Beware the angry leader: Trait anger and trait anxiety as - BORA

19 Pages·2014·0.3 MB·English

by  Leo Kant

TheLeadershipQuarterly24(2013)106–124
ContentslistsavailableatSciVerseScienceDirect
The Leadership Quarterly
journal homepage: www.elsevier.com/locate/leaqua
Beware the angry leader: Trait anger and trait anxiety as predictors of
☆
petty tyranny
Leo Kanta,b,⁎,1, Anders Skogstada,2, Torbjørn Torsheima,3, Ståle Einarsena,4
aDepartmentofPsychosocialSciences,FacultyofPsychology,UniversityofBergen,Christiesgate12,N-5015Bergen,Norway
bFalckNutecAS,Gravdalsveien255,P.O.Box6YtreLaksevaag,NO-5848Bergen,Norway
a r t i c l e i n f o a b s t r a c t
Articlehistory: Drawingonthegeneralaggressionmodelandtheoriesofvictimizationandtemperamental
Received6January2012 goodness-of-fit,weinvestigatedtraitangerandtraitanxietyasantecedentsofpettytyranny:
Revised24August2012 employingamultileveldesignwithdatafrom84seacaptainsand177crewmembers.Leader
Accepted25August2012 trait anger predicted subordinate-reported petty tyranny. Subordinate trait anxiety was
Availableonline19September2012
associatedwithsubordinate-reportedpettytyranny.Theassociationbetweenleadertraitanger
andsubordinate-reported pettytyrannywasstrongestamonglowtraitangersubordinates
Keywords: supportingthetheoryoftemperamentalgoodness-of-fit—orrathermisfit—indyads.Hence,
Pettytyranny leaderanger-generatedpettytyrannyseemstoconstituteitselfbothasanaverageleadership
Abusivesupervision
styleandasbehaviortargetingspecificsubordinates,inthiscaselowtraitangersubordinates.
Traitanger
Inaddition,anxioussubordinatesreportmoreexposuretosuchabusiveleadershipbehaviors
Traitanxiety
Temperamentalgoodness-of-fit irrespectiveoflevelsoftraitangerinthecaptain.Thepracticalimplicationsareaboveallthe
needsfororganizationalandindividualmanagementofleadertraitanger.
©2012TheAuthors.PublishedbyElsevierB.V.Allrightsreserved.
1.Introduction
“Anyonecangetangry,orgiveandspendmoney—theseareeasy;butdoingtheminrelationtotherightperson,inthe
rightamount,attherighttime,withtherightaiminview,andintherightway—thatisnotsomethinganyonecando,noris
iteasy”.Aristotle:Nicomacheanethics(trans.2000,p.35)
Aristotlewasnot,inouropinion,barkingupthewrongtreewhenhedeclaredangertobeatrickybeastinrelationtoover-
the-linebehavior.Ourquestionishowtraitangeroperatesinrelationtoamodernconceptualizationofaspecificover-the-line
behaviorinworkinglife:pettytyrannyamongleadersandmanagers.Theconceptofpettytyrannyortyrannicalleadership(see
alsoEinarsen,Aasland,&Skogstad,2007)wasoriginallyintroducedbyAshforth(1994)asadescriptionofleaderswholordtheir
powersoversubordinates,byself-aggrandizement,belittlingsubordinates,behavinginarbitraryways,showingnon-contingent
punishment,discouraginginitiative,andshowingalackofconsideration.
☆ Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
⁎ Correspondingauthorat:FalckNutecAS,Gravdalsveien255,P.O.Box6YtreLaksevaag,NO-5848Bergen,Norway.Tel.:[phone](mobile),+4755
942000(telephonecompany);fax:[phone].
E-mailaddresses:[email protected],[email protected](L.Kant),[email protected](A.Skogstad),[email protected]
(T.Torsheim),[email protected](S.Einarsen).
1 Tel.:[phone](mobile),[phone];fax:[phone].
2 Tel.:[phone](office);fax:[phone].
3 Tel.:[phone](office);fax:[phone].
4 Tel.:[phone];fax:[phone].
1048-9843/$–seefrontmatter©2012TheAuthors.PublishedbyElsevierB.V.Allrightsreserved.
http://dx.doi.org/10.1016/j.leaqua.2012.08.005
L.Kantetal./TheLeadershipQuarterly24(2013)106–124 107
Theconceptofpettytyrannymayactasanumbrellaconceptforarangeofsimilarbutmorerecentlyintroducedconcepts,such
as destructive leadership (Einarsen, Aasland, & Skogstad, 2007; Schyns & Hansbrough, 2010), generalized workplace abuse
(Rospenda,Richman,Wislar,&Flaherty,2000),workplacebullying(Hoel,Glasø,Hetland,Cooper,&Einarsen,2009),andabusive
supervision (cf. Tepper, 2007 for how some of these concepts interrelate): focusing on the sustained hostile and demeaning
behaviorofsuperiorsagainsttheirsubordinates.Explicitempiricalresearchonsuchdestructiveleadershipbehaviorsisrelatively
recent,with clearinspiration fromotherfields ofresearchsuchasworkplace deviance(Robinson&Bennett, 1995), counter-
productiveworkbehavior(Fox&Spector,2005),hostileworkplace behaviors(seeKeashly&Jagatic,2003,p.33),workplace
bullying (Einarsen, Hoel, Zapf, & Cooper, 2003), and victimization at work (Aquino & Thau, 2009). Empirical studies on the
association between such forms of leadership practices and detrimental outcomes for followers have accordingly been
accumulating(Hershcovis&Barling,2010;Tepper,2007).Yet,fewstudiesexistonthepotentialpredictorsofabusiveandhostile
formsofleadershipbehavior.Theoreticalcontributions,however,suggestarangeofpromisingindividualaswellassituational
predictors (Ashforth, 1994; Tepper, 2007). Among the proposed individual factors, various personality characteristics are
commonlysuggested(Padilla,Hogan,&Kaiser,2007;Tepper,2007),withtraitangerasaparticularlylikelycandidate,indicatedby
boththeory(Hershcovis&Barling,2007)andempiricalevidence(Bettencourt,Talley,Benjamin,&Valentine,2006).Inlinewith
this,traitangerfiguresasakeyfactorinfundamentaltheoriesoninterpersonalaggression,suchasthegeneralaggressionmodel
(Anderson&Bushman,2002),alongwithseveraloverarchingmodelsofcounterproductiveworkbehavior(e.g.Spector&Fox,
2005).Yet,empirically,traitangerasapredictorofpettytyrannyremainstobetested.Thefirstaimofthisstudyisthereforeto
investigatetherelationshipbetweenself-ratedleadertraitanger(Spielberger,1996)andsubordinate-ratedpettytyranny,which
wesuggesttohavebothanindividuallevelinfluence,andagroup-levelinfluenceinlinewithanaverageleadershipstyleapproach.
Secondly,wewishtoinquireaboutexposuretopettytyrannyandpersonalityinrelationtotheotherparty,tothefollower—the
target.InthiswefirstturntothepioneeringworkofOlweus(1978,2003)—furthersupportedbythelaterbodyofresearchon
victimization at work (e.g. Aquino & Thau, 2009)—suggesting the largest group of victims being characterized by anxiety,
self-doubt,andsubmissiveness.Thesevictimsmaybothbetargetedmoreandexperiencetheabusetheyfacedifferentlythando
othervictims.Thesecondaimofthisstudyisthereforetoinvestigatetherelationshipsbetweenself-ratedsubordinatetraitanxiety
(Spielberger,1983)andsubordinate-ratedpettytyranny,whichwesuggesttobeanindividual-levelinfluenceonbehalfofthe
subordinate.
Hershcovis and Barling (2007) call for more investigations of the relationship between actor and target in research on
workplaceaggression:combinationsofleaderandfollowerpersonalitymay,specifically,contributebeyondthemaineffectsof
eitherleaderorfollowerpersonalityonthefollowerratingsofpettytyranny(Bowling&Beehr,2006).Pettytyrannymaytake
placebothinteamsbutalsoindyadicrelationships(cf.Rayner&Cooper,2003),indicatingthatanimportantissuemaybehow
personalities fittogether—ornot(Kristof-Brown,Zimmerman,&Johnson,2005), oftencalled “interpersonalchemistry”in lay
language.Abasictemperamentalgoodness-of-fitisclaimedtodeterminethedevelopmentofparentandchildrelations,rather
thanthetemperamentofthechildperse;poorfitleadingtopoorrelations(Lerner&Lerner,1983).Thethirdaimofthisstudyis
therefore to investigate the qualified, interactive relationship between high self-rated leader trait anger and low self-rated
subordinatetraitangerindeterminingsubordinate-ratedpettytyranny,whichwesuggesttobeadyadic,hencewithin-group
influence.
Thus,thethreeaimsofthestudycorrespondwithmakingconsecutiveacquaintancewith“theleadersandthefollowersboth
separatelyandincombination—thatis,asleaders,followers,andlinkages”(Yammarino&Dansereau,2008,p.136),investigating
therolewhichtraitangerandtraitanxietymayplayinpettytyrannyinthesethreedifferentdimensionsofleadership(Graen&
Uhl-Bien,1995).Theoretically,thestudypromisestoshedlightonsomelikelyantecedentsofpettytyrannyandtheaccording
conceptualimplicationsofthem.Inthis,wehavetheoverarchingassumptionthatpettytyrannymaybothbeastyleaffectingall
subordinates,aswellasbeingdyadicinnature.Thatis,weassumethatabusiveleaderswillbegenerallyabusivewhilesome
subordinatesyetmaybemoreaffectedthanareothers.Methodologically,thestudypromisestojointherelativelyfewstudies
that clarify issues of levels of analysis (Yammarino, Dionne, Uk Chun, & Dansereau, 2005), and that do so while employing
separatedatasourcesinfightingsingle-sourcebiases(Barling,Dupre,&Kelloway,2009).Theobservantreaderwillfurthermore
havenoticedhowthetwotraitschosen,oftenregardedaspartsofaneuroticismdimensionofpersonality,arehypothesizedto
have somewhat different effects for actor and targets, hence following a narrow bandwidth personality approach (Bergner,
Neubauer,&Kreuzthaler,2010),asthebroaderpersonalitydimensionofneuroticismhasyieldedsomewhatcontradictoryand
inconclusiveresultsinthisdomain(e.g.Berry,Ones,&Sackett,2007;Salgado,2002).Forpractice,thestudypromisestosupply
leaders,subordinates,andtheirorganizationswithavantagepointfordevisingcountermeasuresagainstthealreadydocumented
troublesinthewakeofpettytyranny.
2.Antecedentsofpettytyrannyandabusivesupervision
Todate,empiricalinvestigationsofantecedentsofpettytyranny,andtheconceptuallycloseabusivesupervision,rangefrom
macro-tomicro-levelfactorsinwhichharshnessofthermalclimateanddegreeofcollectivewealthonasocietallevelarethe
mostdistantmacro-levelpredictorsinvestigated(cf.VandeVliert,Matthiesen,Gangsøy,Landro,&Einarsen,2010).Furtherdown
the funnel of antecedents investigated, we find perceptions of organizational factors as well as micro-level attitudinal and
state-like characteristics of leaders and subordinates. Hoobler and Brass (2006) for instance found that when university
supervisorsshowedahighhostileattributionbias,andexperiencedahighamountofpsychologicalcontractviolationsfromtheir
108 L.Kantetal./TheLeadershipQuarterly24(2013)106–124
organization, their students reported higher levels of abusive supervision. Intimately linked to behavior yet distal enough to
precedebehavior—hereactsofpettytyranny—liestherealmofrelativelystablepersonalitytraits.
AlthoughalreadysuggestedbyAshforth(1994)inhisseminaltheoreticalstudyaslikelyantecedentsofpettytyranny,stable
anddistalpersonalitypredictorsaresurprisinglyabsentintheempiricalstudiesonantecedentsofpettytyrannyandabusive
supervision,forinstanceasreviewedbyTepper(2007).Accordingly,Tepper(2007)proposespersonalityasimportantforfuture
researchinthisfield.Therearehoweversomeexamplesofrecentcontributionsaddressingpersonalityanddestructiveleadership
intheshapeofleaderMachiavellianism(Kiazad,Restubog,Zagenczyk,Kiewitz,&Tang,2010),andalsoleaderperceptionsof
deep-level dissimilarity between leader and subordinate, relationshipconflict and subordinate performance (Tepper, Moss, &
Duffy,2011).ThesystematicstudyofwhatAshforth(1994)calledtheetiologiesofineffectiveleadershipis,however,stillinits
fledglingstage,particularlywithrespecttotheinfluenceofpersonality.
Theoretically, petty tyranny is (a) an aggressive behavior directed towards other people in a formally unequal power
structure,namelytowardssubordinates,that(b)crossesthelineofwhatmaybeconsideredculturallyacceptablebehavior.The
definition does not prescribe petty tyranny to be either a reactive aggression, which is a “hot-blooded” emotionally driven
aggression,oracoldinstrumentalorproactiveaggression(Anderson&Bushman,2002).Pettytyrannyisdescriptiveofleader
behaviorperse,irrespectiveofwhatdrivesit.Adeeptheoreticalaswellasempiricalsoundingboardforsuchactsamongleaders
maybefoundinthegeneralaggressionmodel,wheretraitsareseenasimportantpersonfactorspredictinghumanaggression
(Anderson&Bushman,2002).The“consistentrelationshipsbetweenpersonalityvariables(e.g.traitanger,negativeaffect)and
workplace aggression” (Barling et al., 2009, p. 677) further support the notion of personality as a likely predictor of petty
tyranny.HerewealsonoteRobinsonandBennett's(1995,p.567)propositionthat“individualvariablesmaybemorelikely
[thanorganizationalvariables]toexplaininterpersonalformsof[workplace]deviance”.Withthestrongestrelationshipfound
between trait negative affectivity and interpersonal forms of deviance (.33, pb0.05), Aquino, Lewis, and Bradfield (1999,
p.1087)concludedtheirstudybasedonstructuralequationmodelingbystatingthat“employeesaremorelikelytorespondto
negative emotional states by exhibiting direct forms of deviance against individuals than they are to act indirectly against
employinginstitutions”.
Advocatesofpersonalityasanimportantantecedentofleadershipbehavior,suchasHoganandKaiser(2005,p.175),claimthat
managerialfailure“isrelatedmoretohavingundesirablequalitiesthantolackingdesirableones,thatis,havingthewrongstuff”.We
willconsiderthis,withoutgoingtotheextremesofpsychopathology,whenwefocusontraitanger:apersonalitytraitthatmaybe
considered undesirable, or as having toxic qualities due to its effects on both individual subordinates and entire workgroups
(Schaubroeck,Walumbwa,Ganster,&Kepes,2007),albeitneitherexclusivelyundesirablenornecessarilybeinganabnormalquality
(Furnham&Crump,2005).Thelatterisimportantasno“compellingevidence”isfoundforthenotionofworkplaceaggressionbeinga
“functionofmentalillness”(Barlingetal.,2009,p.685);instead,wefindtraitanger,alongwithtraitanxiety,tobeapartofthenormal
psychologyoftheuniversalcommoner.Thetwotraitsarefurthermorenarrowenoughtoclarifyconceptuallydifferentinfluences
uponperpetratorandtarget(cf.Bergneretal.,2010,p.196),aswellasforassessingpersonality-basedfitormisfitbetweentheparties
(Kristof-Brownetal.,2005,p.318).
Traitangerhereisdefinedastherelativelystableindividual“dispositiontoperceiveawiderangeofsituationsasannoyingor
frustrating,andthetendencytorespondtosuchsituationswithmorefrequentelevationsinstateanger”(Spielberger,1996,p.1).
Inthisdefinitionanger“referstoanemotionalstatethatcomprisesoffeelingsthatvaryinintensity,frommildannoyanceor
aggravationtofuryandrage”(Spielberger,1996,p.9).Spielbergeremphasizesthedifferencebetweenthestablepronenesstoan
emotional state of anger and the behavior, often labeled as aggression. The latter implies “destructive or punitive behavior”
directedtowardsotherpersons(Spielberger,1996,p.9),inourcasepettytyrannydirectedtowardssubordinates.Traitangeris
thus a personal characteristic likely to predict behavior of an abusive nature. It is furthermore a provocation-sensitive trait
(Bettencourtetal.,2006),whichisrelevanthereaswewishtoinvestigateitinaspecificrelationship;leadershighintraitanger
maybeparticularlyprovokedbytheactionsordemeanorofparticularsubordinates.Furthermore,traitanxiety—heredefinedas
“individual differences in the tendency to perceive a wide range of situations as dangerous or threatening” (Spielberger &
Sydeman,1994,p.294)—isstronglyindicatedasapredictorofvictimization(e.g.Aquino&Thau,2009;Olweus,1978).Bothtrait
angerandtraitanxietyarethereforepersonalcharacteristicsthatarelikelytobeofrelevanceincasesofpettytyranny,albeit
somewhatdifferentlyaswillbetheorizedinthefollowing.
2.1.Leadertraitangerasapredictorofpettytyranny
Westartoutwiththefirstdimensioninleadership—theleader(Graen&Uhl-Bien,1995),incidentallyalsotheaggressorin
pettytyranny.Inthegeneralizedaggressionmodel(Anderson&Bushman,2002),traitangermaybeseenasasignificantperson
inputwhichviaroutesofaffect,cognition,andarousalmayresultinbothimpulsiveandthoughtfulaction.Furthermore,traitanger
may play several roles that may contribute towards leaders displaying the repeated behavior towards subordinates that
constitutespettytyranny.Hightraitangermayforinstancereduceinhibitionsforretaliation,asshownbyitsprovocation-sensitive
nature(Bettencourtetal.,2006),makingtheclassicalhostileorhot-bloodedaggressionmorelikely(Anderson&Bushman,2002).
Ontheotherhand,hightraitangermayalsoallowaleadertomaintainaggressiveintentovertimeandhelpresolveambiguitiesin
vaguesocialsituations,henceenergizingbehavior(Anderson&Bushman,2002),perhapseventothepointwhereitcrossesthe
lineofappropriatebehavior(cf.Geddes&Callister,2007),thusbecomingpettytyranny.
Eveniftraitangerisconsideredanantecedentfactorinmodelsofcounterproductiveworkbehavior(e.g.Spector&Fox,2005),
itremainstobetestedempiricallyasapossibleantecedentofpettytyranny,althoughitisspecificallysuggestedassuchbyTepper
L.Kantetal./TheLeadershipQuarterly24(2013)106–124 109
(2007)inhisreviewofabusivesupervision.Awiderrangeofstudies,however,showstraitangertopredictaggressivebehaviors,
counterproductiveworkbehaviors,andbullying:forinstance,ameta-studyshowedtraitangertobeasignificantpredictorof
interpersonalaggression(Hershcovisetal.,2007).Ithasalsobeenfoundtobeapredictorofaggressionofbothapsychological
andphysicalnature(Douglas&Martinko,2001),particularlyregardingbehaviordirectedtowardssubordinates(Inness,LeBlanc,
&Barling,2008).Miles,Borman,Spector,andFox(2002)showedinaregressionanalysisthattraitangeraloneexplained9%ofthe
varianceincounterproductiveworkbehavior,beyondthe9%ofvarianceexplainedbythreeenvironmentalvariables:workload,
constraints,andinterpersonalconflict.Traitangerhasalsobeenshowntobeassociatedwithbeingaperpetratorofworkplace
bullying(Matthiesen&Einarsen,2007).Furthermore,thetendencytointerpretstimuliasmorenegativeisnoticeablewhentrait
anger is high (Wilkowski & Robinson,2007), potentiallyleading to more hostile behaviors. Hence, we proposethe following
hypothesis:
Hypothesis1. Highlevelsofself-reportedleadertraitangerareassociatedwithelevatedlevelsofsubordinate-reportedpetty
tyranny.
Wheninvestigatingthefirsthypothesisweapplyanaverageleadershipstyleapproach,assumingthattraitangercausessimilar
behaviors towards all subordinates in the team (Yammarino & Dansereau, 2008, p. 136), alternatively causing many team-
memberstoperceivepettytyrannyeveniftheypersonallyarenottargeted(seeFigure1inOlweus&Limber,2010,p.125).Actsof
aggressiondoinfluencebystanders—beitchildrenwitnessingmilitaryviolence(Qouta,Punamäki,Miller,&El-Sarraj,2008),or
family members (Dekel & Monson, 2010) and even therapists (Arvay, 2001) being vicariously or secondarily traumatized by
personswithpost-traumaticstressdisorder.Wesuggestthereachofpettytyrannytobeparticularlysimilartothatofbullying,
withmanypartiesinvolved—evensointhepopularscenarioofasingleperpetratorandasinglevictim(Namie&Lutgen-Sandvik,
2010;Salmivalli,2010).Thebystandersmayrangefromhenchmenandsupporters,toonlookersanddefenders(seeFigure1in
Olweus&Limber,2010,p.125)—allcapableofbeingaffectedbythepettytyranny.Thestatisticalanalysisofourempiricaldatawill
thereforepertaintobetween-groupdifferenceswhentestingthishypothesis.Wethusconsiderthesubordinatesofagivenleader
asasinglebutunifiedgroup.Inthefollowing,wewillconsecutivelyturntotheindividualanddyadiclevel,inordertocoverour
overarchingassumptionofpettytyrannyasaphenomenonworkingonseverallevels.
2.2.Subordinatetraitanxietyasapredictorofpettytyranny
Theseconddimensionofleadershipaddressestheindividualsubordinate(Graen&Uhl-Bien,1995),whichisrepeatedinthe
toxictrianglemodelofleaderderailment(Padillaetal.,2007).Throughtheiractions,character,orappearance,subordinates
may provoke petty tyranny in their superiors. Leaders may react upon certain subordinates more than others through, for
example,anincreaseinpettytyranny(Tepper,Duffy,Henle,&Lambert,2006).Forinstance,theanxioussubmissivevictims
identifiedbyOlweus(1978)“signaltoothersthattheyareinsecureandworthlessindividualswhowillnotretaliateiftheyare
attackedorinsulted”(Olweus,2003,p.66),andaretherebypossiblyseen“asripetargetsforexploitation”(Tepperetal.,2006,
p.105)bypotentialabusers.Theperhapsmostsalientcharacteristicoftargetsofabuse—traitanxiety—wasthusindicatedearly
on by Olweus (1978). It has since been consistently demonstrated in a wide range of victimization research among adults
(cf.Aquino&Thau,2009).Thepersonalitycharacteristicoftraitnegativeaffectivity,includingbothtraitanxietyandtraitanger,
“shows the most consistent relationship to various victimization measures” (Aquino & Thau, 2009, pp. 722–723), as
exemplifiedbytargetsofbullyingcomparedtonon-bulliedpersonsportrayinghigherscoresonallsixfacetsofneuroticism
(Perssonetal.,2009).Inextensiveresearchconductedonschoolchildren,targetsaretypicallyfoundtobehighonanxiety,
resulting in exposure to bullying (Olweus, 2003). Similarly, Matthiesen and Einarsen (2007) found targets of workplace
bullyingtobehighonanxiety.
Anextensionofthevictimprecipitationhasbeendescribedinbullying(Olweus,2003)andvictimization(Aquino&Thau,
2009),anddemonstratedinameta-studyof18longitudinalinvestigationsofpeervictimizationamongchildrenasaviciouscircle
(Reijntjes,Kamphuis,Prinzie,&Telch,2010):Namely,thatprolongedexposuretoabusemayalsoincreasetheleveloftraitanxiety,
besideselicitingtheabuseinthefirstplace.
Finally,subordinateshighintraitanxietymay,ofcourse,bemoresensitivethanthoselowintraitanxiety,andmaytherefore
perceiveleaderbehaviordifferently—apronenesstoperceive“situationsasdangerousorthreatening”isattheverycoreofthe
concept(Spielberger,1983).Whetherleaderbehaviorisperceivedashostileornot,whetheritisperceivednotonlyasexpressed,
but also as improper(cf. Geddes& Callister,2007) maythus to someextentbe influenced by theperceivers personality. The
experienceoneteammemberhasofagivenbehaviormaybequitedifferentfromthatofanotherteammember(Bowling&Beehr,
2006);forexample,whatisperceivedasanironicjokebyonesubordinatemaybeperceivedasanactofsarcastichumiliationby
another.Suchhypersensitivitytonegativeeventsmaybeexemplifiedbysubordinatetraitnegativeaffectivityseemingtoinfluence
theirperceptionoftheirleaders’abusivenessindyadicrelationships(Tepperetal.,2006).
The association of target trait anxiety and perpetrator abusive behavior is hence highly likely, though untried with petty
tyrannyspecifically.Oneobviouspurposeistherebytoreplicatefindingsinadjacentfields:Anotherperhapslessobviouspurpose
istotryvictimtraitanxietyconjunctivelywiththetypicallycovariatetraitangerinordertorevealpossibledifferingeffectsofthe
traitsuponpettytyranny.Wheninvestigatingthesecondhypothesisweapplyawithin-groupapproach,viewingsubordinatesas
independentindividuals(Yammarino&Dansereau,2008,p.137),asdifferentsubordinatesmaycontributetoelicitingpettytyranny
110 L.Kantetal./TheLeadershipQuarterly24(2013)106–124
fromthesameleadertovaryingdegrees,oratleastdifferintheirtendencytoperceivesuchbehaviorintheirleader.Thetrait
anxietyofthevictimsthussubstantiatesourproposalofthefollowinghypothesis:
Hypothesis2. Subordinateshighonself-reportedtraitanxietywillreporthigherlevelsofexposuretopettytyrannythando
subordinateslowonself-reportedtraitanxiety.
2.3.Dyadicinteractionofleadertraitangerandsubordinatetraitangeraspredictorsofpettytyranny
Thethirddimensionofleadershipcoverstherelationshipbetweentheleaderandthesubordinate(Graen&Uhl-Bien,1995),or
thelinkageitself:“Itisonlywhentheleadersandfollowerslinktogetherthatleadershipemerges”(Yammarino&Dansereau,2008,
p.137).Justasamarriagemayworkoutwellwithagoodmatchofpartners,ordisastrouslywithapoormatchofpartners,itis
feasible to find analogously matched entities among organizational members. Studies on leader–member exchange have
demonstratedthispoint,thoughwithconstructiveoreffectiveleadershipasthefocaldependentvariable(cf.Nahrgang,Morgeson,
&Ilies,2009).Yet,sucharelationshipfocusisgenerallylackinginworkplaceaggressionresearchaccordingtoHershcovisand
Barling(2007).RaynerandCooper(2003)andAasland,Skogstad,Notelaers,Nielsen,andEinarsen(2010)haveshownthatthe
largemajorityofleadersareratedasdisplayingbothgoodandbadleaderbehavior,thusdyadicrelationshipswithinthegroupare
likelytovary.Forthiswenowfocusonapotentialleader–subordinatedyadlinkage(Yammarino&Dansereau,2008).Likethe
analysisforthesecondaimofthestudy,theanalysisforthethirdaimwillthereforepertaintowithin-groupdifferences.
Theoretically,wewillarguethatwemayfindparticularlypoorrelationshipswithpettytyrannyinthematchingofhightrait
angerinleadersandlowtraitangerinsubordinates,duetothesimplereasonofamisfitofpersonalities.Traitangerisapartof
temperament, which we humans are in essence born with. From developmental psychology, we will adopt the theory on
temperamentalgoodness-of-fitbetweenparentandchild,whichdeterminesthedevelopmentoftheirrelationsratherthanthe
temperamentofthechildperse(Lerner&Lerner,1983,1987).Similarityinpersonalitymay,inthesamemanner,explaindiffering
qualitiesofrelationshipanddifferingbehaviorinworkinglife(Kristof-Brownetal.,2005).Hence,similarityintraitangermay,
theoretically,increasethelikelihoodofgoodleader–memberexchange(Gerstner&Day,1997).
The opposite of similarity—leadership distance—is by Napier and Ferris (1993) argued to decrease the quality of leader–
memberexchange—anotiongainingnewsupportinrecentlyfounddirecteffectsofleaderperceptionsofdeep-leveldissimilarity
onabusivesupervision(Tepperetal.,2011).AccordingtoAntonakisandAtwater(2002),threedimensionsofleadershipdistance
exist: physical, social, and interaction frequency. The captains and crew members who constitute the sample of this study,
workingonboard smallvessels,are bothphysicallycloseand havea high interaction frequency.This leavesperceived social
distanceintheshapeofpersonality(heretraitanger)asthepossibledistancethatdecreasesthequalityoftheirleader–member
exchange,inourcaseexemplifiedbyreportsofpettytyranny.Kristof-Brownetal.(2005)conceptualizefitatworkintoeither
supplementaryorcomplementaryfit.Theformerbeingwhensimilarpersonalitiesfit,thelatterwhendifferentonesfit.Weargue,
inlinewiththedevelopmentalgoodness-of-fitmodel,thatsimilartraits—asupplementaryfit—mayinfluenceconstructiveleader
behavior. We argue, contrarily, that differences in traits—a complementary fit or rather personality misfit—may influence
destructiveleaderbehavior.Hence,the relationshipbetween highleadertrait anger and petty tyrannyshould bestronger in
dyadswithlowtraitangersubordinatesandweakerindyadswithhightraitangersubordinates.
Alternativetheoreticalframeworksareconceivablewhenlookingatleader–subordinatelinkages,yetbeingmoreambiguous
regarding predictions of dyadic trait anger-fuelled petty tyranny than are predictions following goodness-of-fit theory. For
instance,itwouldbepossibletoarguefortargetselectionbaseduponrationalchoicetheory(Miethe&Rothschild,1994),an
examplebeinghowweightingsoftheratioofeffectversusdangerhavebeenclaimedtoberelevantinexplainingactsofaggression
(Björkqvist,Österman,&Lagerspetz,1994).Suchrationalchoicescouldresultinleadersdeeminglowtraitangersubordinatesto
belessdangeroustargets,andthereforebeingpronetodisplaymorepettytyrannytowardslowthantowardshightraitanger
subordinates,thelatterwhoineffectwoulddeterleadersfrompettytyranny.Targetselectionbaseduponsuchrationalchoicescan
beconsideredamechanismforwhymanyvictimsare foundtobesubmissivevictims(cf. Olweus,2003).However,opposite
directionalitycouldalsobeargued:hightraitangersubordinatescouldbemorelikelytocommittransgressionsintheeyesof
theirhightraitangerleaders,andeithercreategroundsforescalatingactsofaggressionfrombothpartiesoratleastintheeyesof
theirhightraitangerleaders,beingperceivedasaformofprovocativevictim(cf.Olweus,2003).Therefore,althougharelevant
theoreticalframework,therationalchoicetheoryisunnecessarilycomplexinthiscase,as(a)thedirectionofourpredictions
wouldremainambiguous,and(b)thehigherordercognitiveprocessesnecessarilyinvolvedwouldremovefocusfromthetrait
anger-fuelledreactivepettytyrannyweaimtostudy.
Temperamental goodness-of-fit theory, contrarily, provides a basis for unambiguous prediction in our case. Furthermore,
followingthistheory,therecognitionofacounterpart'stemperamentcaneasilybeestablishedwordlesslyandwithoutawareness
ortheneedofhigherordercognitioninthoseinvolved.Suchmirrormechanisms,thatisasysteminvolvingmirrorneuronsin
humanaswellasmonkeybrainsallowing“thatactionsperformedbyothers,afterbeingprocessedinthevisualsystem,aredirectly
mappedontoobservers'motorrepresentationsofthesameactions”(Rizzolatti,Fabbri-Destro,&Cattaneo,2009,p.24).Notably
thesemirrormechanismsarealsoshowntobeinvolvedinthe“remarkableabilitytoreadothers'emotionalstatesfromamere
glanceattheirfaces”(Daprettoetal.,2006,p.30).Suchaffectsharingdoesnotonlyworkfornewbornsandtheircaregivers
(Singer,2006)forthehererelevantestablishmentoftemperamentalgoodness-of-fit(Lerner&Lerner,1987),butisalsosuggested
tohaverelevanceinworkinglifeandleadership(Goleman&Boyatzis,2008).
L.Kantetal./TheLeadershipQuarterly24(2013)106–124 111
Hence,itisenoughthattheleaderandsubordinateobserveeachotherforthemtoknowwhethertheyaretemperamentally
alikeornot,whetherthereisafitoramisfitbetweenthem.Then,iftheleaderastheaggressor,possessesahigherleveloftrait
angertobeprovokedbysomething,notevennecessarilybysubordinatebehavior(Bettencourtetal.,2006),leadertraitanger
wouldexplainevenmoreofpettytyrannyagainstdissimilarsubordinates,thatislowtraitangersubordinates.Aspointedout
above, we expect similarity to lead to benign outcomes—less trait anger-fuelled petty tyranny in our case, while misfit or
dissimilaritywillleadtoworseoutcomesormoretraitanger-fuelledpettytyranny.Again,inthecaseoflowtraitangerleadersand
hightraitangersubordinatestheremaywellbesomereportsofpettytyranny.Althoughtheyhaveamisfitintemperamentwith
hightraitangersubordinates,lowtraitangerleadersarenotlikelytobeprovokedintoreactivepettytyrannyinthefirstplace.Low
traitangerleadersmaystillactwithpremeditatedproactiveaggressivebehavior,butitisoutsidethescopeofthisstudy.
In conclusion, the ontogenetical nature of trait anger and according likelihood of interpersonal problems in the case of
personalitymisfitrepresentingaformofsocialleadershipdistance—leadustoproposethefollowinghypothesis:
Hypothesis3. Higherlevelsofself-reportedleadertraitangerwillinteractwithlowerlevelsofself-reportedsubordinatetrait
angertoincreasesubordinate-reportedpettytyranny.Inshort:therelationshipofleadertraitangerandpettytyrannywillbe
strongestforlowtraitangersubordinates.
Thetheoreticalargumentsweherebyhavepresented,suggestpersonalitytraitsasareasonablepathtofurtherourunderstanding
of reactivepettytyranny,specificallylooking atnarrow bandwidthfacets oftrait angerand traitanxiety.First,welook athow
perpetratorshighontraitangermaydrivereactivepettytyrannyeveninfluencingentireteams.Second,welookathowtargettrait
anxiety,butnottraitanger,may individuallycontributeto exposuretoandexperiencesofpetty tyranny.This,wepredict,will
furthermorecontrasthowtargettraitangerwillcontributeinanoppositedirectionindyadicinteractionswithhightraitanger
leaders.Third,wethereforelookathowtemperamentalmisfitbetweenperpetratorandtargetlevelsoftraitangermayformabasis
forevenhigherlevelsofanger-fuelledpettytyrannyindyadswithhightraitangerleadersandlowtraitangersubordinates.
3.Methods
3.1.Procedure/sample
Totestourhypothesesontrait-basedinfluencesonpettytyranny,weconductedaworkenvironmentsurveyamongcaptains
andcrewmembersofamajortransportcompanyworkingonboardferriesinregularservicealongtheNorwegiancoastline.The
samplingwasnaturally clustered;sinceindividualcrewmembersbelongedtoteamssharingaparticularcaptain,thesample
formedaninherentlyhierarchicalstructurerequiringandallowingamulti-levelstatisticaldesign.
Eachvesselhad3–4teamsworkinginrespectiveshifts:eachteamconsistedofacaptainandacrewof2–10members.The
teamsworkedandlivedtogetherforuptosevendaysaweek,24haday.Ofthe837surveyquestionnairesdistributedtothe
companyferryservices,462werereturned,providingatotalresponserateof55.2%,slightlyabovethemeanfoundinsurveysof
thiskind(cf.Baruch&Holtom,2008).The462respondentsincluded303crewmembersand105captains.For31ofthe303crew
members,teammembershipwasnotproperlyidentified,leaving272crewmembersfrom134teamsintheavailablestudysample.
Nineteenofthecrewmembers(7%)werefemale.Theiraveragetenurewasfouryears.Inthese134teams,informationfrom99
captainswasavailable.Allofthecaptainsweremale.Theiraveragetenurewas5.28years.Duetomissingdata,theeffectivestudy
sampleinthemultilevelanalysisconsistedof177crewmembersand84captains(forfurtherdetailsseeanalysissection.)
3.2.Measures/instruments
3.2.1.Traitanger
WemeasuredtraitangerwiththeState-TraitAngerExpressionInventory(STAXI)(Spielberger,1996)forcaptainsandcrew
membersalike.Thisscaleconsistsof12items(α=.74)withresponsecategoriesrangingfrom1(almostnever)to4(almost
always).Anexampleitemis:“IgetangrywhenI'msloweddownbyothers'mistakes”.
3.2.2.Traitanxiety
WemeasuredtraitanxietywiththeState-TraitAnxietyInventory(STAI)(Spielberger,1983)forcaptainsandcrewmembers
alike.Thisscaleconsistsof20items(α=.88)withresponsecategoriesrangingfrom1(almostnever)to4(almostalways).An
exampleitemis:“Iaminclinedtotakethingshard”.
3.2.3.Pettytyranny
PettytyrannywasmeasuredwithfiveitemsinspiredbyAshforth(1994)andusedinpreviousstudies(Aaslandetal.,2010;
Einarsenetal.,2007).Thescale,presentedtothecrewmembers,consistsoffiveitems(α=.81)describingthebehavioroftheship's
captainwithresponsecategoriesrangingfrom1(notcorrect)to5(entirelycorrect).Theitemshadtheheading“Myimmediate
leader/captainhas”,andwererespectively:(1)“yelledatusinrageorspenttimeactinggrumpy”;(2)“belittledorhumiliatedmeor
otheremployees,ifwefailtoliveuptohis/herstandards”;(3)“keepstrackofthemistakesofothers,whileconsistentlytryingto
coveruphisorherownmistakes”;(4)“justifiedownactionsbyunfairlyblamingothers”;and(5)”spreadfalseinformationaboutyou
oryourco-workers,inordertoharmyour/theirpositioninthefirm”.
112 L.Kantetal./TheLeadershipQuarterly24(2013)106–124
3.2.4.Controlvariables
Inordertoputourhypothesesandstudyvariablesthrougharigoroustest,weincludedarangeofcontrolvariablesforcaptains
andcrewmembersalikepertainingto:(a)demographics,(b)potentialsituationalinfluencesontheenactmentandreportingof
pettytyranny,and(c)associatedpersonalityfacetswithinthebroaderdimensionofneuroticism.
Gender differences amongst perpetrators are well known in aggression research, with males more often than females
committingactsofactiveaggression(Barlingetal.,2009).Males,olderemployees,andthosewithlongertenurehavealsobeen
overrepresentedasvictims,thoughcontraryresultsalsoexist(Aquino&Thau,2009).Shorttenureonboardtheshipshouldalso
be controlled for in order to address “divestiture socialization tactics, negative social communication from insiders that are
designedtodismantletheidentityofnewrecruits(VanMaanen&Schein,1979)”(inTepper,2007,pp.282–283).Weaccordingly
controlledforgender,age,andtenureonboardtheshipsintheanalyses.
Traitangerisaprovocation-sensitivetrait(Bettencourtetal.,2006).Asweprimarilysearchforprovocationsandreactions
withintheleader–subordinaterelationship,weneedtocontrolforpossiblesituationalprovocations(e.g.Anderson&Bushman,
2002;Ashforth,1994;Spector&Fox,2005).Rolestressorsareconsideredsignificantmoderatorsofleadershipbehavioringeneral
(Podsakoff,MacKenzie,Ahearne,&Bommer,1995),andofworkplaceaggression(Hershcovis&Barling,2007),andhavebeen
associatedwithbothenactment(Hauge,Skogstad,&Einarsen,2009)andexposuretoworkplacebullyingandharassmentatwork
(Bowling&Beehr,2006).Threemeta-studiesshowedthatroleambiguityandroleconflictwere:(a)negativelyrelatedtoleader–
memberexchange(Gerstner&Day,1997),(b)relatedtoworkplaceharassment(Bowling&Beehr,2006),and(c)influencinga
rangeofconsequencesdetrimentaltoemployees(Örtqvist&Wincent,2006).Thetworolestressorsmaythusinfluenceleader
behavior,perceptionsofbeingatargetofharassment,leaderandsubordinatetension,subordinateattitudes,andsubordinate
behaviorthatinturnmayelicittyrannicalleaderbehavior:therebycontributingtoexperiencesoffrustrationandprovocationthat
maytriggertraitangertoaggressiveaction(Bettencourtetal.,2006).Hence,wecontrolledforroleconflictandroleambiguity
reported by both captains and crew-members in order to investigate what personality might predict beyond such typical
situationalinfluences.RoleambiguityandroleconflictweremeasuredbythescalesofRizzo,House,andLirtzman(1970).Both
scalesconsistoffiveitems(bothα=.82),withresponsecategoriesrangingfrom1(verytrue)to7(veryfalse).Weformulatedthe
roleambiguityitemsasroleclarity,whichwereversescoredfortheanalyses.Anexampleitemofroleclarityis:“Ifeelcertainabout
howmuchauthorityIhave”,andofroleconflictis:“Ireceiveincompatiblerequestsfromtwoormorepeople”.
Inthisstudywehavechosentofocusonthenarrowbandwidth(cf.Bergneretal.,2010)traitangerandtraitanxiety,respectively,
arguingthatthetwomayrelatedifferentlytoreportsofpettytyranny;traitangerbeingassociatedwiththeperpetrationofabuse
(Anderson&Bushman,2002)andtraitanxietywithvictimization(Aquino&Thau,2009).Furthermore,aswearguetraitangerin
dyadicinteractiontofollowatemperamentalgoodness-of-fitparadigm(Kristof-Brownetal.,2005;Lerner&Lerner,1983),traitanger
mayrevealadifferentinfluencethantraitanxietyinrelationtovictimization.Sincethetwotraitsareknowntocovariatewithina
broaderdimensionofneuroticismornegativeaffectivity,thetwonarrowbandwidthtraitsmayactasconfoundersforeachother.
Therefore,wheninvestigatingtheinfluenceonpettytyrannyofleadertraitanger(Hypothesis1),wecontrolledforleadertrait
anxiety.Wheninvestigatingtheinfluenceofsubordinatetraitanxiety(Hypothesis2),wecontrolledforsubordinatetraitanger.
Wheninvestigatingtheinteractionofleadertraitanger×subordinatetraitanger(Hypothesis3),wecontrolledfortheinteraction
termofleadertraitanger×subordinatetraitanxiety,alltocomposethestrictestpossibletestofourhypotheses.
3.3.Analysis
Anotablefeatureofthesamplingframeisthatallmembersofateamsharedthetargetoftheirratings,namelythecaptain,thus
forminganesteddatahierarchy.Giventhissamplingprocedure,weexpectedteammemberratingsofcaptains'pettytyrannyto
show some dependence. Dependence between observations had implications for data analysis. First, given that dependence
between observations of the same captain couldbe expected, it wasnecessary toassess the levelof consistency across crew
Table1
Disaggregatedcorrelationsbetweenstudyvariables,accountingforteamclustereffectsa.
Variable Mean SD Correlation
1 2 3 4 5 6 7 8
1.Pettytyranny 1.62 0.79
⁎
2.Crewmembertraitanxiety 1.59 0.38 .17
⁎⁎⁎
3.Crewmembertraitanger 1.51 0.32 .13 .27
⁎⁎⁎
4.Crewmemberroleambiguity 2.26 0.87 .05 .37 .07
⁎⁎ ⁎⁎ ⁎⁎
5.Crewmemberroleconflict 3.1 1.27 .27 .15 .22 .27
6.Captaintraitanxiety 1.47 0.35 .05 .03 −.07 .11 .01
⁎ ⁎⁎⁎
7.Captaintraitanger 1.41 0.28 .21 .00 .09 .06 .04 .42
⁎
8.Captainroleambiguity 2.1 0.60 .12 .11 .08 .07 .01 .22 .01
9.Captainroleconflict 3.38 1.28 .12 −.01 .02 .09 .03 .49⁎⁎⁎ .31⁎⁎ .19
a n=169withlistwisedeletion.
⁎ pb.05.
⁎⁎ pb.01.
⁎⁎⁎ pb.001.
L.Kantetal./TheLeadershipQuarterly24(2013)106–124 113
members.Weneededtoaccountforastronglevelofdependenceintheanalysis.Ifwecontrarilyshouldhaveignoreddependence,
standardparametricmodelsondependentdatawouldhavetendedtodeflatestandarderrors,resultinginincreasedlikelihoodof
typeIerror.Second,asthetheoreticalconceptofpettytyranny,asopposedtoabusivesupervision,doesnotexcludeteam-level
influences(Tepper,2007),multilevelanalysisisusefultoilluminatetheconceptbeyondthemerenecessitiesofahierarchical
structureindata.Insum,wechosetouseamultilevelmodeltoaccountforthedependency:adecisionnecessaryduetothe
clusteredsampledesignaswellastheoreticalargumentsrelatingtoHypothesis1investigatingtraitangerasanantecedentof
pettytyrannyasanaverageleadershipstyle,hencebehaviorstobeobservedbymostorevenallmembersoftheteam.
Toanalyzetheroleofpersonalitytraitsoncaptains'behaviortowardscrewmembers,weusedthemany-perceivers,onetarget
design(MP1T),asoutlinedinKenny,Kashy,andCook(2006).IntheMP1Tdesign,theleader(herethecaptain)isthefocalperson,
and multiple partners rate the captain's behavior towards crew members in general. The MP1T can be viewed as a nested
hierarchy,withcrewmemberreportsinthisstudyclusteredwithinteams,withonecaptainratedperteam,andcanbefittedusing
multilevelmodeling(Goldstein,2003).InlinewithKennyetal.(2006),weformulatedatwo-levelanalysis,withcrewmembersas
level1andthefocalpersonaslevel2.BasedontheMP1Tdesign,weassumethatthecharacteristicsofthefocalperson,which
affectallcrewmembers,willcausethecorrelationbetweencrewmemberratingsofthefocalperson.Inamultilevelmodel,we
would expect indications of such a correlation by the extent to which reportsof leadership behaviorvary across leaders but
comparativelylittlebetweencrewmembersratingthesameleader.Thus,asastartingpoint,weconductedamultilevelvariance
componentmodelforpettytyranny,withrandominterceptsattheleaderlevel.Theobjectiveofthisanalysiswastoexaminethe
intra-class correlation for crew member reports, indicating the extent to which crew member's ratings of their leader are
consistent. Consistent reports give reason to believe that crew member reports reflect a consistent, generalized, or average
leadershipstyleofacaptaintowardsthecrewmembersasateam.Bycontrast,lowconsistencyincrewmemberreportsofpetty
tyrannyislikelytoreflectadyadicandspecificinteractionpatternbetweenacaptainandeachindividualcrewmember.
Themainanalysisofthestudywasamultilevelregressionanalysis,regressingcrewmemberreportsofleaderbehavioron
team-levelandindividual-levelcharacteristics.Thekeyobjectiveofthisanalysiswastoisolatetheassociationbetweencaptain
characteristicsandcrewmemberreportsofpettytyranny.Weintendedtoassesstheimpactof,respectively,captainandcrew
memberpersonalitytraitsonpettytyranny,controlledforrolestressorsexperiencedbycaptainsandindividualcrewmembers
alike.Duetotheskeweddistributionofleadershipbehavior,weconductedrobustmaximumlikelihoodestimationsofparameters
andYuan–Bentlercorrectedchi-squarestatisticsusingMplus5.2.
Inthisstudy,missingdatamightoccurattwolevelsofanalysis:theindividuallevelobservations,andthebetweenteam-observations.
Inacompletecaseanalysis,allcaseswithmissingteamlevelvariables(captainreported)ormissingindividual-level(crewmember
reported)informationontheindependentvariableswouldbeexcludedfromtheanalysis.Forteamswithoutcaptainobservationsonthe
independentvariables,allindividualswithintheteamwouldalsobeexcluded.Assumingamissingatrandom(MAR)mechanismwe
thususedfull-informationmaximumlikelihood(FIML)estimationofmissingdatawithinandbetweenteams.Intheregressionmodels
thatincludedcross-levelinteractiontermsofcaptainandcrewmembervariables,thecomputedcross-levelinteractiontermwasbased
onobservedinformationonly.Thus,thereportedanalysiswasbasedonresponsesfrom177crewmembersand84teamswithcaptain
information,althoughthefullpotentialsampleinatwo-levelmaineffectsmodelwouldbe272individualsfrom134teams.
4.Results
Table 1 shows the disaggregate bivariate correlations between study variables.Petty tyranny showedstatistically significant
asociationswithcrewmembertraitanxietyandcrewmemberroleconflict,andwithcaptaintraitanger.Table2showstheresults
fromamultivariatetwo-levelunrestrictedmodel ofrelevantstudyvariables.Itcanbeseenthatatthewithin-team-levelpetty
tyrannycorrelatedmoderatelywithcrewmembersroleconflict(r=0.37),butonlyweaklywithcrewmemberstraitanxietyandtrait
Table2
Estimatedcorrelationsfrommultivariatetwo-levelmodelofrelevantstudyvariablesa.
Variable Mean SD Correlation
1 2 3 4 5
Level1—withinteam
1. Pettytyrannyb 0.00 0.60 (.81)
2. Crewmembertraitanxiety 1.60 0.15 .19 (.88)
3. Crewmembertraitanger 1.51 0.11 .06 .30 (.74)
4. Crewmemberroleambiguity 2.25 0.87 .09 .38 .06 (.82)
5. Crewmemberroleconflict 3.14 1.75 .37 .27 .22 .26 (.82)
Level2—betweenteam
1. Pettytyranny 1.71 0.10
2. Captaintraitanxiety 1.49 0.13 .13
3. Captaintraitanger 1.43 0.08 .50 .40
4. Captainroleambiguity 2.13 0.37 .23 .20 −.05
5. Captainroleconflict 3.46 1.42 .26 .38 .26 .07 −
a Crewmembern=177,captainn=84.Valuesinparenthesesarealphacoefficients.
b Level1—withinteampettytyranny(1)meaniscentered.
114 L.Kantetal./TheLeadershipQuarterly24(2013)106–124
anger.Theteam-levelvariationofcrewmember-reportedpettytyrannywasstronglycorrelatedwithcaptaintraitanger(r=0.50),
butonlyweaklyrelatedtootherteam-levelvariables.
4.1.Variancecomponentsofcrewmemberreportsofpettytyranny
Akeyquestioniswhetherthecrewmemberreportsoftheirleadersareinagreement,asindicatedbythesizeoftheintraclass
correlation. Consistent reports give reason to believe that crew member reports reflect a consistent, generalized, or average
leadershipstyleofacaptaintowardsthecrewmembersasateam.Incontrast,lowconsistencyincrewmemberreportsofpetty
tyrannyislikelytoreflectadyadicandspecificinteractionpatternbetweenacaptainandeachindividualcrewmember.Table3
showstheresultsofatwo-levelvariancecomponentmodel.Thekeyvariableofinterest—pettytyranny—showedasubstantial
between-teamvariancecomponent.TheICC1was0.19,andtheICC2was0.33forthemeanofraters.Whiletheobservedpattern
suggestedstrongdependenceincrewmemberratingsoftheircaptain'spettytyranny,theagreementwastoolowtowarranta
team-levelapproach.TheICC2of0.33suggeststhataggregatedanalysesusingthemeanofcrewmemberratingswouldnotbea
reliablemeasureofleadershipbehaviorunderthecurrentgroupssize.
4.2.Multilevelregressionmodeling
Conceptually, the strong dependence suggests multiple sources of influence on crew member ratings of captains' petty
tyranny.Theindividuallevelcomponentmayreflectindividualdifferencesinperception,orindividualdifferencesinrelatingto
thecaptain.Incontrast,thesharedcomponentmayreflectconsistentteam-levelinfluencesonthecrewmemberreportsofpetty
tyranny.Theobservedpatternindicatesthatapurelyindividuallevelmodelaswellasapurelyaggregatelevelmodelalonewould
failtocapturesignificantinformationabouttheinfluencesoncrewmemberreportsoftheircaptain'spettytyranny.
Tocapturebothwithin-teamandbetween-teampredictorsofpettytyranny,wespecifiedanestedtwo-levelmodel.Wetested
four such nested two-level models in a series, with the intention of first controlling for possible demographic variables and
role-stressors, then proceeding with individual crew member contributions, captains' contributions, and, finally, cross-level
interactioneffectsinordertoinvestigatethecontributionofthecaptain–crewmemberpersonalitylinkage.
Thefirstnestedtwo-levelmodelofpredictorsofpettytyranny,labeled“psychosocialfactors”,includeddemographicvariables
androlestressors:levelonepredictorswereteamtenure,gender,andcrewmember-perceivedroleambiguityandroleconflict;
level two predictors were captain-perceived role ambiguity and role conflict. To test the hypothesis that crew member trait
anxietypredictedhigherpettytyranny(Hypothesis2),thesecondmodel,labeled“crewmemberpersonality”,includedcrew
membertraitangerandtraitanxietyasindependentlevel-onepredictorsofcrewmember-reportedpettytyranny.Totestthe
hypothesisthatcaptainstraitangerpredictedhigherpettytyranny(Hypothesis1),thethirdmodel,labeled“captainpersonality”,
includedcaptaintraitangerandtraitanxietyasasharedlevel-twopredictorofpettytyranny.Totestthehypothesisthatcaptains
traitangerinteractwithcrewmembertraitangerinpredictingthepettytyranny(Hypothesis3),thefourthandfinalmodel,
labeled“dyadicinteraction”, included interaction terms betweencaptaintrait angerand crewmembertrait anger,aswell as
captaintraitangerandcrewmembertraitanxiety.Table4showstheunstandardizedmodelcoefficientsoftheseriesofthefour
nestedmodels.
Thefirstmodel,displayedinthefirstcolumnofTable4,showedamaineffectofcrewmemberroleconflictandgenderonpetty
tyranny.Beingfemalewasassociatedwithlowercrewmember-reportedpettytyranny,indicatinglessreportedpettytyranny.
Higher crew member role conflict was associated with a higher level of crew member-reported petty tyranny. In line with
Hypothesis2thesecondmodelshowedthatcrewmembertraitanxiety—butnottraitanger—wasassociatedwithahigherlevelof
crewmember-reportedpettytyranny.InlinewithHypothesis1,thethirdmodelshowedthatcaptaintraitangerwasassociated
withhighercrewmember-reportedpettytyranny.Finally,inlinewithHypothesis3thefourthmodel(seethelastcolumnof
Table 4) showed a cross-level interaction effect between captain trait anger and crew member trait anger, but not between
captaintraitangerandcrewmembertraitanxiety.
Wepresentthecross-levelinteractioneffectbetweencaptaintraitangerandcrewmembertraitangerinFig.1.Thefigureshows
thattherelationshipbetweencaptaintraitangerandcrewmemberreportsofpettytyrannywasstrongeramongcrewmemberswith
lowtraitangercomparedtocrewmemberswithhightraitanger,inlinewithHypothesis3.Asapost-hocfollow-up,weconducteda
simpleslopeanalysisofcaptaintraitangeratdifferentlevelsofcrewmembertraitanger.Standarderrorsforthesimpleslopeswere
obtainedusingtheprocedureofCohen,Cohen,West,andAiken(2003),basedoninformationfromtheasymptoticcovariancematrix.
Table3
Variancecomponentsofpettytyrannywithin-andbetweenteamsa.
Estimate SE Estimate/SE p
Intercept 1.71 0.06 30.39 0.00
Random
Within 0.57 0.08 7.01 0.00
Between 0.14 0.06 2.18 0.03
ICC1 0.19
ICC2 0.33
a ICC1=intraclasscorrelation1;ICC2=intraclasscorrelation2.
L.Kantetal./TheLeadershipQuarterly24(2013)106–124 115
Table4
Modelsummaryfrommultilevelregressionmodelwithpettytyrannyasdependentvariablea.
Variable M1:psychosocial M2:crewmember M3:captain M4:dyadicinteraction
factors Personality personality
B SE p B SE p B SE p B SE p
Intercept 1.60 0.06 *** 1.60 0.06 *** 1.60 0.06 *** 1.60 0.06 ***
Gender(malesreference) −0.36 0.14 ** −0.35 0.13 ** −0.35 0.13 ** −0.35 0.13 **
Tenureinteam,years 0.00 0.01 0.01 0.01 0.01 0.01 0.01 0.01
Crewmemberroleambiguity −0.02 0.07 −0.08 0.08 −0.07 0.08 −0.06 0.08
Crewmemberroleconflict 0.15 0.06 ** 0.15 0.06 ** 0.15 0.06 ** 0.14 0.06 **
Captainroleambiguity 0.13 0.09 0.12 0.10 0.16 0.09 0.17 0.09
Captainroleconflict 0.06 0.04 0.06 0.04 0.05 0.05 0.05 0.05
Crewmembertraitanxiety 0.36 0.18 * 0.36 0.18 * 0.34 0.18
Crewmembertraitanger −0.11 0.19 −0.16 0.19 −0.05 0.16
Captaintraitanxiety −0.22 0.22 −0.24 0.22
Captaintraitanger 0.61 0.24 * 0.63 0.24 **
Captaintraitanger×crewmembertraitanxiety 0.10 0.55
Captaintraitanger×crewmembertraitanger −1.06 0.53 *
Randomwithinteams 0.47 0.08 0.44 0.08 0.45 0.08 0.44 0.08
Randombetweenteams 0.08 0.06 0.10 0.06 0.07 0.07 0.07 0.07
R-squaredwithin 0.093 0.124 0.121 0.138
R-squaredbetween 0.117 0.099 0.404 0.419
−2(loglikelihood)b 2475.02 2470.20 2464.39 2461.59
Akaikesinformationcriterion 2541.03 2540.20 2538.39 2539.60
Yuan–BentlerscaledLRTtestc 21.64 4.21 4.30 4.66
⁎pb.05.⁎⁎pb.01.⁎⁎⁎pb.001.
a Models1through4,crewmembern=177,captainn=84.
b Underrobustmaximumlikelihood,the−2(loglikelihood)doesnotfollowachi-squaredistribution.
c M1comparedtonullmodel;M2comparedtoM1;M3comparedtoM2;M4comparedtoM3.
At high levels of crew member trait anger, the statistical association between captain trait anger and petty tyranny was not
significantlydifferentfromzero(B=0.13,SE=0.30).Incontrast,atlowlevelsofcrewmembertraitanger,thestatisticalassociation
wasstrongandsignificantlydifferentfromzero(B=0.93,SE=0.26).
5.Discussion
Leaders,includingtheseafaringcaptainsinvestigatedinthisstudy,mayattimesneedtoacttough,evendisplayinganger,but
withoutpassingintotheimproper.Itis,however,easytostepovertheline.Inappropriatelylordingonespoweroverothers—
displayingpettytyranny(Ashforth,1994)—isaconsiderableprobleminworkinglife(cf.Tepper,2007).Personalitytraits,beitthe
leader'sortheirsubordinate's,willinfluencebothindividualbehaviorandperceptionthereof,aswellashowtwoindividualsfit
together.Itisthereforevitaltounderstandtheextenttowhichpersonalitytraitsoftheperpetrator,thevictim,andthefitoftheir
combinedtraitscontributetotheaggressivebehaviorofpettytyranny.
Inthisstudy,weinvestigatedleader(captain)andsubordinate(crewmember)traitangerandtraitanxietyasantecedentsof
pettytyranny.First,onateamlevel,wehypothesizedself-reportedleadertraitangertopredictsubordinate-reportedpettytyranny.
The results supported our first hypothesis (Hypothesis 1). Secondly, on an individual level, we hypothesized that self-reported
subordinatetraitanxietypredictedsubordinate-reportedpettytyranny.Theresultssupportedoursecondhypothesis(Hypothesis2),
althoughtheresultsrevealedaratherweakrelationshipbetweenthetwo.Thirdly,onadyadiclevel,wehypothesizedself-reported
Fig.1.Interactionofcaptaintraitangerandcrewmembertraitanger.