Fleeson, Malanos, Achille, 2002

14 Pages·2002·0.16 MB·English

JournalofPersonalityandSocialPsychology Copyright2002bytheAmericanPsychologicalAssociation,Inc.
2002,Vol.83,No.6,1409–1422 0022-3514/02/$5.00 DOI:10.1037//0022-3514.83.6.1409
An Intraindividual Process Approach to the Relationship Between
Extraversion and Positive Affect: Is Acting Extraverted
as “Good” as Being Extraverted?
William Fleeson, Adriane B. Malanos, and Noelle M. Achille
WakeForestUniversity
This article investigates whether rapid variation within a person in extraversion is associated with
positiveaffectvariationinthatperson.InStudy1,participantsreportedtheirextraversionandpositive
affectevery3hrfor2weeks.Eachparticipantwashappierwhenactingextravertedthanwhenacting
introverted. Study 2(cid:1)s diary methodology replicated the relationship for weekly variations in positive
affect.Study3’sexperimentalmethodologyreplicatedtherelationshipwhenextraversionwasmanipu-
latedwithinafixedsituation.Thus,therelationshipbetweenextraversionandpositiveaffect,previously
demonstrated between persons, also characterizes the internal, ongoing psychological functioning of
individuals and is likely to be explained by something capable of rapid intraindividual variation.
Furthermore, traits and states are at least somewhat isomorphic, and acting extraverted may increase
well-being.
The purpose of this article is twofold. The first purpose is to ship between extraversion and positive affect increases in impor-
investigatewhethertherelationshipbetweenextraversionandpos- tance if it is a principle generally characteristic of psychological
itive affect is evident within person over time; that is, whether functioningratherthanbeinglimitedtoonlythoseindividualswho
short-term,fast-moving,andeverydayvariationsinextraversionin are dispositionally predisposed to benefit from it. Second, it pro-
onepersonareassociatedwithfast-movingvariationsinthatsame vides another domain in which to test whether the content of
person’spositiveaffect.Thispurposeisimportantforatleastthree behavior or its authenticity (concordance with dispositions) is
reasons. First, it determines whether the extraversion–positive moreimportant(Emmons,Diener,&Larsen,1986;McGregor&
affectrelationshipobservedinbetween-personsstudiesislimited Little,1998;Moskowitz&Coˆte´,1995;Roberts&Donahue,1994;
toadescriptionofco-occurrencesofdifferencesbetweenindivid- Ryan&Deci,2001;Sheldon,Ryan,Rawsthorne,&Ilardi,1997).
uals or can be included in the characterization of the ongoing, Third, the larger is the proportion of individuals who show the
internal psychological functioning of individuals. Second, it pro- relationship, the more widely the relationship can potentially be
vides a test of the density distributions model of traits—specifi- applied as an intervention in the service of positive psychology
cally,thatalargepartoftheindividualdifferencesintraitsisthe (Ryan & Deci, 2001; Seligman & Csikszentmihalyi, 2000; Shel-
frequencywithwhichindividualsenactcorrespondingstates(e.g., don&King,2001).
thefrequencyofactingextraverted)andthatwithin-personvaria-
tion in states is meaningful and at least somewhat isomorphic to The Relationship Between Extraversion
variation in traits (Fleeson, 2001). Third, it tests the potential
and Positive Affect
implicationofthebetween-personscorrelationthatindividualscan
become happier by acting more extraverted. That is, if this is a The between-persons extraversion–positive affect relationship
potential route to self-improvement, it must be the case that meansthatmoreextravertedindividualsarehappierthanaremore
changes within a person in extraversion are associated with introverted individuals, on average. Researchers have demon-
changesinthatpersoninpositiveaffect. stratedthisrelationshipbyobtainingreportsofoverallordisposi-
Thesecondpurposeofthisarticleistoidentifytheproportionof tionalextraversionaswellasofoverallpositiveaffect.Resulting
individualswhoshowtherelationshipaswellaswhichindividuals correlationstypicallyvaryinthe.10to.70range(Diener&Lucas,
showtherelationship—thatis,whosevariationsinpositiveaffect 1999; Lucas & Fujita, 2000). This finding has been reasonably
are predictable from their variations in extraversion. This second robust and has been observed with self-reports on a variety of
purposeisimportantforatleastthreereasons.First,therelation- extraversion measures, in spouse reports, after social desirability
was partialed out, in adults of various ages, and in at least 39
differentcountries(Lucas,Diener,Grob,Suh,&Shao,2000).
Theextraversion–positiveaffectrelationshiphasalsobeenone
WilliamFleeson,AdrianeB.Malanos,andNoelleM.Achille,Depart-
of the more important recent findings in personality psychology,
mentofPsychology,WakeForestUniversity.
foratleasttworeasons.First,itaddsvalidityandmeaningtothe
Portionsofthisworkwerepresentedatthe107thAnnualConventionof
extraversiontrait,encouragingitsacceptanceasoneoftheviable
theAmericanPsychologicalAssociation,August1999,Boston,Massachu-
waystostudypersonality.Second,itdemonstratesthatpersonality
setts.WethankBobBeckandJohnSetafortheircomments.
CorrespondenceconcerningthisarticleshouldbeaddressedtoWilliam may be as important to psychological well-being as is circum-
Fleeson, Department of Psychology, Wake Forest University, Winston- stance. Thus, individuals may have the potential to contribute
Salem,NorthCarolina27109.E-mail:[email protected] directlytotheirownwell-being.Inmoregeneralterms,thefinding
1409
1LINELONG
1410 FLEESON,MALANOS,ANDACHILLE
isalsoimportanttothegrowinginterestinpositivepsychology— processes as perception, cognition, and motivation, this includes
thestudynotonlyofhowtoamelioratedetrimentalaspectsoflife individuals’navigationofandinteractionwiththesocialworldas
butalsoofhowtoenhancebeneficialaspectsoflife(Ryan&Deci, wellasself-regulation.
2001; Seligman & Csikszentmihalyi, 2000; Sheldon & King, Beingbasedonbetween-personscorrelations,theextraversion–
2001).Inparticular,thisfindingsuggeststhatonepotentialwayto positive affect relationship describes the covariations of differ-
increasepositiveaffectinlifeistobecomemoreextraverted. ences between people: The happier people are the ones who are
Itisnotknownwhyextravertsarehappierthanintroverts,and more extraverted than others. The between-persons relationship
atleastthreeclassesofexplanationshavebeenoffered.Explana- has been very useful for identifying those who are happier than
tionsfocusingontheapparentlysocialnatureofextraversionhave others, for demonstrating the importance of extraversion, and for
found little support (Emmons & Diener, 1986; McCrae & Costa, provoking creative theorizing about processes that produce posi-
1987).Forexample,Pavot,Diener,andFujita(1990)showedthat tiveaffect.However,makingabetween-personsrelationshipinto
introvertsinfactspendjustaboutasmuchtimewithothersasdo onethatischaracteristicoffast-varyingpsychologicalfunctioning
extravertsandexperiencejustaboutasmuchincreaseinhappiness typicallyrequiresconceptualchangesinthenatureofthevariables
as do extraverts when spending time with others. More recently, andtherelationship,because(a)thevariablesmustvarywithinone
Lucasetal.(2000)showedthatsocializingcorrelatedlessstrongly person rapidly and continuously and (b) such variance must be
withthecoreofextraversionthandidpositiveaffectandthatthis meaningful (Borkenau & Ostendorf, 1998; Brown & Moskowitz,
finding held up across 39 cultures. A second class of explana- 1998; Gable & Reis, 1999; Larsen, 1989; Nesselroade, 1991;
tions—temperamental—proposesthatindividualshavebaselineor Stone, Shiffman, & DeVries, 1999; Tennen, Affleck, Armeli, &
set-point levels of positive affect. These dynamic equilibrium Carney,2000;Valsiner,1986).Sometimessuchconceptualchanges
(Headey&Wearing,1989),set-point(Lykken&Tellegen,1996), are not easily made consistent with current conceptions of the
and affective level (Gross, Sutton, & Ketelaar, 1998) hypotheses variables or of their relationship, and in such cases it is not
suggest that happiness levels are more or less fixed within indi- expected that the between-persons relationship will characterize
viduals at different levels because of structural differences, that ongoingpsychologicalfunctioning.
homeostatic-likeprinciplesbringindividualsquicklybackdownto There are at least three reasons why the relationship between
theirsetpointsafterpositivedisturbancesandquicklybackupto extraversion and positive affect is one of those not expected to
theirsetpointsafternegativedisturbances,andthatextravertsare characterize psychological functioning. First, traits are typically
happier because they have higher set points. The third class of conceivedofasstructural,stable,andnotvaryingrapidlywithinan
explanationsproposesthatextravertsaremorereactiveorsensitive individual (and perhaps not varying much even over years; Mc-
to positive stimuli and events than are introverts. For example, Crae&Costa,1990).Althoughanindividualcanactasthoughhe
Larsen and Ketelaar (1991) showed that extraverts increase posi- orshehasadifferentlevelofextraversion,itisnotclearthatthis
tive affect more than do introverts when watching a comedy or is the same as changing his or her level of extraversion. Second,
viewing pleasant photographs. Whether it explains extraverts’ principles of authenticity and concordance might imply that the
enhanced reactivity as due to cognitive mechanisms (Rusting & contentofanindividual’sbehavior(e.g.,whetheritisextraverted
Larsen, 1998), a stronger behavioral activation system (BAS; or introverted) is less important than whether the behavior is in
Eysenck,1981),ortheratiobetweentheBASandthebehavioral accordancewiththeindividual’svaluesorpreferences.Forexam-
inhibition system (Gray, 1971), this third class of explanations ple, many individuals prefer slow, relaxing activities to loud,
suggeststhatarelativelyfixed,structuraldifferenceinreactivityis exuberantones.
responsible for the positive affect difference between extraverts Third, current explanations for the relationship between extra-
andintroverts(Carver,Sutton,&Scheier,2000;Davidson,1998; version and positive affect rely on fixed, structural, and complex
Depue&Collins,1999;Gable,Reis,&Elliot,2000). differencesbetweenextravertsandintroverts.Becausesuchstruc-
tural, fixed features are the aspect of extraversion that affects
The Extraversion–Positive Affect Relationship and the positive affect and because complex, fixed structures cannot
changerapidlywithinanindividual,rapidvariationinextraversion
Ongoing Psychological Functioning of Individuals
should not affect positive affect. Specifically, temperamental ex-
An important goal of psychology is to elucidate the ongoing, planations posit that extraversion’s role is to determine individu-
internal psychological functioning of individuals (Allport, 1937; als’ set points of happiness, around which other psychological
Epstein, 1983; Lamiell, 1997; Larsen, 1989; Murray, 1938; Nes- processes may operate (Gross et al., 1998; Headey & Wearing,
selroade,1991;Pervin,1994).Thatis,oneofpsychology’scentral 1989; Lykken & Tellegen, 1996). Perturbations in happiness due
tasks is to identify principles that describe the dynamics of how toexternaleventsmaychangetemporarylevelsofpositiveaffect,
psychological elements internal to individuals vary and interact but homeostatic principles will relatively rapidly return positive
witheachother.Whentheelementsofpsychologicalfunctioning affecttothesetpoint.Reactivityexplanationspositthatextraver-
are slow varying, changing over the course of months or years, sion’sroleispriortopsychologicalfunctioning:Itdeterminesthe
these changes are known as development. Developmental pro- strengthofthereactiontopositiveevents.
cesses concern the creation and maintenance of structures and Ofcourse,itmustbethecasethatsomewithin-personprocess
systems.Whentheelementsofpsychologicalfunctioningarefast orotherisresponsiblefortherobustbetween-personsrelationship
varying,changingwithindays,hours,ornanoseconds,thechanges betweenextraversionandpositiveaffect.Thatis,somethingmust
concern the operation of internal structures. That is, fast-varying have been operating within at least some extraverts to produce
psychologicalfunctioningdescribeshowdevelopedstructuresand their higher levels of happiness (or within introverts to produce
systemsoperatewithinoneindividual(i.e.,howpartsofthemind theirlowerlevelsofhappiness).However,currentconceptionsof
interact with each other). In addition to describing such basic extraversion,positiveaffect,andtheirrelationshiparemosteasily
INTRAINDIVIDUALEXTRAVERSIONANDPOSITIVEAFFECT 1411
consistentwiththenotionthatthisprocessisalreadycompleteor That is, if the trait comes into existence through enactment of
involves variables other than rapid within-person variation in relevantstates,thenconsequencesofthetraitmayalsocomeinto
extraversion. Finding, in contrast, that the extraversion–positive existence through enactment of relevant states. Thus, state–trait
affect relationship does characterize the ongoing psychological isomorphism predicts that the extraversion–positive affect rela-
functioning of individuals would broaden considerably the rele- tionshippreviouslyrevealedbetweenpersonsmaybecharacteris-
vance of the relationship, add to its credibility, and change the ticofpsychologicalfunctioning,thatindividuals’highsandlows
understandingofit.Thisfindingwouldbroadentherelevanceby of happiness may be predictable from their highs and lows in
includingtheextraversion–positiveaffectrelationshipintherealm extraversion states, that structural or slow-moving aspects of in-
ofbasicprocessesofpsychologicalfunctioningandbybringingit dividualsarenotneededtoexplaintheextraversion–positiveaffect
intotherealmofself-regulationandnavigationofthesocialworld relationship, and that one principle characterizing everyday navi-
(Carver&Scheier,1999).Thisfindingwouldaddtotherelation- gationoftheworldisthatmoreextravertedbehaviorbringswith
ship’scredibilitybydemonstratingthatitreallyisextraversionthat itmorehappinessthandoesmoreintrovertedbehavior.
is essential to the relationship rather than some fact limited to
extraverts.Andthisfindingwouldchangetheunderstandingofthe How Many and Which Individuals Show an
relationship by demonstrating that rapid variation in extraversion
Extraversion–Positive Affect Relationship?
occurs,ismeaningful,andneedstobereferencedinexplanations
oftherelationship. Thesecondpurposeofthisstudyistodeterminetheproportion
of individuals who show a within-person relationship between
Traits as Density Distributions of States: extraversionandpositiveaffect.Therelationshipmaycharacterize
theongoingpsychologicalfunctioningofsomeindividualsyetnot
State–Trait Isomorphism
characterizeallorevenmostindividuals.
Arecentmodeloftraitsasdensitydistributionsofstates(Flee- Infact,thereareatleastthreeversionsofthegeneralideathat
son,2001)predictsthatthebetween-personsextraversion–positive the content, per se, of behavior is less important than is its
affectrelationshipisdirectlycharacteristicoftheongoingpsycho- relationship to the individual’s dispositional personality. Authen-
logicalfunctioningofindividuals.Thatis,justasvariationsacross ticity(McGregor&Little,1998;Roberts&Donahue,1994;Shel-
people in their dispositional levels of extraversion predict varia- don et al., 1997), situational congruence (Emmons et al., 1986),
tions across people in their levels of positive affect, the density and behavioral concordance (Moskowitz & Coˆte´, 1995) models
distributions model predicts that variations within a person over sharethepropositionsthatbehaviordoesinfluenceaffectbutthat
shortperiodsoftimeinhisorherlevelofextraversiondoinfact individualsdifferinhowanygivenbehaviorinfluencesaffect.In
occur and are predictive of variations in that person’s level of particular, behaviors that are coherent in some form with the
positiveaffect.Thus,anotherreasonforthesestudiesistotestthis actor’s broader personality influence affect most positively. Al-
model. though these models differ importantly in the form of coherence
BuildingontheworkofAllport(1937),BussandCraik(1983), and the reason for the positive consequences of coherence (e.g.,
Cantor(1990),Epstein(1979),Fiske(1961),Larsen(1989),Mis- familiarity,values,self-determination),theyagreethatthewithin-
chel (1968), Moskowitz (1982), Nesselroade (1988, 1991), and person relationship between extraversion and positive affect
others,thestartingassumptionofthedensitydistributionsmodelis shouldbestrongestforthosewhoaredispositionallyextraverted.
thatstatescanbedescribedinthesamewayasaretraits.Thatis, Thecurrentstudiestestwhetherconcordanceorstate–traitiso-
thewayanindividualisactingatthemomentcanbedescribedin morphismbetterdescribesthewithin-personextraversion–positive
thesametermsandwiththesamescalesasaretraits(e.g.,justas affect relationship. Isomorphism is supported to the extent that
individuals can be described as anywhere from low to high in mostindividualsshowtherelationshipandthatthestrengthofthe
conscientiousness, behaviors can be described as anywhere from relationshipisnotpositivelyrelatedtodispositionalextraversion.
low to high in conscientiousness). In three experience-sampling In contrast, authenticity, situational congruence, and behavioral
studies,then,participantsdescribedtheirbehaviorduringthepre- concordance are supported to the extent that fewer and mainly
vioushourseveraltimesperdayfor2to3weeks,usingadjectives extravertedindividualsshowtherelationship.Itisimportantforus
commonly used for describing Big Five traits (Goldberg, 1992) tobeclearthatthisarticledoesnottestthegeneralapplicabilityof
and on 7-point scales. It was found that people routinely and these models—each is likely to be applicable to some cases and
regularlyexpressedalllevelsofalltraits,formingdistributionsof not to others. In fact, a comprehensive description of personality
states covering the entire 7-point scale. For example, the typical necessarily includes a full account of the conditions under which
individualenactedbehaviorsatalllevelsofextraversionwithina states and traits are related according to different principles
matter of days. Thus, extraversion does vary rapidly within a (Funder,2001).Thegoalofthepresentresearchistotestwhether
person,anditvariesenoughtopotentiallysupportcovariationwith state–traitisomorphismappliesinatleastonecase—specifically,
positiveaffectwithinaperson.Itwasalsofoundthatthelocation inthecaseoftheimportantrelationshipbetweenextraversionand
of an individual’s distributions of states for a particular trait was positive affect. If state–trait isomorphism applies in at least one
highlystable,suggestingthatthefrequencieswithwhichindivid- case, this means that it is one of the principles relating states to
ualsenactvariouslevelsofstatesarealargepart(butnotall)ofthe traits.
natureofindividualdifferencesintraits. Therearetwoadditionalreasonstodeterminetheproportionof
Together, these findings suggest the principle of state–trait individualsforwhomtheextraversion–positiveaffectrelationship
isomorphism,thatstatessharemanypropertiesandconsequences characterizes their ongoing psychological functioning. First, the
withtraits.Ifatraitisinlargepartthedistributionofstates,then greateristheproportionofindividualswhoshowtheextraversion–
consequences of the trait may follow when the state is enacted. positive affect relationship, the greater is the credibility of the
1412 FLEESON,MALANOS,ANDACHILLE
relationship. If only a portion show it, the relationship can be The response rate was within normal range for such experience-
dismissednotasageneraldescriptionofpersonalityandtraitsbut sampling studies. For the included 42 participants, the mean number of
asapeculiaritytocertainindividuals(e.g.,thosewhohighlyvalue reportswas50.6of65possible(78%),andthemedianwas53.5reports
socializing). Second, the greater is the proportion of individuals (82.3%),witharangeof23to63reports.Participantshadbeeninstructed
to miss a report if it would be a major inconvenience to complete (e.g.,
whoshowtherelationship,thewideristhepotentialapplicationof
whiletheyweredriving,takinganexam,sleeping).Participantswerealso
therelationshipasanintervention.
toldtheycouldcompleteareportupto3hrlaterthanthescheduledtime
buttononethelessdescribethescheduledhour.Reportswerealsomissed
Overview of Studies because participants forgot or had computer problems. In the interest of
maintaining data quality, we excluded completed reports if they did not
Three studies examine the within-person relationship between meetstrictcriteria.First,reportsthatcontainedfourormoremissingvalues
extraversionandpositiveaffect.Study1usesexperience-sampling or at least 85% identical responses were excluded (e.g., the participant
recorded1sforallresponses).Second,anadvantagetoPalmPilotsisthat
methodology:Participantsreportedtheircurrentlevelofextraver-
theysurreptitiouslyrecordthedateandtimeofcompletion.Thus,allevents
sionandpositiveaffectfivetimesperdayfor13days.Thisstudy
completedatleast1hrearlierormorethan3hrlaterthanthescheduled
investigates whether momentary increases in extraversion are as-
time were excluded, which guaranteed that all reports were completed
sociatedwithmomentaryincreasesinpositiveaffect.Study2uses
closeintimetothedescribedbehavior.Intotal,237ofthe2,126reports
a diary method: At the end of every week for 10 weeks, partici- wereexcludedforoneofthesereasons(11%).
pants reported their levels of extraversion and positive affect Materials. The daily reports were in the same format as traditional
during the previous week. Besides replicating Study 1 with a adjective-basedBigFiveandaffectscales,withtheexceptionthatrather
different methodology, Study 2 addresses whether sustained in- thandescribingthemselvesingeneral,participantsdescribedtheirbehavior
creases in extraversion, in addition to momentary increases, are and emotion during the previous hour (e.g., “During the previous hour,
also associated with increases in positive affect. Study 3 is an how talkative were you?”). Extraversion and positive affect were each
represented by four items (extraversion: “talkative,” “energetic,” “asser-
experiment designed to test whether the effect could be (a) ma-
tive,” “adventurous”; positive affect: “excited,” “enthusiastic,” “proud,”
nipulated and (b) found within a fixed, controlled situation. Par-
“alert”).TheBigFiveareappropriatelyassessablewithalargevarietyof
ticipants were instructed to act extraverted or to act introverted
adjectives(Goldberg,1992);forthisstudy,adjectiveswerechosenthat(a)
duringa10-mindiscussion;afterward,theyreportedtheirlevelsof
loadedonthecorrectfactorinGoldberg(1992)eitheraloneoraspartof
positiveaffectduringthediscussion. abipolaritemorinDeRaad,Hendriks,andHofstee(1994),(b)together
Investigatingtheextraversion–positiveaffectrelationshipacross represented the breadth of a factor, (c) were easily used to describe
threedistinctmethodologiesshouldprovidearelativelydefinitive behavior,and(d)containednoemotionwords(toavoidredundancywith
caseabouttherelationship.Thisisimportantforpersonalitypsy- theaffectscales).Foraffect,eightrepresentativeitemswerechosenfrom
chologyforexaminingwhetherthiscrucialrelationshipischarac- the Positive and Negative Affect Schedule (PANAS; Watson, Clark, &
teristicoftheongoingpsychologicalfunctioningofindividuals,for Tellegen,1988).ItemswerepresentedinGoldberg’s(1992)opaqueorder:
The five traits were cycled through, with one adjective per cycle in the
testingthedensitydistributionsmodeloftraits,andforexploring
above-listed order, followed by alternating negative and positive affect
the practical implications that one can increase positive affect by
adjectives.Alladjectiveswererespondedtoonscalesrangingfrom1to7,
actingmoreextraverted.
withhighernumbersmeaningthattheadjectivewasmoredescriptive.We
computedscalescoresforextraversionandpositiveaffectforeachreport
Study 1 bytakingthemeanofthecorrespondingfouritems.Thus,eachparticipant
producedabout45descriptionsofhowextravertedheorshewasandhow
Method much positive affect he or she experienced in 45 different 1-hr periods.
Participantsalsoindicatedwhichsituationtheywereinthemajorityofthe
Participants. Forty-sixstudentsparticipatedintheexperimentinpar- hour,choosingfromclass,studying,exerciseorsports,workorvolunteer
tialfulfillmentoftherequirementsforanintroductorypsychologycourse. activities,party,dorm,meal,orother.
Four participants provided fewer than 20 valid reports and so were ex- Reliabilitywascalculatedacrossallincludedreportsandwasfoundto
cludedfromallanalyses.PortionsofStudy1havebeenpreviouslypub- besimilartoreliabilityofBigFivetraitsandaffectscalesinpreviouswork:
lished(Fleeson,2001). forextraversion,Cronbach’s(cid:1)(cid:2).72;forpositiveaffect,(cid:1)(cid:2).80.
Procedure. Fivetimesperdayfor13days,participantsdescribedhow At the end of the introductory session, participants also completed a
theyhadbeenactingandfeelingduringtheprevioushour.Thesereports standard assessment of the Big Five, describing what they are like in
werecompletedonaregularschedule,every3hr(noon,3PM,6PM,9PM, general.Thesameadjectivesplus“shy”(reversescored)and“bold”were
and midnight) and took about 1 to 2 min to complete. Reports were usedinthisassessment.
completedonPalmPilots,hand-heldcomputersaboutthesizeofacalcu-
lator. Each question was printed on a small screen, and participants Results and Discussion
respondedbypressinganumberwithaplasticstylus.Toencouragetimely
completion,weaskedparticipantstodownloadtheirdataevery2days,and Descriptives and suitability of within-person data. Each par-
thosewhomissedadownloadwerecontacted. ticipantwasinsomewaysequivalenttoacompletestudy,provid-
The first report occurred during a 45-min introductory session. The ing multiple measurements of extraversion, multiple measure-
procedurewasexplained,questionswereanswered,andparticipantschose
ments of positive affect, and a measure of covariation between
code names to make their anonymity salient. The unique nature of this
them.Thus,itisimportanttoexaminethedataqualityseparately
study,thatitinvestigatedacompletepictureof2weeksofeachindivid-
for each participant, to ensure that the variables are reasonably
ual’slife,wasstressedaswellasthatitwasimportantthattheycomplete
distributed for each individual. For example, Figure 1 shows a
as many reports as honestly as possible. At the end of the introductory
session,participantswereinvitedtowithdrawforpartialcreditiftheyfelt scatterplot of 1 participant’s extraversion and positive affect
thestudywastoointrusive.Participantsalsocompletedastandardassess- scores. Each point represents this individual’s extraversion and
mentoftheBigFiveduringthisintroductorysession. positiveaffectduring1hour.Forextraversionandpositiveaffect,
INTRAINDIVIDUALEXTRAVERSIONANDPOSITIVEAFFECT 1413
Unstandardized betas were calculated for each participant, and
Figure2showsthedistributionofbetasinthesample.Eachpoint
representsoneindividual’srelationshipbetweenextraversionand
positive affect, and the average beta across individuals was .66,
meaningthatthetypicalindividualimprovedinaffectatthesame
times he or she increased in extraversion. A one-sample t test
revealed that this average beta is significantly greater than zero,
t(41) (cid:2) 23.51, p (cid:3) .001, meaning that it is beyond the power of
chancetohaveproducedit.Moreimpressive,however,isthatno
singleparticipanthadanegativerelationship,andthebetaclosest
tozerowas.20.1Thatis,everysingleindividualwashappierwhen
heorsheactedextravertedthanwhenheorsheactedintroverted,
andtheextraversion–positiveaffectrelationshipwascharacteristic
of the ongoing psychological functioning of individuals. Further-
more,thepsychologicalfunctioningofahighproportionofindi-
vidualswascharacterizedbythisrelationship.
Figure1. Oneparticipant’sdistributionofextraversionstatesandposi-
tiveaffectstatesover2weeks.Eachpointrepresents1hr.Theregression Individual differences in the relationship between extraversion
slopeof.63meansthatthisindividual’spositiveaffectincreasedby.63for and positive affect. Although all participants’ ongoing, internal
everypointhisorherextraversionincreased. psychologicalfunctioningwascharacterizedbyapositiverelation-
ship between acting extraverted and feeling positive affect, indi-
this individual’s means were 3.97 and 4.94, and standard devia- viduals differed in the strength of the relationship, as shown in
tions were 1.01 and 0.90, respectively, demonstrating that this Figure2.Weconductedsplit-halfreliabilityanalysestodetermine
individual had enough fast-moving variability in extraversion to whethertheseindividualdifferenceswerestablecharacteristicsof
support the possibility of covariation between extraversion and individuals rather than due to error. Each individual’s data were
positiveaffect.Thisparticipantalsohadnormaldistributions,with divided into two even–odd halves, and extraversion–positive af-
nooutliers. fect betas were calculated for each half. The correlation between
Corresponding analyses conducted separately for each individ- thetwohalveswas.46(p(cid:3).01),indicatingthatthoseindividuals
ual revealed extraversion means between 2.47 and 5.35, positive who demonstrated a stronger than typical relationship between
affectmeansbetween2.24and6.21,extraversionstandarddevia- extraversion and positive affect in one half of the data also dem-
tions between 0.65 and 1.80, and positive affect standard devia- onstratedastrongerthantypicalrelationshipintheotherhalfofthe
tions between 0.58 and 1.73. Thus, no participant suffered from data. The only way to account for this is that individuals differ
floororceilingeffects,allparticipantsvariedsufficientlyinboth reliably in the relationship—that the strength of the relationship
extraversionandpositiveaffect,andmostparticipantsvariedrou- betweenextraversionandpositiveaffectisafeatureofindividuals.
tinely from very introverted to very extraverted. All participants
To test whether concordance applies to this relationship, we
but one showed relatively normal distributions with no extreme
correlated individual differences in strength of the extraversion–
outliers,andthatoneparticipant’soutlierwasremoved.(Addition-
positiveaffectrelationshipwithextraversionscoresfromatypical
ally, the following analyses produced similar results when a
questionnaire assessment. The correlation between dispositional
stricter criterion was used for outliers: removing all reports more extraversion and the beta was r(41) (cid:2) (cid:4).35, p (cid:3) .05. That is,
thanthreestandarddeviationsawayfromthatindividual’smean.)
extraverts do not enjoy acting extraverted any more than do
Within-person relationships between extraversion and positive
introverts. In fact, introverts were the ones who particularly in-
affect. The primary question in Study 1 is whether individuals
creased in positive affect at those times they increased in acting
increase in positive affect, in comparison with themselves, when
extraverted,contrarytopredictionsofauthenticityorconcordance.
theyincreaseinextraversion.Everyindividualhasmomentswhen
Within-situation analyses. In addition to varying in extraver-
heorsheishappierthanatothermoments;thequestioniswhether
sion, individuals also varied in the situational contexts of their
those moments are the same as ones in which the individual is
actions. Individuals are likely to vary both in their extraversion
actingmoreextravertedthanheorshenormallydoes.Theimpor-
level and in their positive affect level across different situation
tant point to be stressed is that each individual is his or her own
types, and such covariation may produce a spurious relationship
comparisonstandard:Dothatperson’srelativelygreatermoments
between extraversion and positive affect. This is not a central
of positive affect correlate with his or her greater levels of
concernofthisarticle,asthemainpointistofollowascloselyas
extraversion?
possiblethebetween-personsfindings,exceptusingwithin-person
Wecalculatedaregressionforeachparticipantseparately,pre-
methodology.
dicting positive affect from extraversion, with the resulting beta
Nonetheless, we have available a crude control for situation.
indicatingthedegreetowhichthatparticipant’soccasionsthathad
Each participant indicated which of the eight situation types de-
greater positive affect were also those with greater extraversion.
scribed his or her current situation at the time of each report.
For example, Figure 1 shows the regression line for one partici-
Regressions were calculated separately for each situation within
pant,whoseunstandardizedbetawas.63,meaningthat,forevery
point that two of his or her occasions differed on extraversion,
thoseoccasionsdiffered.63pointsonpositiveaffect.Thus,thisis 1One participant who completed only 14 valid reports and thus was
one individual for whom the within-person relationship between excludedfromanalysesshowedanegativerelationshipof(cid:4).15between
extraversionandpositiveaffectholds. extraversionandpositiveaffect.
1414 FLEESON,MALANOS,ANDACHILLE
dictedfromstateextraversion,withinterceptsandslopesallowed
tovaryrandomlyacrossindividuals.Stateextraversionhadabeta
of.68(p(cid:3).001),verifyingthatthetypicalindividualexperiences
a moderately strong relationship between his or her current state
positive affect and his or her current state extraversion. Of the
explainablevariationwithinindividualsinlevelsofstatepositive
affect (1.28), state extraversion explained 48% (0.62). Further-
more, variation across individuals in the strength of the relation-
ship was both small and significant (p (cid:3) .01), verifying that
individuals differed reliably in the relationship between extraver-
sionandpositiveaffectbutthatfewindividualsapproachedazero
relationship.
Next,dispositionalextraversionfromthequestionnaireandthe
interaction between dispositional extraversion and state extraver-
sionwereaddedtothemodel.Theinteractionwassignificant((cid:2)(cid:2)
(cid:4).05, p (cid:3) .05), verifying that more extraverted individuals had
weakerwithin-personrelationshipsbetweenstateextraversionand
statepositiveaffect.However,addingdispositionalextraversionto
themodeldidnotexplainanybetween-personsvariationinaver-
age levels of positive affect (in fact, unexplained variation in-
Figure 2. Distribution of within-person betas predicting positive affect
creasedfrom0.30to0.32).Astudyspecificallydesignedtodetect
fromextraversion.Eachparticipantcontributedonebeta,representinghow
between-personsrelationshipsmaywellfindthatsomeofthat0.32
muchhisorherpositiveaffectincreasedwhenhisorherlevelofextra-
versionincreasedinagivenhour.Noparticipanthadanegativebeta,and between-persons variation in positive affect is explained by dis-
theclosesttozerowas.20. positionalextraversion.
Finally,theabilitytoadjustforvaryingreliabilityandprecision
is particularly important in the within-situation analyses because
each participant. Thus, each regression considered only reports those analyses suffered from fewer available reports. Separately
thatoccurredinthesamesituationtypeandevaluatedwhetherone within each situation, multilevel analyses predicted state positive
individual’s positive affect varied with his or her extraversion affectfromstateextraversion,withslopesandinterceptsallowed
variations within that situation type alone. Figure 3 shows the tovaryrandomly.Forallsituationsexceptataparty,theaverage
resultingbetasforeachsituation,averagedacrossallparticipants. individual’sslopewasgreaterthan.60andsignificantatthep(cid:3)
(Regressions were calculated for a given participant for a given .001level(theanalysesfailedtoconvergeforparty).
situationonlyifheorshereportedbeinginthatsituationatleast In sum, the multilevel model provides compelling verification
five times; thus, the number of participants differed across situa- for the results presented earlier as well as adds additional infor-
tiontypes,andtheestimateswerelessreliableforsomeindividuals mationaboutvariationandexplainedvariation.
thanforothers.)Forexample,thebarforinclassindicatesthatthe
typicalindividualhada.75relationshipbetweenextraversionand Study 2
positiveaffectwhenweconsideredonlythosehoursinwhichhe
or she was in class, such that the classes in which he or she was Study 1 shows that the relationship between extraversion and
happier were those classes in which he or she acted more extra- positive affect holds within person as well as between persons.
verted.One-samplettestsrevealedthatallbetasweresignificantly
greaterthanzeroexceptforthebetainexerciseandsportsituations
(p (cid:2) .06). Thus, acting more extraverted was associated with
enhancedpositiveaffectregardlessofwhethertheindividualwas
in class, studying, working, at a party, in a dorm, or at a meal.
However, these conclusions can be considered only tentative be-
causeofthesmallnumberofreportsforsomesituationsforsome
individuals.
Simultaneousmultilevelmodels. Itispossibletoestimatesep-
arately for each individual the precision of his or her estimated
relationship between extraversion and positive affect and to take
this into account when calculating group statistics. This should
result in more reliable estimates of the typical individual’s
extraversion–positive affect relationship and of the effect of dis-
positionalextraversiononthatrelationship.Inaddition,itprovides
estimatesofseveralvariationsandexplainedvariations.Wechose
to estimate by using simultaneous multilevel modeling with re-
Figure 3. Mean betas within each of eight different situation types.
strictedmaximumlikelihoodestimation(Singer,1998). Holding situation type constant by conducting analyses on reports from
Results from such analyses provide verification of the above onlyonesituationtypeatatimerevealedpositivewithin-personrelation-
ordinary least squares results. First, state positive affect was pre- shipsbetweenextraversionandpositiveaffectforeachsituationtype.
INTRAINDIVIDUALEXTRAVERSIONANDPOSITIVEAFFECT 1415
Thus,Study1providesevidence(a)forsomedegreeoftrait–state Wecalculatedaregressionforeachparticipantseparately,pre-
isomorphism and (b) that the extraversion–positive affect rela- dictinghisorherpositiveaffectvarianceacrossthe10weeksfrom
tionship describes not only differences between individuals but his or her extraversion variance across the 10 weeks. Figure 4
also a process that is ongoing throughout daily life within each shows a histogram of resulting betas. Results are very similar to
individual. those of Study 1, in that nearly all participants had a positive
PreciselybecauseStudy1istargetedatthisongoingprocess,it relationship: Weeks in which they were happier were weeks in
investigatesrapid(hourly)changesinextraversionandinpositive which they acted more extraverted. The average beta was .53,
affect. It is possible, however, that hourly changes are too rapid, whichwassignificantlygreaterthanzero,t(56)(cid:2)9.82,p(cid:3).001.
andalthoughshort-termburstsofextraversionmaybeassociated Only6participantsofthe57showedanegativebeta((cid:4).29,(cid:4).12,
with positive affect, such increases in positive affect are neither (cid:4).06, (cid:4).03, (cid:4).02, and (cid:4).02). Given the smaller number of
sustainablenorgenuine.Study2investigatesthesamephenomena occasionsoverwhichtheseregressionswerecalculated,itispos-
but uses a diary methodology to assess extraversion and positive sible that these represent error (i.e., unusual weeks for those 6
affect over the previous week at weekly intervals to test whether participants).
longertermaffectisalsoassociatedwithlongertermstatevaria- In sum, longer term increases in state extraversion were also
tioninextraversion. associated with longer term increases in positive affect. The
within-person relationships are not explainable as short term or
inauthenticflukes.
Method
Split-half reliabilities were used in Study 1 to test whether
Participants. Sixty-three students participated in the experiment in individuals differed reliably in the strength of the relationship
partial fulfillment of the requirements for an introductory psychology betweenextraversionandpositiveaffect,buttheyarenotreason-
course.The57whocompletedall10weeks’reportswereincludedinthe abletocalculateinStudy2,withonly10occasionsperparticipant.
analyses. However,dispositionalextraversionwasfoundtobeunrelatedto
Procedure. Every Monday or Tuesday for 10 consecutive weeks, thestrengthofthewithin-personrelationshipbetweenextraversion
participants reported to the same location and described their affect and andpositiveaffect(r(cid:2).00,ns).
behaviorduringthepreviousweek.Participantswhodidnotcometothe
Becauseofthesmallnumberofreportsperindividual(i.e.,10),
locationwerecalledtocomeinontheimmediatelyfollowingWednesday.
multilevelmodelsthattakeintoaccountprecisionofwithin-person
Duringaninitialsession,participantswereinstructedastotheprocedure
estimates are particularly important. Such analyses verified the
and also completed a measure of the Big Five and the PANAS, both
referring to how they are in general. During the final week, participants above results. First, weekly positive affect was predicted from
completedthegeneralizedBigFiveandPANASmeasuresagain.Comple- weeklyextraversionwithrandomlyvaryinginterceptsandslopes.
tionofmaterialswasexcellent.Fifty-sevenparticipants(93%)completed The beta for extraversion was .59 (p (cid:3) .001), verifying that the
all10weeks’materials. typicalindividualhadapositiverelationshipbetweenextraversion
Materials. Theweeklyreportswereinthesameformatastraditional and positive affect across weeks. The model including disposi-
adjective-basedBigFiveandaffectscales,withtheexceptionthatrather tional extraversion and its interaction with state extraversion did
thandescribingthemselvesingeneral,participantsdescribedtheirbehavior
andemotionduringthepreviousweek.Extraversionwasrepresentedby5
items (“talkative,” “assertive,” “shy,” “bold,” “energetic”), and positive
affect was described by 10 items (“excited,” “interested,” “strong,” “en-
thusiastic,” “proud,” “alert,” “inspired,” “determined,” “attentive,” “ac-
tive”). Participants responded to all adjectives on scales ranging from 1
to7,withhighernumbersmeaningthattheadjectivewasmoredescriptive;
we computed scale scores for extraversion and positive affect for each
reportbytakingthemeanofthecorrespondingitems.
Reliabilitywascalculatedacrossallincludedreportsandwasfoundto
besimilartoreliabilityofBigFivetraitsandaffectscalesinpreviouswork:
forextraversion,Cronbach’s(cid:1)(cid:2).67;forpositiveaffect,(cid:1)(cid:2).88.
Results and Discussion
Eachparticipant’sdatawereexaminedseparatelyforsuitability
for analyses. The average participant’s within-person standard
deviations across the 10 weeks were 0.59 for extraversion
(range(cid:2)0.25to1.24)and0.60forpositiveaffect(range(cid:2)0.17
to1.19),meaningthattherewasenoughvariabilityacrossweeksin
how extraverted each individual acted and how much positive
affect each individual experienced (a few individuals showed
relatively little variability, reducing the likelihood that we would
Figure 4. Distribution of within-person betas predicting positive affect
find a substantial relationship between extraversion and positive
duringaweekfromextraversionduringthesameweek.Eachparticipant
affectforthoseindividuals).Noreportshadextraversionorposi-
contributed one beta, representing how much his or her positive affect
tive affect scores more than 2.5 standard deviations from their increasedwhenhisorherlevelofextraversionincreasedinagivenweek.
reporter’s mean, so no outliers were removed, and the data were Nearly every participant was characterized by a positive relationship
consideredsuitable. across10weeks.
1416 FLEESON,MALANOS,ANDACHILLE
notproduceasignificantinteraction((cid:2)(cid:2).05,p(cid:2).41),suggesting Materials. Itwasimportanttokeepthemanipulatedstatesascloseas
thatextravertsdonotbenefitmorethandointrovertsfromacting possibletotheself-reportedstatesusedinStudies1and2.Thus,instruc-
extraverted.Insum,althougheachparticipantcontributedonly10 tionswerephrasedintermsofspecificmarkeradjectivesofextraversion.
reports,thepowerofthestudycomesfrommultiplyingthatby57 Wemanipulatedextraversionbyinstructingparticipantstoactbold,talk-
ative,energetic,active,assertive,andadventurous.Wemanipulatedintro-
participantsforatotalof570reports.However,replicationofthis
versionbyinstructingparticipantstoactreserved,quiet,lethargic,passive,
findingacrossmoreweeksisneededforaconvincingcase.
compliant,andunadventurous.Inaddition,toremindparticipantsoftheir
capabilityofactingintheseways,weaskedthemtothinkofarecenttime
Study 3 inwhichtheydidactinthosewaysandtoanswerfourquestionsaboutthat
timewithaboutonesentenceeach.(GiventheresultsofStudies1and2,
Studies1and2demonstratethattheextraversion–positiveaffect
wepresumedthatallparticipantshadactedhighlyextravertedandhighly
relationship characterizes ongoing within-person processes that introvertedonatleastoneoccasioneachduringtherecentweek.)
occuroverbothveryrapid(hourly)statefluctuationsandlessrapid In the first discussion, participants rank ordered the usefulness of 10
(weekly) state fluctuations. However, these relationships may be itemsavailableafteranairplanecrashinNorthernMinnesota(astandard
duetovariationsinsituations.Thatis,theserelationshipsmaybe task;e.g.,Haunschild,Moreland,&Murrell,1994).Intheseconddiscus-
byproductsofthefactthatsituationalvariationleadstovariationin sion,participantsrankordered10possiblesolutionstotheparkingproblem
affect and also to variation in extraversion. For example, social oncampus.Bothtaskswereambiguousandinteresting,soastoencourage
involvementanddiscussion.
situationslikelyincreasepositiveaffect(Cantoretal.,1991;Flee-
Aftereachdiscussion,participantsratedtheirownbehaviorandaffect,
son & Cantor, 1995) and also increase extraverted behavior. Al-
their attitudes toward the discussion, and others’ behavior and affect.
though Study 1 shows that the relationship was evident even
Behaviorwasassessedwith29adjectives,8ofwhichassessedextraversion
withineachofseveralsituationtypes,thosewereonlysuggestive
(“energetic,”“assertive,”“lethargic,”“talkative,”“shy,”“bold,”“passive,”
andsupplementaryanalysesratherthanthemainpointofStudy1. and“adventurous”).AdditionaladjectivesfortheotherBigFivetraitswere
Furthermore,itispossiblethatmoresubtlevariationsinsituations includedasdistractors.Foreachadjective,participantsindicatedhowwell
were still present and responsible for the extraversion–positive itdescribedthewaythey“actedduringthediscussion.”Affectwasrated
affectrelationship. onthePANAS(Watsonetal.,1988),withtheinstructionthatparticipants
Thefinalstudyinvestigateswhetherthewithin-personrelation- rate the “extent you have felt this way during the discussion.” We cre-
shipcanbedemonstratedinafixedsituation.Experimentalmeth- ated 11 items to assess participants’ attitudes toward the discussion—
specifically,toassesshowmuchtheyenjoyedthediscussion(e.g.,“Ihad
odologywasused;participantscametothelabandparticipatedin
momentsoffunduringthediscussion,”“Thediscussionwasenjoyable”).
afixedsituation,andtheywereinstructedtoacteitherextraverted
Participantsrespondedtotheseitemson7-pointscales.Ratingsofothers’
or introverted. If positive affect varies as a result of instruction
behaviorandaffectwereidenticaltoratingsofparticipants’ownbehavior
while the situation is held constant, then it is clear that the
andaffect,exceptthatparticipantsfirstindicatedwhichotherpersonthey
extraversion–positive affect relationship is not due to situational wererating(identifiedbytheletterA,B,orConthetable)andthenrated
variation. “howwelltheadjectivedescribesthewaythatdiscussantactedduringthe
discussion” or “to what extent that discussant felt this way during the
Method discussion.”Someparticipantswereconfusedbytheratingprocedure(e.g.,
ratedthesameothertwiceorambiguouslyidentifiedtheother),and36of
Participants. Forty-seven students participated in the experiment in the188setsofobserverratingswerenotincludedbecauseofparticipant
partial fulfillment of the requirements for an introductory psychology mistakes.Thetwoobservers’ratingswereaveragedtoformacomposite
course.Onlyparticipantswhosescoreonamass-testingextraversionscale rating;observeragreementwasreasonablyhigh:r(61)(cid:2).82,p(cid:3).001,
was in the top or bottom 20% of all mass-testing participants (i.e., less across conditions; r(29) (cid:2) .23, ns, in the extraversion condition; and
than4.38orgreaterthan5.62)wereinvitedtoparticipate(oneindividual r(32)(cid:2).68,p(cid:3).001,intheintroversioncondition.
with a score of 5.50 also participated; one individual completed the
experimenttwice,soonlythefirstparticipationwasincluded,reducingthe
Results and Discussion
samplesizefrom48to47).
Design. The study uses a 2 (state extraversion) (cid:5) 2 (dispositional A2(stateextraversion)(cid:5)2(dispositionalextraversion)analy-
extraversion) design. State extraversion was a counterbalanced, within-
sisofvariance(ANOVA)onpositiveaffectduringthediscussion
person variable. Dispositional extraversion was a between-persons revealed a main effect of state extraversion, F(1, 45) (cid:2) 150.11,
variable.
p (cid:3) .001, but no effect of dispositional extraversion, F(1,
Procedure. Participantsarrivedatthelaboratoryingroupsof3,satat
45)(cid:3)1.00,noraninteraction,F(1,45)(cid:3)1.00.Thus,aspredicted,
a small, semicircular table, and were individually, independently, and
randomlyassignedtoactextravertedorintrovertedbyquestionnairedis- acting extraverted increased the amount of positive affect during
tribution.Participantsfirstreadinstructionstoactextravertedorinstruc- the discussion (M (cid:2) 5.16, SD (cid:2) 1.14), as compared with the
tions to act introverted (depending on their assigned condition) during a amount of positive affect participants experienced when acting
followinggroupdiscussion.Participantsthenengagedinthegroupdiscus- introverted(M(cid:2)2.34,SD(cid:2)0.97),andthiseffectdidnotdepend
sionforupto10min.Attheendofthegroupdiscussion,eachparticipant on the dispositional extraversion of the person acting that way.
ratedhisorherownbehaviorandaffectduringthediscussionandattitudes Notethatthiseffectwasovertwoandahalfstandarddeviations.
toward the discussion, followed by ratings of the other participants’ be-
Counterbalancingallowedustotestforordereffects;a2(state
haviorandaffectduringthediscussion.Participantswerethenassignedto extraversion) (cid:5) 2 (dispositional extraversion) (cid:5) 2 (order)
thecomplementarystateconditionandengagedinasecondgroupdiscus-
ANOVAonpositiveaffectrevealedasmallthree-wayinteraction,
sion,followedbyratingsoftheirownbehaviorandaffectandratingsof
F(1,43)(cid:2)8.63,p(cid:3).01.AsplottedinFigure5,follow-uptests
others’behaviorandaffect.Thus,eachparticipantparticipatedinbothan
instructed extraversion and an instructed introversion condition, and the indicatedasignificanttwo-wayinteractionbetweenorderandstate
orderwasrandomlydeterminedforeachparticipantwithoutregardforthe extraversion for dispositional introverts, F(1, 21) (cid:2) 16.67, p (cid:3)
instructionstootherparticipantsinthesamesession. .01, but not for dispositional extraverts, F(1, 22) (cid:3) 1.00. This
INTRAINDIVIDUALEXTRAVERSIONANDPOSITIVEAFFECT 1417
Figure 5. Three-way interaction between dispositional extraversion, instruction to act extraverted or intro-
verted,andorderofinstruction.Dispositionalintrovertsbenefitedfromactingextravertedsomewhatmorein
Session 2, whereas dispositional extraverts benefited from acting extraverted as much in Session 1 as in
Session2.However,thecentralfindingisthatactingextravertedhadaverylargeeffectonpositiveaffectduring
thediscussion.
interaction for dispositional introverts means that they benefited on observer-reported extraversion revealed a main effect of state
somewhatmorefromactingextravertedaftertheyhadfirstacted extraversion,F(1,37)(cid:2)20.29,p(cid:3).001,butnoeffectofdispo-
introverted,whereastheybenefitedsomewhatlesswhentheyhad sitional extraversion, F(1, 38) (cid:3) 1.00, nor an interaction, F(1,
to act extraverted from the beginning. This finding may hint at a 38) (cid:3) 1.00. Thus, observers also reported that participants were
kindofcongruenceeffect,suchthatintrovertsareabletobenefit happierwheninstructedtoactextraverted(M(cid:2)5.28,SD(cid:2)0.98)
asmuchasextravertsfromactingextravertedbutthatitrequiresat as compared with when instructed to act introverted (M (cid:2) 2.63,
leastminimalfamiliaritywiththesituationbeforetheyreceivethe SD(cid:2)1.26).
fullbenefit.Itmayalsobeacontrasteffect,suchthatintrovertsare
able to benefit from extraversion more when they have the clear
General Discussion
contrastofjusthavingactedintroverted.However,Figure5(cid:1)smain
point is that these order effects are small as compared with the Thetypicalinterpretationofthebetween-personsextraversion–
overwhelming finding that acting extraverted had a very large positive affect relationship is correctly comparative across per-
effectonpositiveaffectduringthediscussion. sons:Individualswhoaremoreextravertedthanothersalsohave
Attitudestowardthediscussion. A2(stateextraversion)(cid:5)2 morepositiveaffectthanthoseothers(Lucasetal.,2000).Thatis,
(dispositional extraversion) ANOVA on attitudes toward the dis- a change in extraversion requires going from one person to an-
cussion also revealed a main effect of state extraversion, F(1, other.Translatingtoawithin-personcorrelationwouldrequirethat
45)(cid:2)79.27,p(cid:3).001,butnoeffectofdispositionalextraversion, changes in one individual’s extraversion are associated with
F(1,45)(cid:2)2.65,ns,noraninteraction,F(1,45)(cid:3)1.00.Thatis, changesinthatsameindividual’spositiveaffect.Mathematically,
actingextravertedalsoincreasedparticipants’favorabilitytoward nothingrequiressuchatranslation.Moreimportant,theconceptual
the discussion (M (cid:2) 5.18, SD (cid:2) 1.18) as compared with acting implications of such a translation argue against it for many rea-
introverted(M(cid:2)3.08,SD(cid:2)1.23). sons,includingthat(a)somethingstructuralorfixedaboutextra-
Manipulationcheck. Werepeateda2(stateextraversion)(cid:5)2 vertsseemsresponsiblefortheirincreasedhappiness,(b)becom-
(dispositionalextraversion)ANOVAwithself-ratingsofextraver- ing happier seems more difficult than simply acting more
sion to check whether participants indeed acted in the ways they extraverted, (c) the effect of acting extraverted might seem to
wereinstructedto,andthisANOVArevealedamaineffectofstate depend on personal preferences or values, (d) state extraversion
extraversion, F(1, 45) (cid:2) 195.09, p (cid:3) .001, but no effect of mightbeverydifferentfromtraitextraversion,and(e)variationin
dispositionalextraversion,F(1,45)(cid:2)1.34,ns,noraninteraction, external variables, such as circumstance and fortune, seem so
F(1, 45) (cid:3) 1.00. Thus, individuals acted more extraverted in the important in producing variation in happiness that it is hard to
extravertedcondition(M(cid:2)5.71,SD(cid:2)1.08)ascomparedwithin believethataprimarydeterminantofvariationindailyhappiness
theintrovertedcondition(M(cid:2)2.21,SD(cid:2)0.98). originates internally. Thus, translating this to a within-person
Observerratings. Observerratingswereusedasatestofthe process is typically taken to mean that, at most, changes in one
robustness of the findings both (a) to possible demand character- individual’s trait extraversion over years might change that indi-
isticsand(b)intermsofthefindings’visibilityeventooutsiders. vidual’soveralllevelofpositiveaffect.
A2(stateextraversion)(cid:5)2(dispositionalextraversion)ANOVA However,thereisanentiredomainofpsychologicalfunctioning
onobserver-reportedpositiveaffectrevealedamaineffectofstate left out by this interpretation—the domain of navigation of the
extraversion,F(1,38)(cid:2)97.30,p(cid:3).001,noeffectofdispositional environment, of self-regulation, and of interaction with others.
extraversion,F(1,38)(cid:3)1.00,andnointeraction,F(1,38)(cid:3)1.00. Whenresearchersdolinktheextraversion–positiveaffectrelation-
A2(stateextraversion)(cid:5)2(dispositionalextraversion)ANOVA ship to this domain of psychological functioning, they invoke
1418 FLEESON,MALANOS,ANDACHILLE
proxyprocessessuchaschangesinamountofsocializingorhow meaning of a given state depends on the traits of the actor (Mc-
individuals interpret events (e.g., Gable et al., 2000). That is, Gregor&Little,1998;Roberts&Donahue,1994;Sheldonetal.,
personality is taken as relevant to process only as a predictor of 1997). For example, although both agreeable and disagreeable
process,whereasnonpersonalityvariablesarerecruitedtoconsti- individualsfindbeingdisagreeableunpleasant,disagreeableindi-
tutetheprocessitself(Tennen,Suls,&Affleck,1991).Incontrast, vidualsdonotfinditquiteasunpleasantasagreeableindividuals
straightforwardlytranslatingthecorrelationtoonethatischarac- findit(Coˆte´ &Moskowitz,1998;Moskowitz&Coˆte´,1995).We
teristicofwithin-personvariationwouldmeanthateachindividual arewillingtobetthatisomorphismandcongruenceareeachtrue
changes in extraversion rapidly (e.g., hour to hour) and that pos- underdifferentconditions.Infact,recentsupportbothforconcor-
itive affect changes within that individual in step. The present dance and for isomorphism calls for more research to determine
findingsshowthatthislattertranslationissurprisinglywarranted when and under what conditions each principle is more accurate
and that the extraversion–positive affect relationship is so perva- andtoestablishthepersonalitydynamicsofbehaviorandtraits.
sive that it characterizes even individuals’ ongoing and fast- Thefindingssupportingisomorphismmayalsohaveimportant
varyingpsychologicalfunctioning. clinical implications. At a general level and in line with the
Study 1’s experience-sampling methodology shows that it is “doing” view of personality, as presented by Cantor (1990), the
possibletopredictanindividual’spositiveaffectatalltimesfrom findings suggest that individuals have flexibility and opportunity
the extraversion of his or her current behavior, regardless of toactindifferentwaysandbringaboutpersonallydesiredconse-
whether he or she is purposefully or more naturally acting intro- quences. That is, positive affect was predicted by what the indi-
verted, extraverted, or anywhere in between. Furthermore, every vidualsdidregardlessofwhat(traits)theyhad.Thus,individuals
participant’s psychological functioning was characterized by the arenotlimitedtoasingle,setpersonalityortotheconsequences
covariation of extraversion and positive affect. This finding of of their long-standing traits. At a more specific level, the robust
general applicability corrects the misperception of introverts as between-persons relationship between extraversion and positive
being happiest during nonexuberant activity and also increases affect has always promised a possible clinical intervention for
both the credibility and the importance of the extraversion– increasing positive affect, that individuals can increase their pos-
positiveaffectrelationshipbyshowingthatitisageneralprocess itive affect by increasing their extraversion. The present studies
notlimitedtoextravertsbutcharacteristicofpossiblyallindivid- showforthefirsttimethatarathersimpleinterventionofencour-
uals’ongoingpsychologicalfunctioning.Study2’sdiarymethod- aging individuals to act more extraverted may be particularly
ologyprovidespreliminaryevidencethatthephenomenonistrue successful. Before it can be established as clinically useful, the
notonlyforhourlybutalsoforweeklyupsanddownsofhappiness causality of this relationship and its applicability to the clinical
andthattherelationshipappliestosustainedincreasesinextraver- context certainly need to be established. But the first and critical
sion as well as to short-term bursts. However, replication across step,showingthatthisprocessdoesoccurwithinindividuals,has
more than 10 weeks is suggested. Study 3’s experimental meth- nowbeentaken.Notethatthisinterventionisrelevantnotonlyto
odologydiscountssituationalexplanations,showingthephenom- those suffering from a lack of positive affect (e.g., depressed
enoninafixedandcontrolledsituation. individuals) but also to positive psychology (Sheldon & King,
2001). Less attention has been paid to interventions that enhance
positivewell-beingthantothosethatreducenegativewell-being,
Taking States Seriously
andtheextraversion–positiveaffectrelationshipisonethatwould
States is a word about which personality psychologists have support enhancing positive well-being (Ryan & Deci, 2001; Se-
disagreed:Somehavereservedtheusefortheparticularcontentof ligman&Csikszentmihalyi,2000)
affect, such that states can refer only to affective terms such as Isomorphism may provide many useful opportunities for per-
happy,anxious,orsad.Incontrast,weagreewiththosewhohave sonalitypsychologists.Variationinstatesiseasiertoobserveand
limited the word to refer to formal properties rather than to more rapid than is variation in traits. For example, it would take
content:Statesarepsychologicalentitiesthatareshort-lived,con- many years and many individuals to observe the within-person
tinuous,andconcrete,asopposedtolong-lived,discontinuous,and relationshipbetweentraitextraversionandpositiveaffect;observ-
abstract (Cattell, Cattell, & Rhymer, 1947; Fridhandler, 1986; ingsufficientstatestakesmuchlesstime(althoughitdoesrequire
Nesselroade, 1988; Patrick & Zuckerman, 1977; Spielberger, extensive sampling of each individual’s daily behavior). Thus,
Lushene,&McAdoo,1977).Thus,statescanapplytotraitcontent personality research into the processes involving traits may be
justastraitscan. greatly facilitated by observations of variation in states. Further-
Thecurrentfindingssupportthehypothesisthatstatescantake more, taking states as potentially causal allows personality psy-
onimportantpropertiesoftraits,thatstatesandtraitsareisomor- chologists to focus on variables of interest to personality in the
phic in at least some regards. Indirectly, then, the findings also explanations of process rather than relying on external variables
supportthedensitydistributionsmodeloftraits.Specifically,part such as situations. Finally, a decades-old problem in personality
of having a trait is simply acting that way (slightly) more often, psychologyhasbeenthedifficultyinusingthemethodsofexper-
andactingacertainwayissimilartobeingthatway.Thisisnotto imentalcontrolthatefficientlyenhanceclaimsofinternalvalidity
say that faking extraversion is as good as being extraverted, but (Cronbach,1957;Pervin,1990).Theprimaryreasonforthisisthe
genuinely acting extraverted may be equivalent to being extra- field’ssubstantiveinterestinpersonalityvariables,whichbydef-
verted. In regard to positive affect, there may be nothing special inition arrive at the researcher’s lab with a history, precluding
about being extraverted that one cannot gain simply by acting random assignment. The validity of state–trait isomorphism may
extraverted.Incontrasttoisomorphismandbecauseoftheimpor- allowsomeendtothisproblem.Totheextentthatstateshavethe
tance of personal integrity, familiarity, and comfort, congruence samepropertiesastraits,experimentersshouldbeable,aswedid
theoriessuggestthatstatesarenotequivalenttotraitsandthatthe in Study 3, to randomly assign individuals to states and instruct