Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are encouraged to visit:
http://www.elsevier.com/copyright
ContentslistsavailableatSciVerseScienceDirect
Social Networks
j ou rn a l h o m e pa ge :w w w . e l s e v i e r . c o m / l o c a t e / s o c n e t
A relationship between verbal aggression and personal network size
Koen Vanbrabant
a,∗, Peter Kuppens
a, Johan Braeken
b, Evelien Demaerschalk
a, An Boeren
a, Francis Tuerlinckx
aaDepartmentofPsychology,UniversityofLeuven,Leuven,Belgium
bDepartmentofMethodologyandStatistics,TilburgUniversity,Tilburg,TheNetherlands
a r t i c l e i n f o
Keywords:
Personalnetworksize Overtaggression Verbalaggression
a b s t r a c t
Aggressionhasbeenassociatedwithnegativesocialconsequences.Yet,moreadaptiveviewsofaggres- sionholdthatitcanhavebeneficialcorrelatesaswell.Infourstudies,weexaminedtherelationship betweenaggressionandpersonalnetworksize,apropertyassociatedwithimportantsocialbenefits.The resultspointedtoaconsistentpositiverelationshipbetweenverbalaggressionandsocialnetworksize.
Thisrelationshipremainedaftercontrollingforthirdvariableslikeextraversionandwhenusingdiffer- entmeasuresofnetworksize.Thefourthstudysoughttoexplaintherelationshipsintermsofmutual relationshipswithstatusandpower,butfoundnosupportformediation.
© 2011 Elsevier B.V. All rights reserved.
Aggression is an integral part of human nature. Due to its harmfulconsequencesmuchresearchisdevotedtoitscausesand control. Yet, the fact that aggression has persisted throughout humanhistorysuggeststhatitmayhaveanadaptivesideaswell.
Inthispaper,weexaminetheadaptiveversusmaladaptivenature ofaggressionbyexamininghowitisassociatedwithanimportant socialasset,namelypersonalnetworksize.
1. Aggression:adaptiveormaladaptive?
Alongstandingtraditioninpsychologicalresearchviewsaggres- sionassociallymaladaptive(Smith,2007).Forinstance,Hawley (2003)providesa reviewof researchthat associatesaggression with,among others,peerrejection(CoieandDodge,1998),low educationalachievement(BrooksandNewcomb,1995),andunem- ployment(Caspietal.,1987).Consequently,severaldevelopmental theoriesfocusonthemaladaptivesideofaggressionandexclude theideathataggressioncanbeassociatedwithsocialadaptation (Hawley,2003).
Ontheotherhand,itisnotparticularlydifficulttofindpeo- plewhobehaveaggressivelybutarewelladapted.Wemaythink ofinstancesofaggressioninlargecompanymanagers(e.g.,Steve Jobsallegedaggressivemanagementstyle,seee.g.,Colvin,2007), politics(e.g.,politicianNeilFaragewhoinsultseuropresidentHer- manVanRompuyonFebruary24,2010),orsports(e.g.,theHaka
∗ Correspondingauthorat:CenterforthePsychologyofLearningandExperimen- talPsychopathology,FacultyofPsychologyandEducationalSciences,Universityof Leuven,Tiensestraat102,3000Leuven,Belgium.Tel.:+3216326117;
fax:+3216326099.
E-mailaddress:Koen.Vanbrabant@ppw.kuleuven.be(K.Vanbrabant).
wardanceinrugby).Mostevolutionarytheoriesendorsetheidea thataggressivebehaviourmustbeadaptive(Smith,2007)other- wiseitwouldnothavebeensuchaprominentfeatureofhuman evolutionaryhistory(Buss,2005).Inlinewithanadaptiveviewon aggression,researchbyTiedensandcolleaguesshowedthatthe overtexpressionofaggressionisassociatedwiththeperceptionof statusandsocialpower(Tiedensetal.,2000).Alsootherresearch hasshownthataggressioncanyieldimportantsocialbenefits,such as positive friendshipqualitiesand personal disclosure (Averill, 1983;GrotpeterandCrick,1996)andsocialstatus(Tiedens,2001).
Inthispaper,wefocusontherelationshipbetweenaggressionand aparticularsocialoutcome,namelyaperson’spersonalnetwork size(PNS).
2. Personalnetworksize
Inhumans,personalnetworksseemtobeconstitutedoutofa seriesofhierarchicallevels(Zhouetal.,2005).Thisrangesfrom thesupportclique,fromwhomonewouldseekadviceandemo- tionalsupportintimesofdistress,overthesympathygroup,which consistsofpersonswithwhomonehadcontactthelastmonth,to theactivenetwork.Thislastlevelconsistsofallindividualswith whomonehasapersonalrelationship,andfeelstheneedtokeep incontactwith(DunbarandSpoors,1995;Killworthetal.,1998).
Theactivenetworkconsistsofnotonlypersonsyoufeelclose with,butalsomoredistantacquaintances.Yet,theseweakerties stillhaveimportantbenefits,becausetheyarecrucialinproviding accesstoinformation,ideasandexperiences.Weaktiesaremore numerous,moreheterogeneousandarelesslikelytobeconnected witheachotherthanstrongties(Granovetter,1983).Theseweaker connections play an important role in the life of individuals 0378-8733/$–seefrontmatter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.socnet.2011.10.008
(Christiakisand Fowler,2009).Forexample,researchpointsout thatpeoplewitha largepersonalnetworkhavemorequalityof lifethanpeoplewithasmallpersonalnetwork(Stokes,1983).In conclusion,havingalargepersonal networkisconsideredtobe beneficial.
3. Thisstudy
Thepresentstudyexaminestheadaptiveversusmaladaptive natureofaggressionfromanovelperspectivebyexaminingthe relationshipbetweenindividualdifferencesinaggressionandPNS.
Nopreviousresearchhasdirectlyexaminedthisrelationship,and weadopted aprimarilyexplorative approach (seeRozin, 2009) inwhichwesoughttoexaminewhetheraggressionandPNSare interrelatedandwhich directionthis relationshiptakes. Onthe onehand,themaladaptiveviewofaggressionwouldarguethat highlevelsofaggressionshouldbeassociatedwithnegativesocial consequencesleadingtosmallerPNS.Ontheotherhand,ifaggres- sionservesanadaptivesocialfunctiononecouldexpectapositive associationbetweenaggressionandPNS.
Several techniques have been developed to measure PNS.
Throughoutthefourstudies,wereliedontwoindirectnetwork estimators,thescale-upmethod(Studies1,2and4;Killworthetal., 1998)andthenamegeneratoroverlaptechnique(Study3;Agresti, 1994).
4. Study1
Theaimofthefirststudywasasimpleexplorationoftherela- tionshipbetweenaggressionandPNS.
4.1. Method 4.1.1. Participants
Asampleof178firstyearpsychologystudents(150women;age M=19;SD=2;agerangefrom18to44)participatedinreturnfor coursecredits.
4.1.2. Materialsandprocedure
Aggression.The(Dutchtranslationofthe)Buss–PerryAggression Questionnaire(BPAQ;BussandPerry,1992)hasbeenusedexten- sivelyasameasureofaggressioninpreviousresearch.Itmeasures physicalaggression,verbalaggression,anger,andhostilityvia29 items.Asourfocusisonaggression,weonlyconsideredscoreson theverbalandphysicalaggressionsubscales.
Personalnetworksize.PNSor“degree”wasestimatedforevery participantusingthescale-upmethod(Killworthetal.,1998).A personalnetworkisdefinedasthesetofpeoplethatoneknows (e.g.,Bernardetal.,1989).In ordertodefineandexplaintothe participantswhatismeantbyknowingsomeone,wefollowthe definitionofBernardandcolleagues:personAknowspersonBifA knowsBbynameorbyface;AcancontactBbymailorphone;and AandBhadcontactatleastoncethelasttwoyears.
Thescale-upmethodusesquestionsoftheform“Howmany peopleof[acertainsubpopulationofpeople]doyouknow?”An exampleofsuchaquestionis:“Howmanypeopledoyouknow whogotmarriedinthelast12months?”Ifoneknowsthesizeofthe subpopulation,thisinformationcanbeusedtoestimatethePNSof aparticipant.Forinstance,whenthissurveywasconducted,89,700 Belgianpeoplegotmarriedinthelast12monthswhilethesizeof theentireBelgianpopulationwas10,666,866inhabitants.Ifapar- ticipantknowsfiveindividualsthatgotmarriedinthelasttwelve months,thePNSforanindividualcanbeestimatedasfollows:
5
89,700·10,666,866≈714
10 15 20
34567
Verbal aggression
log Personal Network Size (PNS)
Fig.1.Scatterplotdisplayingtherelationbetweenpersonalnetworksize(onthe abscissa)andverbalaggression(ontheordinate).Thefulllineisalinearregression line;thedottedlineisalowesssemi-parametricregressionline(Study1).
Byaveragingtheresponsesofarespondentformanysubgroupsthe accuracyoftheestimatecanbeincreased.Themultipleresponses needtobecombined:
ˆdi=
K k=1yik K k=1Nk·N (1)
whereyikisthenumberofpeoplethatpersoniknowsinsubpopu- lationk,Nkrepresentsthesizeofthesubpopulationk,andNisthe sizeofthewholepopulation.
In this study a questionnaire with 32 subpopulations was administered.Thesubpopulationconsistedof12firstnamesand 20othersubpopulations(seeAppendix).
4.2. Results
Theaveragesizeof theparticipant’sPNSwas290(SD=241) degrees. Thedistribution ofPNS waspositivelyskewedsothat thesizeofnetworksoftwoindividualscandifferbyanorderof magnitude.Asourstatisticaltoolsassumelinearityandadditivity and these assumptions are not fit for data on a multiplica- tivescale, PNSwas log-transformed(hereafter named logPNS).
There was a positive correlation between logPNS and verbal aggression (r(178)=0.26, p<0.0002). Fig. 1 graphically displays this relationship. The correlationbetween logPNSand physical aggressionwasnon-significant(r(178)=0.09,p=0.12)(verbaland physical aggression correlated 0.47, p<0.0001). In a regression analyses, verbal aggression remained a significant predictor of logPNS(t(174)=3.02,p=0.003,ˇ=0.05)whencontrollingforphys- ical aggression (t(174)=−0.79, p=0.43, ˇ=−0.008) and gender (t(174)=−1.69,p=0.094,ˇ=−0.27).
4.3. Discussion
Study1providedfirstevidencethatverbalaggressionisposi- tivelyassociatedwithPNS.Thefindingsfurthersuggestthatthereis noassociationbetweenphysicalaggressionandPNSandthatgen- derdoesnotseemtoplayaroleinthefoundrelationship.However, aweaknessofStudy1isthatitdidnotcontrolforothervariables
Table1
CorrelationsofallvariablesinStudy2.
1. 2. 3. 4. 5. 6. 7. 8. 9.
1.PNS 1.00**
2.logPNS 0.93** 1.00**
3.Verbalaggression 0.25** 0.26** 1.00**
4.Physicalaggression 0.10 0.07 0.24** 1.00**
5.Neuroticsm −0.01 0.00 −0.11 0.13* 1.00**
6.Extraversion 0.16* 0.21** 0.20** 0.00 −0.47** 1.00**
7.Openness 0.15* 0.11 0.10 −0.17* −0.10 0.00 1.00**
8.Agreeableness −0.06 −0.07 −0.31** −0.40** −0.18* 0.27** 0.26** 1.00**
9.Conscientiousness 0.02 0.01 0.00 −0.18** −0.21** 0.14* 0.13* 0.31** 1.00**
*p<0.05.
**p<0.01.
thatmightpossiblyexplaintherelationbetweenverbalaggression andPNS.ThiswillbeaddressedinStudy2.
5. Study2
Extraversionispositivelyrelatedtoaggressionandtheovert expressionofanger(e.g.,BödekkerandStemmler,2000),andhas beenassociatedwiththepropensitytoconnectwithothers(e.g., Totterdeletal.,2008).
TherelationshipbetweenverbalaggressionandPNSmaythere- foresimplybeduetotheirmutualassociationwithextraversion.In Study2weexaminedtheassociationbetweenaggressionandPNS inadifferentsample,whilecontrollingforextraversion.Wealso controlledfortheotherFiveFactorpersonalitytraits,someofwhich alsodemonstraterelationshipswiththevariablesunderstudy(for instanceagreeablenessandaggression;Burtonetal.,2007).
5.1. Method 5.1.1. Participants
Asampleof187firstyearpsychologystudents(157women;age M=19;SD=2,agerangefrom17to36)participatedinreturnfor coursecredits.
10 15 20 25
4.04.55.05.56.06.57.0
Verbal aggression
log Personal Network Size (PNS)
Fig.2.Scatterplotdisplayingtherelationbetweenpersonalnetworksize(onthe abscissa)andverbalaggression(ontheordinate).Thefulllineisalinearregression line;thedottedlineisalowesssemi-parametricregressionline(Study2).
5.1.2. Materialsandprocedure
Personality.TheDutchversionoftheNEO-FFI(Hoekstraetal., 1996)wasusedtoassesneuroticism,extraversion,opennessto experience,agreeablenessandconscientiousness.Eachscalecon- sistsof12items.
Aggression.TheBPAVwasusedsimilartoStudy1.
Personalnetworksize.LikeStudy1,Study2reliedonthescale- upmethodtoestimatepersonalnetworksize.Thesurveyagain consisted of32 items,however usingdifferentfirst names(see Appendix).
5.2. Results
Thesizeoftheparticipant’spersonalnetworkwas253(SD=149) degrees on average, and was positively skewed. There was a positive correlation between logPNS and verbal aggression (r(187)=0.26,p=0.0002;seeFig.2),whilethecorrelationbetween logPNSandphysicalaggressionwasnon-significant(r(187)=0.07, p=0.46).Onlyextraversionwasasignificantpersonalitycorrelate of logPNS(r(164)=0.19, p<0.005) (see Table 1 for correlations betweenallvariables).
Inaregressionmodel,weincludedverbalaggressionandphys- icalaggressionaspredictorsoflogPNS,whilecontrollingforthe fivepersonalitydimensionsandgender.Thefindingsshowedthat logPNSremainedassociatedwithverbalaggression(t(133)=2.40, p=0.03, ˇ=0.03), while there was again no significant rela- tionwithphysicalaggression(t(133)=−0.49,p=0.62,ˇ=−0.003).
Extraversion (t(133)=3.86, p<0.0001, ˇ=0.03) and neuroti- cism(t(133)=2.11,p=0.04, ˇ=0.01)and gender(t(133)=−2.82, p=0.006,ˇ=−0.35),weretheonlyothersignificantcovariatesin themodel.
5.3. Discussion
TheresultsofthissecondstudyreplicatedthefindingsofStudy 1, even whencontrolling forextraversion(aswell astheother BigFivepersonalityfactors).Weconcludethatwhenthescale-up methodisusedtoestimatesocialnetworksize,thereisastable associationbetweenverbalaggressionandPNS.
Wefoundamaineffectofgenderinthisstudy,withmalesbeing moreaggressivethanfemales,althoughwewouldadvisecaution toover-interpretthereliabilityof thisfinding,duetothesmall numberofmaleparticipants.Thusfar,however,itcannotberuled outthatthefoundrelationshipisspecifictotheusedmethodto estimatePNSinbothstudies.
6. Study3
Study3usedadifferentmeasuretoestimatePNS,namelythe
“namegeneratoroverlap”technique(Killworthetal.,1984).We alsoaimedatreplicatingthefindingthattherelationshipbetween
PNSandverbalaggressionwasnotduetosharedvariancewith personality.
6.1. Method
6.1.1. Participants
Asampleof74firstyearpsychologystudents(62women,age M=19;SD=2,range17–36)participatedinreturnforcoursecred- its.
6.1.2. Materialsandprocedure
Personality.PersonalitywasmeasuredwiththeNEO-FFI.
Aggression.Verbalandphysicalaggressionweremeasuredwith thesameinstrumentasinStudies1and2.
Personal network size. Following the “name generator over- lap” technique, an estimate can be made of the size of the (sub)population on the basis of counting individuals who are namedinmorethanonesample.Forthistechnique,thePetersen estimator(Agresti,1994;Petersen,1896)isthemoststraightfor- wardestimator.ConsidersampleAandsampleB,havingsample sizesnAandnBeachwithanumberofindividualsthatareacquain- tancesofanindividual.Thenumberofindividualsthatappearin bothsamplesareenumerated,nAB.Underindependence,itholds that
nAB
N = nA
N ·nB
N
SolvingfortheunknownNgivesthePetersenestimator ˆNofthe totalnumberofacquaintances:
Nˆ= nA·nB
nAB
ThismethodcanbeappliedtothestudyofPNSasfollows.To generatea first sample of acquaintances(called List A)we, for example,usethereversedsmall-worldtechnique(Bernardetal., 1982)andask“Whichacquaintancescangetyouincontactwiththe primeministerofBelgium?”Theparticipantforexampleresponds withfournames(nA=4;e.g.,MarkB.,MichelleO.,GeorgeL.and MarieV.)inthisfirstnamegeneratormoment.Togenerateasec- ondsampleofacquaintances(calledListB)we,forexampleuse thephonebooktechnique(PoolandKochen,1978).Thepartici- pantreceivesfiverandomchosensurnamesoutofaphonebook, andreports allacquaintanceswiththesesurnames.Inthis sec- ondnamegeneratormoment,theparticipantrecalledforexample sevennamesofacquaintances(nB=7;e.g.,KarelB.,Anne-MarieM., GeorgeL.,BartV.,MarieV., PieterB.,DriesL.and ElsL.).When wecomparethesetwonamegeneratormoments,weseethattwo namesappearinbothlists(nAB=2).Basedonthisoverlapinname
1.5 2.0 2.5 3.0 3.5 4.0 4.5
1.82.02.22.42.62.8
Verbal aggression
log Personal Network Size (PNS)
Fig.3.Scatterplotdisplayingtherelationbetweenpersonalnetworksize(onthe abscissa)andverbalaggression(ontheordinate).Thedottedlineisalinearregres- sionline;thefulllineisalowesssemi-parametricregressionline(Study3).
generatormomentswecanmakeanestimateoftheparticipant’s PNSbyusingtheabovementionedformula:
N =ˆ nA·nB
nAB =4·7 2 =14
The instrument for this study consisted of five naming moments:thephonebooktechniquewith100surnames,thephone booktechniquewith100firstnames,thesmallworldtechnique wasusedtwotimeswithdifferentquestions,andallparticipants hadtolistallofthepersonstheyconsideredtobeintheirsup- portgroup.Withtheoverlapinnamesbetweenthesemoments, wewereabletoestimatethePNSforeveryparticipant.
6.2. Results
Thesizeoftheparticipant’spersonalnetworkwasonaverage 229 (SD=170) degrees, and was positively skewed. After log- transformationtherewasapositivecorrelationbetweenlogPNS andverbalaggression(r(74)=0.32,p=0.003;seeFig.3).logPNSwas notcorrelatedwithphysicalaggression(r(74)=0.08,p=0.249)but waspositivelyrelatedtoextraversion(r(74)=0.34,p=0.002;see Table2forthecorrelationsbetweenallvariables).
A regression analysis was performed that included verbal aggression and physical aggressionas predictors of PNS, while controllingforthefivepersonalitydimensionsandgender.Verbal aggressionremainedasignificantpredictoroflogPNS(t(65)=2.62,
Table2
CorrelationsbetweenmeasuresofStudy3.
1. 2. 3. 4. 5. 6. 7. 8. 9.
1.PNS 1
2.logPNS 0.94** 1
3.Verbalaggression 0.27* 0.32** 1
4.Physicalaggression 0.09 0.08 0.33** 1
5.Neuroticism −0.01 0.00 −0.06 0.11 1
6.Extraversion 0.26* 0.34** −0.06 −0.23 −0.15 1
7.Openness 0.01 −0.04 0.31** 0.10 0.18 −0.21 1
8.Agreeableness −0.19 −0.18 −0.26* −0.55** −0.09 0.34** 0.00 1
9.Conscientiousness −0.20 −0.19 −0.15 −0.27* −0.26* 0.29* −0.06 0.36** 1
*p<0.05.
**p<0.01.
p=0.011, ˇ=0.30), after controlling for physical aggression (t(65)=−1.15,p=0.255,ˇ=−0.14),gender(t(65)=−1.85,p=0.069, ˇ=−0.22),andtheBigFivepersonalitydimensions.Onlyextraver- sionwasasignificantpersonalitycovariate(t(65)=4.35,p<0.0001, ˇ=0.47).
6.3. Discussion
Study3againprovidedevidencethatverbalaggressionisposi- tivelyassociatedwithPNS,evenwhenusingadifferentestimator ofPNS.Thisagainheldaftercontrollingforthemajorpersonal- itydimensionsincludingextraversionandgender.Throughoutthe previousstudies,wehavedemonstratedaconsistentpositiverela- tionbetweenverbalaggressionandPNS.Anattempttoexplainthis relationshipwasmadeinStudy4.
7. Study4
Tiedensetal.(2000)arguethathighstatusisassociatedwith emotionsincludinganger and pride whilelow status is associ- atedwithsadnessandguilt.Theauthorsconcludethatemotions areinferredfromstatus andthat statusinturncanbeinferred fromtheexpressionofthecorrespondingemotions.Also,research byCillessenandMayeux(2004)foundthatadolescentaggression canleadtohigherpeerstatus.Insum,thedisplayofaggressionis relatedtostatus.Highstatus,inturn,maybeassociatedwithan increaseinPNS(Belle,1983).Notonlymightapositionofstatus bringapersonintouchwithmorepeople(Keltneretal.,2003), otherpeoplearealsomotivatedtoincludeapersoninhighstatus positionsintotheirpersonal network,giventheexpectedsocial advantages.
Theabovesuggeststhatthenotionofstatuscouldhelptoexplain therelationshipbetweenPNSandverbalaggression.Specifically, perceivedpowerorstatusmayfunctionasamediatoroftherela- tionshipbetweenverbalaggressionandPNS.Weaimedtoexamine thispossibilityinafourthstudy.Weincludedtwodifferentmea- suresreflectingstatus: onedesigned tomeasure thesubjective senseof personal power and the other reflectingself-reported sociometricstatus.
7.1. Method 7.1.1. Participants
Asampleof52undergraduatepsychologystudents(32women, ageM=21;SD=1,range19–24)participatedinreturnforcourse credit.Oneparticipantwasexcludedfromtheanalysesbecause theparticipant’sPNSestimatewasclearlyanoutlier(>3000).
7.1.2. Materialsandprocedure
Extraversion.TheextraversionscaleoftheDutchversionofthe NEO-FFIwasused.Onlyextraversionwasusedbecausetheother threestudiespointedoutthatthiswastheonlysignificantperson- alitycovariate.
Aggression.TheBPAVwasusedtomeasureverbalandphysical aggression.
Personalnetworksize.PNS wasestimatedusing thescale-up method,identicaltoStudy2(seeAppendix).
Subjective power. Subjective power was measured with a DutchtranslationoftheSubjectiveSenseofPowerQuestionnaire (AndersonandGalinsky,2006).Thequestionnaireconsistsofeight questionsallratedonafivepointLikertscale(e.g.,“IthinkIhavea greatdealofpower”).
Self-reported sociometric status. Sociometric status is usually measuredusingpeer ratingsin a so-calledroundrobindesign, whichwaspracticallyunfeasibleinoursample.For thatreason
12 14 16 18 20 22
4.55.05.56.06.5
Verbal aggression
log Personal Network Size (PNS)
Fig.4.Scatterplotkdisplayingtherelationbetweenpersonalnetworksize(onthe abscissa)andverbalaggression(ontheordinate).Thefulllineisalinearregression line;thedottedlineisalowesssemi-parametricregressionline(Study4).
the existing peerrating questionnaire wasconvertedto a self- reportquestionnaire.Thequestionnaireconsistedofnineitemsall ratedonafivepointLikertscale.Thefirstfiveitemswereinspired byAndersonetal.(2006);theotherfouritemswereinspiredby CillessenandMayeux(2004).WeperformedaPrincipalComponent Analysis(PCA)withcomponent1explaining40%ofthevariance.
Consequently,scoresonthefirstprincipalcomponentwereused asameasureofself-reportedsociometricstatus.
7.2. Results
Asin Study2,thesizeoftheparticipant’spersonalnetwork was249(SD=133)andpositivelyskewed.Afterlog-transformation therewasapositivecorrelationbetweenestimatedPNSandver- balaggression(r(51)=0.37,p=0.004;seeFig.4).Inthisstudythe correlationbetweenPNSandphysicalaggressionwasalsosignifi- cant(r(51)=0.35,p=0.005;seeTable3forcorrelationsbetweenall variables).
Thedatawerefurtheranalysedtotestforthemediationmodel describedabove.Inthismediationmodel,sociometricstatusand subjectivepowerweretreatedaspotentialmediatorsfortherela- tionshipbetweenPNSandverbalaggression.Similartotheother studies, extraversion was included as a covariate. The media- tionanalyseswereconductedaccordingtothemethodproposed by Baron and Kenny (1986).The first step showed that verbal aggression,t(49)=2.80,p=0.007,ˇ=0.06,isasignificantpredic- toroflogPNS,evenwhencontrollingforextraversion,t(49)=1.52, p=0.14, ˇ=0.02. Analyses in a secondstep showedthat verbal aggression, t(49)=2.50, p=0.016, ˇ=0.31, is a significant pre- dictor for subjective power after controlling for extraversion, t(49)=2.24, p=0.03, ˇ=0.14;and verbalaggression,t(49)=2.85, p<0.01,ˇ=0.23isasignificantpredictorofsociometricstatus,after controlling for extraversion, t(49)=3.30, p=0.002, ˇ=0.13. The laststep,however,showedthatsociometricpower,t(47)=−0.722, p=0.47,ˇ=−0.02andstatus,t(47)=0.35,p=0.35,ˇ=0.013arenot significantpredictorsoflogPNSaftercontrollingforextraversion, t(47)=1.40, p=0.17, ˇ=0.02 and verbal aggression, t(47)=2.59, p=0.013,ˇ=0.059.Inthis last model,therelationshipbetween verbalaggressionandlogPNSremainedsignificant,evenaftercon- trollingforstatus,powerandextraversion.SeeFig.5foravisual
Table3
CorrelationsbetweenmeasuresofStudy4.
1. 2. 3. 4. 5. 6. 7.
1.PNS 1.00**
2.logPNS 0.93** 1.00**
3.Verbalaggression 0.32* 0.37** 1.00**
4.Physicalaggression 0.33** 0.31* 0.35** 1.00**
5.Extraversion 0.26* 0.21 0.03 0.14 1.00
6.Power 0.07 0.10 0.33** 0.25* 0.30* 1.00**
7.Status 0.20 0.23 0.35** 0.11 0.41** 0.45** 1.00**
*p<0.05.
**p<0.01.
Fig.5. ThemediationmodeltestedinStudy4withcorresponding regression weights.Extraversionwasincludedasacovariatewhentestingthewholemodel (**p<0.01).
overviewofthemediationanalyseswiththecorrespondingstan- dardisedregressionweights.
7.3. Discussion
VerbalaggressionandPNSwereagainpositivelyassociated,also whencontrollingforextraversion.Anattempttoexplainthefound linkbyincludingmeasuresof powerand statusin a mediation modelfailed:powerandstatusdidnotseemtobeassociatedwith PNS.
8. Generaldiscussion
Findingsfromfourstudiesconsistentlyprovidedsupportfora positiverelationshipbetweenPNSandverbalaggression.Thisrela- tionshipwasobservedwhenusingdifferentmethodstoestimate PNS,and couldnot beexplained throughshared variancewith personalityfactorslikeextraversion.Our findingsindicatedalso thatsubjectivepowerandstatusdonotmediatetherelationship.
Indeed,astrikingobservationwasthatevenafterincludingpower, statusandextraversionintheassociation,astronglinkbetween verbalaggressionandPNSremains.
Threeofthestudiesreliedonthenetworkscale-upmethodfor estimatingPNS.Thesimilaritiesoftheaveragenetworksizes(290, 253,249)tothe290fromtheoriginalscale-upstudieswerestrik- ing,aswasthesimilaritybetweentheseestimatesandtheestimate obtainedthroughadifferentmethod,the“namegeneratoroverlap”
(249).
The previously undocumented relationship between verbal aggressionandPNSraisesseveralinterestingissues.First,itdemon- stratesthataggressionhasadaptivesocialcorrelates.Althoughwe donotclaimacausaldirectionunderlyingtheassociationbetween VAandPNS(whereaswemayhavetacitlyassumedinplacesthat PNSisasocialconsequenceofverbalaggression,ofcourse,ourdata donotruleoutthepossibilitythathavingalargerorsmallerPNS makespeoplemoreorlessverballyaggressive)wecanonlyremark thatifthedisplayofaggressionwasonlydetrimentalforone’ssocial world,verbalaggressionwouldnotbepositivelyrelatedtothesize
ofone’spersonalnetwork.Itispossible,however,thataggression hasotherdetrimentalconsequencesforone’snetworkotherthan onitssize.Forinstance,aggressioncouldaffectthequalityofone’s personalrelationships.Itwouldbeinterestinginfutureresearchto examinealsothequalitynexttothequantityofone’ssocialnetwork intheirrelationshipwithaggression.
Second,itwasonlythemilder,verbalformofaggressionthat yieldsanassociationwithPNS.Itmaybethattheadaptive cor- relatesofaggressionthereforeonlyholdforthismilderformand thatphysicalaggressiondoesnothavethesamesocialimplica- tions.Alternatively,itcannotberuledoutthatarestrictedrange ofphysicalaggressioninoursamplescouldaffecttheassociation withphysicalaggression.
Wedidnotfindaconsistenteffectofgender.Theliteratureon genderdifferencesinaggressionisratherambiguous(e.g.,Crickand Grotpeter,1995;DavidandKistner,2000).Theseambiguitiesseem inparttobeafunctionofthetypeofaggressionthatisbeingcon- sidered,particularlyovertversuscovertforms(Archer,2004).To explorethegendereffectinfutureresearchitcouldbeinteresting toincludeother,morecovertformsofaggression.
9. Limitationsandfutureresearch
Thefindingofthefourstudiesislimitedinasensethattheyall relyonself-reportmeasures.Byusingdifferentmethodsforesti- matingPNSwewereabletoexcludethepossibilitythatourPNS estimation issomehowbiased byself-reportormethodeffects.
Yet,forverbalaggression,thecurrentfindingscouldbestrongly corroboratedbyalsolookingatnonself-reportmeasuresofverbal aggressionsuchaspeerreportsorbehaviouralmeasures.Asecond limitationisthatweonlyexploredtherelationshipbetweenPNS andverbalaggressioncross-sectionally.Experimentalorlongitudi- nalstudiesmayhelptoestablishpossiblecausalpathwaysbetween thesetwovariablesandassuchprovideimportantcluestothepro- cessormechanismunderlyingtherelationshipanditsdirection.An exampleofanexperimentalstudycouldbethat(theexpressionof) verbalaggressionismanipulatedandthatproximityseeking(both byandtowardsthatperson)ismeasuredasadependentvariable.
Longitudinalresearchcouldtaketheformofmeasuringbothverbal aggressionandPNSatdifferentmomentsintimeduringaperiod whereapersonalnetworkisexpectedtochangedramatically(e.g., adolescentsmakingtheswitchfromhighschooltocollege).And examiningthebidirectionallinksbetweenbothvariables.
10. Conclusion
Verballyaggressiveindividuals knowmorepeoplethantheir less verbally aggressive counterparts. This consistent finding clearlyrevealsanadaptivecorrelateofaggression,andidentifies verbalaggressionasanimportantfactorinpersonalnetworks.In linewithWallbott andScherer (1986),weconcludethatverbal aggressionistheexpressionofasocialemotioninwhichpersonal relationswithotherpersonsplayanimportantrole.
Acknowledgements
Thisresearchwas supportedby KULeuvenResearch Council GrantGOA/05/04.WewouldliketothankJeroenStoutenandStijn Decosterfortheirusefulcommentsandpracticalhelp.
Appendix1.
Thequestionedsubpopulationsofthescale-upquestionnaire.
Questionnaire1isusedinStudy1;Questionnaire2inStudies1 and4.
Questionnaire1 Questionnaire2
Mailman Mailman
Centenarians Centenarians
Divorcedindividuals(thepastyear) Divorcedindividuals(thepastyear) Womenwhogavebirth(thelastyear) Womenwhogavebirth(thelast
year)
Individualswhomarriedthelastyear Individualswhomarriedthelastyear
Nurses Nurses
Individualswithlastname“Maes” Individualswithlastname“Maes”
Individualswhohadadeadlyaccident lastyear
Individualswhohadadeadly accidentlastyear
Architects Architects
Individualswithlastname“DeSmet” Individualswithlastname“DeSmet”
Soldiers Soldiers
Prisoners Prisoners
Depressedindividuals Depressedindividuals
Cooks Cooks
Restaurantmanagers Restaurantmanagers
IndividualswithlastnameGoossens IndividualswithlastnameGoossens
Hairdressers Hairdressers
IndividualswithlastnamePeeters IndividualswithlastnamePeeters IndividualswithfirstnameJulia IndividualswithfirstnameJan IndividualswithfirstnameKatrien IndividualswithfirstnameEric/Erik IndividualswithfirstnameGuido Individualswithfirstname
Isabelle/Isabel
IndividualswithfirstnameChristine IndividualswithfirstnameAn/Ann IndividualswithfirstnameAlain Individualswithfirstname
Frederick/Frederik
IndividualswithfirstnameChristophe IndividualswithfirstnameGeert IndividualswithfirstnameNicole Individualswithfirstname
Katrien/Catherine IndividualswithfirstnameSimon Individualswithfirstname
Filip/Philippe
IndividualswithfirstnameKarel Individualswithfirstname Kathleen/Katleen
IndividualswithfirstnameRosa IndividualswithfirstnameClaire IndividualswithfirstnameSteven IndividualswithfirstnameJonas IndividualswithfirstnameLinda IndividualswithfirstnameElke
References
Agresti,A.,1994.Simplecapture–recapturemodelspermittingunequalcatchability andvariablesamplingeffort.Biometrics50,494–500.
Anderson,C.,Galinsky,A.D.,2006.Power,optimism,andrisktaking.EuropeanJour- nalofSocialPsychology36,511–536.
Anderson,C.,Srivastava,S.,Beer,J.S.,Spataro,S.E.,Chatman,J.A.,2006.Knowingyour place:self-perceptionsofstatusinface-to-facegroups.JournalofPersonality andSocialPsychology91,1094–1110.
Archer,J.,2004.Sexdifferencesinaggressioninreal-worldsettings:ameta-analytic review.ReviewofGeneralPsychology8,291–322.
Averill,J.R.,1983.Studiesonangerandaggression:implicationsfortheoriesof emotion.AmericanPsychologist38,1145–1160.
Baron,R.,Kenny,D.,1986.Themoderator–mediatorvariabledistinctioninsocial psychologicalresearch:conceptual,strategic,andstatisticalconsiderations.
JournalofPersonalityandSocialPsychology51,1173–1182.
Belle,D.,1983.Theimpactofpovertyonsocialnetworksandsupports.Marriage andFamilyReview5,89–103.
Bernard,H.R.,Johnsen,E.C.,Killworth,P.D.,Robinson,S.,1989.Estimatingthesize ofanaveragepersonalnetworkandofaneventsubpopulation.In:Kochen,M.
(Ed.),TheSmallWorld.Ablex,Norwood,NJ,pp.159–175.
Bernard,H.R.,Killworth,P.D.,McCarty,C.,1982.Aninformantdefinedexperiment insocialstructure.SocialForces61,99–133.
Bödekker,I.,Stemmler,G.,2000.Whorespondshowandwhentoanger?Theassess- mentofactualangerresponsestylesandtheirrelationtopersonality.Cognition andEmotion14,737–762.
Brooks,J.S.,Newcomb,M.D.,1995.Childhoodaggressionandunconventionality:
Impactonlateracademicachievement,druguse,andworkforceinvolvement.
JournalofGeneticPsychology156,393–410.
Burton,L.A.,Hafetz,J.,Henninger,D.,2007.Genderdifferencesinrelationaland physicalaggression.SocialBehaviorandPersonality35,41–50.
Buss,A.H.,Perry,M.,1992.TheAggressionQuestionnaire.JournalofPersonalityand SocialPsychology63,452–459.
Buss,D.M.,2005.TheHandbookofEvolutionaryPsychology.Wiley,NewYork.
Caspi,A.,Elder,G.H.,Bem,D.J.,1987.Movingagainsttheworld:lifecoursepatterns ofexplosivechildren.DevelopmentalPsychology23,308–313.
Cillessen,A.H.N.,Mayeux,L.,2004.Fromcensuretoreinforcement:developmental changesintheassociationbetweenaggressionandsocialstatus.ChildDevelop- ment75,147–163.
Coie,J.D.,Dodge,K.A.,1998.Aggressionandantisocialbehavior.In:Damon,W.
(SeriesEd.),Eisenberg,N.(Vol.Ed.),HandbookofChildPsychology:Vol.3, Social,Emotional, andPersonalityDevelopment, 5thed. Wiley,NewYork, pp.779–862.
Colvin,G.,2007.SteveJobs’BadBet.Fortune.http://money.cnn.com/magazines/
fortune/fortunearchive/2007/03/19/8402325/index.Htm(retrieved31.03.10).
Christiakis,N.A.,Fowler,J.H.,2009.Connected:TheSurprisingPowerofOurSocial NetworksandHowTheyShapeOurLives.Little,BrownandCompany,Hachette BookGroup,NewYork.
Crick, N.R., Grotpeter, J.K., 1995. Relational aggression, gender, and social–psychologicaladjustment.ChildDevelopment66,710–722.
David,C.F.,Kistner,J.A.,2000.Dopositiveself-perceptionshavea“darkside”?Exam- inationofthelinkbetweenperceptualbiasandaggression.JournalofAbnormal ChildPsychology28,327–337.
Dunbar,R.I.M.,Spoors,M.,1995.Socialnetworks,supportcliquesandkinship.
HumanNature6,273–290.
Granovetter,M.S.,1983.Thestrengthofweakties:anetworktheoryrevisited.Soci- ologicalTheory1,201–233.
Grotpeter,J.K.,Crick,N.R.,1996.Relationalaggression,overtaggression,andfriend- ship.ChildDevelopment67,2328–2338.
Hawley,P.H.,2003.Prosocialandcoerciveconfigurationsofresourcecontrolin earlyadolescence:acaseforthewell-adaptedMachiavellian.Merrill-Palmer Quarterly49,279–309.
Hoekstra,H.A.,Ormel,J.,DeFruyt,F.,1996.NEOPI-R,NEOFFIBigFivePersoon- lijkheidsvragenlijsten:Handleiding(NEOPI-R,NEOFFIBigFivePersonality Questionnaires:Manual).Swets&Zeitlinger,Lisse,TheNetherlands.
Keltner,D.,Gruenfeld,D.H.,Anderson,C.,2003.Power,approach,andinhibition.
PsychologicalReview110,265–284.
Killworth,P.D.,Bernard,R.H.,McCarty,C.,1984.Measuringpatternsofacquaintance.
CurrentAnthropology25,381–397.
Killworth,P.,Johnsen,E.,McCarty,C.,Shelley,G.A.,Bernard,H.R.,1998.Asocial networkapproachtoestimatingseroprevalenceintheUnitedStates.Social Networks20,23–50.
Petersen,C.,1896.TheyearlyimmigrationofyoungplaiceintotheLimfjordfromthe GermanSea.ReportoftheDanishBiologicalStationtotheMinistryofFisheries 6,1–48.
Pool,I.,Kochen,M.,1978.Contactsandinfluence.SocialNetworks1,5–51.
Rozin,P.,2009.Whatkindofempiricalresearchshouldwepublish,fund,and reward? A different perspective. Perspectives onPsychological Science 4, 435–439.
Smith, P.K., 2007. Why hasaggression been thought of asmaladaptive? In:
Hawley, P.H., Little, T.D., Rodkin, P.C. (Eds.), Aggression and Adaptation:
TheBrightSidetoBadBehavior.NewJersey,LawrenceErlbaumAssociates, pp.65–83.
Stokes,J.,1983.Predictingsatisfactionwithsocialsupportfromsocialnetwork structure.AmericanJournalofCommunityPsychology11,141–152.
Tiedens,L.Z.,Ellsworth,P.C.,Mesquita,B.,2000.Stereotypesaboutsentimentsand status:emotionalexpectationsforhighandlowstatusgroupmembers.Person- alityandSocialPsychologyBulletin26,560–575.
Tiedens,L.Z.,2001.Angerandadvancementversussadnessandsubjugation:the effectsofnegativeemotionexpressionsonsocialstatusconferral.Journalof PersonalityandSocialPsychology80,86–94.
Totterdel,P.,Holman,D.,Hukin,A.,2008.Socialnetworkers:measuringandexamin- ingindividualdifferencesinpropensitytoconnectwithothers.SocialNetworks 30,283–296.
Wallbott,H.G.,Scherer,K.R.,1986.Theantecedentsofemotionalexperience.In:
Scherer,K.R.,Wallbott,H.G.,Summerfeld,A.B.(Eds.),ExperiencingEmotion:A Cross-culturalStudy.CambridgeUniversityPress,NewYork,pp.69–83.
Zhou,W.X.,Sornette,D.,Hill,R.A.,Dunbar,R.I.M.,2005.Discretehierarchicalorgan- isationofsocialgroupsizes.ProceedingsoftheRoyalSocietyofLondon272, 439–444.