Are leadership constructs really independent?JensRowoldandLarsBorgmannCentAre leadership - PDF document

Areleadershipconstructsreallyindependent?JensRowoldandLarsBorgmannCenterforContinuingEducation,TUDortmundUniversity,Dortmund,GermanyPurpose–Theaimofthisstudyistocontributetotheconstructvalidityofleadershipstyles.Although Thecurrentissueandfulltextarchiveofthisjournalisavailableatwww.emeraldinsight.com/0143-7739.htm Leadership&OrganizationDevelopmentJournalVol.34No.1,2013pp.20-43EmeraldGroupPublishingLimited ThisresearchwassupportedbytheGermanResearchCouncil(DFG,No.Ro3058/2-1,PrincipalInvestigator:JensRowold).TheassistanceofNicoleDriemeyer,BarbaravonKleist,StefanHessmer,MarkoMuhlena,MareikePfennig,VerenaHehn,KathrinStaufenbiel,HannaFerdinandandStephanieBarthelwithdatacollectionisgratefullyacknowledged. 20LODJ34,1 stronglyandpositivelyrelatedtotransactionalleadership(e.g.Bycioetal.,1995;TepperandPercy,1994).Theseconvergenceswithregardtocontentandempiricalrelationshipsexistnotonlywithinoneleadershiptheory,butalsobetweenleadershiptheories.CampbellandFiske(1959)arguethatthevalidityofaconstructcanbequestionablegivenhighcorrelationswithotherconstructsfromwhichitwasintendedtodiffer.Morespecifically,thestrongpositiverelationshipsaffecttheconvergentanddiscriminantvalidityofthedifferentleadershipconstructs.CampbellandFiske(1959)describeconvergentvalidityasthedegreetowhichmultiplemeasuresofthesameconceptareinagreement.AccordingtoNunnallyandBernstein(1967),measuresshouldhaveconvergentanddiscriminantvalidity.Discriminantvalidityisthedegreetowhichmeasuresofdifferentconceptsaredistinct,andisindicatedbylowcorrelationsbetweenthemeasureofinterestandothermeasuresthataresupposedlynotmeasuringthesameconstruct(HeelerandRay,1972).Forthepurposeofthepresentstudy,discriminantvalidityisoperationalizedasasmall|24(Cohen,1988)correlationbetweenconstructs,whilemedium(i.e.0.24orlarge(||37)correlationsindicateconvergenceoftherespectiveleadershipmeasures.Todate,thediscriminantvalidityofthedifferentleadershipconstructshasnotbeenestablishedappropriately.Asaconsequence,researcherscannotrelyoncompletelyvalidatedtheoriesonleadership.Moreover,thisleadstothequestionofwhethertherespectiveleadershiptheoriesmightincludeconstructsthatsharecommonelementsorarebaseduponcertaindimensionsoftheoreticaloverlap.Morecritically,itmightbespeculatedthattheseconstructswouldbepartiallyredundantbecausetheunderlyingprocesses(e.g.motivatingfollowers)areatleastpartiallythesame.However,withintheoreticalandempiricalliteratureonleadership,thesecriticalquestionshavebeenneglectedsofar.Ingeneral,theconstructvalidityofanygivenconstructisacriticalelementofthescientificapproach.Despitetheimportanceofthisaspectofvalidity,nosystematicresearchonthisissueinthefieldofleadershipyetexists.Consequently,expertsinthefieldofleadershiphavebeencallingforresearchaddressingthisissue(Graenetal.,2010;HouseandAditya,1997;Judgeetal.,2004;Sashkin,2004;Yukl,1999,2002).Thus,thepresentworkaimedatclosingthisgapandcontributingtothequestionoftheconstructvalidityofleadershipconstructs.Thepresentstudyisamongthefirsttoinvestigatetheconvergentanddiscriminantvalidityfordifferentleadershipconstructsacrossleadershiptheories.First,severalmajorleadershiptheoriescurrentlybeingextensivelyresearchedarebrieflydiscussed.Problemswiththeconstructvalidityofthesetheoriesarehighlighted.Next,limitationsoftheempiricalliteraturefocussingontheseleadershiptheoriesarepresented.Inordertoovercometheselimitations,anempiricalstudywasconductedsothatforthefirsttime,resultsregardingtheconvergentanddiscriminantvalidityofmultipleleadershipconstructsfromseveralleadershiptheorieswereavailable.Theanalysisstrategythatformedthecoreofthepresentstudy(multi(MTMM)analysis,basedonstructuralequationmodeling(SEM))allowedforconsiderableconfidenceinthevaliditycoefficientsobtainedandtheconclusionsthatcanbedrawnfromtheresults.Insum,thepresentstudyisthefirstsystematicresearchefforttoexplicitlytesttheconstructvalidityoffivetheoreticallydistinctleadershipconstructs.LeadershiptheoriesandconstructsTransformationalandtransactionalleadershipTheconstructsoftransformationalandtransactionalleadershipwerefirstdevelopedbyBurns(1978).Subsequently,Bass(1985)builtonBurns’workanddevelopedthefull Leadershipconstructs rangeleadershiptheory(FRLT),includingseveralsubconstructsoftransformationalandtransactionalleadership.Currently,transformationalandtransactionalleadershiprepresentthemostresearchedconstructsofleadershiptoday(Antonakisetal.,2003;AvolioandBass,2002;AvolioandYammarino,2002;JudgeandPiccolo,2004;WaldmanandYammarino,1999).Transformationalleadersmotivatetheirfollowerswithapositive,value-basedvisionofthefuture.Followerstrustintheirleader’svisionandaremotivatedtoperformbeyondexpectations(Bass,1985;Podsakoffetal.,1996;ShinandZhou,2003).Onefacetoftransformationalleadershipreferstoindividualizedconsideration,wheretheleadercarefullyevaluates–andactsupon–his/herfollowers’needs(AvolioandBass,1995).Asfortransactionalleadership,thisleadershipstyleisbasedonclearlydefinedtransactionsbetweenleaderandled(Bass,1985).Meta-analyticalempiricalevidenceemergedsupportingthecriterion-orientedvalidityoftransformationalandtransactionalleadershipwithregardtobothsubjective(Dumdumetal.,2002;JudgeandPiccolo,2004;Loweetal.,1996)andobjective(Barlingetal.,1996;Tosietal.,2004)performance.Forexample,inasampleofemployeesandleadersfromaGermanpublictransportcompany,itwasfoundthattransformationalleadershipwaspositivelyrelatedtobranch-levelprofit,aswellastofollowers’jobsatisfaction(RowoldandHeinitz,2007).However,theconstructvalidityofthetransformational-transactionalleadershiptheoryisproblematic.Forexample,onemeta-analysis(JudgeandPiccolo,2004)reportedthecorrectedcorrelationbetweentransformationalandtransactionalleadership(0.80).Thisrelationshipcouldbeinterpretedasalackofdiscriminantvalidity,becausefromtheory,theseconstructswerehypothesizedtobedistinctconstructs(Bass,1985).Furthermore,BrownandKeeping(2005)confirmedthatthemeasurementoftransformationalleadershipishighlyinfluencedbytheaffectratersfeeltowardtheirleader.Thus,thetheoryoftransformationalleadershiphasbeencriticizedforinadequateconstructvalidity(Bycioetal.,1995;HouseandAditya,1997;etal.,1997;Tejedaetal.,2001;TepperandPercy,1994;Yukl,1999).ConsiderationandinitiatingstructureAfterempiricalresearchfailedtoprovidestrongsupportfortherelationshipbetweenleaders’stablecharacteristics(e.g.personalitytraits)andcriteriaofeffectiveleadershipinthe1940s(e.g.Stogdill,1948),leadershipresearchlookedforbehavior-relatedconstructsofeffectiveleadership.Asaconsequence,theleadershipstylesofconsiderationandinitiatingstructureemergedfromnumerousstudiesconductedattheUniversityofMichiganInstituteforSocialResearch(Likert,1955,1961),andtheOhioStateUniversityinthe1940sand1950s(Fleishman,1973).Putsimply,considerationcharacterizesfollower-centeredleadershipbehaviorwheretheleadertakesintoaccounthis/herfollowers’needsandabilities.Incontrast,initiatingstructurereferstostructuringworktasksfortherespectivesubordinatesandsettingdeadlines(Fleishman,1953;SeltzerandNumeroff,1988).Arecentmeta-analysis(Judgeetal.2004)thatwentbeyondlimitationsofearlierresearch(e.g.Stogdill,1948)revealedthatbothconsiderationandinitiatingstructurearepositivelyandstronglyrelatedtosubjectiveindicatorsofperformance.Morespecifically,considerationismoreassociatedwithfollowers’levelsofjobsatisfactionandmotivation,whileinitiatingstructureismorerelatedtovariouscriteriaofperformance.Interestingly,thismeta-analysisalsofoundthatinsomecasesthecorrectedcorrelationsbetweenconsiderationandinitiatingstructurewereashighas0.46.Apparently,thisstrongempiricalconvergenceisnotinlinewiththetheoreticaldistinctivenessofthetwoconstructs. 22LODJ34,1 Severalscholarsnotedaclosesimilaritybetweenconsiderationandindividualizedconsideration,afacetoftransformationalleadership(HouseandAditya,1997;Yukl,2002).Bothleadershipstylesarehighlyfollower-oriented,activeleadershipbehaviors.Inlinewiththisidea,severalempiricalstudiesrevealedanoverlapbetweentheconstructsofconsiderationandtransformationalleadership.Forexample,basedondatafromUSmanagers,SeltzerandBass(1990)reportedhighlypositivecorrelationsbetweensubscalesoftransformationalleadershipandconsideration(0.470.69;0.01).Also,GeyerandSteyrer(1994)foundapositivecorrelationbetweentransformationalleadershipandinitiatingstructure.TheseresultswerereplicatedbyKeller(2006)inasampleofresearchanddevelopmentprojectteammembersintheUSA.Insum,ithasbeenshownthatinitiatingstructureandconsiderationhavesomeoverlapwithtransformationalleadership,althoughmoretheoreticalandempiricalresearchseemsnecessarytoclarifythisissue.Fromatheoreticalperspective,transactionalleadershipandinitiatingstructurehavesomedegreeofoverlap,too.Bothleadershipstylesrefertoshort-term,work-relatedbehaviorssuchasgoalsetting.However,toourknowledge,onlyoneempiricalstudyhasasyetexploredthisissue.BasedonasampleofemployeesofaGermanenergysupplycompany,RowoldandKersting(2008)reportedacorrelationof0.360.01)betweentransactionalleadershipandinitiatingstructure.Thus,giventhisscarcityofempiricalresearch,furtherexplorationregardingthisissueseemstobewarranted.Leader-Member-Exchange(LMX)LMXwasdefinedasapositive,mutuallytrustfulrelationshipbetweenleadersandled(GraenandUhl-Bien,1995;LidenandGraen,1980).Thisleadershiptheorywasdesignedinordertoaddresssomeshortcomingsofearlierleadershiptheories.Forexample,whilepriortheoriesimplicitlyassumedthatleadersexhibitidenticalbehaviortowardeachoftheirfollowers,LMXexplicatedforthefirsttimethatleaderstreateachfollowerdifferently.Morespecifically,ineachteam,leadersdistinguishbetweenfollowersfromtheirin-groupandtheirout-group.Thus,leadersformcloserelationswithfollowersfromtheirin-group(andexchangemoreresources),whiletheyhavemoreformalrelationshipwithmembersfromtheout-group(and,consequently,exchangelessresources).SeveralempiricalstudiesfoundpositiverelationshipsbetweenLMXandperformance(e.g.Erdoganetal.,2004).Also,aliteraturereviewrevealedthatLMXhaspositiverelationshipswithseveralperformanceindicatorsontheindividuallevel(e.g.jobsatisfaction,jobperformance;cp.GerstnerandDay,1997).Fromatheoreticalperspective,bothtransformationalleadershipandLMXelicittrustinfollowers.Thus,itmightbearguedthatthesetwoleadershipconstructsrelyatleastpartiallyonidenticalprocesses(seeGraenetal.,2010).Inlinewiththisidea,empiricalresearchfoundstrongpositivecorrelationsbetweentheconstructsofLMXandtransformationalleadership.BasedonastudyofvariousindustriesinChina,Wangetal.(2005)reportedacorrelationof0.71(0.01)betweenLMXandtransformationalleadership.ComparableresultswereobtainedbyHowellandHall-Merenda(1999),basedonasampleofbankemployees(0.53;Interestingly,thesamestudyreportedapositiverelationshipbetweenLMXandafacetoftransactionalleadership,namely,contingentreward(0.05).ItmightbearguedfromtheorythatbothLMXandtransactionalleadershiprelyonapositive,mutuallytrustfulrelationshipbetweenleaderandled(Bass,1985;Graen,1995). Leadershipconstructs However,besidestheHowellandHall-Merenda(1999)study,moreempiricalresearchseemsnecessarybeforefirmconclusionscanbedrawn.Also,noempiricalresearchyetexiststhatexplorestherelationshipsbetweenLMX,andconsiderationandinitiatingstructure,respectively.TheoreticaloverlapbetweenleadershiptheoriesOverall,theconvergencesoftheleadershipstylesdescribedaboveindicatemeaningfultheoreticaloverlapandimplytheexistenceofunderlyingdimensions.Thereviewofintegrativeleadershipliterature(Graenetal.,2010;HouseandAditya,1997;Yukl,1999;Yukl,2002;AntonakisandHouse,2002;Sashkin,2004;HuntandConger,1999)confirmssimilaritiesanddifferencesbetweenleadershipconstructsaswell.Itseemsthatthereare“coredimensions”whichareeitherexplicitlyorimplicitlypartofeveryleadershiptheory.ThesedimensionsoftheoreticaloverlaparesummarizedinTableIandarediscussedinturn.Itshouldbenotedthatthesedimensionsbynomeansrepresentanexhaustiveormutuallyexclusivelist.TableIillustratesthatleadershipconstructshavemoremeaningfulsimilaritiesthandifferences.Thismakessensebecauseeachleadershipconstructwasdevelopedforthesamepurposes,namelyaccountingforleaders’behaviorsatworkandexplainingvarianceinfollowers’motivation.Thisconvergenceinsubstanceoftheleadershipconstructshasbeenignoredsofar.Thepresentstudyhypothesizesconceptualoverlapbetweeneachofthesixleadershipconstructsanddescribestheconvergentanddiscriminantvaliditiesoftherespectiveleadershipconstructs.LimitationsofleadershipresearchThereareafewstudiesthatinvestigatedtheconstructvalidityofaleadershipstyleintheformofdiscriminantandconvergentvalidity.Alban-MetcalfeandAlimo-Metcalfe(2000),Alimo-MetcalfeandAlban-Metcalfe(2001)investigatedtheconstructvalidityoftransformationalleadership.Thestudyprovidedevidencefortheconvergentanddiscriminantvalidityofdifferentaspectsoftransformationalleadership.RaffertyandGriffin(2004)providedinitialsupportforthediscriminantvalidityofthefivesubdimensionsoftransformationalleadershiptheyidentified.OtherstudieshaveinvestigatedtheconstructanddiscriminantvalidityoftheMLQ.Whileseverallarge-samplestudiesfoundsupportforadequatefactorialvalidity(e.g.Antonakisetal. LeadershipconstructsDimensionsofleader’sbehaviorTFTALFCISLMXLevelofactivityþþþþþþBuildingtrustinfollowersþþþRolemodelingþþþExpressionofemotionsþþControllingfollowersþþMotivatingfollowersintrinsicallyþþFocusonstrategyþþFollowers’workfacilitationandfeedbackþþTF,transformationalleadership;TA,transactionalleadership;LF,laissez-faireC,consideration;IS,initiatingstructure;LMX,Leader-Member-Exchange;,theoreticallyrelevantforleadershipconstruct;,irrelevanttoleadershipconstruct TableI.Dimensionsofpotentialtheoreticaloverlapbetweenleadershipconstructs 24LODJ34,1 2003),otherempiricalstudiesfoundvaryingmodelsoftheMLQ-5Xfactorstructure(e.g.GeyerandSteyrer,1994;Tejedaetal.,2001),questioningthefactorialvalidityoftheMLQ-5X.Forexample,DenHartogandVanMuijen(1997)testedthefactorstructureoftheMLQandreportedastructurecomprisingatransformational,atransactional,andalaissez-fairefactor,butnoseparatedimensionsoftransformationalandtransactionalleadership.Duetotheoreticalargumentsandempiricalresults,Carless(1998)criticizedtheweakdiscriminantvalidityandtheuseoftheMLQ-5X.Thesestudieshaveincommon,thattheyfocusononlyoneleadershipparadigm,whereastheconvergencesbetweendifferentleadershipstylesandtheoriesremainunclear.Thisbriefreviewofpriorresearchreportinginformationabouttheconvergentanddiscriminantvalidityofleadershipconstructsrevealedthatinvirtuallyallempiricalstudiesconductedsofar,therespectiverelationshipsbetweenleadershipconstructscouldbeinterpretedasconvergent.Thisisbothsurprisingandproblematic,becausefromtheory,theseconstructswereexplicitlydesignedtobeindependentfromeachother(SeltzerandBass,1990).Despitetheseconvergencesinsubstance,researchondiscriminantandconvergentvalidityfocussedonlyononegivenleadershiptheory.Asdemonstratedabove,fewempiricalstudiesexistthathintatthepossibilitythattheseleadershipconstructsareempiricallyinterrelated.Anotherlimitationoftheleadershipliteratureisthatforsomeoftheconvergences(e.g.relationshipbetweenLMXandconsideration),noempiricalresearchyetexists.Asaconsequence,anygivenleadershipconstructcannotbecomparedwithalltheotherconstructsintermsofconvergentanddiscriminantvalidity.Thislimitsourunderstandingofthenomologicalnetworkofleadershipconstructs.Clearly,considerablymoreempiricalresearchisnecessarytocompleteourknowledgeaboutboththeconvergentanddiscriminantvaliditiesofleadershipconstructsaswellasthenomologicalnetworkofleadership.Onepossibleexplanationforthehithertoobservedpositiveinterrelationshipsbetweenleadershipconstructsisthattheseconstructswereassessedbythesameratingsource.Morespecifically,in“thetypicalleadershipstudy”(Hunteretal.,2007),thevariousleadershipconstructs(e.g.transformationalandtransactionalleadership)areassessedfromthesamesource(Avolioetal.,1991).Thatis,thesubordinateassessestheseleadershipbehaviorsofhis/herrespectivesupervisor.Thus,theintercorrelationsbetweenleadershipconstructsmaybeinflatedduetosame-sourcebias.Thepresentstudyaimedataddressingthisgap.Consequently,theleadershipconstructswereassessedbymultipleratingperspectivesinordertoexcludesame-sourcebiasasapotentialexplanationfortheproblematicconvergentvaliditiesoftheleadershipconstructs.TheMTMMapproachSincethelate1950s,MTMMhasbeenappliedtotestconstructvalidity.Especially,theconvergentanddiscriminantvalidityofconstructshavebeenexplored(CampbellandFiske,1959).ThetermtraitisoftenusedasasynonymforconstructintheMTMMliterature.InMTMMresearch,severalconstructsareassessedbyseveralmethods.Consequently,theinterrelationshipsbetweenthestudy’sconstructsareutilizedtoanalyzeconvergentanddiscriminantvalidity.Forexample,regardlessofthemethodapplied,thestudyvariablesthatrepresentonespecificconstructshouldbehighlyconvergent.Incontrast,differentconstructsshouldnotbeintercorrelated,regardlessofthemethod,thusyieldingsupportfordiscriminantvalidity.TheadvantageofMTMM Leadershipconstructs researchisthatduetotheintegrationofdifferentassessmentmethodologies,accurateestimationsofconvergentanddiscriminantvaliditiescanbeobtained.MTMManalysesimproveresearchbecauseamorethoroughunderstandingofthepsychometriceffectsofonerespectiveconstructisgiven.Also,sinceseveralconstructsareincludedinoneMTMManalyses,theinterrelationshipsbetweenconstructscanbeestimatedwithmethodologicalrigor.SinceMTMManalyseshavetheseadvantages,MTMMhasbeenappliedtosupporttheconstructvalidityofconstructsfromvariousfieldssuchaspersonality(BiesanzandWest,2004)andjobperformance(Conway,1996).Surprisingly,MTMMhasbeenappliedveryrarelytothefieldofleadership.AnexampleisthestudyofRowoldandHeinitz(2007)conductingacorrelation-basedMTMM.Theyfoundtransformationalleadershiptobeconvergenttocharismaticleadership,asassessedwiththeCongerandKanungoScales(CongerandKanungo,1998).Theyalsoreportedanunexpectedconvergencebetweentransactionalandcharismaticleadership.However,severallimitationsofthisstudyshouldbenoted.First,RowoldandHeinitz’(2007)analysesreliedonsingle-source(i.e.subordinate’sperspective)data.Theyproposedthattwosubordinateswithonecommonsupervisorcouldbeviewedastwoindependentmethods.However,itiswellknownfromleadershipresearchthatsubordinatesworkinginoneteamareheavilyinterdependent(Graen,1995).Also,relyingononesinglemethodforMTMManalysesseemstorepresentanoversimplification(Podsakoffetal.,2003).Third,theMTMMreliedonanalysesofzero-ordercorrelations.Severallimitationsofthisapproachhavebeendiscussedinthemethodologicalliterature(Marsh,1993;Marshetal.,1992).Thecriteriaarebasedonthepremisesthatnocorrelationsbetweentraitandmethodfactorsexist,alltraitsareequallyinfluencedbymethodfactorsandthemethodfactorsareuncorrelated(SchmittandStults,1986).Additionally,noprecisestandardsareprovidedfordetermininghowwellthecriteriaaremet(Bagozzietal.,1991).AnadvancementintheMTMMapproachthataddressestheproblemsdescribedabove,hasbeentheapplicationofSEMtoMTMM.TheuseofSEMprovidesamorepowerfulmethodforaddressingconstructvaliditythantheapproachesusedbefore.Confirmatoryfactoranalysis(CFA)usingSEMmakesfewerassumptionsandprovidesmorediagnosticinformationaboutreliabilityandvaliditythantheapproachofCampbellandFiske(1959).Incontrasttotheclassicalprocedure,theCFAmodelallowsmethodstoaffectmeasuresoftraitstodifferentdegreesandtocorrelatefreelyamongthemselves.SEMprovidesadvantageslike:measuresoftheoveralldegreeoffitareprovided,usefulinformationissuppliedonwhetherandhowwellconvergentanddiscriminantvalidityareachieved(i.e.throughdifferencetests,thesizeoffactorloadingsfortraits,andtheestimatesfortraitcorrelations).UsingtheCFAapproach,itispossibletopartitionthevarianceintotrait,method,anderrorcomponents(i.e.throughsquaredfactorloadingsanderrorvariance)(Bagozzietal.,1991).Thus,giventhisscarcityofMTMMleadershipresearchaswellasitsmethodologicallimitations,andgiventhelimitationsofleadershipresearchsummarizedabove,additionalMTMMresearchfocussingonleadershipconstructsseemswarranted.StudygoalsandhypothesisThepresentstudyexaminedtheconvergentanddiscriminantvalidityoftheleadershipconstructsoftransformationalandtransactionalleadership,considerationandinitiatingstructure,andLMX.Asnotedabove,onepossibleexplanationforthestrongpositiveintercorrelationsoftheseleadershipconstructsmaybethatempiricalresearch 26LODJ34,1 assessedtheseconstructswiththesamemethod.Thus,utilizingtheMTMMapproach,thepresentstudytestedtheconstructvalidityofleadershipconstructswithmethodologicalrigorbyincludingsubordinateandself-ratingsoftherespectiveleadershipconstruct.Formorepreciseestimatesoftheinterrelationshipsbetweenleadershipconstructs,SEMwasutilized.Foramorethoroughunderstandingofthevalidityofeachoftheseleadershipconstructs,theamountofvarianceduetotheconstructitself,themethodsapplied,anderrorwascalculated.Itshouldbenotedthatforseveraloftheleadershipconstructsincludedinthepresentstudy,noresearchexistedtestingtheirrespectiveinterrelationshipswithcertainotherconstructs(e.g.LMXandinitiatingstructure).Thus,thepresentstudyaimedatprovidinginformationthatcouldhelptocompleteourunderstandingofthenomologicalnetworkofleadershipconstructs.ItwasexpectedthatintheMTMManalysis,theleadershipconstructswouldexhibitdiscriminantvalidity.Thereasonforthisliesinthefactthateachoftheconstructswereexplicitlyconceptualizedasdiscriminantconstructs.Forexample,Bass(1985)developedhistheoryoftransformationalandtransactionalleadershipontheideathattheseleadershipconstructswouldbenecessaryforourcurrentunderstandingofeffectiveleadershipinorganizations,beyondthehithertoestablishedconstructsofconsiderationandinitiatingstructure.Also,Bass(1985)proposedtheconstructoftransformationalleadershiptobeindependentoftransactionalleadership.Hearguedthatthefrequencyofoneoftheseconstructswouldnotdependonthefrequencyoftherespectiveotherconstructbutonsituational(e.g.work-related)demands:Theconstructsoftransformationalandtransactionalleadership,consideration,initiatingstructure,andLMXhavediscriminantvalidity.MethodsSamplesandproceduresTheparticipantsofthepresentstudywererecruitedviaanewspaperarticlepublishedinamajorbusinessmagazineinGermany.Thisarticleincludedalinktotheonlinesurvey.Eachleaderwhocompletedthesurveyreceivedastandardizedreportabouthis/herleadershipprofile(i.e.strengthsandweaknesses).Also,theleaderwastoinviteasubordinatetotakepartinthissurvey.Thisprocedureyieldedatotalsampleof148pairsofleadersandtheirrespectivesubordinates.Leaders’meanagewas44.55years(SD8.62);77.2percentweremale,22.8percentwerefemale.Theleadershadameanjobtenureof10.79years(SD7.59).Altogether,11.3percenthadajuniorhighschool,29.0percentahighschoolgraduation,and60.0percentauniversitydiploma.Theleadersampleconsistedof76.6percentmiddleortopmanagersand23.4percentfirst-levelsupervisors.Themeanageofthesubordinateswas37.9years11.90)andtheaveragetenurewas16.59years(SD9.34).Asforgender,49.7percentofthesubordinatesweremaleand50.3%werefemale.41.1.2%hadajuniorhighschool,19.4percentahighschoolgraduation,and39.5percentauniversitydiploma.MeasuresTransactionalandtransformationalleadership.Fortheassessmentoftransactionalleadership,fouritemsfromaGermanvalidatedversion(HeinitzandRowold,2007)oftheTransformationalLeadershipInventory(TLI;cf.Podsakoffetal.,1990; Leadershipconstructs Podsakoffetal.,1996)wereutilized(sampleitem:“[]providesmewithpositivefeedbackifIperformwell”).TheTLIwasalsoimplementedtoassesstransformationalleadership(22items,sampleitem:“[]hasinspiringplansforthefuture”).Theinternalconsistencyofthetransformationalleadershipsubscalewas0.88fortheself-ratingand0.96forsubordinateperspective.Thetransactionalleadershipscale’sreliabilitywas0.77forself-ratingand0.91forassessmentbysubordinates.Considerationandinitiatingstructure.Theleadershipstyleofconsiderationwasassessedwith22itemsfromaGermanvalidatedversion(Fittkau-GartheandFittkau,1971)oftheSBDQ(Fleishman,1953)(sampleitem:“[]showsinterestintheindividualwell-beingofhis/hersubordinates”).Initiatingstructurewasassessedwith12itemsfromthesamequestionnaire(sampleitem:“[]assignsspecifictaskstohis/hersubordinates”).TheCronbach’sfortheself-assessmentofconsiderationwas0.82andforthesubordinate’srating0.95.Initiatingstructurehadareliability0.56fortheself-perspectiveandan0.84forthesubordinateperspective..FortheassessmentofLMX,aGermanvalidatedversion(Schyns,2002)ofGraenandUhl-Bien’s(1995)LMXscalewasused(sevenitems,sampleitem:“Itrustmyleaderenoughtodefendhis/herdecisions”).Thereliabilitywas0.68fortheself0.94forthesubordinateperspective.Withtheexceptionofinitiatingstructure,allreliabilitieswerewithintheacceptablerange(0.680.94).Thus,thatthereliabilityofinitiatingstructure(self-assessment)wasslightlybelowthisthresholdshouldbekeptinmindwheninterpretingtheresults.DataanalysisForthepurposeofestablishingthemeasurementmodelandsupportadequatefactorialvalidityoftheinstruments,CFAwereutilized.Wetestedwhetherthemodelwithfiveleadershipconstructs(i.e.targetmodel)fittedthedatareasonablywell.Giventhehighnumberofindicatorsforthefirst-ordermodel,theitemswerecombinedintotwoparcelsperleadershipconstruct(Bandalos,2002;Landisetal.,2000;etal.,2002).Thatis,withinonerespectiveleadershipconstruct,thevariousitemsfortheassessmentofthisconstructformedoneoftwoindicatorsforthisscale.Theseparationofeachtotalscoreintotwoparcelsforeachleadershipstyleresultedin10measures.Theloadingsofthetwoparcelsoftherespectiveleadershipstylewereconstrainedtobeequal(i.e.equivalence);thisprocedureyieldedanover-identifiedmodelwith150degreesoffreedom.Figure1providesanoverviewoftheprinciplesthatwerepartofthestatisticalproceduresdescribedabove.Forthesakeofcomprehensibility,onlythreeleadershipstyles(i.e.transformational,transactional,andlaissez-faireleadership)wereincludedinFigure1.Itshouldbenotedthatintheactualanalyses,allfiveleadershipconstructswereincludedsimultaneously.Toassesstheconditionofmultivariatenormalityofthedata,anomnibustestbasedonSmall’sstatistics(seeLooney,1995)wasperformed.Theresultsshowedasignificantviolationofthemultivariatenormality(211.38,df0.001).Toaccountforthemissingmultivariatenormality,theparametersoftheproposedmodelareestimatedusingtheunweightedleastsquares(ULS)discrepancyfunction(Byrne,2001).ULSwasappliedbecauseitisrobustforusewithdatanotnormallydistributed,withrelativelysmallsamplesizes(Ximenez,2006).Finally,bootstrapping(seeHancockandNevitt,2001)wasusedtogenerateandparameterestimates,standarderrorsofparameterestimates,andsignificancetestsforindividualparameters. 28LODJ34,1 Toassessthemodelfit,severalfitindiceswerecomputed(seeKline,2005).Firstthevalueswerecalculated.Next,thegoodness-of-fit(GFI)andtheadjustedGFI(AGFI)wereutilizedtotestmodelfit.Amongothers,HuandBentler(1999)postulateavalueof0.90asminimumforappropriatefit(seeByrne,2001;Hairetal.,1998;HuandBentler,1995).Thestandardizedrootmeansquareresidual(SRMR)wascalculatedaswell.Valuesbelow0.08indicategoodfit(HuandBentler,1999).Forthepurposeofcomparingrivalingmodels,twocriteriawereutilized.Asafirststep,the-differencetestwasapplied.Thewascalculatedbysubtractingtherespectivemodelfromthatofthefive-factormodel;likewise,dfwascalculatedbysubtractingmodeldffromthatofthefive-factormodel(Kline,2005).However,the-differencetesthasbeencriticizedbecauseitmightbebiasedbysamplesize.Thus,asasecondstep,theAGFI,andtheSRMRwerecomputed.AsdescribedbyCheungandRensvold(2002),theseindicesofdifferencesshouldexceedvaluesof0.01beforedifferencesbetweentherespectivecompetingmodelscanbeestablished.ResultsFactorialinvarianceBeforeconductingtheMTMManalysis,evidenceforfactorialvalidityofthemeasurementmodelandforfactorialinvariancehastobeestablished.Thetwomeasurementmodels,onefortheself-ratingperspectiveandoneforthesubordinateperspective,weretestedseparately.Asdescribedabove,twoparcelsweremodeledasindicatorsforonerespectiveleadershipconstruct,withequivalentloadings.TableIIrevealsthatboththeselfandthesubordinateperspectivemodelshowedaclosefittothedata.Next,factorialinvariancewastested.Factorialinvariancecanassumedifferentforms(SteenkampandBaumgartner,1998;VandenbergandLance,2000).Themostbasicformisconfiguralinvariance.Whetherthespecifiedmodelhasthesamenon-zeroandzerofactorloadingsforthetworatingformsistobetested.Also,allfactorloadingsshouldbesufficientlyhigh.Establishingconfiguralinvariancewouldimplythattheleadershipitemsmeasurethesameleadershipconstructsacrossratingperspectives.Datafromthetworatingperspectiveswerecombinedintoone(multi-group)CFAtotesttheconfiguralinvariance. TF, transformational leadership; TA, transactional leadership; LF, laissez-faireS, self-rating; F, follower’s ratingSelfFollowerTF.STF.STF.FTF.FTA.STA.STA.FTA.FLF.SLF.SLF.FLF.FParcel 1Parcel 1Parcel 1Parcel 1Parcel 1Parcel 1Parcel 2Parcel 2Parcel 2Parcel 2Parcel 2Parcel 2TALF Figure1.OverviewofprinciplesunderlyingtheCTUMmodel Leadershipconstructs AscanbeseenfromTableII,themodelfittedthedatareasonablywell.Also,allfactorloadingswerepositiveandstrong(0.76;SD0.13;minimummaximumThenextformoffactorialinvarianceismetricinvariance,wherethefactorloadingsarerequiredtohavethesameloadingsacrossratingperspectives.AscanbeseenfromTableII,theAGFIwasslightlytoolowandtheSRMRwastoohighforestablishingmetricinvariance.Itshouldbenoted,however,thatconfiguralinvariancewasasufficientconditionforthefollowingMTMManalyses.Insum,theseresultswereinterpretedasevidenceforadequateconfiguralinvariance.MTMManalysesInMTMMresearch,severalSEMapproachesarefeasible(MarshandBailey,1991;etal.,1992).Thesemodelshaveincommonthattheindicatorsrepresentitemsdesignedtoassessonesingletraitwhichismeasuredwithonerespectivemethod.Forexample,oneindicatorisoneparcelofitemsdesignedfortheassessmentoftransformationalleadership,asobservedfromthesubordinateperspective.TheseMTMMmodelsdifferinthewayinterrelationshipsbetweentraitsorbetweenmethodfactorsareallowedornot.First,acorrelatedtraitcorrelatedmethods(CTCM)modelexists.Theindicatorsloadontheirrespectivetraitaswellasontheirrespectivemethodfactors.Also,intheCTCMmodel,thetraitsareallowedtocorrelate,asarethemethodfactors.Next,theuncorrelatedtraitcorrelatedmethod(UTCM)modelisbasicallythesameastheCTCMmodel,withtheexceptionthatthetraitsarenotallowedtocorrelate.Third,thecorrelatedtraituncorrelatedmethods(CTUM)modelisalsoverysimilartotheCTCMmodel,withtheexceptionthatthemethodsarenotallowedtobeintercorrelated.Fourth,intheuncorrelatedtraituncorrelatedmethod(UTUM)model,neitherthetraitsnorthemethodsareallowedtointercorrelate.InapplicationsofthesefourMTMMmodels,ithasoftenbeenobservedthatthesemodelsyieldproblemswithidentifications(e.g.underidentification),noconvergence,ornomeaningfulresults(e.g.correlations1.0;cf.Marshetal.,1992).Thus,inordertoovercometheseproblems,acorrelateduniqueness(CU)modelhasbeenproposed,whereinsteadofmethodfactors,correlationsbetweenerrortermswithinone dfGFIAGFISRMRStep(1):invarianceanalysesIndependencemodel1,244.111900.220.140.20Measurement–subordinateperspective30.00250.980.960.06Measurement–self-perspective30.56250.980.950.02Configuralinvariancemodel33.46600.980.960.04Metricinvariancemodel93.68650.940.890.08Step(2):MTMManalyses149ModeldoesnotconvergeUTCM40.501590.980.970.03CTUM18.811500.990.990.02UTUM56.571600.980.970.0470Modelisnotidentified80ModelisnotidentifiedGFI,goodness-of-fitindex;AGFI,adjustedGFI;SRMR,standardizedrootmeanresidual TableII.Resultsoftheconfirmatoryfactoranalyses 30LODJ34,1 respectivemethodareallowed.TheseCUsrepresentthemethodfactors.AsinCTCM,thetraitsareallowedtocorrelateinthecorrelatedtraits,correlateduniqueness(CTCU)model.Finally,intheuncorrelatedtraits,correlateduniqueness(UTCU)model,thetraitsarenotallowedtobeintercorrelated,whiletheuniquenessshouldbeintercorrelatedwithinonerespectivemethod.ItshouldbenotedthatinMTMMresearch,itiswellknownthateachofthesesixpossiblemodelsmightyieldproblemswithidentification,etc.(Eid,2000;Marshetal.,1992).AsnopriorresearchhadbeenconductedontheissueofMTMMofleadershipstyles,eachofthesesixpossibleMTMMmodelswastested.TheresultsaresummarizedatthebottomofTableII.Itwasfoundthatonlythreemodelsconvergedandyieldedmeaningfulresults,i.e.theUTCM,theCTUM,andtheUTUM.AcloserinspectionoftheabsolutefitindicesofthesethreeMTMMmodelsrevealedthattheCTUMshowedthebestfittothedata.Thatis,theGFIandAGFIwerehighestforthismodel,whiletheSRMSandthe,respectively,werelowestforthismodel.Especially,theAGFIoftheCTUMwas0.02betterthananyothermodel.Thismeetsthecriterionfordistinguishingfitindicesofcompetingmodels(CheungandRensvold,2002).Thus,itwasconcludedthattheCTUMmodelwasmostappropriateforthepresentstudy.Figure1depictsmaincharacteristicsofthismodel.FactorloadingsFurtheranalysesoftheCTUMmodelrevealedinsightintothefactorloadingsofthetraitfactorsaswellasthemethodfactors.Proportionsoftraitvarianceforeachleadershipconstructwerecomputedbysquaringfactorloadings,andthesewereaveragedtoobtainasummaryproportionoftraitvarianceforeachtrait(i.e.leadershipstyle).ThelastrowinTableIIIrevealedthatthelowestvarianceexplainedwasforinitiatingstructure,andthehighestfortransactionalleadership.Insum,allfiveleadershipconstructsexplainedconsiderablevarianceinthedata.Itshouldbenotedthatmanyoftheindicatorshadhighloadingsonthetwomethodfactors,indicatingstrongmethodeffects.Infact,assummarizedinthelastrowofTableIII,15percentofvarianceintheindicatorsofself-ratingswasexplainedbythemethod,while55percentofvarianceintheindicatorsofsubordinateratingswasexplainedbythemethod.Apparently,aconsiderableportionofvariancewasduetothemethodsappliedinthisstudy.VariancecomponentsInordertofurtherexploretheissueofvariance,thevariousvariancesourceswereanalyzed.AssummarizedinTableIV,foreachoftheindicators,theamountofvarianceduetotrait,method,anderrorwascalculated.Thedifferentamountsofvarianceforeachleadershipconstructwerecomputedbysquaringthefactorloadings.Forexample,thetraitvarianceofTFself(parcel1)wascalculatedbysquaringtherespectivefactorloadingof0.96(seeTableIII),resultinginavalueof0.91.TableIVrevealsthatonaverage,thetraitsaccountedfor36percentofthevarianceinthedata.Also,asimilaramountofvarianceinthedatawasexplainedbythetwomethodfactors(i.e.37percent).Interestingly,only10percentofthevarianceinthedatawasduetoerror.Incombination,theseanalysesregardingfactorloadings,sourcesofvariance,andcomponentsofvariancerevealedthatonaverage,boththeleadershipconstructsandthemethodsappliedaccountedforapproximatelyequalamountsofvarianceinthedata.Thus,whiletheconvergentvalidityofthefiveleadership Leadershipconstructs TFTACISLMXSelf-ratingSubordinate-ratingTFself0.96*(0.78*)0.16(0.37)TFsubordinate0.21(0.17)0.89*(0.89*)TAself0.84*(0.86*)0.09(0.10)TAsubordinate0.90*(0.88*)0.03(Cself0.77*(0.89*)0.05(0.31)Csubordinate0.19(0.21)0.84*(0.91*)ISself0.39(0.43)0.06(0.89*)ISsubordinate0.23(0.16)0.89*(0.16)LMXself0.64(0.51)0.51(0.40)LMXsubordinate0.19(0.18)0.91*(0.86*)Varianceexplained0.400.760.370.100.250.150.55Valuesindicatethestandardizedfactorloadingsforparcel1oftherespectivetrait,valuesinbracketsforparcel2.TF,transformationalleadersTA,transactionalleadership;C,consideration;IS,initiatingstructure;LMX,Leader-Member-Exchange.* TableIII.factorloadingsintheCTUMmodel 32LODJ34,1 constructscouldbesupportedbythedata,alsoconsiderablemethodeffectsexistedduetoselfandsubordinateratings.DiscriminantvalidityInordertoexploretheinterrelationshipsofthefiveleadershipstyles(i.e.hypothesizeddiscriminantvalidity),thecorrelationsbetweenthelatentsecond-orderconstructs(i.e.leadershipconstructs)couldbeutilized(seeTableV).Overall,theleadershipstyleswerehighlyinterrelated(mean|0.51).Thus,,i.e.thattheleadershipconstructswouldbediscriminant,hadtoberejected.Althoughthecorrelationsdidnotreachlevelsofstatisticalsignificance,theirabsolutevaluescouldbeinterpretedaslargeeffectsizes(Cohenetal.,2002).DiscussionUtilizingadvancedSEM-basedMTMMmethodology,thepresentstudywasthefirstempiricalworktoexploretheconvergentanddiscriminantvalidityoffiveleadershipconstructs.TheresultsoftheMTMManalysescanbediscussedfromamethodologicalperspective:Itwasfoundthatapproximatelyequalamountsofvarianceintheleadershipdatawereduetothemethodsapplied(i.e.selfandsubordinateratings)andtotheconstructs(i.e.fiveleadershipconstructs).Forthefirsttime,duetothemethodologicalrigoroftheMTMManalysesapplied,therelativecontributionof MeasuresTraitMethodErrorTFself0.91(0.61)0.03(0.13)0.01(0.06)TAself0.71(0.73)0.01(0.01)0.14(0.12)Cself0.62(0.78)0.00(0.09)0.07(0.02)ISself0.15(0.18)0.00(0.80)0.15(0.00)LMXself0.41(0.26)0.26(0.16)0.05(0.13)TFsubordinate0.04(0.03)0.79(0.79)0.09(0.14)TAsubordinate0.82(0.77)0.00(0.00)0.09(0.11)Csubordinate0.03(0.04)0.71(0.83)0.19(0.07)ISsubordinate0.05(0.03)0.50(0.79)0.14(0.11)LMXsubordinate0.03(0.03)0.84(0.74)0.10(0.17)Mean0.360.370.10Valuesindicatethestandardizedfactorloadingsforparcel1oftherespectivetrait,valuesinbracketsforparcel2.TF,transformationalleadership;TA,transactionalleadership;C,consideration;IS,initiatingstructure;LMX,Leader-Member-Exchange TableIV.SourcesofvariancesintheCTUMmodel FactorTFTACISLMXTransformationalleadership(TF)Transactionalleadership(TA)0.51Consideration(C)0.410.29Initiatingstructure(IS)0.860.660.50Leader-Member-Exchange(LMX)0.730.450.370.48Allcorrelationswerenotsignificant, TableV.EstimatedcorrelationsamongtraitfactorsintheCTUMmodel Leadershipconstructs methodandconstructvarianceinleadershipresearchcouldbeestimated.Apparently,aconsiderablemethodbiasexistsforthefiveleadershipconstructsoftransformationalandtransactionalleadership,considerationandinitiatingstructure,andLMX.Itappearsthatinempiricalstudies,datafortheseconstructsisasstronglyinfluencedbythemethodappliedasitisinfluencedbytheleadershipconstructitself.Stateddifferently,whenstudyingthephenomenonofleadership,itisasimportanttoask“Whoobservesleadership?”asitistoask“Whichkindofleadershipisdisplayed?”Thisresultisinsharpcontrasttothevastmajorityofleadershipresearchthathasbeenconductedsofar.Priorresearchfocussedheavilyonthedescriptionanddefinitionofvariousleadershipconstructs.Alimitationinthislineofresearchwasthatithadbeenimplicitlyassumedthattheunderlyingleadershipbehaviorsoftheseconstructscouldbeobservedregardlessoftheratingperspective.However,theresultsofthepresentstudydemonstratethattheconstructsthemselvesareheavilybiasedbytheratingperspective.Itshouldbenotedthattheassessmentofleadershipconstructswasonlymarginally(i.e.10percentonaverage)biasedbyerror.Inadditiontothismethodologicalperspective,theresultscanbeinterpretedwithregardtothenomologicalnetworkofleadershipconstructs.Thepresentstudywasthefirsttosystematicallyexploreconvergentanddiscriminantvaliditiesoffiveleadershipconstructs.Thereby,insightintotheproblemofpotentialoverlapbetweendifferentleadershiptheorieswasmadepossible.Thatis,animportantlimitationofpreviousleadershipliteraturewasthatwhileitcouldhavebeenassumedimplicitlythatdifferenttheoriesweretosomedegreeoverlapping,thisideahasneverbeenaddressedexplicitlywithinasystematicresearcheffort.Overall,thediscriminantvaliditieshypothesizedfromtheorycouldnotbesupportedinthepresentstudy.Allleadershipconstructsweretosomedegreeconvergent,despitethefactthatmultipleratingperspectivesfortheassessmentoftheseconstructswereapplied.Indetail,itwasfoundthattransformationalandtransactionalleadershipcorrelatedpositively,Incomparisontothehighcorrectedcorrelationof0.80reportedbyJudgeandPiccolo(2004),thisobservedMTMMcorrelationisrelativelylow.ApossibleexplanationforthisdifferenceinconvergenceisthatincontrasttovirtuallyallpriorstudiesthatformedthebasisforJudgeandPiccolo’s(2004)meta-analysis,thepresentstudyreliedonmultiple-perspectivedataassessment.Thus(mono-)methodeffectsaccountforthehithertoobservedhighintercorrelationsbetweenleadershipconstructs.Thisexplanationseemstoholdfortheobservedcorrelationbetweentransformationalleadershipandconsideration(0.41).ThisMTMM-basedcorrelationissomewhatlowerthancorrelationsreportedinpriorstudies(e.g.SeltzerandBass,1990;0.01).However,severalcorrelationsbetweenleadershipconstructsresultingfromthisMTMMstudywerecomparabletocorrelationsfrompriorstudies.First,thepresentstudyfoundacorrelationof0.73betweentransformationalleadershipandLMX.Thisisinlinewithpriorresearch,whichreportedcorrelationsof0.71(Wangetal.,2005)and0.53(HowellandHall-Merenda,1999),respectively.Second,whileRowoldandKersting(2008)reportedacorrelationof0.36betweentransactionalleadershipandinitiatingstructure,thepresentstudyfoundanevenstrongerconvergence(0.66).Third,theconvergence(i.e.0.45)betweentransactionalleadershipandLMXfoundinthepresentstudyisidenticaltotheresultsreportedbyWangetal.0.05).Insum,thesecomparisonsrevealthatdespitetheMTMMmethodologyappliedinthepresentstudy,theconvergencesbetweenleadershipconstructswereasstrongastherelationshipsreportedinstudieswhichmightbeaffectedbymono-methodbiases. 34LODJ34,1 ImplicationsfortheoryWhatimplicationsdotheseresultshaveforleadershipresearchingeneral?First,sinceaccordingtotheresultsofthepresentstudyleadershipconstructsareheavilybiasedbytherespectiveratingperspectiveapplied,leadershiptheoriesshouldbespecifiedwithregardtovariousratingperspectives.Sofar,leadershiptheorieshavegenerallyreliedontheassessmentofleadershipbehaviorfromthesubordinateperspective(e.g.BassandAvolio,2000).However,theratingperspectivesofthetargetleader,his/hersupervisors,andhis/hercolleagueshaveauniquepointofviewontheleadershipprocess(seeMountetal.,1998).Forexample,thetargetleader’ssupervisorhasaccesstostrategicandvisionaryplanningofthetargetleader.Also,thetargetleaderhasinsightintohis/herdecisionmakingandothercognitiveprocessesrelevanttoleadership,etc.Morecompletetheoriesofeffectiveleadershipintoday’sorganizationsshouldexplicitlydescribewhichleadershipbehaviorscanbeobservedfromtheabovementionedratingperspectives.Also,itshouldbearticulatedwhichleadershipprocessesareresponsibleforrating-perspective-specificeffectsoftheseleadershipprocesses.Forexample,intransformationalleadership,leadersarticulateavisionwhichtheythencommunicate.Whichpartofthisvisionarycommunicationcanbeperceivedfromtheself,thefollowers,thecolleagues,andtheleader’sboss’perspective,andwhy?Asaspeculation,onlysomepartofaleader’svisionwouldbecommunicatedtotheleader’sboss,becausethisparthasimplicationsfortheoverallstrategyoftheleader’scompany.Otherpartsofthevisionwouldbecommunicatedtoothermembersoftheleaders’network,creatingopportunitiesforuniqueassessments.Whyarethefiveleadershipconstructshighlyconvergent?Onepossibleexplanationforthisresultmightlieinthefactthattheleadershiptheoriesincludedinthepresentstudysharecommonelementsorarebasedonunderlyingdimensions(seeTableI).Infact,aclosereadingoftheintegrativeleadershipliterature(e.g.HouseandAditya,1997;RowoldandHeinitz,2007;Yukl,1999,2002)whichrevealedthedimensionsthatwerecommontoallfiveleadershipconstructs,mightexplaintheirhighconvergences.First,allleadershipconstructsrefertotheleader’slevelofactivitydirectedtoalargerdegreetowardtheirrespectivesubordinates(Bass,1985;Antonakisetal.,2003).Forexample,whileconsiderationisaclassofbehaviorsthatreferstosubordinate-centeredwell-beingandafriendship-likerelationship,initiatingstructurereferstodefiningtasksandsettingdeadlineswiththesubordinate.Althoughbothconstructswerehypothesizedtobedistinct,theysharetheprocessofaleader-ledinteraction.Asecondreasonforthehighconvergencebetweenleadershipconstructsisthatseveralconstructsareimplicitlybasedoncontrollingfollowers.Thatis,atleasttheconstructsoftransactionalleadership(Bass,1985)andinitiatingstructure(Fleishman,1973)relyonassigningtaskstothesubordinateandconsequently,controllingwhetherthesetasksareaccomplished(cf.TableI).Whileanexhaustivetheoreticalcomparisonofallleadershipconstructswasbeyondthescopeofthispaper,thesetwoexamplesofoverlapincontentoftherespectivedefinitionsoftheleadershipconstructsdemonstratethattheseconstructshavesomedegreeofredundancy.Thus,leadershipresearchcanbecriticizedbecausethisredundancyhasnotbeenarticulatedexplicitlybytherespectiveauthorsoftheleadershiptheories.Neitherdoesacriticalcomparisonofleadershiptheoriesexist.Bothissueshavebeenarticulatedbyleadershipscholars(e.g.Judgeetal.,2004;Yukl,2002).Consequently,futuretheoreticalworkshouldanalyzetheconceptualoverlap,inordertosortouttheunderlyingmechanismsandultimately,toprovideacondensedsetofclassesofleadershipbehaviors(seeRowoldandHeinitz,2007).Accordingtothe Leadershipconstructs resultsofthepresentstudy,itmightbeexpectedthatlessthanfiveleadershipconstructswouldbesufficienttoexplainleadershipbehavior.However,thispropositionremainstobetestedbyempiricaldata.ImplicationsforpracticePractitionersareoftenconfrontedwithseveralpossibleinstrumentsfortheassessmentofleadershipconstructs.Oneimplicationofthepresentstudyforpracticeisthatfromaconstructvalidityperspective,transformationalandtransactionalleadership,consideration,initiatingstructure,andLMXaresimilar.Thus,practitionersshouldbasetheirdecisionaboutassessmenttoolsforleadershipconstructsontherespectivecriterionvalidity.Whilebothtransformationalleadershipaswellasconsideration/initiatingstructurehavesimilarlevelsofstrongcriterionvalidity,economiccriteriasuchasnumberofitemspersurveycanbeutilizedtoselectaninstrumentfortheassessmentofeffectiveleadership.Morespecifically,incomparisontotheMLQ-5X,Podsakoff’sTLIisfreelyavailable,hasfeweritems,andincludesmoretransformationalleadershipfacets.Also,withregardtoratingperspective,theresultsofthepresentstudyhaveimplicationsforpractitioners.Itwasarguedabovethatinthedomainofleadership,theratingperspectiveappliedisasimportantastheleadershipconstructassessed.Thus,practitionersshouldincludeseveralratingperspectivesintheirleadershipprogramsandbalancetheprosandconsofeachoftheseperspectives(i.e.360feedback).Forexample,itiswellknownthatwithinself-ratingsofleadership,leaderstendtopresentanoverlypositivepictureoftheirownbehaviors.Bydefinition,followersaretherecipientsofleadershipbehaviorandaresupposedtoactonthisbehavior(e.g.enhancedlevelsoffollowers’motivation).Moreover,thetargetleader’ssupervisorsmightbeabletoseebehaviorsthatotherratersprobablywouldnot(e.g.engaginginstrategicplanning).Acarefulevaluationoftheseprosandconsofeachratingperspectiveisrecommendedbeforespecificassessmentstrategiesarechosen.However,asaminimum,bothself-ratingsandfollowerratingsshouldbeincludedininstrumentsforleader’sdevelopment.LimitationsanddirectionsforfutureresearchAswitheverystudy,thepresentstudyhassomelimitations.AlthoughthepresentstudyrepresentsthefirstMTMMeffortonleadershipwithtworatingperspectives,futureresearchshouldcapitalizeontheMTMMapproachbyincludingadditionalratingperspectives.Whileitwasimportanttoincludeselfandsubordinateratingsasthetwomostcommonlyimplementedperspectivesinthefieldofleadership,additionalratingperspectivescouldbevaluableforourunderstandingofmethodeffectsinleadershipresearch.Also,from360feedbackresearchtheratingperspectivesofself(e.g.targetleader),leader’ssupervisor,colleague,externalstakeholders(e.g.customers),andsubordinatehavebeensuggestedasbeingusefultoobtaininformationaboutthetargetleader’sbehavior.Next,itisawell-researchedfactthatpersonalityconstructsinfluencedimensionsofleadership(BonoandJudge,2004;etal.,2002).Thus,futureresearchshouldincludepersonalitytraitsofleaderssuchasextroversioninordertoallowforamorethoroughunderstandingoftheleadershipprocess.Second,althoughthepresentstudyincludedfiveleadershipconstructs,itwouldbeinterestingtoincludeadditionalconstructs.Asthesefiveleadershipconstructsyieldedconvergences,itmightbeimportanttochoosemorediscriminantleadershipconstructs 36LODJ34,1 suchasabusivesupervision(Tepperetal.,2007),authenticleadership(Walumbwaetal.,2008),andfactorsofsubstitutesforleadershiptheory(Dionneetal.,2002).Onlybyincludingdiscriminantconstructs,willfutureresearchbeabletoyieldamorecompleteandbalancednomologicalnetworkofleadershipconstructs.Anothersuggestionforfutureresearchmaybetheexaminationoftheconvergenceofconstructsderivedfromleadershipbehavior,suchastheleadershipbehaviorofaninnovationchampionorthebehaviorofachangechampion(Howelletal.,2005;HowellandBoies,2004).Championinnovationbehaviorisproposedtostemfromleadershipbehavior,thusitispossibletoexaminewhethertheclassicleadershipstylesinvestigatedarerelatedtochampionbehaviorornot.Third,otheraspectsofvalidityshouldbetakenintoaccountaswell.Forexample,theresultsfrommeta-analysesprovideinsightintotherespectivecriterionvalidityofleadershipconstructs(Judgeetal.,2004;JudgeandPiccolo,2004).Especiallyforpractitioners,criterion-orientedvalidityisoneofthemostimportantaspectsofvalidity.Fromatheoreticalperspective,itwouldbeinterestingtoknowtherelativecriterionvalidityofleadershipconstructs.Thatis,whilepriormeta-analyticresearchfoundcriterionvaliditiesforsetsoftwoorthreeleadershipstyles,itwouldbeimportanttoexploretherelativecriterion-orientedvalidityofmore(e.g.five)leadershipconstructs.Therecommendationtousetheratingperspectivesofself(e.g.targetleader),leader’ssupervisor,colleague,externalstakeholders(e.g.customers)mustbeevaluatedinpractice,especiallyintermsofatimeandfinancialpointofview.Cost-utilityanalysiscanhelptoinvestigatethefeasibilityofthisapproachandmakethecost-usabilitybalancecleartodecisionmakers.Thepresentstudyreliedonaself-assessmentformofinitiatingstructurethatwasbelowcommonstandardsforinternalconsistency.Aspriorstudiesutilizingthissurveyreportedsatisfactorylevelsofreliability,itisspeculatedthatitwasthesamplecharacteristicsofthepresentstudythatyieldedtheunsatisfactoryinternalconsistencyestimate.Nevertheless,futurestudiesshouldaddressthisissuebyincludingmorereliableinstrumentsfortheassessmentofinitiatingstructurefromtheself-perspective.Also,thequestionremainswhetherculturemightmoderatetherelationshipsbetweenleadershipconstructs.Itshouldbenotedthatthereisnowempiricalevidencethatasregardsleadershipandculture,resultsfromGermanycanbecomparedtothoseobtainedintheUSA(i.e.bothbelongtothe“western”culture,e.g.highindividualism;cp.Atwateretal.,2005;Brodbecketal.,2002;Kuchinke,1999;RowoldandKersting,2008).Nevertheless,itwouldbeinterestingtoexplorethenomologicalnetworkofleadershipconstructsinotherculturessuchasChina,Taiwan,andIndia(WalumbwaandLawler,2003).ReferencesAlban-Metcalfe,R.J.andAlimo-Metcalfe,B.(2000),“Ananalysisoftheconvergentanddiscriminantvalidityofthetransformationalleadershipquestionnaire”,InternationalJournalofSelectionandAssessment,Vol.8No.3,pp.158-75.Alimo-Metcalfe,B.andAlban-Metcalfe,R.J.(2001),“Thedevelopmentofanewtransformationalleadershipquestionnaire”,JournalofOccupational&OrganizationalPsychology,Vol.74No.1,pp.1-27.Antonakis,J.andHouse,R.J.(2002),“Thefull-rangeleadershiptheory:thewayforward”,inAvolio,B.J.andYammarino,F.J.(Eds),TransformationalandCharismaticLeadership:TheRoadAhead,JAI,Amsterdam,pp.3-34. Leadershipconstructs Antonakis,J.,Avolio,B.J.andSivasubramaniam,N.(2003),“Contextandleadership:anexaminationofthenine-factorfull-rangeleadershiptheoryusingthemultifactorleadershipquestionnaire”,LeadershipQuarterly,Vol.14No.3,pp.261-95.Atwater,L.,Waldman,D.A.,Ostroff,C.,Robie,C.andJohnson,K.M.(2005),“Self-otheragreement:comparingitsrelationshipwithperformanceintheUSandEurope”,InternationalJournalofSelectionandAssessment,Vol.13No.1,pp.25-40.Avolio,B.J.andBass,B.M.(1995),“Individualconsiderationviewedatmultiplelevelsofanalysis:amulti-levelframeworkforexaminingthediffusionoftransformationalleadership”,LeadershipQuarterly,Vol.6No.2,pp.199-218.Avolio,B.J.andBass,B.M.(2002),DevelopingPotentialAcrossaFullRangeofLeadershipLawrenceErlbaumAssociates,Mahwah,NJ.Avolio,B.J.andYammarino,F.J.(2002),TransformationalandCharismaticLeadership:TheRoad,JAI,Amsterdam.Avolio,B.J.,Yammarino,F.J.andBass,B.M.(1991),“Identifyingcommonmethodvariancewithdatacollectedfromasinglesource:anunresolvedstickyissue”,JournalofManagementVol.17No.3,pp.571-87.Bagozzi,R.P.,Youjae,Y.andPhillips,L.W.(1991),“AssessingconstructvalidityinorganizationalAdministrativeScienceQuarterly,Vol.36No.3,pp.421-58.Bandalos,D.L.(2002),“Theeffectsofitemparcellingongoodness-of-fitandparameterestimatebiasinstructuralequationmodeling”,StructuralEquationModeling:AMultidisciplinaryJournal,Vol.9No.1,pp.78-102.Barling,J.,Weber,T.andKelloway,E.K.(1996),“Effectsoftransformationalleadershiptrainingonattitudinalandfinancialoutcomes:afieldexperiment”,JournalofAppliedPsychologyVol.81No.6,pp.827-32.Bass,B.M.(1985),LeadershipandPerformanceBeyondExpectations,FreePress,NewYork,NY.Bass,B.M.andAvolio,B.J.(1990),“Theimplicationsoftransactionalandtransformationalleadershipforindividual,team,andorganizationaldevelopment”,ResearchinOrganizationalChangeandDevelopment,Vol.4No.2,pp.231-72.Bass,B.M.andAvolio,B.J.(2000),MLQMultifactorLeadershipQuestionnaire,MindGarden,RedwoodCity.Biesanz,J.C.andWest,S.G.(2004),“Towardsunderstandingassessmentsofthebigfive:multitrait-multimethodanalysesofconvergentanddiscriminantvalidityacrossmeasurementoccasionandtypeofobserver”,JournalofPersonality,Vol.72No.4,pp.845-76.Bono,J.E.andJudge,T.A.(2004),“Personalityandtransformationalandtransactionalleadership:ameta-analysis”,JournalofAppliedPsychology,Vol.89No.5,pp.901-10.Brodbeck,F.C.,Frese,M.andJavidan,M.(2002),“LeadershipmadeinGermany:lowoncompassion,highonperformance”,AcademyofManagementExecutive,Vol.16No.1,pp.16-30.Brown,D.J.andKeeping,L.M.(2005),“Elaboratingtheconstructoftransformationalleadership:theroleofaffect”,LeadershipQuarterly,Vol.16No.2,pp.245-72.Burns,J.M.(1978),,Harper&Row,NewYork,NY.Bycio,P.,Hackett,R.D.andAllen,J.S.(1995),“FurtherassessmentsofBass(1985)conceptualizationoftransactionalandtransformationalleadership”,JournalofApplied,Vol.80No.4,pp.468-78.Byrne,B.M.(2001),StructuralEquationModelingwithAmos:BasicConcepts,Applications,andProgramming,LawrenceErlbaumAss,NewYork,NY.Campbell,D.T.andFiske,D.W.(1959),“Convergentanddiscriminantvalidationbythemultitrait-multimethodmatrix”,PsychologicalBulletin,Vol.56No.2,pp.81-105. 38LODJ34,1 Carless,S.A.(1998),“AssessingthediscriminantvalidityoftransformationalleaderbehaviorasmeasuredbytheMLQ”,JournalofOccupational&OrganizationalPsychology,Vol.71No.4,pp.353-8.Cheung,G.W.andRensvold,R.B.(2002),“Evaluatinggoodness-of-fitindexesfortestingmeasurementinvariance”,StructuralEquationModeling,Vol.9No.2,pp.233-55.Cohen,J.(1988),StatisticalPowerAnalysisfortheBehavioralSciences,Erlbaum,Hillsdale,NY.Cohen,P.,Cohen,J.,West,S.G.andAiken,L.S.(2002),AppliedMultipleRegression/CorrelationAnalysisfortheBehavioralSciences,LawrenceErlbaumAssociates,NJ.Conger,J.A.andKanungo,R.N.(1998),CharismaticLeadershipinOrganizations,Sage,ThousandOaks,CA.Conway,J.M.(1996),“Analysisanddesignofmultitrait-multiraterperformanceappraisalJournalofManagement,Vol.22No.1,pp.139-62.DenHartog,D.N.andVanMuijen,J.J.(1997),“Transactionalversustransformationalleadership:ananalysisoftheMLQ”,JournalofOccupational&OrganizationalPsychology,Vol.70No.1,pp.19-34.Dionne,S.D.,Yammarino,F.J.,Atwater,L.E.andJames,L.R.(2002),“Neutralizingsubstitutesforleadershiptheory:leadershipeffectsandcommon-sourcebias”,JournalofAppliedPsychology,Vol.87No.3,pp.454-64.Dumdum,U.R.,Lowe,K.B.andAvolio,B.J.(2002),“Ameta-analysisoftransformationalandtransactionalleadershipcorrelatesofeffectivenessandsatisfaction:Anupdateandextension”,inAvolio,B.J.andYammarino,F.J.(Eds),TransformationalandCharismaticLeadership:TheRoadAhead,JAI,Amsterdam,,pp.35-66.Eid,M.(2000),“Amultitrait-multimethodmodelwithminimalassumptions”,PsychometrikaVol.65No.2,pp.241-61.Erdogan,B.,Kraimer,M.L.andLiden,R.C.(2004),“Workvaluecongruenceandintrinsiccareersuccess:thecompensatoryrolesofleader-memberexchangeandperceivedorganizationalsupport”,PersonnelPsychology,Vol.57No.2,pp.305-32.Fittkau-Garthe,H.andFittkau,B.(1971),FragebogenzurVorgesetzten-Verhaltens-Beschreibung(FVVB)[SurveyfortheAssessmentofLeader’sBehavior],Hogrefe,GoFleishman,E.A.(1953),“Thedescriptionofsupervisorybehavior”,JournalofAppliedPsychologyVol.37No.1,pp.1-6.Fleishman,E.A.(1973),“Twentyyearsofconsiderationandstructure”,inFleishman,E.A.andHunt,J.G.(Eds),CurrentDevelopmentsintheStudyofLeadership,SouthernIllinoisUniversityPress,Carbondale&Edwardsville,pp.1-40.Gerstner,C.R.andDay,D.V.(1997),“Meta-analyticreviewofleader-memberexchangetheory:correlatesandconstructissues”,JournalofAppliedPsychology,Vol.82No.6,pp.827-44.Geyer,A.L.J.andSteyrer,J.M.(1994),“TransformationaleFuhrung,klassischeFuhrungstheorienundErfolgsindikatorenvonBankbetrieben(Transformationalleadership,classicleadershiptheories,andindicatorsofsuccessinbanks)”,ZeitschriftfuBetriebswirtschaftslehre,Vol.64No.4,pp.961-79.Graen,G.(1995),“Relationship-basedapproachtoleadership:developmentofleader-memberexchange(LMX)theoryofleadershipover25years:applyingamulti-levelmulti-domainperspective”,TheLeadershipQuarterly,Vol.6No.2,pp.219-47.Graen,G.,Rowold,J.andHeinitz,K.(2010),“Issuesinoperationalizingandcomparingleadershipconstructs”,LeadershipQuarterly,Vol.21No.3,pp.563-75.Graen,G.B.andUhl-Bien,M.(1995),“Relationship-basedapproachtoleadership:developmentofleader-memberexchange(LMX)theoryofleadershipover25years:applyingamulti-domainperspective”,LeadershipQuarterly,Vol.6,pp.219-47. Leadershipconstructs Hair,J.F.,Anderson,R.E.,Tatham,R.L.andBlack,W.C.(1998),MultivariateDataAnalysis5thed.,PrenticeHall,UpperSaddleRiver,NJ.Hancock,G.R.andNevitt,J.(2001),“Performanceofbootstrappingapproachestomodelteststatisticsandparameterstandarderrorestimationinstructuralequationmodeling”,StructuralEquationModeling,Vol.8No.3,p.353.Heeler,R.M.andRay,M.L.(1972),“Measurevalidationinmarketing”,JournalofMarketingResearch(JMR),Vol.9No.4,pp.361-70.Heinitz,K.andRowold,J.(2007),“GutekriterieneinerdeutschenAdaptationdesTransformationaleLeadershipInventory(TLI)vonPodsakoff(PsychometricpropertiesofaGermanadaptationoftheTransformationalLeadershipInventory)”,ZeitschriftfuArbeits-undOrganisationspsychologie,Vol.51No.1,pp.1-15.House,R.J.andAditya,R.M.(1997),“Thesocialscientificstudyofleadership:QuoVadis?JournalofManagement,Vol.23No.3,pp.409-73.Howell,J.M.andBoies,K.(2004),“Championsoftechnologicalinnovation:theinfluenceofcontextualknowledge,roleorientation,ideageneration,andideapromotiononchampionemergence”,TheLeadershipQuarterly,Vol.15No.1,pp.123-43.Howell,J.M.andHall-Merenda,K.E.(1999),“Thetiesthatbind:theimpactofleader-memberexchange,transformationalandtransactionalleadership,anddistanceonpredictingfollowerperformance”,JournalofAppliedPsychology,Vol.84No.5,pp.680-94.Howell,J.M.,Shea,C.M.andHiggins,C.A.(2005),“Championsofproductinnovations:defining,developing,andvalidatingameasureofchampionbehavior”,JournalofBusinessVenturing,Vol.20No.5,pp.641-61.Hu,L.T.andBentler,P.M.(1995),“Evaluatingmodelfit”,inHoyle,R.H.(Ed.),StructuralEquationModeling:Concepts,IssuesandApplications,Sage,ThousandOaks,CA,pp.33-67.Hu,L.T.andBentler,P.M.(1999),“Cutoffcriteriaforfitindexesincovariancestructureanalysis:conventionalcriteriaversusnewalternatives”,StructuralEquationModeling,Vol.6No.1,pp.1-55.Hunt,J.G.andConger,J.A.(1999),“Fromwherewesit:anassessmentoftransformationalandcharismaticleadershipresearch”,LeadershipQuarterly,Vol.10No.3,pp.335-43.Hunter,S.T.,Bedell-Avers,K.E.andMumford,M.D.(2007),“Thetypicalleadershipstudy:assumptions,implications,andpotentialremedies”,LeadershipQuarterly,Vol.18No.5,pp.435-46.Judge,T.A.andPiccolo,R.F.(2004),“Transformationalandtransactionalleadership:ameta-analytictestoftheirrelativevalidity”,JournalofAppliedPsychology,Vol.89No.1,pp.755-68.Judge,T.A.,Piccolo,R.F.andIlies,R.(2004),“Theforgottenones?Thevalidityofconsiderationandinitiatingstructureinleadershipresearch”,JournalofAppliedPsychology,Vol.89No.5,pp.36-51.Judge,T.A.,Bono,J.E.,Ilies,R.andGerhardt,M.W.(2002),“Personalityandleadership:aqualitativeandquantitativereview”,JournalofAppliedPsychology,Vol.87No.4,pp.765-80.Keller,R.T.(2006),“Transformationalleadership,initiatingstructure,andsubstitutesforleadership:alongitudinalstudyofresearchanddevelopmentprojectteamperformance”,JournalofAppliedPsychology,Vol.91No.1,pp.202-10.Kline,R.B.(2005),PrinciplesandPracticeofStructuralEquationModeling,GuilfordPress,NewYork,NY.Kuchinke,K.P.(1999),“Leadershipandculture:work-relatedvaluesandleadershipstylesamongonecompany’sUSandGermantelecommunicationemployees”,HumanResourceDevelopmentQuarterly,Vol.10No.2,pp.135-54. 40LODJ34,1 Landis,R.S.,Beal,D.J.andTesluk,P.E.(2000),“Acomparisonofapproachestoformingcompositemeasuresinstructuralequationmodels”,OrganizationalResearchMethodsVol.3No.2,pp.186-207.Liden,R.C.andGraen,G.B.(1980),“Generalizabilityoftheverticaldyadlinkagemodelofleadership”,AcademyofManagementJournal,Vol.23No.3,pp.451-65.Lievens,F.,VanGeit,P.andCoetsier,P.(1997),“Identificationoftransformationalleadershipqualities:anexaminationofpotentialbiases”,EuropeanJournalofWorkandOrganizationalPsychology,Vol.6No.4,pp.415-30.Likert,R.(1955),“Developingpatternsinmanagement”,GeneralManagementSeriesNo.178,AmericanManagementAssociation,NewYork.Likert,R.(1961),NewPatternsinLeadership,McGraw-Hill,NewYork,NY.Little,T.D.,Cunningham,W.A.,Shahar,G.andWidaman,K.F.(2002),“Toparcelornottoparcel:exploringthequestion,weighingthemerits”,StructuralEquationModeling,Vol.9No.2,pp.151-73.Looney,S.W.(1995),“Howtousetestsforunivariatenormalitytoassessmultivariatenormality”,AmericanStatistician,Vol.49No.1,p.64.Lowe,K.B.,Kroeck,K.G.andSivasubramaniam,N.(1996),“Effectivenesscorrelatesoftransformationalandtransactionalleadership:ameta-analyticreviewoftheMLQliterature”,LeadershipQuarterly,Vol.7No.3,pp.385-425.Marsh,H.W.(1993),“Multitrait-multimethodanalyses:inferringeachtrait-methodcombinationwithmultipleindicators”,AppliedMeasurementinEducation,Vol.6No.1,pp.49-81.Marsh,H.W.andBailey,M.(1991),“Confirmatoryfactoranalysesofmultitrait-multimethoddata:acomparisonofalternativemodels”,AppliedPsychologicalMeasurement,Vol.15No.1,pp.47-70.Marsh,H.W.,Byrne,B.M.andCraven,R.(1992),“OvercomingproblemsinconfirmatoryfactoranalysesofMTMMdata:thecorrelateduniquenessmodelandfactorialinvariance”,MultivariateBehavioralResearch,Vol.27No.4,pp.489-507.Mount,M.K.,Judge,T.A.,Scullen,S.E.,Sytsma,M.R.andHezlett,S.A.(1998),“Trait,raterandleveleffectsin360-degreeperformanceratings”,PersonnelPsychology,Vol.51No.3,pp.557-76.Nunnally,J.C.andBernstein,I.H.(1967),PsychometricTheory,McGraw-Hill,NewYork,NY.Podsakoff,N.P.,MacKenzie,S.B.,Moorman,R.H.andFetter,R.(1990),“Transformationalleaderbehaviorsandtheireffectsonfollowers’trustinleader,satisfaction,andorganizationalcitizenshipbehaviors”,LeadershipQuarterly,Vol.1No.2,pp.107-42.Podsakoff,P.M.,MacKenzie,S.B.andBommer,W.H.(1996),“Transformationalleaderbehaviorsandsubstitutesforleadershipasdeterminantsofemployeesatisfaction,commitment,trust,andorganizationalcitizenshipbehaviors”,JournalofManagementVol.22No.2,pp.259-98.Podsakoff,P.M.,MacKenzie,S.B.,Lee,J.Y.andPodsakoff,N.P.(2003),“Commonmethodbiasesinbehavioralresearch:acriticalreviewoftheliteratureandrecommendedremedies”,JournalofAppliedPsychology,Vol.88No.5,pp.879-903.Rafferty,A.E.andGriffin,M.A.(2004),“Dimensionsoftransformationalleadership:conceptualandempiricalextensions”,LeadershipQuarterly,Vol.15No.3,pp.329-54.Rowold,J.andHeinitz,K.(2007),“Transformationalandcharismaticleadership:assessingtheconvergent,divergentandcriterionvalidityoftheMLQandtheCKS”,Quarterly,Vol.18No.2,pp.121-33.Rowold,J.andKersting,M.(2008),“Theassessmentofcharismaticleadership:validityofaGermanversionoftheConger-KanungoScales”,EuropeanJournalofPsychological,Vol.24No.2,pp.124-30. Leadershipconstructs Sashkin,M.(2004),“Transformationalleadershipapproaches:areviewandsynthesis”,inAntonakis,J.,Cianciolo,A.T.andSternberg,R.J.(Eds),TheNatureofLeadership,Sage,ThousandOaks,CA,pp.171-96.Schmitt,N.andStults,D.M.(1986),“Methodologyreview:analysisofmultitrait-multimethodAppliedPsychologicalMeasurement,Vol.10No.1,pp.1-22.Schyns,B.(2002),“UberprufungeinerdeutschsprachigenSkalazumLeader-Member-Exchange-Ansatz(TestingaGermanScalefortheLeader-Member-Exchangeapproach)”,rDifferentielleundDiagnostischePsychologie,Vol.23No.2,pp.235-45.Seltzer,J.andBass,B.M.(1990),“Transformationalleadership:beyondinitiationandJournalofManagement,Vol.16No.4,pp.693-703.Seltzer,J.andNumeroff,R.E.(1988),“Supervisoryleadershipandsubordinateburnout”,AcademyofManagementJournal,Vol.31No.2,pp.439-46.Shin,S.J.andZhou,J.(2003),“Transformationalleadership,conservation,andcreativity:evidencefromKorea”,AcademyofManagementJournal,Vol.46No.6,pp.703-14.Steenkamp,J.E.M.andBaumgartner,H.(1998),“Assessingmeasurementinvarianceincross-nationalconsumerresearch”,JournalofConsumerResearch,Vol.25No.1,pp.78-89.Stogdill,R.M.(1948),“Personalfactorsassociatedwithleadership:asurveyoftheliterature”,JournalofPsychology,Vol.25No.1,pp.35-71.Tejeda,M.J.,Scandura,T.A.andPillai,R.(2001),“TheMLQrevisited–psychometricpropertiesandrecommendations”,LeadershipQuarterly,Vol.12No.1,pp.31-52.Tepper,B.J.andPercy,P.M.(1994),“Structuralvalidityofthemultifactorleadershipquestionnaire”,EducationalandPsychologicalMeasurement,Vol.54No.3,pp.734-44.Tepper,B.J.,Moss,S.E.,Lockhart,D.E.andCarr,J.C.(2007),“Abusivesupervision,upwardmaintenancecommunication,andsubordinatespsychologicaldistress”,AcademyofManagementJournal,Vol.50No.5,pp.1169-80.Tosi,H.L.,Misangyi,V.F.,Fanelli,A.,Waldman,D.A.andYammarino,F.J.(2004),“CEOcharisma,compensation,andfirmperformance”,LeadershipQuarterly,Vol.15No.3,pp.405-20.Vandenberg,R.J.andLance,C.E.(2000),“Areviewandsynthesisofthemeasurementinvarianceliterature:suggestions,practices,andrecommendationsfororganizationalresearch”,OrganizationalResearchMethods,Vol.3No.1,pp.4-70.Waldman,D.A.andYammarino,F.J.(1999),“CEOcharismaticleadership:levels-of-managementandlevels-of-analysiseffects”,AcademyofManagementReview,Vol.24No.2,pp.266-85.Walumbwa,F.O.andLawler,J.J.(2003),“Buildingeffectiveorganizations:transformationalleadership,collectivistorientation,work-relatedattitudesandwithdrawalbehaviorsinthreeemergingeconomies”,InternationalJournalofHumanResourceManagementVol.14No.7,pp.1083-101.Walumbwa,F.O.,Avolio,B.J.,Gardner,W.L.,Wernsing,T.S.andPeterson,S.J.(2008),“Authenticleadership:developmentandvalidationofatheory-basedmeasure”,JournalofManagementVol.34No.1,pp.89-126.Wang,H.,Law,K.S.,Hackett,R.D.,Wang,D.andChen,Z.X.(2005),“Leader-memberexchangeasamediatoroftherelationshipbetweentransformationalleadershipandfollowers’performanceandorganizationalcitizenshipbehavior”,AcademyofManagementJournalVol.48No.3,pp.420-32.nez,C.(2006),“AMonteCarlostudyofrecoveryofweakfactorloadingsinconfirmatoryfactoranalysis”,StructuralEquationModeling,Vol.13No.4,pp.587-614. 42LODJ34,1 Yukl,G.(1999),“Anevaluativeessayoncurrentconceptionsofeffectiveleadership”,EuropeanJournalofWorkandOrganizationalPsychology,Vol.8No.1,pp.33-48.Yukl,G.(2002),LeadershipinOrganizations,5thed.,PrenticeHall,UpperSaddleRiver,NJ.FurtherreadingBrowne,M.W.andCudeck,R.(1993),“Alternativewaysofassessingmodelfit”,inBollen,K.A.andLong,J.S.(Eds),TestingStructuralEquationModels,SagePublications,NewburyPark,CA,pp.136-62.Frese,M.,Beimel,S.andSchoenborn,S.(2003),“Actiontrainingforcharismaticleadership:twoevaluationsofstudiesofacommercialtrainingmoduleoninspirationalcommunicationofavision”,PersonnelPsychology,Vol.56No.3,pp.671-98.Heck,R.H.(1998),“Factoranalysis:exploratoryandconfirmatoryapproaches”,inMarcoulides,G.A.(Ed.),ModernMethodsforBusinessResearch,Erlbaum,Mahwah,NJ,pp.177-215.n,L.K.andMuthen,B.O.(2007),Mplus:StatisticalAnalysiswithLatentVariables:User’s,Muthen&Muthen,LosAngeles,CA.Podsakoff,P.M.andDalton,D.R.(1987),“Researchmethodologyinorganizationalstudies”,JournalofManagement,Vol.13No.2,pp.419-41.Podsakoff,P.M.,MacKenzie,S.B.,Ahearne,M.andBommer,W.H.(1995),“Searchingforaneedleinahaystack:tryingtoidentifytheillusivemoderatorsofleadershipbehaviors”,JournalofManagement,Vol.21No.3,pp.422-70.West,M.A.,Borill,C.S.,Dawson,J.F.,Brodbeck,F.,Shapiro,D.A.andHaward,B.(2003),“Leadershipclarityandteaminnovationinhealthcare”,LeadershipQuarterly,Vol.14No.4,pp.393-410.CorrespondingauthorJensRowoldcanbecontactedat:jens.rowold@tu-dortmund.de Topurchasereprintsofthisarticlepleasee-mail:Orvisitourwebsiteforfurtherdetails:www.emeraldinsight.com/reprints Leadershipconstructs