Emotion Differentiation Moderates Aggressive Tendencies in Angry People

12 Pages·2012·0.48 MB·English

Emotion ©2011AmericanPsychologicalAssociation
2012,Vol.12,No.2,326–337 1528-3542/12/$12.00 DOI:10.1037/a0025762
Emotion Differentiation Moderates Aggressive Tendencies in Angry People:
A Daily Diary Analysis
Richard S. Pond, Jr. Todd B. Kashdan
UniversityofKentucky GeorgeMasonUniversity
C. Nathan DeWall Antonina Savostyanova
UniversityofKentucky GeorgeMasonUniversity
Nathaniel M. Lambert Frank D. Fincham
BrighamYoungUniversity TheFloridaStateUniversity
Anger is commonly associated with aggression. Inefficient anger-coping strategies increase negative
affectanddepletetheregulatoryresourcesneededtocontrolaggressiveimpulses.Factorslinkedwith
betteremotionregulationmaythenweakentherelationshipbetweenangerandaggression.Thecurrent
work explored one factor associated with emotion regulation—differentiating one’s emotions into
discretecategories—thatmaybufferangrypeoplefromaggression.Threediarystudies(N(cid:1)628)tested
thehypothesisthatemotiondifferentiationwouldweakentherelationshipbetweenangerandaggression.
InStudy1,participantshighinemotiondifferentiationreportedlessdailyaggressivetendencieswhen
angry,comparedtolowdifferentiators.InStudy2,comparedtolowdifferentiators,highdifferentiators
reportedlessfrequentprovocationindailylifeandlessdailyaggressioninresponsetobeingprovoked
and feeling intense anger. Study 3 showed that high daily emotional control mediated the interactive
effect of emotion differentiation and anger on aggression. These results highlight the importance of
considering how angry people differentiate their emotions in predicting their aggressive responses to
anger.
Keywords:emotion,affect,categorization,anger,aggression
Thepotentialforangerpervadesdailylife.Driverspoundtheir become aggressive while others do not. The current research
hornsinsteadofpolitelyhonking,coworkersmakebitingremarks focuses on one factor—the ability to differentiate one’s negative
instead of offering constructive criticism, and close relationship emotional experiences into discrete categories—that might make
partners criticize their loved ones instead of supporting them. people resilient to aggression following anger provocation. We
Anger has long been associated with aggressive behavior propose that emotion differentiation improves coping and allows
(Berkowitz, 1983, Berkowitz, 1990; Dollard, Doob, Miller, forgreaterregulatorycontroloverone’semotionalstate,duetoan
Mower, & Sears, 1939). Moreover, angered people often aggress enhanced capacity for understanding, clarifying, and describing
when doing so can change their mood (Bushman, Baumeister, & what one feels at any point in time. We predict that such better
Phillips,2001).Despitewidespreadacceptanceoftheroleofanger emotional control would have implications for buffering angry
in predicting aggression, it is unclear why some angry people peoplefromaggressivebehavior.
Anger and Poor Emotion Regulation
ThisarticlewaspublishedOnlineFirstOctober24,2011. Whenpeoplefeelangryorupset,theyaremotivatedtoimprove
Richard S. Pond, Jr., Department of Psychology, University of Ken- their mood (Morris & Reilly, 1987). Indeed, emotion regulation
tucky;ToddB.Kashdan,DepartmentofPsychology,GeorgeMasonUni- effortsoccursorapidlyafteranupsettingeventthatitisdifficult
versity; C. Nathan DeWall, Department of Psychology, University of tokeepparticipantsinabadmood(Isen,1984,Isen,1987;Taylor,
Kentucky; Antonina Savostyanova, Department of Psychology, George 1991;Worth&Mackie,1987).However,angerislinkedtopoor
MasonUniversity;NathanielM.Lambert,DepartmentofFamilyScience, strategiesforemotionregulation.Forinstance,angrypeopleoften
Brigham Young University; Frank D. Fincham, Family Institute, The
ruminate over the experience that provoked them; they turn their
FloridaStateUniversity.
attention inward, either reliving the angry experience itself or
The authors thank Gregory T. Smith and Peggy S. Keller for their
focusingattentiononhowbadtheyfeel(Caprara,1986;Lyubomir-
helpfulcommentsonpreviousdraftsofthisarticle.Theauthorsalsothank
sky&Nolen-Hoeksema,1995).Angerruminationisnotaneffec-
JohnB.Nezlekforhisstatisticaladvice.
CorrespondenceconcerningthisarticleshouldbeaddressedtoRichard tive coping strategy because it ultimately increases aggression,
S. Pond, Jr., Department of Psychology, University of Kentucky, Kastle evenagainstinnocentbystanders(Bushman,2002;Bushman,Bo-
Hall0004,Lexington,KY40506-0044.E-mail:[email protected] nacci, Pedersen, Vasquez, & Miller, 2005; Denson, Pedersen,
326
EMOTIONDIFFERENTIATIONBUFFERSAGGRESSION 327
Friese, Hahm, & Roberts, 2011). For instance, anger rumination Taylor,2004).Peoplewhohavedifficultyidentifyingfeelingsare
playsamajorroleindisplacedaggression(e.g.,roadrage;Denson, alsopronetoengageinimpulsiveaggression(Fossatietal.,2009;
Pedersen, & Miller, 2006). Rumination processes also increase Teten, Miller, Bailey, Dunn, & Kent, 2008). However, this re-
angryaffect,physiologicalarousal,andaggressivecognition(Ped- search is difficult to interpret due to the reliance on people’s
ersenetal.,2011).Angerruminationalsodepletespreciousregu- responsestoglobalquestionnairesontheirmetacognitiveabilityto
latoryresourcesneededtosuppressaggressiveimpulses(Denson, differentiate emotions. People asked to respond to broad state-
2009;Densonetal.,2011). ments (e.g., “I have feelings that I can’t quite identify”) may
Venting is another poor emotion regulation strategy linked to provide answers based on limited insight, influenced by social
anger.Ventinginvolvesactingoutaggressiveimpulsesonpeople desirability. Some novel approaches from neuroimaging (e.g.,
orinanimateobjects(e.g.,punchingapillow,screaming,hittinga Newetal.,2002)andinterviewtranscriptanalyses(Villemarette-
punchingbag).Angeredpeopleexpecttofeelbetterafterventing. Pittman,Stanford,&Greve,2003)havealsoshownthatdeficitsin
Infact,ifangeredpeoplebelievetheirmoodisunchangeable,they the ability to recognize and articulate emotional experiences also
aremuchlesslikelytobehaveaggressively(Bushmanetal.,2001). predictedimpulsiveaggression.
Although some people take pleasure in venting their frustrations, Emotion differentiation may be better captured with a skills-
ventingincreasesaggressionbecauseitcuesaggressivecognition based measurement, such as observing how people discriminate
and behavior (Bushman, 2002; Bushman, Baumeister, & Stack, their emotional experiences over multiple occasions (Dunning,
1999;Hornberger,1959).Ineffect,ruminationandventingrepre- Heath, & Suls, 2004). Only a few studies have examined daily
sentpoorangerregulationstrategies,ifone’sgoalistoreduceor variations in emotions as they relate to daily experiences (e.g.,
evenpreventaggressivebehavior. Kashdan et al., 2010; Swendsen et al., 2000; Tennen, Affleck,
Armeli, & Carney, 2000). To better understand how emotional
experiencesinfluencethetendencytoaggress,aswellaspossible
Emotion Differentiation as a Marker for Resiliency
bufferstothisrelationship,itiscrucialtoemployamethodology
The preceding research implies that anger is associated with thatcapturesfeelingsandeventsacrosstimeinaperson’snatural
ineffectiveemotionregulationstrategiesthatincreaseaggression. environment.Fromthisperspective,emotiondifferentiationshows
Traitsassociatedwitheffectiveemotionregulationmaytherefore promise as a component of resilience. For example, Ong and
buffer angered people from behaving aggressively. Emotion dif- Bergeman (2004) found that higher levels of trait resilience pre-
ferentiation is a promising candidate for an individual difference dicted greater emotion differentiation and a higher co-occurrence
factor associated with effective emotion regulation, which may ofpositiveandnegativeemotionsovera30-daysamplingperiod.
have implications for weakening the relationship between anger Kang and Shaver (2004) found high emotion differentiators had
andaggression. better interpersonal relationships (both self- and partner-rated).
Emotion differentiation, also known as emotional granularity, Ong, Bergeman, and Boker (2009) argued that “emotional com-
refers to how much a person is aware of and able to classify plexity may be a resource that resilient people draw upon during
experiences into discrete emotion categories (Barrett, Gross, times of challenge and adversity” (p. 1784). Consistent with this
Christensen,&Benvenuto,2001;Kashdan,Ferssizidis,Collins,& argument,thecurrentinvestigationaddresseshowemotiondiffer-
Muraven, 2010; Tugade, Fredrickson, & Barrett, 2004). People entiationmaybufferangrypeopleagainstaggressiveresponding.
whoshowadiminishedcapacityforunderstanding,clarifying,and
describingwhattheyarefeelingatagivenpointoftime(i.e.,low Emotional Control Underlies Emotion
emotion differentiators) have difficulty discerning more detail Differentiation Effects
about an emotional state beyond the presence of positive or neg-
Whymightemotiondifferentiationbufferangeredpeoplefrom
ative valence (i.e., feeling “good” or “bad”). This lack of under-
aggression? Better emotional control is one possibility. People
standingmayleadlowemotiondifferentiatorstomisinterpretand
whoarehighinemotiondifferentiation,particularlythedifferen-
ruminateaboutnegativearousalstates.Moreover,lowdifferentia-
tiationofintensenegativeemotions,havemoreemotionalintelli-
torsmaybecomesusceptibletobingeeating,substanceabuse,and
gence, a greater sensitivity to internal and situational cues, and a
other self-destructive strategies to down-regulate negative emo-
largerinventoryofadaptiveemotionregulationstrategiesattheir
tions (Kashdan et al., 2010; Taylor, Bagby, & Parker, 1997).
disposal (Barrett & Gross, 2001; Barrett et al., 2001; Kashdan &
Parker, Taylor, and Bagby (1998) suggest that people with little
Rottenberg,2010).Infact,peoplewhoarebetteratdifferentiating
abilitytodistinguishtheiremotionshavelesscapacityforemotion
their emotional experience also report less self-distraction and
regulation.Ontheotherhand,peoplewhofarebetterindiscerning
moreengagementduringtimesofstress,aswellasagreaterability
emotionsaremoreadeptatregulatingemotionsandpossessfewer
to think carefully about their behavioral options in the stressful
negative attitudes and less distress about intense emotions; thus,
situation(Tugadeetal.,2004).Thus,highemotiondifferentiators
theymaybemoreresilientwhenconfrontingstress(Barrettetal.,
mayspendlesstimeandself-regulatoryresourcesoncounterpro-
2001;Kang&Shaver,2004;Kashdanetal.,2010;Tugadeetal.,
ductive anger-regulation strategies that ultimately increase angry
2004).
affect and aggression (Bushman, 2002; Bushman et al., 1999;
Preliminary evidence for the association between low emotion
Densonetal.,2011).
differentiation and aggression stems from the literature on alex-
ithymia. People with alexithymia—who are less able to identify
The Present Research
and describe emotions, and less able to discern emotions from
bodily sensations—report similar or more intense physiological The present research tested the dual hypothesis that emotion
arousal to unpleasant stimuli (e.g., Luminet, Rime, Bagby, & differentiation moderates aggressive responding in angry people,
328 PONDETAL.
andthatgreateremotionalcontrolmediatesthiseffect.Toprovide & Perry, 1992), which was modified for daily use. This measure
converging evidence in support of our hypothesis, we conducted was used because self-reported propensities toward physical and
threediarystudiesexaminingnegativeemotiondifferentiationand verbal aggression are strongly related to behavioral aggression
aggression. We predicted that on days when people feel more (Giancola & Parrott, 2008). Responses across the items were
intense anger, high emotion differentiators will report less daily summedtoformacompositemeasureofdailyaggressivetenden-
aggression,comparedtolowdifferentiators.InStudy3,wetested cies.Highernumbersindicatedgreateraggression.
whether greater emotional control mediates the interactive effect
ofemotiondifferentiationandangeronaggression. Procedure
Participants were given a URL to record their feelings and
Study 1: Emotion Differentiation Predicts Less Daily
behaviorsthreetimeseachweekfor25days,whichincludedthe
Aggressive Tendencies in Angry People
measuresofnegativeaffect,angerintensity,andaggression.Par-
Study 1 provided an initial test of the hypothesis that emotion ticipantswereinstructedtocompletetheirdailysurveysattheend
differentiation predicts less aggression in people who feel angry. ofeachdaybeforemidnight.Toincreasecompliance,researchers
Participants completed daily measures of negative emotion and stressed that receiving full participation credit was contingent on
aggressivetendenciesthreetimesaweekfora25-dayperiod(total timelyreporting,andthatatime-datestampwouldberecordedon
of 10 waves). We predicted that high emotion differentiators each log. All information submitted via online survey was confi-
would be less susceptible to aggression when they felt angry, dentialandstoredonasecureserver.Adebriefingfollowed.
compared to people who were less adept at differentiating their
negativeemotions. Results
Method Preliminary Analyses
Ourmainpredictionwasthatamongpeoplewhoreportedfeel-
Participants
ing more intense anger, those who were better at differentiating
among their negative emotions would be less susceptible to ag-
Participants consisted of 199 undergraduate students (76%
gression. Because the data violated the assumption of indepen-
women)atalarge,Southern-Atlanticuniversity.Withameanage
of 19.42 (SD (cid:1) 1.54), 70.4% of participants were Caucasian, dence in ordinary least squares regression (i.e., daily measures
nested within individual participants), we used multilevel model-
13.1% were Hispanic, 2.0% were Asian American, 18.2% were
ing techniques to account for their nested structure, using HLM
AfricanAmerican,and9.3%ofparticipantsreportedtheirraceas
Version6.08(Nezlek,2001;Raudenbush&Bryk,2002;Rauden-
“other.”Studentsreceivedresearchcreditforparticipation.
bush,Bryk,Cheong,&Congdon,2000).Atotalof1,928daysof
data were provided by 199 participants (M (cid:1) 9.69). Participants
Measures
followedproperprotocolfortimelydailyresponsesfor94.4%of
Negative emotion differentiation. Participants completed the entries (i.e., completed entries at the end of each day before
thenegativeaffectsubscaleofthePANAS(NA;Watson,Clark,& midnight).
Tellegen,1988),whichassessedtheintensityofnegativeemotions Theaverageintraclasscorrelationforthedifferentiationofneg-
(e.g.,nervous,distressed,scared,hostile)participantsexperienced ative emotions was .84 (SD (cid:1) .47), indicative of an acceptable
eachday.Eachitemwasmeasuredona5-pointscale(fromvery levelofvariability.Toimprovetheinterpretationoffindings,we
slightlyornotatalltoextremely).Anindexofnegativeemotion reverse-scoredthescaleofemotiondifferentiation(i.e.,multiplied
differentiation was computed by calculating average intraclass scores by (cid:2)1), so that larger scores reflected greater emotion
correlations with absolute agreement between the negative emo- differentiationinsubsequentanalyses.
tion adjectives across the assessment period for each participant Ourinitialanalysesfocusedonthereliabilityofthedailymea-
(Barrett,1998;Kashdanetal.,2010;Tugadeetal.,2004).Larger sures for anger intensity and aggressive tendencies. We followed
correlation scores indicated less differentiation of emotions. A procedures recommended by Nezlek (2007, 2011) for assessing
Fisherrtoztransformationwascomputedonallintraclasscorre- scalereliabilitywithnesteddata.Usingathree-levelunconditional
lationsbeforeanyadditionalanalyses.1 modelwithitems(Level1)nestedwithindays(Level2)anddays
Dailyintensityofangerexperience. Participantscompleted nestedwithinpeople(Level3)(seeRaudenbush&Bryk,2002for
fouritemsmeasuringhowmuchwithinthepast24hoursaclose formal rationale), analyses showed that the four items for daily
friendhasmadethemfeelangry,frustrated,provoked,andhostile, intensityofexperiencedangerandthefouritemsfordailyaggres-
relativetootherdays.Theitemswereassessedona9-pointscale sive tendencies had adequate reliability (0.91 and 0.66, respec-
from (cid:2)4 (Far less than usual) to (cid:3)4 (Far more than usual) and
weresummedtoformacompositeofdailyangerintensity.Higher
1Studies 1 and 3 also included measures of positive emotion (PA:
numbersindicatedgreaterfeelingsofanger.
Watson,Clark,&Tellegen,1988),whichpermittedustocreatepositive
Daily aggressive tendencies. Participants completed an ab-
emotion differentiation indices. Neither Study 1 nor Study 3 showed
breviated form of the physical (e.g., “Given enough provocation significant positive emotion differentiation (cid:4) daily anger interactions to
today, I might hit another person”) and verbal (e.g., “If people predictaggressivetendencies(p’s(cid:1).30and.90,respectively).Addition-
wereannoyingmetoday,IwouldtellthemwhatIthinkofthem”) ally, there were no main effects for positive emotion differentiation in
aggression subscales of the Aggression Questionnaire (AQ; Buss Studies1and3(p’s(cid:1).07and.65,respectively).
EMOTIONDIFFERENTIATIONBUFFERSAGGRESSION 329
tively). The intraclass correlation coefficient (ICC) for daily ag- negative emotion differentiation did not reach significance, b (cid:1)
gressive tendencies was 0.52, suggesting that 48% of the 0.27,t(197)(cid:1)0.52,p(cid:1).61.Themaineffectforangerintensity
variabilityinaggressiverespondingwaswithin-person. was significant, such that, on average, participants who reported
Next,priortoourprimaryanalyses,weexaminedwhetheranger feelingmoredailyangerfromaclosefriendalsoreportedgreater
intensityvariedasafunctionofemotiondifferentiation(grand-mean dailyaggressivetendencies,b(cid:1)0.07,t(197)(cid:1)2.54,p(cid:1).01.
centered).Theseanalysesconsistedofatwo-levelmodelwithdays Toevaluatethenatureofourinteractioneffect,weexaminedthe
(Level1)nestedwithinpeople(Level2).Themaineffectfornegative association between daily anger intensity and daily aggressive
emotiondifferentiationdidnotreachsignificance,b(cid:1)0.41,t(197)(cid:1) tendencies among participants relatively low ((cid:2)1 SD) and high
0.41,p(cid:1).68,suggestingthatdailyangerintensitydidnotvaryasa ((cid:3)1SD)inemotiondifferentiation(Aiken&West,1991).Among
functionofemotiondifferentiation.
participants low in emotion differentiation, anger intensity pre-
We then constructed a multilevel model to test our primary dictedincreaseddailyaggressivetendencies,b(cid:1)0.20,t(cid:1)2.74,
hypothesis, with negative emotion differentiation as a Level 2 p (cid:1) .007. However, among participants who expressed a greater
predictor,dailyangerintensityasaLevel1predictor,across-level
tendencytodifferentiatetheiremotions,therewasnotasignificant
interaction term between negative emotion differentiation and
association between anger intensity and aggression, b (cid:1) (cid:2)0.05,
daily anger intensity, and daily aggressive tendencies as the out-
t (cid:1) (cid:2)0.77, p (cid:1) .44. On days defined by extreme anger, people
comeofinterest.Intheseanalyses,theLevel2predictor(negative
high in emotion differentiation reported 17% fewer aggressive
emotiondifferentiation)wasgrand-meancentered(Aiken&West,
tendenciesthandidpeoplelowinemotiondifferentiation.
1991). Daily anger intensity was group-mean centered (i.e.,
person-centered),therebyeliminatingtheinfluenceofperson-level
differences on parameter estimates of mean daily anger (Nezlek, Discussion
2001). This meant that in within-person analyses, an individual’s
coefficientdescribedtherelationshipbetweendeviationsfromhis Study1offersinitialevidencethatidentifyingandrespondingto
or her mean anger intensity level and deviations from his or her negative emotions in distinct ways is a protective factor against
meandailyaggressionlevel. aggressivebehavior,evenwhenonereportsfeelingangry,acom-
mon elicitor of aggression (Berkowitz, 1983, Berkowitz, 1990).
Moderating Effect of Negative Emotion Differentiation Participants’ capacity for emotion differentiation weakened the
on the Association Between Anger and Aggressive relationship between daily anger provoked by a close friend and
Tendencies daily aggressive tendencies. That is, daily anger predicted fewer
aggressive tendencies among participants high in emotion differ-
Asexpected,analysesrevealedasignificantNegativeEmotion
Differentiation (cid:4) Anger Intensity interaction, b (cid:1) (cid:2)0.13, entiation than participants low in emotion differentiation. Al-
t(197) (cid:1) (cid:2)1.97, p (cid:1) .05 (see Figure 1). The main effect for thoughStudy1showsthatgreateremotiondifferentiationweakens
the relationship between anger and aggression, it is unclear
whether emotion differentiation moderates the effect of anger on
aggression in response to direct provocation. We explore this
effectinStudy2.
Study 2: Emotion Differentiation Predicts Less
Aggression in Response to Direct Provocation
Study 1 showed that people who uniquely clarify and distin-
guish among their felt emotional experiences are less susceptible
todailyaggressionwhentheyfeelangry.Wesoughttoexpandon
the association between emotion differentiation and aggressive
behavior in Study 2 by exploring how emotion differentiation
alters behavioral responses in circumscribed social contexts. Un-
likeStudy1,Study2participantsreportedwhetherornotsomeone
hurt their feelings (i.e., provocation) and if so, whether they
behaved aggressively. This can be construed as an explicit, con-
textualized approach to studying the potential value of emotion
differentiation in understanding the link between anger and ag-
gression. To provide a more sensitive measure of the association
betweenemotiondifferentiationandaggression,participantscom-
pletedmeasuresofnegativeemotionandaggressivebehavioreach
day for a 3-week period (total of 21 waves). We predicted that,
comparedtolowdifferentiators,highdifferentiatorswouldbeless
Figure 1. Interactive effect between daily anger intensity and negative susceptible to aggression in response to being provoked to feel
emotiondifferentiationondailyaggressivetendencies,Study1. angrybyanotherperson.
330 PONDETAL.
Method were given a secure website link for an online survey to be
completed nightly. For the next 21 days, they were instructed to
complete the online survey after 6 p.m. To increase compliance,
Participants
researchers stressed that receiving full participation credit was
Participants consisted of 186 undergraduate students (62% contingentontimelyreportingandwerethenshowntheautomatic
women) at a large, mid-Atlantic university. With a mean age of time-and-datestampingofeachdayofdata.Allinformationsub-
23.99(SD(cid:1)9.14),53.7%ofparticipantswereCaucasian,11.8% mitted via online survey was confidential and stored on a secure
were Hispanic, 11.3% were Asian American, 7.0% were African server.Adebriefingfollowed.
American, 1.6% were Middle Eastern, 1.1% were American In-
dian,and6.5%participantsreportedtheirraceas“other.”Students Results
receivedresearchcreditforparticipation.
Preliminary Analyses
Measures
Weagainusedmultilevelmodelingtoaccountedforthedata’s
Negativeemotiondifferentiation. Participantsreportedhow nestedstructure(HLMVersion6.08,Raudenbush&Bryk,2002;
muchtheyfeltembarrassed,disappointed,bored,anxious/nervous, Raudenbush et al., 2000). Participants provided a total of 3,790
and sad each day. Each item was measured on a 7-point scale days of data (M (cid:1) 20.38). Participants followed proper protocol
(from not at all to very much). A measure of negative emotion fortimelydailyresponsesfor93.7%ofentries(i.e.,after6p.m.on
differentiationwascomputedusingthesameprocedureasinStudy thesamedayandbefore9a.m.thenextday).
1.Largercorrelationscoresindicatedlessdifferentiationofemo- Theaverageintraclasscorrelationfordifferentiationofnegative
tions.AFisherrtoztransformationwascomputedonallintraclass emotionswas.39(SD(cid:1).25),indicativeofanacceptablelevelof
correlationsbeforefurtheranalysis. variability. To better interpret subsequent findings, we again
Whereas Study 1 used a measure of negative affect circum- reverse-scored the scale (multiplying scores by (cid:2)1), such that
scribedtohighenergynegativeemotionadjectives,Study2mea- largerscoresreflectedgreateremotiondifferentiation.
sured negative emotion using adjectives that spanned both of the AswithStudy1,ourinitialanalysesfocusedonthereliabilityof
negative valence quadrants in the circumplex model of emotion thedailymeasureforaggression.Usingathree-levelunconditional
(i.e., low vs. high energy; e.g., Barrett, 1998; Russell, 1980; modelwithitems(Level1)nestedwithindays(Level2)anddays
Watson & Clark, 1997). Several of the items were negative in nested within people (Level 3), analyses showed that the three
valence and high in energy (embarrassed, anxious/nervous) and items for daily aggression had adequate reliability (0.81). The
otheritemswerenegativeinvalenceandlowinenergy(i.e.,bored, intraclass correlation coefficient (ICC) for daily aggression was
sad). For this reason, we expected that the average intraclass 0.33,indicatingthat67%ofthevariabilityinaggressivereactions
correlationinStudy2wouldbelowerthanthatfromStudy1. iswithin-person.
Dailyintensityofanger. Eachday,participantswereasked Next, prior to our primary analyses, we conducted analyses to
if someone or something caused them to feel angry and then examinewhetherthepresenceofprovocationepisodesvariedasa
participants responded to how angry they felt. Participants com- function of emotion differentiation (grand-mean centered). These
pletedaone-itemmeasurethatassessedhowangrytheyfeltona analysesincludedpeoplewhoreportedepisodesofbeingprovoked
7-pointscale(fromnotatalltoextremely).Ifnothingcausedthem by another person and those who did not, with days (Level 1)
to feel angry, they received the lowest possible response when nestedwithinpeople(Level2).Theprimedependentmeasurewas
askedabouttheintensityoftheiranger(i.e.,scoreof1reflecting dichotomous: Did a participant have an episode where they were
notatall). provoked by another person on a day or not? These data were
Daily aggression. Participants were given a probe asking analyzed with a nonlinear (Bernoulli) multilevel model that is
them if someone provoked them, hurting their feelings. If they sometimes referred to as multilevel logistical regression. These
answered yes, subsequent questions were asked about their reac- analyses estimate two types of coefficients: a unit-specific and a
tion. Participants completed three items each day that assessed population average. We used unit-specific estimates instead of
their aggressive behavior when provoked (i.e., “if someone an- population averages, but it should be noted that the two sets of
noyedme,Iwasaggressivetowardthem,”“Isaidnastyorcritical coefficientsprovidedverysimilarresults.Theanalysesestimated
thingswhenIwasupset,”and“Ididthingslikeslammedthedoor alog-oddsforeachpersonthatwasconvertedtoaprobability.We
when I was upset”). Each item was measured on a 7-point scale found that greater emotion differentiation was associated with a
(from not at all to very much). Responses across the three items lower probability of being provoked by another person in daily
were summed to form a composite measure of daily aggression, life.
withhighernumbersindicatinggreateraggressivebehavior. Theseanalysesresultedinaninverserelationshipbetweenneg-
ativeemotiondifferentiationscoresandfrequencyofprovocation
Procedure episodes (b (cid:1) (cid:2).16, t(184) (cid:1) (cid:2)2.15, p (cid:1) .03). This provided a
startingpointtoexaminingwhetheremotiondifferentiationaltered
Participants were given a URL to record their feelings and behavioralresponsestoprovocation.
behaviors each day for three weeks. Following an initial training Next, following procedures in Study 1, we employed a multi-
sessioninwhichparticipantslearnedhowtocompletetheirdaily levelmodeltotestourprimaryhypothesis,withnegativeemotion
questionnaires, participants completed self-report questionnaires differentiation as a Level 2 predictor, daily anger intensity as a
reflecting demographic information and personality traits, and Level1predictor,across-levelinteractiontermbetweennegative
EMOTIONDIFFERENTIATIONBUFFERSAGGRESSION 331
emotiondifferentiationanddailyangerintensity,anddailyaggres- 43% less daily aggression when they were angry, compared to
sion as the outcome of interest. In these analyses, the Level 2 peoplewithlowemotiondifferentiation.
predictor(negativeemotiondifferentiation)wasgrand-meancen-
tered (Aiken & West, 1991). Daily anger intensity was group- Discussion
mean centered (i.e., person-centered), thereby eliminating the in-
fluenceofperson-leveldifferencesonparameterestimatesofmean The results of Study 2 replicate and extend our previous find-
dailyanger(Nezlek,2001). ings,suggestingthatpeoplewhodifferentiateamongtheirnegative
emotionsarerelativelyresilienttotheriskofprovocation.People
withgreaterprecisionindescribingandclarifyingfeltexperiences
Moderating Effect of Negative Emotion Differentiation
report less frequent provocation by other people in daily life.
on the Association Between Anger and Aggression
Moreover, people who fail to adequately differentiate their emo-
Our main prediction was that among people who feel greater tionalstatesarepronetoaggressioninthespecificsituationwhere
anger, being more adept at differentiating negative emotional theyareprovokedbyanotherpersonandexperienceintenseangry
stateswouldreducetheiraggressivebehavior.Asexpected,anal- emotions.
yses revealed a significant Negative Emotion Differentiation (cid:4) Although Studies 1 and 2 showed that emotion differentiation
AngerIntensityinteraction,b(cid:1)(cid:2)0.38,t(184)(cid:1)(cid:2)2.68,p(cid:1).009 moderated the association between anger and aggression, it re-
(see Figure 2). Main effects for negative emotion differentiation, mainsunclearwhatmaymediatethismoderatingeffect.Emotion
b (cid:1) (cid:2)1.29, t(184) (cid:1) (cid:2)2.56, p (cid:1) .01, and anger, b (cid:1) 0.84, differentiation is associated with adaptive coping strategies, such
t(184)(cid:1)17.94,p(cid:5).001,werealsosignificantsuchthatpartici- as better regulation of intense emotions (Barrett et al., 2001;
pants showing greater differentiation among their negative emo- Kashdanetal.,2010).Highemotiondifferentiatorsmaythenhave
tional states reported behaving less aggressively when provoked morecontroloftheiremotionalandbehavioralresponsesinstress-
on average, and participants showing greater anger reported fulsituations.Ourfinalstudyexaminedthispossibility.
greateraggressionwhenprovokedonaverage.
Next, we examined the association between anger and self- Study 3: Emotional Control Mediates the Moderating
reported aggressive behavior among participants relatively low Effect of Emotion Differentiation on Aggressive
((cid:2)1 SD) and high ((cid:3)1 SD) in emotion differentiation (Aiken & Tendencies
West, 1991). Among participants low in emotion differentiation,
therewasastrongpositiveassociationbetweenangerandaggres- InStudy3,weexploredthemechanismunderlyingthemoder-
sion, b (cid:1) 1.28, t (cid:1) 7.14, p (cid:5) .001. Among participants who atingeffectofemotiondifferentiationontherelationshipbetween
expressed greater emotion differentiation, there was a weaker, anger and aggression. Specifically, we examined whether rela-
positiveassociationbetweenangerandaggression,b(cid:1)0.40,t(cid:1) tively high levels of emotional control mediated the interactive
2.21, p (cid:1) .03. People with high emotion differentiation reported effectofemotiondifferentiationandangerintensityonaggression.
Becausehighemotiondifferentiatorsaremoreeffectiveatemotion
regulation, compared to low differentiators, they may experience
greater control of their emotions under conditions of high anger.
Participantscompleteddailymeasuresofnegativeemotion,emo-
tionalcontrol,andaggressivetendenciesthreetimesaweekfora
25-dayperiod(totalof10waves).Wepredictedthatpeoplewho
differentiated more among their negative emotions would be less
susceptible to aggression when they felt angry, compared to low
differentiators, and that this interaction would be mediated by
higherlevelsofdailyemotionalcontrol.
Method
Participants
Participants consisted of 243 undergraduate students (78%
women)atalarge,Southern-Atlanticuniversity.Withameanage
of19.34(SD(cid:1)2.27),79%ofparticipantswereCaucasian,13.8%
were Hispanic, 2.0% were Asian American, 14.2% were African
American, .4% were American Indian or Alaskan Native, and
4.4% of participants reported their race as “other.” Students re-
ceivedresearchcreditforparticipation.
Measures
Figure 2. Interactive effect between daily anger and negative emotion Negative emotion differentiation. As in Study 1, partici-
differentiationondailyaggression,Study2. pantscompletedthenegativeaffectsubscaleofthePANAS(Wat-
332 PONDETAL.
son et al., 1988). The same procedure for calculating negative cies had adequate reliability (0.55). The intraclass correlation
emotiondifferentiationinStudies1–2wasused.Smallerintraclass coefficient (ICC) for daily aggressive tendencies was 0.68, sug-
correlationsamongdiscreteaffectsindicatedgreateremotiondif- gesting that 32% of the variability in aggressive tendencies is
ferentiation. A Fisher r to z transformation was computed on all within-person.
intraclasscorrelationsbeforeanyadditionalanalyses. Next, prior to our primary analyses, we conducted analyses to
Daily anger intensity. Participants completed a one-item examine whether anger intensity varied as a function of emotion
measure that assessed how angry they felt that day on a 7-point differentiation(grand-meancentered).Theseanalysesconsistedof
scale(fromnotatalltoextremely). atwo-levelmodelwithdays(Level1)nestedwithinpeople(Level
Dailyemotionalcontrol. Participantscompletedaone-item 2). The main effect for negative emotion differentiation signifi-
measureassessinghowmuchcontroltheyhadovertheiremotions cantlypredicteddailyangerintensity,b(cid:1)(cid:2)0.37,t(242)(cid:1)(cid:2)4.00,
thatday.Participantswereaskedtoindicatehowmuchtheyagreed p (cid:5) .001, such that participants showing greater differentiation
with the statement “My emotions got the best of me today” on a among their negative emotional states generally reported less
7-point scale (from not at all to extremely). Responses were intensedailyanger.
reverse-scoredsuchthatlargernumbersindicategreaterlevelsof FollowingproceduresinStudies1–2,weemployedamultilevel
emotionalcontrol.2 model to test our primary hypothesis, with negative emotion dif-
Daily aggressive tendencies. Participants completed the ferentiationasaLevel2predictor,dailyangerintensityasaLevel
same abbreviated form of the physical and verbal aggression 1predictor,across-levelinteractiontermbetweennegativeemo-
subscales of the Aggression Questionnaire (AQ; Buss & Perry, tiondifferentiationanddailyangerintensity,anddailyaggressive
1992) that was used in Study 1. The items needed rescaling at tendenciesastheoutcomeofinterest.Intheseanalyses,theLevel
Level 1, in order to achieve approximately equal error variances 2 predictor (negative emotion differentiation) was grand-mean
(Raudenbush & Bryk, 2002). Thus, the items were standardized centered(Aiken&West,1991).Dailyangerintensitywasgroup-
priortoformingthecompositeofaggressivetendencies. mean centered (i.e., person-centered), thereby eliminating the in-
fluenceofperson-leveldifferencesonparameterestimatesofmean
Procedure dailyanger(Nezlek,2001).
Participants were given a URL to record their feelings and
Moderating Effect of Negative Emotion Differentiation
behaviorsthreetimeseachweekfor25days,whichincludedthe
negativeemotion,emotionalcontrol,andaggressionitems.Partic- on the Association Between Anger and Aggressive
ipantswereinstructedtocompletetheirdailysurveysattheendof Tendencies
each day before midnight. To increase compliance, researchers
Asexpected,analysesrevealedasignificantNegativeEmotion
stressed that receiving full participation credit was contingent on
Differentiation (cid:4) Anger Intensity interaction, b (cid:1) (cid:2)0.22,
timelyreporting,andthatatime-datestampwouldberecordedon
t(241) (cid:1) (cid:2)2.56, p (cid:1) .01 (see Figure 3). The main effect for
each log. All information submitted via online survey was confi-
negative emotion differentiation was marginally significant, b (cid:1)
dentialandstoredonasecureserver.Adebriefingfollowed.
(cid:2)0.58, t(241) (cid:1) (cid:2)1.89, p (cid:1) .06, such that participants showing
greater differentiation among their negative emotional states re-
Results
portedfeweraggressivetendenciesonaverage.Themaineffectfor
daily anger intensity was significant, b (cid:1) 0.58, t(241) (cid:1) 10.62,
Preliminary Analyses p (cid:5) .001, such that participants reporting greater anger also
reportedmoreaggressivetendenciesonaverage.
Ourmainpredictionwasthatpeoplewhoshowmoredifferen-
We next examined the association between anger and daily
tiation among their negative emotional states, compared to low
aggressive tendencies among participants relatively low ((cid:2)1 SD)
differentiators, will be less susceptible to aggressive tendencies
and high ((cid:3)1 SD) in the tendency to differentiate their emotions
whentheyareangry,whichwouldbemediatedbyhigherlevelsof
(Aiken&West,1991).Amongparticipantslowinemotiondiffer-
emotionalcontrol.Weagainusedmultilevelmodelingtoaccount
forthedata’snestedstructure(Nezlek,2001;Raudenbush&Bryk,
2002; Raudenbush et al., 2000). Participants provided a total of 2We conducted additional analyses to establish the validity of the
1,864 days of data (M (cid:1) 7.67). Participants followed proper emotioncontrolmeasureinStudy3.Participantscompletedadailymea-
protocol for timely daily responses for 93.3% of the entries (i.e., sure of cognitive reappraisal, which is one type of emotion regulation
completedentriesattheendofeachdaybeforemidnight). strategy(Gross,2008).Participantswereasked,“Howmuchdidyouthink
Theaverageintraclasscorrelationfordifferentiationofnegative aboutthepositiveaspectsofnegativeeventsorsituationsinyourlifesince
emotionswas.84(SD(cid:1).62),whichindicatesanacceptablelevel the last log?” This item was measured on a 1 (Not at all) to 5 (I was
ofvariability.AswithStudies1–2,wereverse-scoredthescaleof constantly thinking these thoughts) scale. Our emotional control item
emotion differentiation (i.e., multiplied scores by (cid:2)1), so that significantly related to this type of cognitive reappraisal, b (cid:1) 0.05, t (cid:1)
2.84, p (cid:1) .005, such that participants who reported greater control over
largerscoresreflectedgreateremotiondifferentiation.
theiremotionsalsoreportedthinkingmoredailyaboutthepositiveaspects
As in Studies 1 and 2, our initial analyses focused on the
of the negative events in their life. Additionally, a subsample of 164
reliabilityofthedailymeasureforaggressivetendencies.Weagain
participantscompletedtheBriefCOPE(Carver,1997)beforetakingpartin
usedathree-levelunconditionalmodelwithitems(Level1)nested Study 3. Negative coping strategies at the trait level, such as venting,
within days (Level 2) and days nested within people (Level 3). self-distraction,anddenial,werenegativelyrelatedtoourdailyemotional
Analyses showed that the four items for daily aggressive tenden- controlitem(p’s(cid:5).0001).
EMOTIONDIFFERENTIATIONBUFFERSAGGRESSION 333
Aspredicted,analysesrevealedasignificantNegativeEmotion
Differentiation(cid:4)AngerIntensityinteraction,b(cid:1)0.15,t(241)(cid:1)
2.68, p (cid:1) .008 (see Figure 4). Main effects for negative emotion
differentiation,b(cid:1)0.34,t(241)(cid:1)3.48,p(cid:1).001,andanger,b(cid:1)
(cid:2)0.45,t(241)(cid:1)(cid:2)12.85,p(cid:5).001,werealsosignificantsuchthat
participants showing greater differentiation among their negative
emotional states reported greater emotional control, and partici-
pantsshowinggreaterangerreportedlessemotionalcontrol.
We next examined the association between anger and daily
emotionalcontrolamongparticipantsrelativelylow((cid:2)1SD)and
high ((cid:3)1 SD) in the tendency to differentiate their emotions
(Aiken&West,1991).Amongparticipantslowinemotiondiffer-
entiation, anger intensity predicted less daily emotional control,
b(cid:1)(cid:2)0.55,t(cid:1)(cid:2)11.60,p(cid:5).0001.Whereas,amongpeoplewho
expressed a greater tendency to differentiate their emotions, the
association between anger and emotional control was a weaker,
negativeassociation,b(cid:1)(cid:2)0.36,t(cid:1)(cid:2)6.82,p(cid:5).0001.
Wenexttestedwhetherdailyemotionalcontrolpredicteddaily
aggressive tendencies, controlling for the main and interactive
effects of emotion differentiation and daily anger. As predicted,
theassociationbetweendailyemotionalcontrolanddailyaggres-
sive tendencies was significant, such that people who reported
greater daily emotional control also reported fewer aggressive
Figure 3. Interactive effect between daily anger and negative emotion tendenciesonaverage,b(cid:1)(cid:2)0.30,t(242)(cid:1)(cid:2)5.59,p(cid:5).001.
differentiationondailyaggressivetendencies,Study3. Last,wetestedformediatedmoderationbyestimatingthe95%
confidence interval of the indirect effect using the empirical-M
test.Traditionalapproachesfortestingmediationareoftenunder-
powered and suffer from inflated Type-I error rates because they
entiation,angerintensitypredictedincreaseddailyaggression,b(cid:1)
incorrectlyassumethattheproductofcoefficients(operationalized
0.80, t (cid:1) 7.69, p (cid:5) .0001. Among participants who expressed a
asab)comprisingtheindirecteffectisnormallydistributed(Bol-
greater tendency to differentiate their emotions, the association
len & Stine, 1990; MacKinnon, Lockwood, & Williams, 2004;
betweenangerandaggressionwasaweaker,positiveassociation,
Preacher&Hayes,2008;Shrout&Bolger,2002).Moreover,such
b (cid:1) 0.37, t (cid:1) 3.76, p (cid:1) .0002. Participants high in emotion
methods are inappropriate for testing mediation with nested data
differentiationreported16%lessdailyaggressivetendenciescom-
pared to participants low in emotion differentiation, when they
wereangry.
Mediated Moderation Analyses
Next, we investigated whether higher levels of emotional con-
trolmediatedtheinteractiveeffectofemotiondifferentiationand
angerintensityondailyaggressivetendencies.Becausewealready
showedasignificantmoderatingeffectofemotiondifferentiation
on the relationship between anger and aggressive tendencies, our
next step was to show that emotion differentiation also signifi-
cantly moderated the relationship between anger and emotional
control (Baron & Kenny, 1986). We used multilevel modeling
procedurestoaccountforthenestedstructureofthedata(Nezlek,
2001;Raudenbush&Bryk,2002;Raudenbushetal.,2000).
To examine the mediating effect of daily emotional control on
the moderating effect between negative emotion differentiation
andangeronaggressivetendencies,wefirstemployedamultilevel
modelwithnegativeemotiondifferentiationasaLevel2predictor,
dailyangerintensityasaLevel1predictor,across-levelinterac-
tiontermbetweennegativeemotiondifferentiationanddailyanger
intensity,anddailyemotionalcontrolastheoutcomeofinterest.In
these analyses, negative emotion differentiation was grand-mean
centeredanddailyangerintensitywasgroup-meancentered(i.e., Figure 4. Interactive effect between daily anger and negative emotion
person-centered). differentiationondailyemotionalcontrol,Study3.
334 PONDETAL.
(Bauer, Preacher, & Gil, 2006). On the other hand, the bitionsagainstviolence(Anderson&Bushman,2002;Berkowitz,
empirical-Mtestgeneratesconfidenceintervalsfortheindirect 1983,Berkowitz,1990).Angerisaparticularlyhazardousemotion
effectfromcriticalvaluesobtainedbyempiricallysimulatinga because people often regulate it through rumination and venting,
series of distributions for the product of two normal random whichultimatelyleadstomoreaggression(Bushman,2002;Bush-
variables (MacKinnon et al., 2004). The empirical-M test does man et al., 2001). People require considerable self-regulatory
not impose the assumption of normality on the distribution of resources to regulate anger in a way that avoids conflict, which
the product. Thus, compared to traditional methods, the may leave them particularly vulnerable to aggressive retaliation
empirical-M test provides more power and more accurate (Denson,2009).Therefore,individualdifferencefactorsassociated
Type-I error rates for single-level (MacKinnonetal.,2004)and with effective emotion regulation may weaken the link between
multilevel (Pituch, Stapleton, & Kang, 2006) designs, and it is angerandaggression.
recommendedfortestsofmediationfornestedmodelsinwhichat In this vein, we explored a novel factor that may increase
leastonevariablewasassessedattheupperlevel(level2)(Pituch resiliency to aggression when people are provoked to feel anger.
&Stapleton,2008).Toconducttheempirical-Mtest,weusedthe Specifically,wehypothesizedthatangrypeoplewhotendtoiden-
computer program PRODCLIN, which provided the confidence tify and clarify their discrete negative emotional experiences in
interval of the indirect effect (MacKinnon, Fritz, Williams, & nuanced ways will be more resilient to aggression, compared to
Lockwood,2007). angry people who differentiate among their negative emotions
Aspredicted,theindirectpaththroughdailyemotionalcontrol relativelyless.Priorworkhasshownthatpeoplewhoarebetterat
wasstatisticallysignificant,asthe95%confidenceintervaldidnot differentiating their emotions are also better at regulating those
includezero((cid:2)0.09to(cid:2)0.01).Theinteractiveeffectofemotion emotions(Barrettetal.,2001;Kashdanetal.,2010;Tugadeetal.,
differentiationandangeronaggressionwaspartiallymediatedby 2004).Becausepeoplewhoarebetteratdifferentiatingtheirneg-
daily emotional control (see Figure 5). Thus, participants who ativeemotionsaremoreemotionallyintelligent,moresensitiveto
tendedtodifferentiatemoreamongtheirnegativeemotionsexhib- internal and situational cues, and have a greater inventory of
ited a weaker relationship between anger and aggression, in part effective emotion regulation strategies (Barrett & Gross, 2001),
becauseoftheirgreatercontrolovertheirdailyemotionalexperi- theymayhavebettercontroloftheiremotionsandtheirbehavior
ences. intheheatofstressfulsituations,comparedtolowdifferentiators.
They may be less likely to use maladaptive strategies for coping
Discussion with anger, compared to low differentiators, which may have
implicationsfortheirlowerlevelsofaggression.
Study 3 provides converging evidence that negative emotion The current work extends previous research by showing that
differentiation moderates aggressive responses to anger. In addi- theseregulatoryadvantagesofemotiondifferentiationhelpbuffer
tion, Study 3 offers direct evidence in support of our conceptual angry people from potentially harmful conflicts. The results of
modelthatemotiondifferentiationpredictsfewerdailyaggressive three diary studies provided converging evidence that emotion
tendencies because people who are better differentiators have differentiation moderates the relationship between anger and ag-
better emotional control under anger-provoking conditions. gression. In Study 1, higher dfferentiation of negative emotions
Whereaspeoplewhoarelessadeptatdifferentiatingtheirnegative served as a resilience factor against aggressive tendencies for
emotionssufferfromlessemotionalcontrolinagivenday,which angry participants. Study 2 replicated this effect, using different
puts them at an increased risk of aggressing against others when measuresofemotiondifferentiationandaggressioninresponseto
theyfeelangry. provocation. Study 3 replicated and extended Studies 1–2 by
giving support to a mechanism underlying this effect. Emotion
General Discussion differentiationmoderatedaggressiveresponsesinangrypeople,in
part,becauseofitsrelationtobetteremotionalcontrol.Thismay
Angerisintimatelylinkedtoaggressionbecauseofitsabilityto
have freed up regulatory resources that would normally be used
increase arousal and hostile cognitions, while also reducing inhi-
toward venting, anger-rumination, and other ineffective coping
strategiesforimprovingone’smood.Insupportofourconceptual
model,Study3showedthatpeoplewhotendtodifferentiatetheir
negativeemotionsreportedgreatercontroloftheirdailyemotions
when they were angry, compared to low emotion differentiators.
Thisgreateremotionalcontrolplayedasignificantmediatingrole
in reducing aggression in angered people who are high emotion
differentiators.
Self-control processes are important for inhibiting aggression
(DeWall&Anderson,2011;Finkel,2007;Slotter&Finkel,2011).
Internal characteristics and situational factors that bolster self-
controlprotectthenecessarycognitiveresourcesneededforover-
riding aggressive impulses. People with low internal self-control,
orwhosufferfromregulatorydepletion,aremorelikelytoimpul-
Figure5. Mediatingeffectofdailyemotionalcontrolontheinteractive sivelyactouttheiraggressionbecausetheyhavefewerresources
effectbetweendailyangerandnegativeemotiondifferentiationondaily forthecognitivereappraisalprocessesrequiredforignoringthose
aggressivetendencies,Study3. aggressive urges (DeWall, Finkel, & Denson, 2011). Our results
EMOTIONDIFFERENTIATIONBUFFERSAGGRESSION 335
builduponrecenttheoriesbyexposingemotiondifferentiationas oftenincludeillegalbehaviors(e.g.,assaultingorharmingothers
one potential internal factor that reduces aggression, in part, by with a weapon), which pose problems in getting institutional
protectingthosenecessaryregulatoryresources. reviewboard(IRB)approvaltomeasurethosebehaviors.Aswell,
measuringtheincidenceofviolenceposesethicalityissues,suchas
whethertoreportparticipantswhoareacleardangertoothersto
Limitations and Future Directions
a law-enforcement agency. Nonetheless, the dynamics between
The current investigation provided consistent support for the generalaggressionandviolencearesimilar,andwewouldexpect
hypothesis that emotion differentiation moderates aggressive re- asimilarmoderatingeffectofemotiondifferentiationandangeron
sponding among angry people. Despite the consistency of our violence.
findings, several limitations may prove beneficial in generating A final direction for future research concerns how emotion
futureresearchideas.First,weshowedthatemotiondifferentiation differentiationmayinfluenceprosocialoutcomes.Angerislinked
moderatesthelinkbetweenangerandaggression,inpart,because toaggressionthroughthemaladaptivestrategiespeopleuse(e.g.,
itisassociatedwithincreasesinemotionalcontrol.However,this venting, rumination) to deal with anger, which exhaust precious
leaves room for other potential mediators. One possibility is that regulatoryresources(Bushman,2002;Bushmanetal.,2001;Den-
people who tend to identify and respond to their emotions in son, 2009). The current work posits that emotion differentiation
uniquewaysarealsomoremindfuloftheirconsciousstate.Mind- buffers angry people from aggression because greater emotion
fulnessislinkedtoagreaterawarenessofone’semotions,which differentiation is linked to the availability of adaptive coping
hasimplicationsforemotionregulationandself-controlprocesses strategies, which decreases the probability of using maladaptive
(Brown & Ryan, 2003), particularly in response to threats to the ones.Yet,inthepursuitofmoodenhancementeffects,peoplewho
self(Niemiecetal.,2010).Peoplewhoaregoodatdifferentiating feel bad may be motivated to act prosocially. For instance, the
theirnegativeemotionsmaybemoremindfulatthetraitlevel.As researchliteratureonhelpingindicatesthatsometimessadpeople
well,theprocessofdifferentiatingone’semotionsonaday-to-day willhelpothersinanefforttomakethemselvesfeelbetter(Cial-
basismayalsohaveimplicationsforexercisingdailymindfulness. dini, Darby, & Vincent, 1973; Manucia, Baumann, & Cialdini,
Futureresearchwouldbenefitfromexploringtraitanddailymind- 1984).Ifhighemotiondifferentiatorsengageinfewerbehaviorsto
fulness as a mediator of the interactive effect of emotion differ- change their mood in response to upsetting events, they may
entiationandangeronaggression. engageinlessprosocialbehaviorwhendistressed.Incontrast,low
Asecondpossiblemediatormightberelatedtoincreasedwork- differentiators,whoexperiencedistressingeventsassimply“bad,”
ingmemorycapacity.Barrettandcolleagues(2001,2004)suggest may be most likely to help when they believe that doing so may
thatpeoplewhohaveagreatercapacityforworkingmemoryhave improve their mood. Thus, emotion differentiation may predict
more executive resources for attending to and identifying emo- less aggressive responses when people are angered, but emotion
tional cues, managing emotional knowledge, and coping with differentiation may predict less prosocial behavior when people
stressful emotions. Recent neuroimaging studies lend initial sup- aredistressed.Thispossibilityawaitsfutureresearch.
port to this hypothesis, showing that brain areas associated with
workingmemoryareinvolvedinemotionregulationandpsycho- Conclusion
logicalwell-being(Ochsneretal.,2004;Urryetal.,2004).Thus,
futureresearchwouldalsobenefitfromexploringworkingmem- Feeling angry is an inevitable part of life. Yet how people
ory capacity as a mediator of the interactive effect of emotion behavewhileangrydifferswidelybetweenindividuals.Thepres-
differentiationandangeronaggression. entinvestigationexaminedtheutilityofemotiondifferentiationin
A second limitation relates to our study being correlational in buffering aggression in people’s naturalistic environment on a
nature.Astrengthofourmethodologywasthatwecouldexamine daily basis. This research builds on prior work by helping us to
people’sfeelingsacrosstimeandwithintheirownenvironments. understandwhysomeangrypeopledonotlashoutatothers,while
However,ourfindingsmaynotrevealcausalrelationships.More- others are at an increased risk of such behavior. Aggression is
over,duetoourdailydiarymethod,ourstudiesexhibitedaheavy costly in terms of the toll it takes on our social relationships,
relianceonsingle-itemmeasures,whichrisklowreliability.Future psychological health, and even our own safety. This research
experimentalworkwouldbenefitfrommanipulatingemotiondif- program sheds light on a novel internal factor that may help to
ferentiation and anger in the laboratory and then measuring ag- preventthenegativeconsequencesofaggression.
gressionbehaviorally.
Athirdlimitationinvolvesthedifferencebetweenviolenceand References
aggression.Ourstudyconcernedthetendencytowardgeneraldaily
aggression (i.e., harming someone who would be motivated to Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
avoid that harm; Baron & Richardson, 1994). Violence, on the interpretinginteractions.London,UK:Sage.
other hand, is intended to cause extreme physical harm (e.g., Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual
Review of Psychology, 53, 27–51. doi:10.1146/annurev.psych
injury, death). All violent acts are aggressive, but not all acts of
.53.100901.135231
aggression are instances of violence (Bushman & Huesmann,
Baron,R.A.,&Richardson,D.R.(1994).Humanaggression(2nded.).
2010).Thebaserateofviolenceislow,duetoitsextremenature;
NewYork:PlenumPress.
therefore, we chose general daily aggression as our outcome of
Baron,R.M.,&Kenny,D.A.(1986).Themoderator-mediatorvariable
interest. We expected more variability in reports of daily aggres- distinction in social psychological research: Conceptual, strategic and
sion across each diary period, which would make it a more statisticalconsiderations.JournalofPersonalityandSocialPsychology,
sensitive measure. Second, items that constitute violent behavior 51,1173–1182.doi:10.1037/0022-3514.51.6.1173