggests that narcissists are likely to H-3b:Narcissismwillbenegativelyrelatedtootherratingsof engageinaggressivebehavior,especiallywhentheirself-concept individuals’contextualperformance. NARCISSISM 765 H-4a: Narcissism will be positively related to self-ratings of oftenunrecognized,thenarcissist’sperceivedstandingislowered taskperformance. fromacomparativesense. H-4b:Narcissismwillbenegativelyrelatedtootherratingsof H-5: Narcissism will predict more strongly and negatively individuals’taskperformance. otherratingsofcontextualperformancethanotherratingsof taskperformance. Relationship of Narcissism to Contextual Versus Task Inflation Versus Enhancement Performance In investigating the degree to which narcissism is related to Althoughintheprevioussectionwearguedthatnarcissismwill self-ratingsandotherratingsofvariouscriteria,andthedifferences be negatively related to other ratings of individuals’ task and intheserelationships,weneedtocommentonwhatourstudydoes contextual performance, there are reasons to believe that this notaddress.Specifically,wearenotstudyinginflationinratingsin negative relationship will be stronger with one of these criteria. thesensethatwearenotconcernedwithwhetherindividualsover- First,toalargeextent,narcissistictendenciesappeartobethevery and underestimate their performance relative to how others esti- oppositeofthosethatreflectaneffectivecontextualperformer.As mate their behavior. Rather, we are interested in whether narcis- noted previously, narcissists tend to lack empathy, engage in sismleadspeopletoevaluatethemselvesmorepositivelyandalso aggressivebehavior,andhaveself-servingmotives.Accordingly, whether narcissism leads to evaluations by others that are less narcissistsshouldbeespeciallyunlikelytocontributepositivelyto positive. Specifically, do those people who score high on narcis- the organization’s social and psychological climate by helping sismratethemselvesmorefavorablythandothosewhoscorelow others, being courteous and a good sport, and going above and on narcissism, and are those who score high on narcissism rated beyondthecallofdutyforthegreatergood.Althoughnarcissistic less favorably by others than are those who score low on narcis- tendencies should detract from task performance as well for rea- sism?Thus,wearenotinterestedinself-otherdifferences(differ- sonsstatedearlier,thecorrespondenceofthesetendencieswiththe ence scores) in criterion ratings and whether narcissism predicts behavioralelementsofcontextualperformanceappearstobemuch these differences. Rather, our concern is the degree to which stronger. narcissism is associated with enhanced self-ratings (more favor- A second reason why we expect narcissism to more strongly able self-ratings than would otherwise be the case [for someone predict other ratings of contextual performance than task perfor- highvs.lowonnarcissism]),aswellastherelationshipofnarcis- mance is that the behavioral activities that comprise contextual sismtootherratingsofthesamecriteria,andacomparisonofthese performance are more discretionary and less explicitly rewarded relationships. thanthebehavioralactivitiesthatcomprisetaskperformance(Bor- man & Motowidlo, 1997). Accordingly, a behavioral predisposi- Role of Big Five Traits tion such as narcissism should play a stronger role in predicting contextual performance because there is less information and Finally, there are two reasons why it may be important to fewerincentivesinthe“situation”regardingwhichspecifictypes consider whether narcissism adds to the prediction of leadership of behavior to enact. In essence, contextual performance consti- ratings and job performance over and above the Big Five traits. tutes the “weaker” criterion in the sense that individuals should First, the Big Five traits are thought to constitute the majority of perceive more freedom to act, and in this type of situation, indi- the domain of personality, and several Big Five traits predict vidual differences in personality are more likely to influence leadership(Judgeetal.,2002)andperformance(Barrick&Mount, behavior(Mischel,1997;Weiss&Adler,1984).Indeed,research 1991) ratings. Second, narcissism itself is related to some of the has consistently demonstrated that personality traits are stronger Big Five traits, so there is the question of concept redundancy. predictorsofcontextualperformancethantaskperformance(e.g., Specifically, narcissism correlates with Extraversion positively, Hattrup,O’Connell,&Wingate,1998;LePine&VanDyne,2001; andAgreeablenessandNeuroticismnegatively,thoughthesecor- Motowidlo & Van Scotter, 1994; Van Scotter & Motowidlo, relations are not especially strong (Graziano & Tobin, 2001). 1996). Thus, in the same way that personality is more likely to Moreover, the trait that is the best Big Five predictor of job predicttaskperformancewhenindividualshaveautonomyinhow performance, and one of the best predictors of leadership (i.e., to go about their work (Barrick & Mount, 1993), due to its Conscientiousness), is generally unrelated to narcissism (Ruiz, discretionarynature,contextualperformancegivesindividualsau- Smith, & Rhodewalt, 2001). For these reasons, we expect that tonomyindecidingwhethertoperformthebehaviors,suggesting controlling for the Big Five will not undermine the effect of that narcissists’ selfish desires will be more evident with contex- narcissism. Nevertheless, we do control for the Big Five traits in tualthantaskperformance. theanalyses. Finally, narcissists are more likely to engage in behaviors that theyperceiveasservingtheirowngoalsratherthanothers’goals Method Overview (Morf & Rhodewalt, 2001). Narcissists will be more likely to direct effort toward activities that reflect task performance given In order to test our hypotheses, we conducted two studies. In thatthesebehaviorsaremorelikelytoberecognizedanddirectly Study 1, we used narcissism to predict self- and other ratings of rewardedthancontextualperformance.Moreover,actsofcontex- leadership. In Study 2, we used narcissism to predict self- and tual performance may be viewed by narcissists as self-defeating otherratingsofleadership,workplacedeviance,contextualperfor- because in contributing to others’ welfare through acts that are mance,andtaskperformance. 766 JUDGE,LEPINE,ANDRICH Study 1 second-order factor analysis to determine whether a higher-order hierar- chicalmeasurementapproach(wherebynarcissismcanbeanalyzedatboth Method thefacetandoveralldimensionlevels)wasjustified.Inthesecond-order factor analysis, the first-order factors consisted of the four narcissism Setting and Participants dimensions,whichwereindicatedbycorrespondingparcels.Thesecond- orderfactorwastheoverallnarcissismfactor,whichwasindicatedbythe Participantsweremaster’sdegreecandidatesatalargeuniversityinthe fournarcissismdimensions(firstorderfactors).Ifthishierarchicalmodel southeastern United States. Roughly one third of the participants were fits the data well, and the first- and second-order factor loadings are studentsinaMasterofBusinessAdministration(MBA)program;theother significant,thensupportforthehierarchicalmeasurementapproachwould twothirdswerestudentsinaMasterofScienceinManagement(MSM) beprovidedbytheresults.Indeed,resultsindicatedthatthehierarchical program. Of the students in the MBA program, 24% were female, 84% model fit the data reasonably well ((cid:1)2 (cid:1) 83.89; Root-mean-square 50 wereWhite,and16%weremembersofaminoritygroup(AfricanAmer- Residual [RMR] (cid:1) .07; Root-mean-square error of approximation ican,HispanicAmerican,orNativeAmerican).TheaverageMBAstudent [RMSEA](cid:1).07;Comparativefitindex[CFI](cid:1).90),thefourfirst-order had 4.3 years of work experience. Of the students in the latter (MSM) factorswereindicatedbytheparcels((cid:2)¯ (cid:1).57,¯t(cid:1)4.26,allp(cid:2).01),and y program,51%werefemale,68%wereWhite,9%wereAfricanAmerican, theoverall(second-order)narcissismfactorwassignificantly(¯t(cid:1)4.43,all 13%wereAsian,and10%wereofotherinternationalorigins.Typically, p(cid:2).01)indicatedbythefourfirst-ordernarcissismdimensions((cid:3)¯(cid:1).77). studentsintheMSMprogramhadonlylimitedfull-timeworkexperience. Theseresultssupporttheaggregationofthedimensionsintoahigher-order factor.Thereliabilityofoverall37-itemnarcissismscalewas(cid:4)(cid:1).90. Procedure BigFivepersonalitytraits. TheBigFivetraitsweremeasuredwiththe 60-itemNEO-FFI(Costa&McCrae,1992).Usinga0(cid:1)Stronglydisagree Inordertogaininsightintotheircapabilitiesasmanagersandleaders, to 4 (cid:1) Strongly agree scale, individuals evaluated their Neuroticism, individualsinbothprogramswereassessedontheirpersonality(BigFive Extraversion, Openness, Agreeableness, and Conscientiousness with 12- personality traits, narcissism) and leadership. Individuals self-reported itemscales.Thefivescaleswerecomputedbysummingtheresponsesto theirpersonalitiesandtheirownleadershipbehaviors.Separately,individ- eachoftheseitems.ThereliabilitiesoftheBigFivescaleswere:Neurot- uals’leadershipbehaviorsalsowereconfidentiallyevaluatedby3–6other icism,(cid:4)(cid:1).86;Extraversion,(cid:4)(cid:1).74;Openness,(cid:4)(cid:1).76;Agreeableness, raters. These other raters were instructed to be people with whom the (cid:4)(cid:1).69;Conscientiousness,(cid:4)(cid:1).81. individual has “worked most closely—supervisors, coworkers, or fellow Leadership. We measured leadership with the Leadership Practices students with whom they have worked on team projects.” (We did not Inventory(LPI;Posner&Kouzes,1993),a30-itemmeasurethatcontains obtaindataontherolesoftheseotherraters,thoughwebelievemostof fivesubscales(Challenging,Inspiring,Enabling,Modeling,Encouraging). them to be “peers” [coworkers and fellow students]). Individuals were PastresearchhasusedtheLPIasameasureoftransformationalleadership promisedthattheirresultswerecompletelyconfidential.Allsurveyswere (Fields&Herold,1997).Moreover,researchsuggeststhattheLPIassesses returneddirectlytothefirstauthor,bybothparticipantsandtheratersof anoverarchingconstructoftransformationalleadership(Carless,2001).As thoseparticipants,intheirrespectivesealedenvelope(s)toensureconfi- previouslynoted,individualsratedthemselvesusingtheLPIandwerealso dentiality. As an incentive and a benefit, individuals were promised and ratedbythreeothers.Accordingly,weaveragedthe30itemsforbothself- given separate feedback reports that summarized their personality and andotherreports.Fortheself-reportedLPI,thereliabilityofthe30-item leadership scores. The reports were provided after the personality and scalewas(cid:4)(cid:1).92.FortheotherLPIreports,theaveragereliabilityofthe leadershipsurveyswerecompleted.Becausethesereportsweretheonly 30-item scale was (cid:4)¯ (cid:1) .95. There was significant agreement among the benefit to participating, and participation was not mandatory, it seems otherratersintermsoftheirleadershipratings(ICC(1)(cid:1).28,p(cid:2).05),so unlikely that individuals completed the other surveys themselves. In all, weaveragedacrosstheraterstoformasingletransformationalleadership 139ofthe159individualshadcompletesetsofsurveys,foraresponserate scale.TheinterraterreliabilityoftheotherratingswasICC(2)(cid:1).54.The of87%. self-othercorrelationwasrˆ.Althoughthiscorrelationisrelativelymodest, self-ratings tend to display weak correlations with supervisor, peer, and Measures followerratingsofperformanceandleadershipcriteria(Brett&Atwater, 2001). Narcissism. Participants’ narcissism was assessed with Emmons’ amendmentoftheNarcissisticPersonalityInventory(NPI;Raskin&Hall, Results 1979,1981).TheabridgedNPIwasderivedfromEmmons’(1987)factor analyticstudyofthe54-itemRaskinandHall(1979)instrument.TheNPI CorrelationsanddescriptivestatisticsforStudy1variablesare hasbeenshowntobeareliableself-reportinventorymeasuringindividual presentedinTable1.Totestnotonlytheeffectofnarcissismon differencesinnarcissismasapersonalitytrait(Emmons,1987;Rhodewalt &Morf,1998).TheNPI,consistingof37forcedchoicedichotomous(1(cid:1) self- and other ratings, but also the differences in the relative yesor0(cid:1)no)items,hasshownconsiderableevidenceofconstructvalidity effects (whether narcissism is more strongly related to self- vs. andinternalconsistency(forreviews,seeEmmons,1987;Raskin&Terry, otherreports),weutilizedmultivariateregression(Edwards,1995; 1988). Greene, 1990, pp. 509–519). The advantages of multivariate re- PriorconfirmatoryfactorstudiesbyEmmons(1984,1987)indicatethat gression are twofold. First, it provides separate estimates of the theNPIcomprisesanoverallmeasurethat,inturn,iscomposedoffour effectoftheexplanatoryvariable(inthiscase,narcissismandthe factors:Leadership/Authority(e.g.,“Iwouldprefertobealeader,”“Ilike Big Five traits) on self- and other ratings. Second, it provides a having authority over people”); Self-absorption/Self-admiration (e.g., “I statistic,intheformofWilks’(cid:2),thatteststheequivalenceofthe think I am a special person,” “I like to look at myself in the mirror”); effect of the explanatory variable across the two (self and other) Superiority/Arrogance (e.g., “I always know what I am doing,” “People equations. A significant (cid:2)(which is distributed as an F-statistic) canlearnagreatdealfromme”);andExploitiveness/Entitlement(e.g.,“I indicatesthattheeffectoftheexplanatoryvariableisdifferentin find it easy to manipulate people,” “I expect a great deal from other people”).Pastresearchgenerallyhastreatednarcissismbothasageneral thetwoequations.Becauseitisimportanttoshowuniqueeffects factor (e.g., Morf & Rhodewalt, 1993), as well as a multidimensional ofspecifictraitsbeyondthefive-factormodeltraits,wecontrolled construct(e.g.,Watson&Biderman,1993).Accordingly,weconducteda fortheBigFivetraitsinallregressions.Finally,forinformational NARCISSISM 767 Table1 Means(M),StandardDeviations(SD),andIntercorrelationsAmongStudy1Variables M SD 1 2 3 4 5 6 7 8 1.Narcissism 0.52 0.15 (.90) 2.Neuroticism 18.91 8.03 (cid:3).11 (.86) 3.Extraversion 30.64 5.53 .36** (cid:3).40** (.74) 4.Opennesstoexperience 29.18 6.60 .04 (cid:3).01 .09 (.76) 5.Agreeableness 30.80 5.45 (cid:3).24** (cid:3).17* .25** (cid:3).01 (.69) 6.Conscientiousness 33.53 6.13 (cid:3).10 (cid:3).22** .14 (cid:3).05 .23** (.81) 7.Leadership—self 7.33 0.91 .35** (cid:3).22** .37** .25** .19* .23** (.92) 8.Leadership—other (peer) 7.74 0.85 .20* (cid:3).10 .15 .10 .09 (cid:3).01 .15 (.95) Note. ListwiseN(cid:1)134.Reliability((cid:4))estimatesarelistedonthediagonal. *p(cid:2).05. **p(cid:2).01. purposes, we also report standardized effect size estimates in the Individuals for whom complete survey responses were unavailable were formofpartialeta-squared((cid:5)2)fromthemultivariateregressions excludedfromthestudy.ThisresultedinafinalsamplesizeofN(cid:1)131. and standardized regression ((cid:6)ˆ) estimates from univariate OLS regressions. Measures Results of this analysis are presented in Table 2. As the table shows,narcissismsignificantlyandpositivelypredictedbothself-((cid:6)ˆ Narcissism. As in Study 1, narcissism was assessed with the 37-item Narcissistic Personality Inventory (NPI; Raskin & Hall, 1979, 1981). The (cid:1) .36, p (cid:2) .01) and other ((cid:6)ˆ (cid:1) .19, p (cid:2) .05, p (cid:2) .05) ratings of participant was asked, using the same dichotomous (yes-no) scale that has leadership,supportingH-1abutnotsupportingH-1b.Althoughnar- been used in past research, whether each of the 37 NPI items described cissism was positively related to both self- and other ratings of themselves.Thereliabilityofthe37-itemscalewas(cid:4)(cid:1).87.AsinStudy1,we leadership, the coefficient was significantly stronger for self-ratings conductedasecond-orderfactoranalysistodeterminewhetheritisappropriate thanforotherratingsofleadership(F(cid:1)11.00,p(cid:2).01).Thestronger totreatnarcissismasbothamultidimensionalandaggregateconstruct.The effectofnarcissismonself-ratingsofleadershipisfurtherillustrated hierarchicalmodel,whichusedthreeparcelsasindicatorsofeachofthefour bythestandardizedeffectsizeestimates,whichshowthatnarcissism narcissism dimensions, and then used the four narcissism dimensions as explainsmoreuniquevarianceinself-ratingsofleadership((cid:5)2(cid:1).12) indicatorsofthesecond-orderoverallfactor,fitthedatareasonablywell((cid:1)520 than in other ratings ((cid:5)2 (cid:1) .04). Thus, though narcissism did not (cid:1)98.55;Root-mean-squareResidual[RMR](cid:1).07;Root-mean-squareerror ofapproximation[RMSEA](cid:1).08;Comparativefitindex[CFI](cid:1).92),the negatively predict other ratings of leadership as expected, it did fourfirst-orderfactorswereindicatedbytheparcels((cid:2)ˆ (cid:1).63,¯t(cid:1)6.29,allp positively predict self-ratings, and moreover did to a significantly (cid:2).01),andtheoverall(second-order)narcissismfactorywassignificantly(¯t(cid:1) greaterdegreethanotherratings.Finally,weshouldnotethatthreeof 6.54, all p (cid:2) .01, all p (cid:2).01) indicated by the four first-order narcissism theBigFivetraitswererelatedtoself-butnototherleadershipratings. dimensions((cid:3)¯ (cid:1).72).Thus,weutilizedanoverallmeasureofnarcissismby Specifically,OpennesstoExperience,Agreeableness,andConscien- averagingresponsesacrossthe37items. tiousness were more strongly and more positively related to self- BigFivepersonalitytraits. TheBigFivepersonalitytraitswereself- reportsofleadershipthantootherreports.1 reportedbyparticipantsusingthe44-itemBigFiveInventory(BFI;John, Donahue, & Kentle, 1991). This measure was constructed to allow for efficientassessmentoftheBigFivedimensionsofNeuroticism,Extraver- Study 2 sion,OpennesstoExperience,Agreeableness,andConscientiousnesswhen thereisnoneedformoredifferentiatedmeasurementofindividualfacets. Method The BFI does not use single adjectives as items. Instead, one or two Sample and Procedures prototypicaltraitadjectivesservedastheitemcoretowhichinstructive, clarifying,orcontextualinformationwasadded.Forexample,theExtra- Thesampleconsistedof143maleandfemalemembersofabeachpatrol versionadjectiveenergeticservedasthebasisfortheBFIitem“Iamfull locatedwithinthemid-AtlanticregionoftheUnitedStates.Approximately ofenergy,”andtheOpennessadjectiveimaginativebecametheBFIitem 21%(N(cid:1)30)oftheseparticipantswerefemale;participantsrangedinage “Ihaveanactiveimagination.”Thus,theBFIitemsareshortandavoid from18to48years.Participationinthestudywasvoluntary;however,in complexsentencestructures,retainingtheadvantagesofadjectivalitems exchangefortheirparticipation,beachpatrolmemberswhoreturnedsur- (brevityandsimplicity)whileavoidingsomeoftheirpitfalls(ambiguousor veys received $10. The response rate for full-time employees was 96%. multiple meanings and salient desirability) (John & Srivastava, 1999). Data were collected from three sources. First, a confidential packet was handdeliveredtoallemployees.Thispacketincludedaquestionnaireto measure self-perceptions of personality (Big Five and narcissism) and 1Giventhepotentiallymultidimensionalnatureofnarcissism(Raskin& leadershipbehavior.Significantotherscompletedameasureofemployees’ Hall,1981),inadditiontoself-versusotherviewsbeinginfluencedbythe narcissismandwereprovidedwithaconfidentialreturnenvelope.Finally, overall narcissism factor, the dimensions of narcissism may be relevant. immediatesupervisorsprovideddataonemployees’leadership,workplace Accordingly, we repeated the analysis reported in Table 2 utilizing the deviance,andtaskandcontextualperformance.Allsurveyswerereturned overall narcissism factor, with an analysis utilizing the four narcissism directly to the third author, by both participants and the raters of those dimensions.Theresultsindicatedthatnoneofthenarcissismdimensions participants,intheirrespectivesealedenvelope(s)toensureconfidentiality. significantlypredictedself-orotherreportsofleadership. 768 JUDGE,LEPINE,ANDRICH Table2 MultivariateRegressionsPredictingSelf-RatingandOther(Peer)Ratings ofLeadership(Study1) Self-Rating(S) OtherRating(O) S-O Difference Partial Partial IndependentVariable Bˆ (cid:5)2 (cid:6)ˆ Bˆ (cid:5)2 (cid:6)ˆ Wilks’(cid:2)(F) Neuroticism (cid:3).01 .00 (cid:3).06 (cid:3).01 .00 (cid:3).06 0.41 Extraversion .02 .01 .12 .00 .00 .03 0.85 Opennesstoexperience .03** .08** .24** .01 .01 .08 5.61** Agreeableness .03* .04* .20* .02 .02 .15 3.72* Conscientiousness .03** .05** .21** .00 .00 (cid:3).05 3.52* Narcissism 2.02** .12** .36** 1.16* .04* .19* 11.00** Note. Bˆ (cid:1)unstandardizedregressioncoefficientfrommultivariateregression.Partial(cid:5)2(cid:1)uniquevariance explained by independent variable. Wilks’ (cid:2)(cid:1) difference between coefficient estimates for self- and other ratings(distributedasF-statistic).(cid:6)ˆ (cid:1)standardizedregressioncoefficientfromunivariateOLSregression. *p(cid:2).05. **p(cid:2).01. Despiteitsbrevity,theBFIdoesnotsacrificeeithercontentcoverageor recorded on a five-point scale ranging from 1 (Not at all) to 5 (Fre- psychometric properties. For example, the eight-item Extraversion scale quently,ifnotalways).Participantsratedtheirowncontextualperfor- includesitemsfromatleastfourofthesixfacetspostulatedbyCostaand manceusingthe24-itemscale.Participants’supervisorsindependently McCrae(1992)—namely,gregariousness,activity,assertiveness,andpos- ratedthemusingthesamesetofitems.Wenotethatthesportsmanship itiveemotions.Responsestoeachitemwererecordedonafive-pointscale itemswerereverse-scoredsothathigherscoresindicatehighersports- ranging from 1 (strongly disagree) to 5 (strongly agree). Scales were manship. Reliabilities of the five factors ranged from .83 to .94. computedbyaveragingparticipants’responsesfortheitems.Reliabilities Research indicates that the behavioral domains of organizational citi- of the five factors—ranging in this study from .80 to .85—proved to be zenshipbehaviorandcontextualperformanceoverlaptheoreticallyand consistentwithpriorresearchthathasdemonstratedalphareliabilitiesof empirically (LePine, Erez, & Johnson, 2002). Accordingly, we aggre- theBFIscalestypicallyintherangeof.75to.90. gatedthefivedimensionsintoanoverallcontextualperformancefactor. Leadership. Leadership behavior was measured with 12 leadership Supportingthisdecision,thesecond-orderfactorloadingswerestrong itemsfromBassandAvolio’s(1990)MultifactorLeadershipQuestionnaire ((cid:3)ˆ (cid:1) .81) and significant (¯t(cid:1)9.13, p (cid:2) .01). Moreover, the five (MLQ). We utilized these items because research repeatedly has shown dimensions displayed a similar pattern of correlations with narcissism that these dimensions load the highest on the overall transformational andtheBigFivetraits.Specifically,acrossbothself-andotherreports leadershipconstruct(Avolio,Bass,&Jung,1999).Participantswererated bytheirimmediatesupervisoronthefrequencywithwhichtheparticipant ofthefournarcissismfacets,andself-andother(supervisor)reportsof engagesineachofthebehaviorsfrom1(Notatall)to5(Frequently,ifnot contextual performance, the average difference in the correlations of always).Participantsalsoratedtheirowntransformationalleadershipbe- eachfacetwiththefiveOCBdimensionswasonly.05.Somenarcissism haviors on the same scale. Self- and other scales were computed by facetsweremorevariableintheirrelationtotheOCBdimensionsthan averagingthe12itemsforeach(self-andother)measure.Thereliabilityof others. The superiority/arrogance narcissism facet, for example, was thismeasurewas(cid:4)(cid:1).85forparticipants’self-reportsand(cid:4)(cid:1).96forthe more variable in its relationship with the OCB dimensions than the other(supervisor)reports. othernarcissismfacetsortheoverallnarcissismconstruct.Eveninthis Workplacedeviance. Weusedthe24-itemscaledevelopedbyBennett case,though,thestandarddeviationincorrelationswasrelativelysmall and Robinson (2000) to assess workplace deviance. In a factor analytic (i.e.,thehighestwasSD(cid:1).09).Thus,thedecisiontoaggregatethefive studyusingself-ratingsoftheseitems,BennettandRobinsonfoundtwo OCB dimensions into a single overall measure appeared to be well correlated, but distinct, scales reflecting the distinction in the intended justified. Accordingly, we computed an overall measure of contextual target of the harmful behavior. Therefore, both interpersonal deviance performancebyaveragingthe24itemsforeachreport.Thereliability (harmfultopeopleintheorganization)andorganizationaldeviance(harm- ofthismeasurewas(cid:4)(cid:1).86forparticipants’self-reportsand(cid:4)(cid:1).94 fultotheorganization)wereassessed.However,pastresearchhasfound fortheother(supervisor)reports. thesetwodimensionstobehighlyrelated(Lee&Allen,2002).Thus,we Taskperformance. Ratingsoftaskperformancewereassessedusing optedtotreatthisasasingleworkplacedeviancevariablebyaggregating Williams and Anderson’s (1991) measure of in-role behaviors. Wil- acrossitems((cid:4)(cid:1).91and(cid:4)(cid:1).82forself-andother[supervisor]ratings, liams and Anderson have shown that their In-Role Behavior scale respectively). Immediate supervisors rated their subordinates on the fre- measuresaconstructdistinctfromcontextualperformance.Participants quencywithwhichtheyengageinworkplacedeviancefrom1(Never)to ratedthemselvesonasix-itemin-roleperformancescalewhichaskedto 5(Daily).Participantsalsoratedtheirownworkplacedevianceonthesame what degree, from 1 (Not at all) to 5 (Frequently, if not always), did scale.Exampleitemsincluded,“Playedameanprankonsomeoneatwork” and“Takenpropertyfromworkwithoutpermission.” theymeettheformalrequirementsoftheirjob.Exampleitemsincluded, Contextualperformance. Participants and supervisors provided rat- “Performs the tasks that are expected as part of this job” and “Ade- ings of contextual performance using the 24-item scale developed by quately completes responsibilities.” Supervisors rated their subordi- Podsakoff, MacKenzie, Moorman, and Fetter (1990). This scale was natesusingthesamesix-itemscale.Self-andother(supervisor)scales originallydesignedtoassessfivedimensionsoforganizationalcitizen- were computed by averaging the six items for each (self- and other) shipbehaviorsuggestedbyOrgan(1988):altruism,conscientiousness, measure.Thereliabilityofthismeasurewas(cid:4)(cid:1).83forparticipants’ sportsmanship, courtesy, civic virtue. Responses to each item were self-reportsand(cid:4)(cid:1).94fortheotherreports. NARCISSISM 769 Results 4 4) 1 9 (. Themeans,standarddeviations,andzero-ordercorrelationsfor arellliaSbtuildityie2s fvoarriathbelesmaeraesusrheoswanreinprTesaebnleted3.oInntethrnealdicaognosnisatle.nAcys 13 (.83).09 shown in Table 1, the correlation between task and contextual * performanceisquitehigh(self-ratings,rˆ(cid:1).59;other[supervisor] 12 94)1679* ratings,rˆ(cid:1).79),which,likeStudy1,dwarfsthedifferentsource (... correlationsofthesamevariable(e.g.,self-andother[supervisor] * ratings of contextual performance correlated ˆr (cid:1) .19). Although 11 86)19*59*18* this is a common finding (Brett & Atwater, 2001), it does not (.... necessarilymeanthetwovariableshavenodiscriminantvalidity. * * Forexample,LePineandVanDyne(1998)foundastrongcorre- 10 82)1453*1136* lationbetweentaskandcontextualperformance(ˆr(cid:1).69),yetthey (.(cid:3).(cid:3).(cid:3).(cid:3). alsofoundthatcontextualperformanceexplainedadditionalvari- * * anceinjobperformanceaftertheeffectofin-roleperformancewas 9 1)0*4*25*5 924131 takenintoaccount. (..(cid:3).(cid:3).(cid:3).(cid:3). As in Study 1, we used the same analytical procedure (multi- variate regression) to assess our hypotheses, controlled for the 8 6)51**20**25** influenceoftheBigFivetraits,andreportthesameinformation.In (.9(cid:3).1(cid:3).4.1.8.1.6 Study2,therearefourcriteriaweinvestigated:leadership,work- placedeviance,contextualperformance,andtaskperformance. 7 5)66**23**7*7**2 For leadership, results in Table 4 indicate that narcissism 81206131 (..(cid:3)...... significantly and positively predicted self-ratings of leadership ((cid:6)ˆ (cid:1) .22, p (cid:2) .01), but significantly negatively predicted other * * * * ratings of leadership ((cid:6)ˆ (cid:1) (cid:3).20, p (cid:2) .05). Thus, both H-1a and 6 80)40*1440*0767*1460*12 (...(cid:3).(cid:3)..... H-1bweresupportedbytheresults.Moreover,thesignificanceof theWilks’(cid:2)statistic(F(cid:1)7.09,p(cid:2).01),andthedirectionofthe ** * * * effects, indicates that the effects of narcissism on self- and other 5 2)0*5*87*59*60*6 8530304040 ratingsofleadershiparesignificantlydifferent.Resultsalsoindi- s (....(cid:3).(cid:3)....(cid:3). e catedthatthreeoftheBigFivetraits—Neuroticism,Opennessto bl a * * * Eenxhpaenrcieendcele,aadnedrshCiponrasctiinegnsti.oSupsneceisfsi—calallys,oNweuerroetiacsissmociwataesdnwegitah- Vari 4 (.83).15.10.40*.12(cid:3).19*(cid:3).10.34*.10.23*(cid:3).03 2 tivelyrelatedtoself-(butnotother)ratingsofleadership,whereas y d **** * * Openness and Conscientiousness were positively related to self- Stu 3 85)22*26*30*42*10051230*0422*06 (butnotother)ratingsofleadership. g (......(cid:3)...... al. n n Forworkplacedeviance,resultsshowninTable5indicatethat o o narcissismwaspositively((cid:6)ˆ (cid:1).13,p(cid:1).06)relatedtoself-ratings Am 2 1)4**46**6**3**4141*61*8 diag of workplace deviance, which fails to support H-2a. Narcissism ons (.8(cid:3).2(cid:3).0(cid:3).2(cid:3).2(cid:3).3(cid:3).1.1.1(cid:3).2(cid:3).1(cid:3).2(cid:3).0 the was positively related to other ratings of participant workplace ati on drsrCrsioncnmawspneaeiiionefaeagosoovtpdavrngdtTimnnnnrtceoicnitaekcaissaiaecorgtrftsabpctlnxiehnnisosstvi(llc.ct.eetieeaFntueaesoeNhonscTltnmaxfayhfetealcwt(cid:1)ti6(tfaluwor(cid:6)ordˆyfwpptocaeuteaneorehinlsa5scntssev(cid:1)orreueanstst.ddkfifps5rxrlieaoipotcssee3tcs7ntsr.imluinhfro2sm,tgacsffeag—efca4sceeonpdalweoni,ecriBsnnwrfmndmtmaaoip(cid:2)cdaitcefsngtoaeoaresoahrcdwinr(cid:2)fivne.nrneFgei(0ciosa(cid:6)trnˆnais1oetlsttsvaiyna.ihi)ttg0sfressh(cid:1)hrnce(ietnk1cmnfeasceorltiai)euorrnaartsf,(cid:3)eicagpcnntislsrosecoiiashiegtsiat.unsdangemtu2srltmilgWnflyi—np5tvttfnophtssoi-p,elgfraitearyOololimoteprssnltykcenrefeppdeassat-donls(cid:2)roieigt’nnfwnefufianr-atvncftlgtp(cid:2)rlonetoadesegyoc.ipr0ter.vreireHeo(koosrss1tveFdnehuotsspIrf)-lihnltiteyst,a2lasttheca(cid:1)nioobmtsHreocneeoTc.ufernpodE-ertfapt4h3eMtrmtrhexbeptdh.epeeatceo6plxhodeoepro)ote3et.asvrrtsirorruwct,reneertirirH6oiastaeolnteepoe.fls,fnpfsvgonp-dotcorrponew(cid:2)corrreeerHoerefaertgpe,srtnrssst.-wsvrofaucee0ag3teoenirhllls1oaolbsstftfrrdykesss)--f--f-,, Table3Means(M),StandardDeviations(SD),andIntercorrel MSD1 1.Narcissism0.590.19(.87)(cid:3)2.Neuroticism2.310.78.103.Extraversion3.790.75.31**4.OpennesstoExperience3.870.67.20*(cid:3)5.Agreeableness4.100.65.03(cid:3)6.Conscientiousness4.260.61.017.Leadership—self3.780.59.34**(cid:3)8.Leadership—other3.720.93.089.Workplacedeviance—self1.590.49.1210.Workplacedeviance—other1.230.28.24**11.Contextualperformance—self4.090.47.12(cid:3)12.Contextualperformance—other4.010.74.1413.Taskperformance—self4.660.51.05(cid:3)14.Taskperformance—other4.360.79.03 (cid:1)(cid:4)Note.NListwise131.Reliability()estimatesarelisted(cid:2)(cid:2)pp*.05.**.01.Inallcases,othersaresupervisors. 770 JUDGE,LEPINE,ANDRICH Table4 MultivariateRegressionsPredictingSelf-RatingandOther(Supervisor)RatingsofLeadership (Study2) S-O Self-Rating(S) OtherRating(O) Difference Partial Partial Wilks’(cid:2) IndependentVariable Bˆ (cid:5)2 (cid:6)ˆ Bˆ (cid:5)2 (cid:6)ˆ (F) Neuroticism (cid:3).14** .05** (cid:3).21** (cid:3).10 .01 (cid:3).09 3.51** Extraversion .09 .02 .11 .10 .01 .08 1.45 OpennesstoExperience .26** .13** .29** .21 .03 .16 10.22** Agreeableness .11 .02 .12 (cid:3).01 .00 (cid:3).01 1.15 Conscientiousness .22** .06** .23** .09 .00 .06 4.09* Narcissism .68** .07** .22** (cid:3).96* .04* (cid:3).20* 7.09** Note. Bˆ (cid:1)unstandardizedregressioncoefficientfrommultivariateregression.Partial(cid:5)2(cid:1)uniquevariance explained by independent variable. Wilks’ (cid:2)(cid:1) difference between coefficient estimates for self- and other ratings(distributedasF-statistic).(cid:6)ˆ (cid:1)standardizedregressioncoefficientfromunivariateOLSregression. *p(cid:2).05. **p(cid:2).01. performance.2Thus,H-4wasnotsupportedbytheresults.Results ratings of leadership and other ratings of contextual performance didindicatethatopennesswaspositivelyrelatedtoself-andother in Study 2, and positively related to other ratings of workplace ratings of contextual performance, though the former effect was deviance in Study 2. It is noteworthy that these results were significantlystrongerthanthelatter(F(cid:1)8.93,p(cid:2).01).Consci- observedwithindependentsources(self-reportsofnarcissismand entiousnesswaspositivelyrelatedtoself-butnototherratingsof otherreportsofthecriteria). bothcontextualandtaskperformance.3 Oneinconsistencyintheresultswasthat,inStudy1,narcissism was positively related to other reported leadership, whereas in Differential Effect of Narcissism on Other Ratings of Study 2, narcissism, as hypothesized, was negatively related to Task Versus Contextual Performance other reported leadership. In hindsight, these results appear to be consistent with previous theorizing in the narcissism literature. In H-5, we predicted that narcissism would be more strongly Specifically, Robins and Beer (2001) hypothesize that, because negativelyrelatedtoother(supervisor)reportsofcontextualthan narcissists emphasize immediate personal gain and self- totaskperformance.Inamultivariateregressionpredictingother aggrandizementoverlong-termrelationships,narcissismhasshort- reports of task and contextual performance, the results revealed termbenefitsbutlong-termcosts.Becausemostoftheotherraters that narcissism significantly negatively predicted contextual per- inStudy1werefellowclassmatesandteammembers,itislikely formance ((cid:6)ˆ (cid:1) (cid:3).25, p (cid:2) .01), whereas it did not predict task performance((cid:6)ˆ (cid:1)(cid:3).11,ns).Moreover,thesecoefficientestimates they had known the individuals for a shorter period of time than were significantly different (F (cid:1) 5.65, p (cid:2).01). Thus, H-5 was the other raters in Study 2. Thus, the more negative leadership perceptions others held of narcissists in Study 2 might be ex- supportedbytheresults. Discussion 2As in Study 1, we repeated the analyses with the overall narcissism factor,relatingthefournarcissismdimensionstothefourStudy2criteria. Resultspresentedinthismanuscriptsuggestthatnarcissismmay The results indicated that, when entered into multivariate regressions differentially influence self- and other perceptions of behavior as together, none of the narcissism dimensions significantly predicted self- reflected in four important organizational criteria. Specifically, ratings of the four criteria. However, when predicting other/supervisor controlling for the Big Five traits, narcissism had different asso- reports, one narcissism dimension—self-absorption/self-admiration—sig- ciationswithself-andotherratingsofanumberofcriteria(lead- nificantlynegativelypredictedthreeofthecriteria:leadership,contextual ership,workplacedeviance,contextualperformance,andtaskper- performance,andtaskperformance. formance), albeit the specific patterns of relationships varied by 3In both studies, we also collected “significant other” measures of thecriterionandbythesourceoftheratings.Overall,despitethe narcissism, whereby 1-2 individuals who knew the participants well paucity of organizational empirical research on the topic (for (spouseorpartner,familymember,closefriend)ratedparticipantsusing exceptions, see Penney & Spector, 2002; Soyer et al., 1999), it thesameversionoftheNPI(withappropriatemodificationininstructions) appearsthatnarcissismmayberelevanttounderstandingratingsof as participants used to evaluate themselves. In order to investigate the relativeeffectoftheotherreportsofnarcissism,weenteredtheotherreport severalimportantorganizationalbehaviors.Turningtotheresults (measuringnarcissismoverall)intothemultivariateregressioncontrolling more specifically, two key findings are particularly worthy of fortheoverallself-reportmeasureofnarcissism.Acrossallcriteriainboth discussion. studies,whenaddingtheothermeasureofnarcissismtothemultivariate First, narcissism was generally negatively related to other/su- regressionmodelincludingtheself-reportofoverallnarcissism,theother pervisorreportsofthecriteria.Specifically,thoughnarcissismwas report was associated with neither self- nor other reports of the criteria. notnegativelyrelatedtootherreportsofleadershipinStudy1,nor Thus,itdoesnotappearthatsignificantothers’perceptionsofparticipants’ totaskperformanceinStudy2,itwasnegativelyrelatedtoother narcissismaddedbeyondself-reports. NARCISSISM 771 Table5 MultivariateRegressionsPredictingSelf-and(Other)SupervisorRatingsofWorkplaceDeviance (Study2) Self-Rating(S) OtherRating(O) S-O Difference Partial Partial IndependentVariable Bˆ (cid:5)2 (cid:6)ˆ Bˆ (cid:5)2 (cid:6)ˆ Wilks’(cid:2)(F) Neuroticism .00 .00 (cid:3).02 .06† .03† .15† 1.55 Extraversion .07 .01 .09 .04 .01 .11 1.01 OpennesstoExperience (cid:3).14* .05* (cid:3).20** (cid:3).06† .01† (cid:3).14† 3.80* Agreeableness (cid:3).17* .04* (cid:3).19* .01 .00 .01 2.89† Conscientiousness (cid:3).23** .07** (cid:3).32** (cid:3).02 .00 (cid:3).06 4.76** Narcissism .37† .02† .13† .38** .06** .24** 4.63** Note. Bˆ (cid:1)unstandardizedregressioncoefficientfrommultivariateregression.Partial(cid:5)2(cid:1)uniquevariance explainedbyindependentvariable.Wilks’(cid:2)(cid:1)differencebetweencoefficientestimatesforself-andotherrating (distributedasF-statistic).(cid:6)ˆ (cid:1)standardizedregressioncoefficientfromunivariateOLSregression. †p(cid:2).10. *p(cid:2).05. **p(cid:2).01. plainedbylengthanddepthofknowledge.Consideringtheroleof indeed,acrossthefivetraitstheuncorrectedcorrelationsinTable timeandexperienceinmodelsofnarcissismmaybeaninteresting 3differfromHurtzandDonovan’scorrespondingcorrelationsby areaforfutureresearch. only, on average, .05. Whether subsequent studies would find A second key finding was that, with a sole exception (task higher correlations for the Big Five relative to narcissism is a performance), across the criteria and the studies, narcissism was questionthatshouldbeaddressedinfutureresearch. relatedtoanenhancedself-viewinthatthemultivariateregression Itisnoteworthythat,alongwithnarcissism,bothOpennessand resultsrevealedthatnarcissismmorepositivelypredicted(orinthe Conscientiousness were more strongly associated with self- than caseofworkplacedeviance,lesspositivelypredicted)self-reports other ratings of the criteria. Frankly, we find the results for ofthecriteriathanotherreports.Thus,consistentwiththehypoth- Openness befuddling. Although research suggests that Openness eses,narcissismappearstoreflectagrandioseself-view. does predict leadership (Judge et al., 2002), we are aware of no OnemightwonderwhytheBigFivetraitswerenotsignificant research suggesting that Openness leads to self-enhancement in predictors of task and contextual performance. Although the cor- leadershipratingsorevaluationsofothercriteria.Thus,itremains relations reported in Table 3 are not strong, they are consistent tobeseeniffutureresearchreplicatesthisresult,andcanprovide withpriorresearchonthevalidityoftheBigFivetraits.Specifi- aconceptualrationalefortherelationship.TheresultsforConsci- cally, the correlation between Conscientiousness and task perfor- entiousnessareeasiertoexplain.Conscientiousnessisamongthe mance in this study is not strong (ˆr (cid:1) .11). However, it is very moresociallydesirableBigFivetraits(McFarland&Ryan,2000; close to the average uncorrected correlation between direct mea- Sto¨ber, 2001), and conscientious individuals are more likely to suresofConscientiousnessandperformance(¯r(cid:1).14)reportedby engageinself-deception(Lee&Klein,2002;Martocchio&Judge, HurtzandDonovan(2000).Astheseauthorsnote,thevaliditiesof 1997; Sto¨ber, Dette, & Musch, 2002). Thus, Conscientiousness theBigFivetraits“tendtobelowtomoderateinmagnitude”(p. maybemorestronglyrelatedtoself-thanotherratingsofcriteria 876).Ourresultsareconsistentwiththeirmeta-analyticfindings; duetoself-deception.Onemightwonderwhetherthepsycholog- Table6 MultivariateRegressionsPredictingSelf-andOther(Supervisor)RatingsofContextual Performance(Study2) Self-Rating(S) OtherRating(O) S-O Difference Partial Partial IndependentVariable Bˆ (cid:5)2 (cid:6)ˆ Bˆ (cid:5)2 (cid:6)ˆ Wilks’(cid:2)(F) Neuroticism (cid:3).01 .00 (cid:3).01 (cid:3).11 .01 (cid:3).11 0.90 Extraversion .00 .00 (cid:3).01 .01 .00 .01 0.01 OpennesstoExperience .18** .11** .25** .19* .04* .17* 8.93** Agreeableness .12* .04* .19** (cid:3).02 .00 (cid:3).02 2.74 Conscientiousness .42** .33** .56** .10 .01 .09 29.35** Narcissism .17 .01 .05 (cid:3).99** .07** (cid:3).25** 5.53** Note. Bˆ (cid:1)unstandardizedregressioncoefficientfrommultivariateregression.Partial(cid:5)2(cid:1)uniquevariance explainedbyindependentvariable.Wilks’(cid:2)(cid:1)differencebetweencoefficientestimatesforself-andotherrating (distributedasF-statistic).(cid:6)ˆ (cid:1)standardizedregressioncoefficientfromunivariateOLSregression. †p(cid:2).10. *p(cid:2).05. **p(cid:2).01.