Selective Exposure to Deserved Outcomesrepository.essex.ac.uk › 17767 › 1 ›...

Post on 04-Jul-2020

4 views 0 download

Transcript of Selective Exposure to Deserved Outcomesrepository.essex.ac.uk › 17767 › 1 ›...

SelectiveExposure1

SelectiveExposuretoDeservedOutcomes

AnnelieJ.Harvey

DepartmentofPsychology,AngliaRuskinUniversity,Cambridge,UnitedKingdom

MitchellJ.Callan*

DepartmentofPsychology,UniversityofEssex,Colchester,UnitedKingdom

RobbieM.Sutton

SchoolofPsychology,UniversityofKent,Canterbury,UnitedKingdom

TomFoulsham

DepartmentofPsychology,UniversityofEssex,Colchester,UnitedKingdom

WilliamJ.Matthews

DepartmentofPsychology,UniversityofCambridge,Cambridge,UnitedKingdom

*Correspondingauthor

Authors’note:

A.HarveyandM.Callancontributedequally.WethankRaelDawtryforhishelpfulcommentsonapreviousdraft.

SelectiveExposure2

Abstract

Researchhasshownthatpeopleoftenreinterprettheirexperiencesofothers’harmandsufferingto

maintainthefunctionalbeliefthatpeoplegetwhattheydeserve(e.g.,byblamingthevictim).Rather

thanfocusingonsuchreactiveresponsestoharmandsuffering,across7studiesweexamined

whetherpeopleselectivelyandproactivelychoosetobeexposedtoinformationaboutdeserved

ratherthanundeservedoutcomes.Weconsistentlyfoundthatparticipantsselectivelychosetolearn

thatbad(good)thingshappenedtobad(good)people(Studies1to7)—thatis,theyselectively

exposedthemselvestodeservedoutcomes.Thiseffectwasmediatedbytheperceived

deservingnessofoutcomes(Studies2and3),andwasreducedwhenparticipantslearnedthat

wrongdoersotherwisereceived“justdeserts”fortheirtransgressions(Study7).Participantswere

notsimplyselectivelyavoidinginformationaboutundeservedoutcomesbutactivelysought

informationaboutdeservedoutcomes(Studies3and4),andparticipantsinvestedeffortinthis

patternofselectiveexposure,seekingoutinformationaboutdeservedoutcomesevenwhenitwas

moretime-consumingtofindthanundeservedoutcomes(Studies5and6).Takentogether,these

findingscastlightonamoreproactive,anticipatorymeansbywhichpeoplemaintainacommitment

todeservingness.

Keywords:informationseeking;selectiveexposure;deservingness;beliefinajustworld

SelectiveExposure3

SelectiveExposuretoDeservedOutcomes

Alongtraditionofresearchintothepsychologyofjusticehasdemonstratedthatpeoplecare

abouttherelationshipbetweenthevalueofpeopleandthevalueoftheiroutcomes—thatis,they

careaboutdeservingness(Sabbagh&Schmitt,2016).Judgmentsofwhatisdeservedorundeserved

generallyfollowfromthesubjectiveperceptionoftherelationbetweenthevalueofpeople’sactions

andthevalueoftheiroutcomes,suchthatagood(bad)personreceivinganegative(positive)

outcomeisperceivedasundeserved,whereasthesamegood(bad)personreceivingapositive

(negative)outcomeisseenasdeserved(seeFeather,1999;Hafer,2011;Lerner,Miller,&Holmes,

1976).

Lerner(1977)arguedthatpeoplearemotivatedtobelievethattheyliveinaworldwhere

peoplegenerallygetwhattheydeserve,becausedoingsoenablesthemtocommittolong-term

goalswithconfidence.Becausebelievingina“just-world”isfunctional,peopleoftenreinterpret

theirexperiencesofunjusteventstomaintainperceptionsofdeservingness(forreviews,seeEllard,

Harvey,&Callan,2016;Hafer&Bègue,2005).Theclassicexampleofthisprocessisthederogation

ofinnocentvictims(Lerner&Simmons,1966),butrecentresearchhascastlightonavarietyof

otherwayspeoplemaintainacommitmenttojusticeanddeservingness(Callan&Ellard,2010),

includingmisrememberingdetailsofpastinjustices(Callan,Kay,Ellard,&Davidenko,2009;Marsh&

Greenberg,2006),perceivingfuturebenefitsforavictim’ssuffering(Hafer&Gosse,2011;Harvey&

Callan,2014),andofferingtohelpvictims(Bal&vandenBos,2015;Harvey,Callan,&Matthews,

2015).

Onefeaturecuttingacrossthesestrategiesformaintainingacommitmenttojusticeand

deservingnessisthattheyinvolvepeople’sreactiveresponsestoharmandsuffering.Take,for

example,immanentjusticereasoning,whichinvolvesbelievingthatabadoutcomewascausedby

someone’spriorimmoralbehavior,howeverphysicallyimplausiblesuchacausalconnectionmight

be(Callan,Sutton,Harvey,&Dawtry,2014).HarveyandCallan(2014)foundthatparticipants

causallyattributedafreakcaraccidenttothevictim’spriorconducttoagreaterextentwhenhe

SelectiveExposure4

previouslystolefromchildren(vs.didnotsteal).Further,participants’beliefsaboutwhathe

deservedmediatedthesecausalattributions—badoutcomeshappentobadpeoplebecausethey

deservethem.Throughimmanentjusticereasoning,then,peopleare,inretrospect,makingsenseof

arandombadoutcomebylocatingits“cause”inthepriormisdeedsoftheunfortunatevictim.

Butaconcernfordeservingnessmayalsoestablishanactive,anticipatorypreferencetosee

deservedoutcomes.Itiswell-documentedthatduringinformationseeking,peopletendto

selectivelyexposethemselvestoinformationthatisconsistentratherthaninconsistentwiththeir

attitudes,beliefs,anddecisions(forreviews,seeFrey,1986;Hart,Albarracín,Eagly,Brechan,

Lindberg,Lee,&Merrill,2009;Smith,Fabrigar,&Norris,2008).Inatypicalselectiveexposure

experiment,participantsareaskedtocommittoanattitude,opinion,ordecisionregardinganissue

andthenaregiventheopportunitytoreceiveadditionalinformationconcerningtheissue.The

additionalinformationisusuallypresentedasalistofshortstatements,commentaries,orabstracts

summarizingopposingperspectivesontheissue(ostensiblyfrompreviousparticipants,experts,

newsarticles,etc.).Forexample,Jonas,Schulz-Hardt,Frey,andThelan(2001)foundthat

participantstendedtochooseadditionalinformationthatsupportedratherthanconflictedwith

theirinitial“policy”decisionconcerningwhetherthegovernmentshouldfundalternativehealing

methodsoronlytraditionalmedicine.Thistendencyforpeopletoseekoutconfirmatoryinformation

hasbeenfoundinavarietyofdomains,includingsocialstereotypes(e.g.,Johnston,1996),smoking

activity(e.g.,Canon&Matthews,1972),investmentdecisions(Jonas&Frey,2003),attitudes

towardstoilettraining(Maccoby,Maccoby,Romney&Adams,1961),attitudestowardcapital

punishment(Smith,Fabrigar,Powell,&Estrada,2007),andreligiousbeliefs(McFarland&Warren,

1992).

Basedonlyontheknowledgeofanotherperson’smoralcharacterorconduct,observers

maybesimilarlybiasedtowardreceivingoutcomeinformationthatisconsistentwithwhatthat

persondeserves.Forexample,peoplemightprefertolearnthataserialrapistwascrippledinacar

crashmorethanlearningthathewonalottery,andtheymightgotosomelengthstodoso,

SelectiveExposure5

presumablybecauserapistsaremoredeservingofbeinginjuredthantheyareofwinninglotteries.

Likewise,peoplemightprefertolearnthatacharityworkerwonalotterymorethanlearningthathe

wascrippledinacarcrash,againbecauseoftheirconcernsaboutdeservingness.Suchselective

exposuretobad(good)outcomesforbad(good)peoplepointstoamoreproactive,anticipatory

routeforpeopletomaintainasenseofjusticeanddeservingnessthanhasbeenpreviously

recognized.Specifically,selectiveexposuretodeservedoutcomesmighthelppeoplenavigate

throughtheworldinawaythatsustainstheassumptionthatitisajustandfairplacewherepeople

getwhattheydeserve.Indeed,ifpeopleselectivelychoosetolearnaboutdeservedratherthan

undeservedoutcomes,thentheycanshieldthemselvesfromthepotentiallyunsettlingprospectthat

theworldisnotsofair,just,andnon-randomafterall.

Basedondissonancetheory(Festinger,1957),themainexplanationforselectiveexposure

tocongenialinformationisdefensemotivation,orthedesiretodefendone’sbeliefs,attitudes,or

decisions(seeHartetal.,2009).Accordingtothisaccount,peopleselectivelyexposethemselvesto

informationcongenialtotheirpriorattitudesanddecisionstoreduceoravoidthepotentialconcern

associatedwiththepossibilitythattheymightbewrong.Intheirmeta-analysis,Hartetal.(2009)

foundthattheeffectsofselectiveexposuretoconfirmatoryinformationincreasedasafunctionof

factorsthatincreasedefensemotivation,suchasmakingdecisionsunderhigh(vs.low)choice,

dedicatingtimeandefforttomakeadecision,justifyingdecisionstoothers,andreportinghigh

commitmenttoabeliefordecision.Whatweareproposinghere—thatpeopleselectivelyexpose

themselvestodeservedoutcomes—issimilarinsofaraspeoplearemotivatedtodefendthebelief

thatpeoplegetwhattheydeserve(Lerner,1980).Indeed,believinginajustworldmightbe

importantenoughtopeoplethatsimplypassivelyreceivinginformationaboutanotherperson’s

moralcharacterorconductmaybesufficienttoinstigateanactivesearchfordeservedoutcomes

whenthereisanopportunitytodoso.

SelectiveExposure6

Buildingonproceduresandmeasuresfromtheselectiveexposureliterature,wetestedthe

generalhypothesisthatpeoplewillselectivelychoosetolearnthatbad(good)thingshappenedto

bad(good)people.Theresultsofarecenteye-trackingstudylendweighttothishypothesis:Callan,

Ferguson,andBindemann(2013)foundthatthegoodorbadmoralconductofcharactersportrayed

withinaudio-visualscenesbiasedparticipants’anticipatorygazepreferencestoimagesofgoodor

badoutcomesjustbeforetheactualoutcomeswererevealed.Theseresultssuggestthatpeople

expect,viatheireye-movements,bad(good)thingstohappentobad(good)people.Butwhether

peopleselectivelychoosetolearnabout,andwillconsciouslyandactivelysearchfor,outcomesthat

areevaluativelyconsistentwiththemoralcharacterorconductofothershasyettobeexamined.

Specifically,althoughimpliedintheirwork,Callanetal.didnotdirectlyshowthatpeople’seye-gaze

preferencesforgood(bad)outcomesoccurringtogood(bad)peoplewerespecificallyduetotheir

concernsaboutdeservingness,nordidtheireye-trackingapproachallowthemtestpredictions

aboutwhetherpeoplemightincursomecosttoselectivelychoosetoreceiveadditionalinformation

aboutdeservedoutcomes.Wethereforeextendedpreviousresearchbyexamining(a)theactual

choicespeoplemakewhenfacedwiththeknowledgethatsomeoneismorallygoodorbad;(b)

whetherpeoplemightincursomecost,bywayoftheirtime,toreceivedeservingness-congruent

outcomeinformation(Studies4to6;cf.Frey,1981);and,importantly,(c)whethertheseselective

exposureeffectsoccurbecauseofpeople’sconcernsaboutdeservingness(Studies2,3,and7).

Sampling

Acrossstudiestheminimumrequiredsamplesizeswerefixedaheadofdatacollection;

however,thefinalsamplesizeswerenotcompletelypredeterminedduetotheremovalofsome

participants(e.g.,forfailingstorycomprehensioncheckquestions;seebelow).Poweranalyses

showedthatoursampleshadatleast80%power(usuallymuchhigher)todetect“medium”effect

sizes(e.g.,dz=.50forwithin-subjectscontrasts;two-tailed,α=.05).Wereportallmeasures,

manipulations,andexclusionsinthesestudies.

SelectiveExposure7

STUDY1

InStudy1,wepresentedparticipantswithaseriesofshortnarrativesdescribingdifferent

peopleengagingineithermorallygoodorbadbehavior.Wethenaskedparticipantstoratethe

extenttowhichtheywantedtolearnaboutdifferentpossiblegoodandbadoutcomesoccurringto

thetargetindividuals.Onthebasisoftheforegoinganalysis,wepredictedthatparticipantswould

wanttoreadmoreaboutthedeservingness-congruentoutcomesthanthedeservingness-

incongruentoutcomes.

Method

Participants

ParticipantsfromtheUnitedStateswererecruitedthroughAmazon’sMechanicalTurk(N=

48;54.2%females;Mage=36.52;SDage=12.80).

Materialsandprocedure

Wetoldparticipantsthestudywasabout“investigatingtheprocessingofnarrative

information”andthattheywouldrate(cf.Brannon,Tagler,&Eagly,2007;Lowin,1969)theextentto

whichtheywantedtoreaddifferentpossibleconclusionstoanumberofshortstories.

AdaptedfromCallanetal.(2013),andemployingafullywithin-subjectsdesign,we

presentedparticipantswith4shortstories(seeonlinesupplementalmaterialsforallofthescenarios

anditemsweusedacrossstudies):2describingagoodperson(e.g.,Jennysavedadrowningpuppy)

and2describingabadperson(e.g.,Sallystolefromacharitycollectionbox).Forexample,foroneof

thestoriesparticipantsread:

Aweekago,JennywaswalkingalongtheRiverWyewhenshespottedapuppydrowningintheriver.Riskingherownlife,Jennydivedintotheriverandsavedthepuppyfromdrowning.

Followingeachstory,participantsreadtwosentencesrepresentingadditionalpiecesof

informationaboutthetargetindividuals.Onesentencerepresentedagoodoutcome(e.g.,“One

weeklater,Jennywassittinginherlivingroomwhenshereceivednewsthatshehadwonanewcar

inasweepstakeshehadentered”)andtheotherrepresentedabadoutcome(e.g.,“Oneweeklater,

SelectiveExposure8

Jennywassittinginherlivingroomwhenshereceivednewsthatherhusbandwasinaterriblecar

accident”).1Participantswereaskedtoratetheextenttowhichtheywantedtoreadtheseendings

tothestories(1=Idonotwanttoreadthesedetailslateroninthesurveyto7=Iwanttoreadthese

detailslateroninthesurvey),ostensiblybecausetheirratingswoulddeterminetheconclusionsto

thestoriesthattheywouldactuallyreadandevaluate.Thestorieswerepresentedtoparticipantsin

afullycounter-balancedrandomorder.

ResultsandDiscussion

Participants’ratingsofwantingtheoutcomeinformationweresubmittedtoalinearmixed

effectsmodelusingthelme4package(Bates,Maechler,Bolker,&Walker,2015,version1.1-10)inR

(RCoreTeam,2015,version3.2.0).Thisformofregressionallowsustomodeleachparticipant’s8

separateresponsesasafunctionoffixedandrandomeffects,ratherthanrequiringustoaverage

thetwoexamplesofgood/badpeopletoformasingleobservationpercellofthedesign,asin

traditionalANOVA.ThemodelincludedfixedeffectsforPerson(goodvs.bad,coded+1and-1),

Outcomeratings(goodvs.bad,coded+1and-1)andthePersonXOutcomeinteraction.We

includedrandominterceptsforparticipantsandscenarios,andrandomslopesbyparticipantsforthe

effectsofPerson,Outcome,andPersonXOutcomeandbyscenariosfortheeffectofOutcome.That

is,weallowedbothmaineffectsandtheinteractiontovaryacrossparticipants,andallowedthe

effectsofOutcometovaryacrossscenarios.Notethat,becauseeachscenarioisonlyeverassociated

withonetypeofperson,wedidnotincludeby-scenariorandomslopesforPersonortheinteraction.

Randomeffectswereuncorrelated(Barr,Levy,Scheepers,&Tily,2013);includingthecorrelation

termsledtooverfittingandafailuretoconverge.WeusedSatterthwaiteapproximationsto

calculatep-valuesusingthelmerTestpackage(Kuznetsova,Brockhoff&Christensen,2015).Analyses

revealedasignificantPersonXOutcomeinteraction,b=0.52;95%CIof0.31,0.73;t=4.77,p<.001

(seeFigure1).Neithermaineffectachievedstatisticalsignificance(bothps>.25).Analysingthedata

1Inaseparatevalidationstudy(N=49),wefoundthatforeachofthescenariosweusedinStudy1,thegood(bad)targetcharacterswereperceivedasmoredeservingofthegood(bad)outcomes(allps<.001,dzs>.68).Detailsandfullstatisticalreportingofthisvalidationstudyarepresentedinthesupplementarymaterials.

SelectiveExposure9

withaconventional2x2ANOVAaftertakingthemeanofthetwoscenariosforagivenPersontype

yieldedexactlythesameconclusions.

Figure1.Theextenttowhichparticipantswantedtoreadoutcomeinformationasfunctionofthevalueofthetargets’moralvalueandthevalueoftheoutcomeinformation(Study1).Errorbarsshow95%CIsofthemeans.

Follow-upanalysesrevealedthatparticipantswantedtoreadthegoodoutcomeinformation

morethanthebadoutcomeinformationafterfirstreadingaboutagoodperson,B=0.513;95%CI

of0.305,0.757;t=6.18,p<.001(rbetweendependentmeasures=-.09).Whenreadingaboutabad

person,participantswantedtoreadthebadoutcomeinformationmorethanthegoodoutcome

information,B=-0.503;95%CIof-0.742,-0.263;t=-5.75,p<.001(rbetweendependentmeasures

=.01).Therefore,participantswantedtoreadconclusionstothenarrativesthatwereconsistent

withwhatthetargets’deserved—bad(good)thingshappentobad(good)people.

SelectiveExposure10

STUDY2

ThepurposeofStudy2wastwo-fold:(a)toreplicateourStudy1findingsusingabetween-

ratherthanwithin-subjectsdesign(withtheformerbeinglesssusceptibletocarry-overor

participantexpectationeffects),and(b)toexaminethemediatingrolethatperceiveddeservingness

playsintheeffectsofatargetperson’smoralvalueonparticipants’selectiveexposuretogoodand

badoutcomeinformation.

Method

Participants

ParticipantsfromtheU.S.A.wererecruitedonlinethroughMTurk(N=138;31.2%females;

Mage=30.45;SDage=10.07).Toensureindependenceofthedata,twoadditionalparticipantswere

notincludedinanalysesbecauseofduplicateIPaddresses(weretainedthedataforonlythefirst

occurrenceofeachduplicateIP).

Materialsandprocedure

Study2wassimilartoStudy1butweadoptedabetween-subjectsdesignsuchthateach

participantonlyreadandrespondedtooneshortstory.Participantsfirstreadaboutaman,named

Geoff,whoeitherthrewapuppyinariver(badperson)orsavedapuppyfromdrowninginariver

(goodperson):

(goodperson)Aweekago,GeoffwaswalkingalongtheRiverWyewhenhespottedapuppydrowningintheriver.Riskinghisownlife,Geoffdivedintotheriverandsavedthepuppyfromdrowning.(badperson)Aweekago,GeoffwaswalkingalongtheRiverWyewhenhespottedapuppyalongthebankoftheriver.Withnoregardforitslife,Geoffpickedupthepuppyandthrewitintheriver.Participantswerethenaskedtoanswerquestionsabout“thepossibleconclusionstothis

narrative”.ParticipantsfirstratedthedegreetowhichtheybelievedGeoffdeservedtoexperience

eachoftwooutcomes:“TowhatextentdoyoubelieveGeoffdeservestowinanewcarina

sweepstakeheenters?”and“TowhatextentdoyoubelievethatGeoffdeservestobeinaterrible

caraccidentthatleaveshiminhospitalinaseriouscondition?”(1=Notatalldeservingto7=Very

SelectiveExposure11

deserving).Next,mirroringtheoutcomesforthedeservingnessratings,participantsratedtheextent

towhichtheywantedtoreadaboutGeoffreceivingtwooutcomes:winningacarinasweepstake

andbeinginaterriblecaraccident(1=Idonotwanttoreadthesedetailslateroninthesurvey,7=I

wanttoreadthesedetailslateroninthesurvey).

ResultsandDiscussion

Becausethereisonlyoneobservationperconditionfromeachparticipant,andonlyone

scenario,weanalysedthedatawithaconventionalANOVAratherthanattemptingtofitamixed-

effectsmodel.A2(person:badvs.good)by2(outcomeinformation:badvs.good)mixedANOVA

withPersonasthebetween-subjectsfactorrevealedasignificantinteractionforparticipants’

judgmentsofdeservingness,F(1,135)=252.91,p<.001,ω2=.63(oneparticipantdidnotanswer

thedeservingnessquestions).

Figure2.Perceptionsofdeservingness(leftpanel)andratingsofwantingtoreadoutcomeinformation(rightpanel)asfunctionofthevalueofthetargets’moralvalueandthevalueoftheoutcomeinformation(Study2).Errorbarsshow95%CIsofthemeans.

ShowninFigure2(leftpanel),participantsinthegoodpersonconditionratedGeoffasmore

deservingofthegoodoutcomethanthebadoutcome,t(69)=17.55,p<.001,dz=2.10(rbetween

SelectiveExposure12

repeatedmeasures=-.034).Conversely,participantsinthebadpersonconditionratedGeoffas

moredeservingofthebadoutcomethanthegoodoutcome,t(69)=-7.37,p<.001,dz=-0.90(r

betweenrepeatedmeasures=-.306).TherewerealsosignificantmaineffectsofPersonand

OutcomeInformation(bothps<.01).

Therewasalsoasignificantinteractionforparticipants’ratingsofhowmuchtheywantedto

readthepotentialoutcomeinformation,F(1,136)=83.01,p<.001,ω2=.37.ShowninFigure2

(rightpanel),participantsinthegoodpersonconditionwantedtoreadthegoodoutcome

informationtoagreaterextentthanthebadoutcomeinformation,t(69)=6.52,p<.001,dz=0.78(r

betweenrepeatedmeasures=-.344).Participantsinthebadpersonconditionwantedtoreadthe

badoutcomeinformationmorethanthegoodoutcomeinformation,t(67)=-6.46,p<.001,dz=-

0.78(rbetweenrepeatedmeasures=.211).Neithermaineffectachievedstatisticalsignificance(ps

>.37).

Toexaminewhetherparticipants’perceptionsofdeservingnessmediatedtheeffectofthe

target’smoralworthontheirwantingtoreadgoodversusbadoutcomeinformation,wefirst

computeddifferencescoresforbothparticipants’ratingsofdeservingnessandtheirratingsof

wantingoftheoutcomeinformation;positivevaluesrepresentparticipants’beliefthatGeoff

deservedagoodoutcomemorethanabadoutcomeandthattheywantedtoreviewthegood

outcomeinformationmorethanthebadoutcomeinformation.Thesedifferencescoreswerehighly

correlated,r=.74,p<.001(pooledacrossconditions),suchthatthemoreparticipantsbelieved

Geoffdeservedagood(vs.bad)outcome,themoretheywantedtoreviewgood(vs.bad)outcome

informationlaterinthesurvey.

Bootstrappedanalyses(Preacher&Hayes,2008;10,000resamples)revealedthatperceived

deservingnessmediatedtheeffectofthetarget’smoralworth(good=1vs.bad=-1)ontherelative

ratingsofwantinggoodvs.badoutcomeinformation(indirecteffect=1.696,95%BiasCorrected

andAcceleratedConfidenceInterval[BCaCI]of1.147,2.24;seeFigure3).Theseresultssuggestthat

SelectiveExposure13

oneofthereasonswhyparticipantswantedtoreviewgood(bad)outcomeinformationforagood

(bad)personwasbecausetheybelievedhedeservedit.

Figure3.Theinfluenceofthemoralworthofthetargetcharacteronselectiveexposureofthegood(vs.bad)outcomesthroughperceiveddeservingnessofthegood(vs.bad)outcomes.Valuesshowunstandardizedregressioncoefficients.*p<.05

STUDY3

Studies1and2foundthatparticipantswantedtoexposethemselvestooutcomeinformationthat

wasevaluativelycongruentwithwhatotherpeopledeserved.InStudy3,weaimedtoreplicate

thesefindingsusingadifferentselectiveexposureparadigm.Specifically,ratherthanhaving

participantsratetheextenttowhichtheywantedtoreceivegoodandbadoutcomeinformation,we

askedthemtochooseamongseveraldifferentgoodandbadoutcomestoreview(cf.Jonas,etal.,

2001).Wealsoexploredtheinterplaybetweenselectiveexposureandselectiveavoidancebyasking

participantstoseparatelychoosetheoutcomestheywantedtoreadandtheoutcomestheydidnot

wanttoread(cf.Frey&Wicklund,1978,Rhine,1967).Onepossibilityisthatratherthanselectively

seekingoutcomeinformationthatiscongruentwithwhatothersdeserve(e.g.,thatacharityworker

wonthelottery),peoplemightselectivelyavoidoutcomeinformationthatconflictswiththeirneed

toseethatpeoplegetwhattheydeserve(e.g.,thatacharityworkerwascrippledinacarcrash).Of

course,selectiveexposureandselectiveavoidancecouldbeoppositesidesofthesamecoininsofar

asbothenablepeopletomaintaintheassumptionthatpeoplegetwhattheydeserve.LikeStudy2,

SelectiveExposure14

wealsoaskedparticipantstoratetheextenttowhichthetargetpersondeservedvariousgoodand

badoutcomesasapotentialmediatorofselectiveexposureandselectiveavoidance.

Method

Participants

ParticipantsfromtheU.S.A.wererecruitedonlinethroughMTurk(N=137;46%females,

0.7%unreported;Mage=33.57;SDage=11.68).Fouradditionalparticipantswerenotincludedin

analysesbecauseofduplicateIPs(n=2)orfailingasimplestorycomprehensioncheck(“Inthestory

youreadatthebeginningofthesurvey,whatdidChrisdotothepuppy?”;n=2).

Materialsandprocedure

LikeStudy2,participantsreadastoryaboutaperson(thenameChriswasusedinStudy3)

whoeitherdrownedapuppyinariver(badperson)orsavedapuppyfromdrowninginariver(good

person).Afterreadingthestory,participantswereasked,“TowhatextentdoyoubelieveChris

deservesto…”andthensawalistof8outcomesthattheywereaskedtorateona7-pointscale(1=

Notatalldeservingto7=Agreatdealdeserving).Ofthe8outcomes,4werebadoutcomes(α=

.98):“…beinjuredinacaraccident”,“…befiredfromhisjob”,“…contractamajorillness”and“have

hisapartmentdestroyedbyaflood”.Theother4outcomesweregoodoutcomes(α=.98):“…win

$100,000playingascratchcardlotteryticket”,“…havehisstocksandsharesskyrocket”,“…begivena

promotionatwork”and“…winaluxurycruisetrip”.

Participantsthensawalistof8possibleoutcomestothestoryaboutChris,whichwere

identicaltotheoutcomesparticipantsratedintermsofhisdeservingness(e.g.,injuredinacar

accident).Weinstructedparticipantstoselectonly2oftheoutcomestheywould“DEFINITELY”want

toreadlaterinthesurveyandonly2outcomesthattheywould“DEFINITELYNOT”wanttoread

(i.e.,weimposedinformationlimits,seeFischer,Jonas,Frey,&Schulz-Hardt,2005).Werandomized

theorderofthequestionssoparticipantswereeitheraskedtochooseamongthe“definitelywant”

outcomesfirstorthe“definitelynotwant”outcomesfirst.

SelectiveExposure15

ResultsandDiscussion

TherewasasignificantPersonXOutcomeInformationinteractionforparticipants’

judgmentsofdeservingness,F(1,134)=306.35,p<.001,ω2=.69.Participantsinthegoodperson

conditionratedChrisasmoredeservingofthegoodoutcomes(M=5.08,SD=1.67)thanthebad

outcomes(M=1.27,SD=0.74),t(68)=16.23,p<.001,dz=1.95(rbetweenrepeatedmeasures=-

.193).ParticipantsinthebadpersonconditionratedChrisasmoredeservingofthebadoutcomes

(M=4.50,SD=2.31)thanthegoodoutcomes(M=1.33,SD=0.83),t(67)=-9.79,p<.001,d=-1.19

(rbetweenrepeatedmeasures=-.28).Neithermaineffectachievedstatisticalsignificance(ps>.11)

Weanalysedthenumberofbadoutcomesparticipantswantedtoread(whichcouldrange

from0to2)betweenthebadandgoodpersonconditions.Becausewefixedthetotalnumberof

choicesparticipantscouldmaketo2,andtherewereanevennumberofgoodandbadoutcome

options,theresultsareidenticalusingthenumberofgoodoutcomesparticipantschoseasthe

dependentvariable(exceptforoppositesign).Thus,wereportonlytheresultsforthenumberof

badoutcomeparticipantswantedtoread(and,perbelow,onlythegoodoutcometheydidnotwant

toread).

Participantsinthebadpersonconditionchosetoreadmorebadoutcomeslaterinthe

survey(M=1.75,SD=0.56)thanparticipantsinthegoodpersoncondition(M=0.26,SD=0.56),

t(134.99)=15.61,p<.001,d=2.66(hereandthroughout,degreesoffreedomwereWelch-

correctedwherenecessary).Participantsinthebadpersonconditionalsochosenottoreceivemore

goodoutcomesonaverage(M=1.59,SD=0.78)thanparticipantsinthegoodpersoncondition(M=

0.20,SD=0.53),t(118.16)=12.17,p<.001,d=2.08.Theresidualcomponentsfromtheseanalyses

werenotnormallydistributed,butnon-parametrictests—specifically,Mann-WhitneyUtestsand

percentilebootstrapconfidenceintervals(CI)ofthemeandifferences(5,000samples)—yieldedthe

samepatternofresultsfortheeffectofthemoralvalueofthetargetonthebadoutcomes

participantswantedtoread,Z=9.36,p<.001,95%bootstrapCIof1.30and1.66,andthegood

outcomesparticipantsdidnotwanttoread,Z=8.42,p<.001,95%bootstrapCIof1.16and1.60.

SelectiveExposure16

Participants’choicesofthebadoutcomestheywantedtoreadandthegoodoutcomestheydidnot

wanttoreadwerehighlycorrelated,r=.87,p<.001.

FollowingStudy2,weexaminedwhetherparticipants’perceptionsofdeservingness

mediatedtheeffectofthetarget’smoralworthontheirchoicestoreadandnottoreadgoodand

badoutcomeinformation.Wefirstcomputeddifferencescoresforparticipants’ratingsof

deservingness(meanofthedeservingnessratingsforthebadoutcomesminusthemeanforthe

goodoutcomes).Thesescorescorrelatedhighlywithparticipants’choicestoreceivebadoutcome

informationandnotreceivinggoodoutcomeinformation(rs=.80and.76,ps<.001,respectively).

ShownifFigure4,bootstrappedanalyses(Preacher&Hayes,2008;10,000resamples)revealedthat

perceiveddeservingnessmediatedtheeffectofthetarget’smoralworth(good=1vs.bad=-1)on

wantingtoreceivebadoutcomeinformation(totaleffect=-0.74;indirecteffect=-0.33,95%BCaCI

of-0.485,-0.175)and,inaseparateanalysis,notwantingtoreceivegoodoutcomeinformation

(totaleffect=-0.69;indirecteffect=-0.42,95%BCaCIof-0.589,-0.245).

SelectiveExposure17

Figure4.Theinfluenceofthemoralworthofthetargetcharacteronselectiveexposuretobadoutcomes(topsection)andselectiveavoidanceofgoodoutcomes(bottomsection)throughperceiveddeservingnessofthegood(vs.bad)outcomes.Valuesshowunstandardizedregressioncoefficients.*p<.05

Study4

InStudy3wefoundthatparticipantsselectivelysoughtdeserved,andselectivelyavoided

undeserved,outcomes.OneissuewithourStudy3designisthataskingparticipantstoprovideboth

theoutcomestheywantedtoreadandtheoutcomestheydidnotwanttoreadmayhave

introducedsomeuncertaintyaboutwhichoutcomestheywouldactuallyreadaftertheymadetheir

selections.Assuch,participantsmayhavechosentoreaddeservingness-congruentoutcomesnot

becausetheywantedtoreadthempersebuttoincreasetheprobabilitythattheywouldnotbe

exposedtodeservingness-incongruentoutcomes(orviceversa).InStudy4,wedisentangled

selectiveavoidanceandselectiveexposurebyincludingevaluatively-neutraloutcomeoptionsand

SelectiveExposure18

askingparticipantstoonlychoosetheoutcomestheywantedtoread.Inthisdesign,participants

demonstrateselectiveexposureiftheychoosetoreadcongruentinformationmorethan

incongruentorneutralinformation.Forexample,Jang(2014)foundthatparticipantsselectively

chosetoreadnewsarticlesthatwereconsistentwiththeirpoliticalattitudes(e.g.,“10Reasonstobe

Pro-Choice”forsomeonewithapositiveattitudetowardpro-choice)morethannewsarticlesthat

wereeitherinconsistentwiththeirattitudes(e.g.,“AbortionHarmfultoMentalHealth”)orwere

neutral(e.g.,“AbortionIssueArisesinBudgetDebate”).Therewasalsonodifferencebetween

participants’choicesofnewsarticlesthatwereneutralorinconsistentwiththeirattitudes.Thus,in

Study4,ifparticipantsarenotselectivelyexposingthemselvestodeservingness-congruent

informationthenthereshouldbenoselectiondifferencebetweendeservingness-congruentand

neutraloutcomes.Inotherwords,includingneutraloutcomeinformationintroducesacontrolto

testwhetherparticipantsareselectivelyexposingthemselvestoinformationthatisspecifically

deservingness-congruentandnotsimplyavoidingdeservingness-incongruentinformation.

InStudy4wealsodidnotimposeanylimitsonthenumberofoutcomesparticipantscould

chosetoread,includingtheoptiontoreadnooutcomeswhatsoever.Doingsoeffectivelyintroduced

acosttoinformationseeking(cf.Frey,1981)—choosinganyoutcomesatallwouldmean

participantswouldhavetoforegosomeoftheirfreetimetofurtherreadandevaluatetheirchosen

outcomesratherthansimplyendingthesurvey.

Method

Participants

Participantswererecruitedonline(N=151;41.1%females;Mage=34.26;SDage=11.76)using

MTurk.ElevenadditionalparticipantswerenotincludedinanalysesbecauseofduplicateIPs(n=7)

orfailingasimplestorycomprehensioncheck(n=4;“Inthestoryyoureadatthebeginningofthe

survey,whatdidSallydoatthecornerstore?”).

Materialsandprocedure

Participantswereinvitedtotakepartinastudy“investigatingtheprocessingofnarrative

SelectiveExposure19

information”.TheyfirstreadaboutSally,whowasbuyingbreadandmilkatacornerstore.Halfof

theparticipantsreadthat,afterpaying,Sally“stoleallthechangefromacharitycollectionthatwas

ondisplayatthecounter”(badperson).TheotherhalfreadthatSally“putallhersparechangeinto

thecharitycollectionthatwasondisplayatthecounter”(goodperson).

Afterreadingthestory,participantswerepresentedwithalistof9possibleoutcomesfor

Sally.Ofthe9outcomes,3werebad(“Sallywasinjuredinacaraccident”,“Sally’sground-floor

apartmentwasflooded”,“Sallycamedownwithaseriousillness”),threeweregood(“Sally’swon

$1,000playingascratchcardlotteryticket”,“Sally’sstocksandsharesskyrocket”,“Sallywasgivena

majorpromotionatwork”)andthreewereneutral(“Sallywenttoaconcert”,“Sallystartedwritinga

newblog”,and“Sallytidiedupheroffice”).Participantsweretoldthattherewerenolimitsonthe

numberofoutcomestheycouldchoosetoread(“Whichoftheseevents,ifany,wouldyouliketo

readmoreaboutconcerningSally'slifesoonaftertheincidentatthecornerstore?”).

ResultsandDiscussion

Overall,alargemajorityofparticipants(94.7%)chosetoreceiveatleastoneoftheoutcome

informationoptions(Mode=1,M=2.58,SD=1.95).Participants’choicesfortheoutcomesthey

wantedtoreadweresubjectedto2(Person:goodvs.bad)X3(Outcomes:goodvs.neutralvs.bad)

mixedANOVAwithrepeatedmeasuresonthesecondfactor.Analysesrevealedsignificantmain

effectsforPerson,F(1,149)=12.82,p<.001,ω2=.07,andOutcome,F(1.77,264.04)=12.82,p<

.001,ω2=.07(Greenhouse-Geissercorrected).Moreimportantly,analysesrevealedasignificant

interaction,F(1.77,264.04)=23.63,p<.001,ω2=.12(seeTable1).Theresidualcomponentsfrom

thisanalysiswerenotnormallydistributed.Wethereforesupplementedconventionalfollow-up

pairedsamplest-testswithnon-parametrictests—specifically,WilcoxonSignedRankstestsand

percentilebootstrapconfidenceintervals(CI)ofthemeandifferences(5,000samples).

Follow-upanalysesrevealedthat,withinthebadpersoncondition,participantschosemore

badoutcomesonaveragethanneutraloutcomes,t(73)=4.65,p<.001,WilcoxonSignedRankstest

Z=4.20,p<.001,95%bootstrapCIof0.386and0.965(rbetweenrepeatedmeasures=.06).There

SelectiveExposure20

wasnosignificantdifferencebetweentheneutralandgoodchoiceswithinthebadpersoncondition,

t(73)=-0.66,p=.501,Z=-.90,p=.369,95%bootstrapCIof-0.206and0.107(rbetweenrepeated

measures=.60).Withinthegoodpersoncondition,participantschosemoregoodoutcomeson

averagethanneutraloutcomes,t(76)=-6.33,p<.001,WilcoxonSignedRankstestZ=5.21,p<

.001,95%bootstrapCIof-1.10and-0.57(rbetweenrepeatedmeasures=.25).Therewasno

significantdifferencebetweentheneutralandbadchoiceswithinthegoodpersoncondition,t(76)=

0.82,p=.415,Z=.92,p=.356,95%bootstrapCIof-0.185and0.445(rbetweenrepeatedmeasures

=-.01).

Table1.Theeffectofthemoralvalueofthetargetonparticipants’choicesoftheoutcomeinformationtheywantedtoreadlaterinthesurvey. ValueofOutcome

Bad Neutral Good

ValueofPerson Bad(n=74)

1.11a(1.05) 0.43b(0.74) 0.49b(0.80)

Good(n=77) 0.84b(1.05) 0.71b(0.90) 1.56a(0.98)

NoPersonInformation(N=101)

0.69a(0.87) 0.58a(0.74) 1.00b(0.82)

Note.Valueswithincellsshowmeans(standarddeviations)ofthenumberofchoices.Meanswithinthegoodandbadpersonconditionsandfortheseparate“nopersoninformation”samplethatdonotshareacommonsubscriptacrossrowsaresignificantlydifferent(p<.05).

Whatoutcomesdopeopletendchosetolearnaboutwhentheyarenotgivenany

informationaboutthetargetcharacter’spriormoralconduct?Toaddressthisquestion,werecruited

aseparatesampleofonlineparticipants(N=101;n=1additionalparticipantremovedforhavinga

duplicateIPaddress;55.9%female;Mage=36.15;SDage=12.00)andaskedthemtochooseamong

thesame9outcomeoptionsfor“Sally”asinthemainstudy,butwegavethemnoinformation

aboutherpriormoralconduct.Specifically,theyread:“Belowyouwillseeseveraleventsthat

happenedinthelifeofawomannamedSally.AllthatyouknowaboutSallyishername.Which

SelectiveExposure21

eventswouldyouliketoreadmoreabout?”.ShowninthebottomsectionofTable1,therewasa

tendencyforparticipantstochosetoreadaboutthegoodoutcomesmorethantheneutral

outcomes,t(100)=3.81,p<.001,WilcoxonSignedRankstestZ=3.55,p<.001,95%bootstrapCIof

0.21and0.63,andthegoodoutcomesmorethanthebadoutcomes,t(100)=2.71,p=.008,

WilcoxonSignedRankstestZ=2.50,p=.013,95%bootstrapCIof0.079and0.525.Therewasno

significantdifferencebetweenparticipants’choicesofbadversusneutraloutcomes,t(100)=0.91,p

=.364,WilcoxonSignedRankstestZ=1.04,p=.30,95%bootstrapCIof-0.128and0.347.Thus,

absentanyinformationaboutthetargetperson’spriormoralconduct,participantstendedto

choosetoreceivegoodoutcomeinformationmorethanbadandneutraloutcomeinformation.This

isconsistentwiththeoriesthatsuggestpeoplearePollyanish(e.g.,Taylor&Brown,1988)or

Panglossian(Kayetal.,2007)–thatis,generallybiasedtowardsperceivingtheirsocialenvironments

inamorepositivethannegativelight.Morespecifically,itisconsistentwiththeoreticalmodelsthat

castselectiveexposureasamoodregulatingprocess,inwhichabsentothermotives,peopleprefer

hedonicallypositiveovernegativeinformation(Knobloch&Zillman,2002;Oliver,2003;Zillman,

1988).GiventhepatternofresultsfromourmainsampleinStudy4(topsectionofTable1),this

tendencyseemstoshiftwhenparticipantsdohaveknowledgeofthetargetperson’smoralconduct:

evenwhentheyhadtheopportunitytochoosenooutcomeinformationatall,participants

selectivelyexposedthemselvestobad(good)outcomesforabad(good)person.Giventhatthere

werenodifferencesbetweenthechoicesofneutralandgoodoutcomeswhenthetargetpersonwas

badandneutralandbadoutcomeswhenshewasgood,thesefindingsalsosuggestthatparticipants

areprimarilyselectivelyseekingdeservingness-congruentinformationratherthanselectively

avoidingdeservingness-incongruentinformation.

STUDY5

InStudies5and6weexaminedwhetherparticipantsmightexpendsomeefforttoreceive

outcomeinformationthatisconsistentwithwhatothersdeserve.InStudy5,wemadeitrelatively

easyordifficultforparticipantstoselecttheoutcomeinformationtheywantedtoreceive.Wedidso

SelectiveExposure22

byhavingparticipantssearchwithinvisualarraysforshapesassociatedwithgoodorbadoutcomes,

whichwereeitherrelativelyeasyorhardtofind.Wepredictedthatbecausepeoplewanttoseethat

othersgetwhattheydeserve,participantswouldchoosetoreadgood(bad)outcomeinformation

forgood(bad)peopleevenwhenitwasrelativelymoredifficulttodoso.

Method

Participants

ParticipantsfromtheU.S.A.wererecruitedonlinethroughMTurk(N=49;57.10%female;

Mage=35.94;SDage=9.59).

Materialsandprocedure

Usingafullywithin-subjectsdesign,participantsinStudy5werepresentedwiththesame

shortstories(2describingagoodpersonand2describingabadperson)andchoicesofadditional

goodorbadoutcomeinformationthatweusedinStudy1(seesupplementarymaterials).For

example,foronestoryparticipantsread:

StevewasridingontheLondonUndergroundtoSt.James’sparktomeethisgirlfriendforapleasantoutdoorpicnicinthepark.Whileatastop,afrailoldladyenteredthesamecarriageoccupiedbySteve.Insteadofgettingupandofferinghisseat,Stevescowledattheoldladyandrefusedtogiveuphisseat.Followingeachstory,participantswerepresentedwithtwosentencesthatrepresentedtwo

piecesofadditionalinformationaboutthestory;onedescribingagoodoutcome(e.g.,“Jennywas

sittinginherlivingroomwhenshereceivednewsthatshehadwonanewcarinasweepstakesshe

hadentered”)andonedescribingabadoutcome(e.g.,“Jennywasinaterriblecaraccidentthatleft

herinhospitalinseriouscondition”).Eachoutcomewaspairedwithacoloredshape;weasked

participantstodecidewhichpieceofinformationtheywantedtoreadlaterinthesurveyby

searchingforandclickingontheshapeassociatedwiththatoutcomewithinanarrayofshapesthat

wouldbeshownonthenextpage.

Onthenextpageparticipantsperformedasearchtaskwheretheysawanarrayofdifferent

shapes,includingthetwotargetshapestheyjustlearnedwereassociatedwiththegoodandbad

SelectiveExposure23

outcomeinformationtheycouldreceive(seeFigure5).

Figure5.AnexamplearrayofshapesusedinStudy3.Inthisexample,theredequilateraltriangle(centeroffigure)representedthedeservingnesscongruentinformationandthebluerhombusrepresentedthedeservingnessincongruentinformation.

Wetoldparticipantstofindandclickontheshapeinthearraythatrepresentedthe

informationtheywantedtoreadlaterinthesurvey.Wedesignedeacharraywithreferenceto

performanceinvisualsearchexperiments,andthetheoriesthathavebeenadvancedtoexplainthis

performance(e.g.,FeatureIntegrationTheory:Triesman&Gelade,1980;GuidedSearch:Wolfe,

1994).Inessence,thesetheoriesproposethatitismoredifficult(lessefficient)tofindtargetitems

thatsharefeatureswithdistractoritems.Targetswhicharedefinedbyasingleuniquefeature(e.g.,

theblueiteminFigure4)seemto“pop-out”andarefoundquicklyregardlessofthenumberof

surroundingitemsandwithouthavingtoallocatefocusedattention.Targetswhicharedefinedbya

conjunctionoffeatures(e.g.,theredequilateraltriangleinFigure4)takemuchlongertofindand

arethoughttoinvolveamorelaborioussearchwhereattentionismovedseriallyfromitem-to-item.

Inthepresentstudy,theshaperepresentingthedeservingness-congruentinformation(e.g.,

agoodoutcomeforagoodperson)wasalwaysthesamecolorasthedistractorshapes(e.g.,thered

SelectiveExposure24

equilateraltriangleinFigure5),andwasthereforelesseasytofind.Theshaperepresentingthe

deservingness-incongruentinformationwasalwaysa“pop-out”targetwithadifferentcolortoall

theothershapesinthearray(e.g.,thebluerhombusinFigure5),andwouldthereforebefound

“preattentively”withlittleeffort.Inotherwords,wemadeitmoredifficultforparticipantstofind

informationthatwascongruent(vs.incongruent)withwhatthetargetsdeserved.Weasked

whetherparticipantswouldstillchoosecongruentinformationevenwhentheirvisualattentionwas

capturedbytheincongruentoutcome,makingitamuchquickeroptiontochoose.

Results

Weusedmixedeffectslogisticregressiontoanalyzeparticipants’choicesofoutcomes

acrossthescenarios.ThedependentvariablewaswhetherparticipantschoseaGoodoutcome

(coded1)oraBadoutcome(coded0).ThepredictorvariablewasPerson(good,coded1,vs.bad,

coded-1).Weincludedrandominterceptsforparticipantsandscenarios,andrandomslopesby

participantsfortheeffectofPerson(withcorrelatedrandomeffects).Therewasanoveralltendency

tofavourGoodoutcomes,Bintercept=3.42,Z=3.83,95%CI1.67and5.18,p<.001(cf.Study4);more

importantly,participantsweremorelikelytochooseagoodoutcomeforagoodperson(73%)than

forabadperson(34%),Bperson=4.15,Z=4.71,95%CIof2.43and5.88,p<.001.

STUDY6

Evenwhenitwasrelativelydifficulttodoso,participantsinStudy5tendedtosearchforand

chooseoutcomesconsistentwithwhatthetargetpersonsdeserved.OurassumptioninStudy5was

thatparticipants’attentionwasinitiallydrawntothedistractorshape,andthatparticipantsthen

hadtodisengageandavoidthetemptationtochoosethisshapebyactivelysearchingfortheshape

associatedwiththedeservingness-congruentoutcome.Astrongertestofthisassumptionisto

examinedifferencesinresponsetimesduringvisualsearchforeachtypeofshape.Whenashapeis

bothvisuallysalient(e.g.,islargerandadifferentcolorthanothershapeswithinanarray)and

associatedwiththeinformationthatparticipantsgenerallywanttoview(i.e.,thedeservingness-

congruentinformationiseasiertofind),responsesshouldbequick.Incontrast,whenashapeisnot

SelectiveExposure25

visuallysalient(e.g.,issmallerandthesamecolorasothershapes)andisassociatedwithwhat

participantswanttoview(i.e.,thedeservingness-congruentinformationishardertofind),then

responsesshouldbeslower,assumingparticipantsareengaginginamoreactive,elaboratesearch

forthisdesiredinformation.

Method

Participants

ParticipantsfromtheU.S.A.wererecruitedonlinethroughMTurk(N=176;46%females;

Mage=33.47;SDage=11.15).Thirtyadditionalparticipantswereexcludedfromanalysesbecauseof

duplicateIPaddresses(n=6),failingasimplemultiplechoicestorycomprehensioncheck(n=8),

skippingthesearchtaskaltogether(n=8),selectingbothshapes(n=7),orhavingnotimingdata

registered(n=1).

Materialsandprocedure

Wetoldparticipantsthattheywouldreadoneshort“incomplete”storyaboutanindividual

goingabouthisdailylife.Participantswerethentoldtheywouldbeshowntwoshapes,each

representingtwodifferentoutcomesfortheprotagonist.Theirtaskwastofindandclickonthe

shapewithinanarrayofshapesthatrepresentedtheoutcomeinformationtheywantedtoread.

Allparticipantsreadthesameshortstory,whichdescribedamannamedGeoffwhothrewa

puppyintoariver(wefocusedonabadpersoninStudy6tosimplifythedesign).Participantswere

thenpresentedwithtwopiecesofadditionalinformation,representedbytwoseparatesentences.

OnesentencedescribedGeoffwinninganewcarinasweepstake(goodoutcome);theother

sentencedescribedGeoffbeinginvolvedinaterriblecaraccidentthatlefthiminhospitalinserious

condition(badoutcome).

Twoshapeswereused:alargeorangecircleandasmallbluestar.ShowninFigure6,the

orangecirclewasthelargestiteminthedisplayandwasdefinedbyauniquecolor,makingita“pop-

out”target.Thebluestarwassmallerandthesamecolorasmultipledistractors,sowepredicted

thatparticipantswouldbemuchlessefficientandsloweratfindingit.Theshapeassociatedwith

SelectiveExposure26

eachoutcomewasvariedbetweensubjectssuchthatthecongruentoutcomewaseitherharder

(bluestar)oreasier(orangecircle)tofind.

Afterthesearchtask,participantsansweredastorycomprehensioncheckitem:“Inthebrief

storyyouread,theman(Geoff):(a)divedintotheriverandsavedthepuppy,(b)threwthepuppy

intotheriver,or(c)whatstory?”.Next,theyanswered‘yes’or‘no’toaquestionaboutwhether

theyselectedtheshapetheyinitiallywanted(“Duringthesearchtask,didyouendupfindingand

selectingtheshapeyouinitiallywantedtofind?”),and,ifno,whytheydidnotselecttheshapethey

initiallywantedtosearchfor(theoptionswere“Icouldn’tfindit”,“Ithoughtitprobablywasn’t

actuallyinthearrayofshapes”and“other,pleasespecify”).

Figure6.ThearrayofshapesusedinStudy4.Theorangecirclewastheeasy-to-findshapeandthebluestarwasthehard-to-findshape.Participantsweretoldeitherthatdeservingness-consistentorinconsistentinformationwasassociatedwithoneoftheseshapes.

ResultsandDiscussion

Agreaterproportionofparticipantschosethedeservingnesscongruentoutcome(72%)than

theincongruentoutcome,χ2contingencytest=32.82,p<.001.Thistendency,however,differedas

afunctionofhoweasyorharditwastofindthecongruentoutcome,χ2contingencytest=26.11,p<

.001,suchthatalargerproportionofparticipantschosetheshapeassociatedwiththecongruent

SelectiveExposure27

outcomewhenitwaseasiertofind(89%)thanwhenitwashardertofind(54%).Moreimportantly,

analysisofparticipants’log-transformedsearchtimes2revealedthatparticipantstooklongerto

chooseashapewhenthedeservingness-congruentinformationwasharder(M=2.73,SD=1.01)

thanwhenitwaseasier(M=1.70,SD=1.15)tofind,t(171.92)=6.29,p<.001,d=.95.

Ancillarysub-groupanalysesshowedthatparticipantswhochosethecircle(theeasier-to-

findoption)tooklongertodosowhenitwasassociatedwiththeincongruent(good)outcome(n=

40;M=2.23,SD=1.16)thanwiththecongruent(bad)outcome(n=79;M=1.50,SD=1.04),

t(71.03)=3.39,p=.001,d=.66.Therewerenosignificantdifferencesinsearchtimesamong

participantswhochosethestar(theharder-to-findoption)whenitwasassociatedwiththe

incongruentoutcome(n=10;M=3.25,SD=0.82)versusthecongruentoutcome(n=47;M=3.14,

SD=0.62),t(11.27)=-0.40,p=.70,d=-0.15.Thedifferenceofthesedifferenceswassignificant,F(1,

172)=4.86,p=.029.Further,20participantsreportednotultimatelychoosingtheshapethey

wantedtochoose,andalloftheseparticipantschosetheeasier-to-findcircle(80%reportedthey

couldnotfindthebluestar;20%thoughtitactuallywasn’tinthearray).Amongparticipantswho

chosethecircle,asignificantlygreaterproportionreportedselectingtheshapetheydidnotwantto

selectwhenthecirclewasassociatedwiththeincongruentoutcome(14/40,35%)thanwhenitwas

associatedwiththecongruentoutcome(6/79,8%)χ2=14.26,p<.001.Theseresultssuggeststhat

eventhoughmanyparticipantsultimatelychosetheeasier-to-findoptionwhenitwasassociated

withthedeservingness-incongruentoutcome,theywereactivelysearchingforthedeservingness-

congruentoutcome,becausetheytooklongertomaketheirchoiceinsteadofimmediatelyselecting

theeasy-to-findshape.

Study7

Weproposedthataconcernfordeservingnessisoneofthereasonswhypeoplemight

selectivelyexposethemselvestobad(good)thingshappeningtobad(good)people,andtheindirect

2Forpurposeofanalysis,searchtimewaslog-transformedtohelpsymmetrizethedata.Theresultsaresimilarusingrawsearchtimes(Ms=22.87vs.10.74seconds),t(161.84)=4.36,p<.001,d=.66.

SelectiveExposure28

effectsofperceiveddeservingnesswefoundinStudies2and3suggestthatthisisthecase.

However,oneissuewiththesemediationfindingsisthatperceiveddeservingnesswasmeasured

ratherthanmanipulated,soitscausalinfluenceisunclear(i.e.,deservingnessmightbeajustification

for,ratherthanacauseof,selectiveexposuretogoodandbadoutcomes).Study7,then,was

designedtoprovidefurtherevidencefortheideathataconcernfordeservingnessunderlies

selectiveexposuretooutcomes.Wedidsobyadoptingamoderation-of-processdesign(Spencer,

Zanna,&Fong,2005).Specifically,drawingonresearchshowingthat“affirmationsofjustice”can

reducepeople’stendenciestoengageinstrategiestomaintainacommitmenttodeservingness(e.g.,

immanentjusticereasoning;seeCallanetal.,2014),participantsinStudy7learnedaboutbad

peoplewhodidordidnotreceive“justdeserts”fortheirtransgressionsbeforeweaskedthemto

ratehowmuchtheywantedtoreceivegoodandbadoutcomeinformation.Ifaconcernfor

deservingnessunderliestheseselectiveexposureeffects,thenlearningthatabadpersonalready

gotwhattheydeserved—thatis,receivedtheir“justdeserts”—shouldreducethenecessityfor

participantstoselectivelyexposethemselvestobadoutcomeinformation.

Method

Participants

ParticipantsfromtheU.S.A.wererecruitedonlinethroughAmazon’sMechanicalTurk(N=

77;45%female;Mage=35.98;SDage=12.60).Datafrom6additionalparticipantswerenotincludedin

analysesbecausetheyeitherincorrectlyansweredasimple,multiple-choicestorycomprehension

question(n=5;“InthestoryyoureadaboutSally,whatdidshedoatthecornerstore?”)ordidnot

answeralloftheitems(n=1).

Materialsandprocedure

Weinformedparticipantsthattheywouldreadshortandincompletenarratives,before

beingaskedtoratehowmuchtheywantedtoreaddifferentpiecesofadditionalinformationabout

thestories.Weinformedparticipantsthisadditionalinformationwouldbeshowntotheminfullat

theendofthesurvey.

SelectiveExposure29

Weusedtwostories,eachdescribingabadperson(Sallystolefromacharitycollectionand

Stevewasmeantoanelderlyperson).Participantsreadandrespondedtoeachofthestories,but

oneofthestoriesendedwiththetarget—SallyorSteve—receivingjustdesertsforhis/her

transgression.FortheSallystory,halftheparticipantslearnedthat“onherwayoutofthestore,

Sallywasapproachedbyamanwhomuggedandassaultedher,smashingherfacetothegroundand

stealingherpursecontaininghercellphoneandothervaluablepossessions.”Forthestoryabout

Steve,halftheparticipantsreadthathewas“crossingthestreetafterleavingthesubwaywhenhe

wasstruckbyataxirunningaredlight.Stevesurvivedtheincidentbutlosttheuseofhislegs.”Thus,

participantsreadtwostories,eachdescribinga“bad”person,oneofwhichconcludedwitha

deservingnessaffirmation.WhetherparticipantslearnedthatSallyorStevereceivedjustdesertswas

determinedrandomlybetweenparticipantsandthetwostorieswerepresentedtoparticipantsina

randomorder.

LikeStudy1,participantswerethenpresentedwithtwopiecesofadditionalinformation

(eachsummarizedinasentence)thatdescribeaneventoccurringtoSteve/Sallyaftertheincident(s)

describedinthestories.Participantswereaskedtoratethedegreetowhichtheywantedtoread

abouttheseoutcomeslaterinthesurvey.Onesentencerepresentedagoodoutcome(Sallywona

luxurycruise;Stevewonalottery)andtheotherdescribedabadoutcome(Sallycontractedaserious

illness;Steve’sapartmentwasdestroyedbyflooding).Participantsratedtheextenttowhichthey

wantedtoreadeachoftheoutcomes(1=Idonotwanttoreadthesedetailslateroninthesurveyto

6=Iwanttoreadthesedetailslateroninthesurvey).

ResultsandDiscussion

Participants’ratingsofhowmuchtheywantedtoreadabouttheoutcomeswereaveraged

acrossthetwoscenariosandsubjectedtoa2(OutcometoRead:goodvs.bad)X2(Justice

Affirmation:affirmationvs.noaffirmation)fullywithin-subjectsANOVA.Analysesrevealeda

significantmaineffectofOutcome,F(1,76)=9.98,p=.002,whichwassignificantlymoderatedby

SelectiveExposure30

JusticeAffirmation,F(1,76)=7.90,p=.006,ω2=.08(seeFigure7).3Follow-upanalysesshowedthat

participantswantedtoreadaboutbadoutcomestoagreaterextentthangoodoutcomeswhen

therewasnojusticeaffirmation,t(76)=3.99,p<.001(rbetweenrepeatedmeasures=-0.37).There

was,however,nosignificantdifferencebetweenratingsofwantingthegoodandbadoutcomes

whenjusticewasaffirmed,t(76)=1.48,p=.144(rbetweenrepeatedmeasures=-0.21).Linear

mixedeffectsmodellingproducedthesameresults,butsufferedsomeproblemswithestimation.

Thus,selectiveexposuretodeservedoutcomesisreducedwhendeservingnessisotherwise

affirmed,lendingfurtherweighttotheideathatthesortsofeffectswehaveexaminedaredueto

people’sconcernsaboutseeingthatpeoplegetwhattheydeserve.

Figure7.Theeffectofjusticeaffirmationontheextenttowhichparticipantswantedtoreadgoodandbadoutcomeinformationforbadpeople.Errorbarsshow95%CIsofthemeans.

3Addingabetween-subjectsfactorthatindicatedwhichscenarioincludedajusticeaffirmationdidnotrevealanysignificantmainorinteractioneffectsofscenario(allps>.29).

SelectiveExposure31

GeneralDiscussion

Employingarangeofstimuliandtasks,thepresentstudiesprovideconsistentsupportfor

thegeneralhypothesisthatpeoplechoosetobeexposedtoinformationaboutdeservedratherthan

undeservedoutcomes.Thiseffectwasmediatedbytheperceiveddeservingnessofoutcomes

(Studies2and3),andwasnotevidentwhenparticipantsknewthatwrongdoershadalready

receivedjustdesertsfortheirtransgressions(Study7).Participantswerenotsimplyselectively

avoidinginformationaboutundeservedoutcomesbutactivelysoughtinformationaboutjust

outcomes(Studies3and4).Participantsinvestedeffortinthispatternofselectiveexposure,seeking

outinformationaboutdeserved(vs.undeserved)outcomesevenwhenitwasmoredifficulttodoso

(Studies5and6).Further,responsetimedatashowedthatparticipantstooklongertosearchfor

informationaboutdeservedoutcomes,andsuggestedthatevenparticipantswhochosetoview

informationaboutundeservedoutcomeshadfirstsearchedforinformationaboutdeserved

outcomes(Study6).

Itwouldseemfunctionallyimportantforpeopletotakeabalanced,utilitarianapproachto

seekinginformationaboutthegoodandbadthingsthatcanhappen.Thiswouldenablepeopleboth

tohopeforandworktowardsthebestwhileavoidingandpreparingfortheworst.Peopleclearly

deviatefromthisaccuracymotivationinthepresentstudies.Ofnote,inStudy4,participantssought

outpositiveratherthannegativeoutcomesinthelifeofapersonwhosemoralstatustheydidnot

know.Thispositivitybiasisconsistentwiththeoreticalmodelsofselectiveexposureasamood

regulatingprocessthathelpspeopleconstruct“positiveillusions”abouttheworld(Oliver,2003;

Zillman,1988;moregenerally,Taylor&Brown,1988).Ofmoreinterest,however,isthatthis

tendencywasaccentuatedwhenparticipantsthoughttheoutcomeshappenedtogoodpeople,but

reversedwhentheythoughttheyhappenedtobadpeople.Thus,thepreferencetolearnabout

hedonicallypositiveoutcomeswastrumpedbythedesiretolearnaboutdeservedoutcomes:

SelectiveExposure32

horrificcarcrashesandterminalillnessesweremoreattractivethandazzlingstrokesofluckand

socialtriumphs,solongastheyweremoredeserved.

Thepresentfindingsbuildupontheeye-trackingfindingsofCallanetal.(2013).The

anticipatorybiasofparticipants’eye-movementstheyfoundcanbeinterpretedasapreferenceto

seedeservedratherthanundeservedoutcomes,broadlyconsistentwiththepresentresults.

However,theycanalsobeinterpretedasapreconsciousexpectationthattheoutcomewillbejust,

consistentwithhowsuchpredictiveeye-movementsareinterpretedintheliteratureonreadingand

storycomprehension.Moreover,Callanetal.werenotabletodifferentiatebetweenpeople’s

expectationsofdeservedoutcomesandthedeliberatechoicespeoplemaketoreceiveinformation

consistentwiththeviewthatpeoplegetwhattheydeserve.Thepresentresultsthereforeprovide

thefirstunambiguousevidencethatparticipants,evenattheexpenseoftheirtime,activelyand

deliberatelychoosetoencounterinformationconsistentwithwhatisdeserved.

Thepresentfindingsuncoveratheoreticallyimportant,hithertounexploredmeansbywhich

peoplepreservethebeliefthattheworldisajustplacefromdisconfirmatoryevidence.Otherwell-

knownstrategiessuchasimmanentjusticereasoning,ultimatejusticereasoning,andvictim

derogationhavebeenshowntoplayanimportantroleinpreservingthepsychologicalbenefitsof

just-worldbeliefs(Ellard,Harvey,&Callan,2016),includingtheabilitytodelaygratification(Callan,

Harvey,&Sutton,2014;Callan,Harvey,Dawtry,&Sutton,2013).However,thesestrategiesinvolve

processingofinformationafterithasbeenencountered,andrunintoimportantpsychological

constraints.Forexample,derogatinginnocentvictimsofmisfortunemayruncountertopeople’s

moralstandards(Hafer&Bègue,2005),andimmanentjusticereasoningrunscountertoreality

constraintsbecauseitisincompatiblewithpeople’sknowledgeofhowthephysicalworldoperates

(Callanetal.,2014).

Incontrast,selectiveexposureallowspeopletoexposethemselvestobiasedsamplesof

outcomeinformationinamannerthatisfreefromtheseconstraints.Inprinciple,evenifpeople

reasoninanunbiasedmanner,throughselectiveexposuretheymaydrawthebiasedconclusionthat

SelectiveExposure33

theworldisarelativelyfairplaceinwhichpeoplegetwhattheydeserve.Suchselectiveexposureto

deservedoutcomesmighthaveimportantimplicationsforhowpeoplesustainandcultivatebeliefs

aboutdeservingnessandcommunicatethosebeliefstoothers.Forexample,wecanexpectthatif

peopleselectivelyexposethemselvestodeservedmorethanundeservedoutcomes,thentheymight

bemorelikelyremembereventsasmorejustandfairthanabalanced,unbiasedassessmentofthe

objectivecircumstancesmighthaveindicated(cf.Callanetal.,2009).Thatis,ifpeopleselectively

exposethemselvestoinformationthatelaboratesondeservedoutcomes,thenthelogicaloutcome

isthattheyhaveanopportunitytorehearsethatinformation.Suchmemorybiasesmighthave

furtherconsequencesforinformationretransmission(Cappella,Kim,&Albarracín,2015)—people

mightcommunicatetoothersthateventswerejustandfairpreciselybecausetheyselectivelychose

andremembertheminthatway.Exploringthisinterplaybetweenselectiveexposure,memory

biases,andsocialcommunicationremainsanimportantavenueforfutureresearch.

OnelimitationofthepresentresearchisourrelianceonsamplesfromAmazon’sMechanical

Turk.ResearchhashighlightedsomeofthestrengthsofMTurkcomparedtotraditionalsampling,

includingMTurkparticipantsbeingmoredemographicallydiversethanstandardundergraduate

samples,therapidandinexpensivenatureofrecruitment,andthesometimessuperiorqualityof

data(e.g.,Buhrmester,Kwang,&Gosling,2011;Clifford,Jewell,&Waggoner,2015;Hauser&

Schwarz,2016).AlthoughMTurksamplesaremorediversethantraditionalsamples,Paolacciand

Chandler(2014)warnedthattheyarenotrepresentativeofthegeneralpopulation.Theyalso

highlightedthatparticipantswhofrequentlyuseMTurkmaybecomefamiliarwithcommonlyused

procedures,materials,andmeasuresandthereforetheirresponsesmaynotbe“organic,”whereas

otherworkersmightnotbefullyattentiveorrespondhonestly.Wetriedtolimittheseconcernsby

usingnovelmaterials,screeningparticipantsformultipleresponses,andremovingparticipantswho

wereclearlynotattendingtoandreadingthematerials.Nonetheless,futureresearchshould

considerreplicatingandextendingourfindingsusingmorerepresentativeandnon-Western

samples.

SelectiveExposure34

Conclusions

Lerner’s(1980)justworldtheorywasimpelledbyaparadox:despitealltheevidencetothe

contrary,peopleappeartobelievethattheworldisajustplace.Itproposedtworesolutionsofthis

paradox.First,itportrayedthe‘justworld’asafictionconstructedanddefendedinthemindofthe

perceiver.Thus,whentheyencounterinjustice,peoplederogatevictimsandfindothercognitive

waysofpreservingtheirfaithinjustice.Second,itframedthe‘justworld’asanidyllicblueprintthat

motivatespeopletobehaviorallyredressinnocentsuffering,andsoreducethedisparitybetween

theiridyllandreality.Thepresentstudiesprovideacomplementaryperspective.Theysuggestthat

the‘justworld’doesnotonlyexistinthemindoftheperceiver,butcanalsobeunderstoodasa

handpickedregionoftheobjectiveworld.Within(vs.beyond)itsdeliberatelylimitedhorizons,

injusticeisrareandjusticecommonplace,suchthatevenanunbiasedobservermightfindjust-world

beliefstobeempiricallywarranted.Indeed,researchinformedbythecognitive-ecologicalapproach

tosocialcognition(Fiedler,2000)hasshownthatsamplingfrombiasedinformationmaylead

unbiasedobserverstodrawbiasedconclusionsaboutthefairnessoftheirsociety(Dawtry,Sutton,&

Sibley,2015).Thepresentstudiesshowthatpeoplearepreparedtoinvesttimeandeffortto

constraintheirexperienceinthisway.Bydoingso,theymaysavethemselvesconsiderableeffort

anddiscomfortinthelongerrun—Inaworldthatseldomcontainsinjustice,cognitiveandbehavioral

strategiestominimizeinjusticeareseldomrequired.

SelectiveExposure35

References

Bal,M.,&vandenBos,K.(2015).Puttingthe“I”and“Us”injustice:Derogatoryandbenevolent

reactionstowardinnocentvictimsinself-focusedandother-focusedindividuals.SocialJustice

Research,28,274-292.

Barr,D.J.,Levy,R.,Scheepers,C.,&Tily,H.J.(2013).Randomeffectsstructureforconfirmatory

hypothesistesting:Keepitmaximal.JournalofMemoryandLanguage,68,255–278.

Bates,D.,Maechler,M.,Bolker,B.,&Walker,S.(2015).FittingLinearMixed-EffectsModelsUsing

lme4.JournalofStatisticalSoftware,67(1),1-48.

Brannon,L.A.,Tagler,M.J.,&Eagly,A.H.(2007).Themoderatingroleofattitudestrengthin

selectiveexposuretoinformation.JournalofExperimentalSocialPsychology,43,611–617.

Buhrmester,M.,Kwang,T.,&Gosling,S.D.(2011).Amazon'sMechanicalTurkanewsourceof

inexpensive,yethigh-quality,data?PerspectivesonPsychologicalScience,6,3-5.

Callan,M.J.,&Ellard,J.H.(2010).Beyondderogationandblameofvictims:Justworlddynamicsin

everydaylife.InD.R.Bobocel,A.C.Kay,M.P.Zanna,&J.M.Olson(Eds.).Thepsychologyof

justiceandlegitimacy:TheOntarioSymposium(Vol.11,pp.53-77).NewYork:Psychology

Press.

Callan,M.J.,Ferguson,H.J.,&Bindemann,M.(2013).Eyemovementstoaudiovisualscenesreveal

expectationsofajustworld.JournalofExperimentalPsychology:General,142,34-40.

Callan,M.J.,Harvey,A.J.,&Sutton,R.M.(2014).Rejectingvictimsofmisfortunereducesdelay

discounting.JournalofExperimentalSocialPsychology,51,41-44.

Callan,M.J.,Harvey,A.J.,Dawtry,R.J.,&Sutton,R.M.(2013).Throughthelookingglass:Long-term

goalfocusincreasesimmanentjusticereasoning.BritishJournalofSocialPsychology,52,377-

385.

Callan,M.J.,Kay,A.C.,Davidenko,N.,&Ellard,J.H.(2009).Theeffectsofjusticemotivationon

memoryforself-andother-relevantevents.JournalofExperimentalSocialPsychology,45,

614-623.

SelectiveExposure36

Callan,M.J.,Sutton,R.M.,Harvey,A.J.,&Dawtry,R.J.(2014).Immanentjusticereasoning:Theory,

research,andcurrentdirections.InJ.M.Olson&M.P.Zanna(Eds.),AdvancesinExperimental

SocialPsychology(Vol.49,pp.105-161).London:AcademicPress.

Cappella,J.N.,Kim,H.S.,&Albarracín,D.(2015).Selectionandtransmissionprocessesfor

informationintheemergingmediaenvironment:Psychologicalmotivesandmessage

characteristics.MediaPsychology,18,396-424.

Canon,L.K.,&Matthews,K.E.(1972).Concernoverpersonalhealthandsmoking-relevantbeliefs

andbehavior.ProceedingsoftheAnnualConventionoftheAmericanPsychological

Association,7,271–272.

Clifford,S.,Jewell,R.M.,&Waggoner,P.D.(2015).AresamplesdrawnfromMechanicalTurkvalid

forresearchonpoliticalideology?ResearchandPolitics,2,1-9.

Dawtry,R.J.,Sutton,R.M.,&Sibley,C.G.(2015).Whywealthierpeoplethinkpeoplearewealthier,

andwhyitmatters:Fromsocialsamplingtoattitudestoredistribution.Psychological

Science,26(9),1389-1400.

Ellard,J.H.,Harvey,A.J.,&Callan,M.J.(2016).Thejusticemotive:History,theory,andresearch.In

C.Sabbagh&M.Schmitt(Eds.),HandbookofSocialJusticeTheoryandResearch(pp.127-143).

NewYork:Springer.

Festinger,L.(1957).Atheoryofcognitivedissonance.Stanford,CCA:StanfordUniversityPress.

Fiedler,K.(2000).Bewareofsamples!Acognitive-ecologicalsamplingapproachtojudgment

biases.PsychologicalReview,107,659–676.

Fischer,P.,Jonas,E.,Frey,D.,&Schulz-Hardt,S.(2005).Selectiveexposuretoinformation:The

impactofinformationlimits.EuropeanJournalofSocialPsychology,35,469–492.

Frey,D.(1986).Recentresearchonselectiveexposuretoinformation.InL.Berkowitz(Ed.),Advances

inexperimentalsocialpsychology(Vol.19,pp.41–80).NewYork:AcademicPress.

Frey,D.,&Wicklund,R.(1978).Aclarificationofselectiveexposure:Theimpactofchoice.Journalof

ExperimentalSocialPsychology,14,132–139.

SelectiveExposure37

Hafer,C.L.(2011).Thepsychologyofdeservingnessandacceptanceofhumanrights.InE.Kals&J.

Maes(Eds.),Justiceandconflicts:Theoreticalandempiricalcontributions(pp.407–427).

Hafer,C.L.,&Bègue,L.(2005).Experimentalresearchonjustworldtheory:Problems,

developments,andfuturechallenges.PsychologicalBulletin,131,128–166.

Hafer,C.L.,&Gosse,L.(2011).Predictingalternativestrategiesforpreservingabeliefinajust

world:Thecaseofrepressivecopingstyle.EuropeanJournalofSocialPsychology,41,730–

739.

Hart,W.,Albarracín,D.,Eagly,A.H.,Brechan,I.,Lindberg,M.,Lee,K.,&Merrill,L.(2009).Feeling

validatedversusbeingcorrect:Ameta-analysisofselectiveexposuretoinformation.

PsychologicalBulletin,135,555–588.

Harvey,A.J.,&Callan,M.J.(2014).Getting"justdeserts"orseeingthe"silverlining":Therelation

betweenjudgmentsofimmanentandultimatejustice.PLoSONE,9,e101803.

Harvey,A.J.,Callan,M.J.,&Matthews,W.J.(2014).Howmuchdoeseffortfulthinkingunderlie

observers’reactionstovictimization?SocialJusticeResearch,27,175-208.

Hauser,D.J.,&Schwarz,N.(2016).AttentiveTurkers:MTurkparticipantsperformbetterononline

attentionchecksthandosubjectpoolparticipants.BehaviorResearchMethods,48(1),400–

407.

Kuznetsova,K.,Brockhoff,P.B.,&Christensen,R.H.B.(2016).lmerTest:TestsinLinearMixed

EffectsModels.Rpackageversion2.0-30.http://CRAN.R-project.org/package=lmerTest

Johnston,L.(1996).Resistingchange:Information-seekingandstereotypechange.EuropeanJournal

ofSocialPsychology,26,799-825.

Jonas,E.,&Frey,D.(2003).SearchingforinformationaboutfinancialdecisionsinEuroversusDM.

EuropeanPsychologist,8,92-96.

Jonas,E.,Schulz-Hardt,S.,Frey,D.,&Thelen,N.(2001).Confirmationbiasinsequentialinformation

searchafterpreliminarydecisions:Anexpansionofdissonancetheoreticalresearchon

SelectiveExposure38

“selectiveexposuretoinformation.”JournalofPersonalityandSocialPsychology,80,557–

571.

Kay,A.C.,Jost,J.T.,Mandisodza,A.N.,Sherman,S.J.,Petrocelli,J.V.,&Johnson,A.L.(2007).

Panglossianideologyintheserviceofsystemjustification:Howcomplementarystereotypes

helpustorationalizeinequality.InM.P.Zanna(Ed.),Advancesinexperimentalsocial

psychology(Vol.38,pp.305–358).SanDiego,CA:AcademicPress.

Knobloch,S.,&Zillmann,D.(2002).Moodmanagementviathedigitaljukebox.Journalof

Communication,52,351-366.

Lerner,M.J.(1977).Thejusticemotive.Somehypothesesastoitsoriginsandforms.Journalof

Personality,45,1–32.

Lerner,M.J.(1980).Thebeliefinajustworld:Afundamentaldelusion.NewYork:PlenumPress.

Lerner,M.J.,&Simmons,C.H.(1966).Observer’sreactiontothe‘‘innocentvictim”:Compassionor

rejection?JournalofPersonalityandSocialPsychology,4(2),203–210.

Lowin,A.(1969).Furtherevidenceforanapproach–avoidanceinterpretationofselectiveexposure.

JournalofExperimentalSocialPsychology,5,265–271.

Maccoby,E.E.,Maccoby,N.,Romney,A.K.,&Adams,J.S.(1961).Socialreinforcementinattitude

change.JournalofAbnormalandSocialPsychology,63,109–115.

McFarland,S.G.,&Warren,J.C.(1992).Religiousorientationsandselectiveexposureamong

fundamentalistChristians.JournalfortheScientificStudyofReligion,31,163-244.

Marsh,D.P.,&Greenberg,M.S.(2006).Theinfluenceofeyewitnesssimilaritytoacrimevictimand

culpabilityonwitness’recall.AppliedPsychologyinCriminalJustice,2,43–56.

Oliver,M.B.(2003).Moodmanagementandselectiveexposure.InJ.Bryant,D.Roskos-Ewoldsen,&

J.Cantor(eds.),Communicationandemotion:EssaysinhonorofDolfZillmann(pp.85–106).

Mahwah,NJ:LawrenceErlbaumAssociates.

Paolacci,G.,&Chandler,J.(2014).InsidetheTurk:UnderstandingMechanicalTurkasaparticipant

pool.CurrentDirectionsinPsychologicalScience,23,184–188.

SelectiveExposure39

Preacher,K.J.,&Hayes,A.F.(2008).Asymptoticandresamplingstrategiesforassessingand

comparingindirecteffectsinmultiplemediatormodels.BehaviorResearchMethods,40,879–

891.

Rhine,R.J.(1967).The1964presidentialelectionandcurvesofinformationseekingandavoiding.

JournalofPersonalityandSocialPsychology,5,416–423.

Sabbagh,C.,&Schmitt,M.(Eds.).(2016).Handbookofsocialjusticetheoryandresearch.New

York:Springer.

Smith,S.M.,Fabrigar,L.R.,&Norris,M.E.(2008).Reflectingonsixdecadesofselectiveexposure

research:Progress,challenges,andopportunities.SocialandPersonalityPsychologyCompass,

2,464–493.

Spencer,S.J.,Zanna,M.P.,&Fong,G.T.(2005).Establishingacausalchain:Whyexperimentsare

oftenmoreeffectiveinexaminingpsychologicalprocessthanmeditationalanalyses.Journalof

PersonalityandSocialPsychology,89,845–851.

Smith,S.M.,Fabrigar,L.R.,Powell,D.M.,&Estrada,M.(2007).Theroleofinformationprocessing

capacityandgoalsinattitude-congruentselectiveexposureeffects.PersonalityandSocial

PsychologyBulletin,33,948–960.

Taylor,S.E.,&Brown,J.(1988).Illusionandwell-being:Asocialpsychologicalperspectiveonmental

health.PsychologicalBulletin,103,193-210.

Treisman,A.,&Gelade,G.(1980).Afeatureintegrationtheoryofattention.CognitivePsychology,

12,97-136.

Wolfe,J.M.(1994).Guidedsearch2.0:Arevisedmodelofvisualsearch.PsychonomicBulletin&

Review,1,202-238.

Zillmann,D.(1988a).Moodmanagementthroughcommunicationchoices.AmericanBehavioral

Scientist,31,327–341.

SelectiveExposure40

SupplementaryMaterials

Study1

ThefourshortstoriesandtheadditionaloutcomeinformationusedinStudy1.Goodoutcomeinformationisnumbered1andbadoutcomeinformationisnumbered2.

Goodperson BadpersonLastSummer,LukewasonholidayintheCaribbean.Hedecidedtohavelunchinarestaurantnearhishotelrecommendedbylocals.Althoughtherestaurantwasverybusy,Lukethoughttheservicewasexcellentandgreatlyenjoyedthefood.Despitehiswaiterbeingtiedupwithanothercustomer,Lukedecidedtowaituntilhewasavailabletoofferhimgreatpraiseandalargetip.Lateronthatevening,Lukedecidedgoforaswimintheocean.

StevewasridingontheLondonUndergroundtoSt.James’sparktomeethisgirlfriendforapleasantoutdoorpicnicinthepark.Earlierthatday,theweatherforecasthadwarnedofa50/50chanceofthunderstorms.Whileatastop,afrailoldladyenteredthesamecarriageoccupiedbySteve.Insteadofgettingupandofferinghisseat,Stevescowledattheoldladyandrefusedtogiveuphisseat.

Additionaloutcomeinformation:1.Whileswimming,Lukefounda$100billburiedinthesandatthebeach.2.Whileswimming,thecurrentcarriedLukeawayandhedrowned.

Additionaloutcomeinformation:1.WhenSteveemergedfromtheUnderground,helookedupandobservedthattheskywasclearandthesunwasshining.2.WhenSteveemergedfromtheUnderground,helookedupandobservedagrey,cloudyskyfromwhichrainwaspouringdown.

Aweekago,JennywaswalkingalongtheRiverWyewhenshespottedapuppydrowningintheriver.Riskingherownlife,Jennydivedintotheriverandsavedthepuppyfromdrowning.

Lastweek,Sallywasinagreengrocersbuyingfruitandvegetables.Afterpaying,shemadesurethatnoonewaslooking,andstoleallthechangefromacharitycollectionthatwasondisplayatthecounter.

Additionaloutcomeinformation:1.Oneweeklater,Jennywassittinginherlivingroomwhenshereceivednewsthatshehadwonanewcarinasweepstakeshehadentered.2.Oneweeklater,Jennywassittinginherlivingroomwhenshereceivednewsthatherhusbandwasinaterriblecaraccident.

Additionaloutcomeinformation:1.Onherwayoutofthegreengrocers,Sallywasapproachedbyamanwhoofferedherfreesamplesofthefoodonofferinhisdelicatessenacrosstheroad.2.Onherwayoutofthegreengrocers,Sallywasapproachedbyamanwhomuggedher,stealingherbagcontainingherpurseandphoneamongotherpossessions.

SelectiveExposure41

TableS1.RatingsofdeservingnessofoutcomesbyaseparatesampleofMTurkparticipants(N=49,n=1removedforhavingaduplicateIPaddress;36.70%females;Mage=34.55;SDage=9.62)foreachoftheoutcomesbyscenarioweusedinStudy1(presentedinarandomorder).Participantsmaderatingsforeachoftheoutcomesandscenarios(cf.Study1inthemaintext),andtheirratingsofdeservingnessweremadeona7pointscale(e.g.,“TowhatextentdoyoubelieveJennydeservestowinanewcarinasweepstakesheenters?”;1=notatalldeservingto7=verydeserving).

Scenario GoodOutcomeM(SD)

BadOutcomeM(SD) t(p) dz

Lukeleftatip 4.39(1.38) 1.31(0.98) 11.61(<.001) 1.66

Jennysavedpuppy 4.69(1.49) 1.41(1.12) 11.68(<.001) 1.67

Stevemeantolady 1.98(0.99) 4.57(1.51) -8.98(<.001) -1.28

Sallystolechange 1.60(1.07) 3.13(1.75) -4.76(<.001) -0.68

SelectiveExposure42

Study2

TheshortstoriesandtheadditionaloutcomeinformationusedinStudy2.Goodoutcomeinformationisnumbered1andbadoutcomeinformationisnumbered2.

Goodperson BadpersonAweekago,GeoffwaswalkingalongtheRiverWyewhenhespottedapuppydrowningintheriver.Riskinghisownlife,Geoffdivedintotheriverandsavedthepuppyfromdrowning.

Aweekago,GeoffwaswalkingalongtheRiverWyewhenhespottedapuppyalongthebankoftheriver.Withnoregardforitslife,Geoffpickedupthepuppyandthrewitintheriver.

Additionaloutcomeinformation:1.Geoffwassittinginhislivingroomwhenhereceivednewsthathehadwonanewcarinasweepstakehehadentered.2.Geoffwasinaterriblecaraccidentthatlefthiminhospitalinaseriouscondition.

SelectiveExposure43

Study3

TheshortstoriesandtheadditionaloutcomeinformationusedinStudy3.

Additionaloutcomeinformation:

Goodperson BadpersonAweekago,ChriswaswalkingalongtheRiverWyewhenhespottedapuppydrowningintheriver.Riskinghisownlife,Chrisdivedintotheriverandsavedthepuppyfromdrowning.

Aweekago,ChriswaswalkingalongtheRiverWyewhenhespottedapuppyalongthebankoftheriver.Withnoregardforitslife,Chrispickedupthepuppyandforcefullydrowneditintheriver

Goodoutcomes Badoutcomes

• Chris’sstocksandsharesskyrocketed

• Chriswon$100,000playingascratchcardlotteryticket

• Chriswasgivenamajorpromotionatwork

• Chriswonaluxurycruisetrip

• Chriswasinjuredinacaraccident

• Chriswasfiredfromhisjob

• Chriscontractedamajorillness

• Chris’sground-floorapartment

wasdestroyedbyflooding

SelectiveExposure44

Study4

TheshortstoriesandadditionaloutcomeinformationusedinStudy4.

Goodperson BadpersonLastweek,Sallywasinacornerstorebuyingbreadandmilk.Afterpaying,shemadesurethatnoonewaslookingandputallofhersparechangeintoacharitycollectionthatwasondisplayatthecounter.

Lastweek,Sallywasinacornerstorebuyingbreadandmilk.Afterpaying,shemadesurethatnoonewaslookingandstoleallthechangefromacharitycollectionthatwasondisplayatthecounter.

Additionaloutcomeinformation:

Goodoutcomes Badoutcomes Neutraloutcomes

• Sally’sstocksandsharesskyrocketed

• Sallywon$1,000playingascratchcardlotteryticket

• Sallywasgivenamajor

promotionatwork

• Sallywasinjuredinacar

accident

• Sally’sground-floorapartmentwasflooded

• Sallycamedownwitha

seriousillness

• Sallywenttoa

concert

• Sallystartedwritinganewblog

• Sallytidiedup

heroffice

SelectiveExposure45

Study5

TheshortstoriesandadditionaloutcomeinformationusedinStudy5.Goodoutcomeinformationisnumbered1andbadoutcomeinformationisnumbered2.

Goodperson BadpersonLastSummer,LukewasonholidayintheCaribbeanhavinglunchinarestaurantnearhishotel.Althoughtherestaurantwasverybusy,Lukethoughttheservicewasexcellentandgreatlyenjoyedthefood.Despitehiswaiterbeingtiedupwithanothercustomer,Lukedecidedtowaituntilhewasavailabletoofferhimgreatpraiseandalargetip.

StevewasridingontheLondonUndergroundtoSt.James’sparktomeethisgirlfriendforapleasantoutdoorpicnicinthepark.Whileatastop,afrailoldladyenteredthesamecarriageoccupiedbySteve.Insteadofgettingupandofferinghisseat,Stevescowledattheoldladyandrefusedtogiveuphisseat.

Additionaloutcomeinformation:1.Whileswimming,Lukefound$100billburiedinthesandatthebeach.2.Whileswimming,thecurrentcarriedLukeawayandhedrowned.

Additionaloutcomeinformation:1.WhenSteveemergedfromtheUnderground,helookedupandobservedthattheskywasclearandthesunwasshining.2.WhenSteveemergedfromtheUnderground,helookedupandobservedagray,cloudyskyfromwhichrainwaspouringdown.

Aweekago,JennywaswalkingalongtheRiverWyewhenshespottedapuppydrowningintheriver.Riskingherownlife,Jennydivedintotheriverandsavedthepuppyfromdrowning.

Lastweek,Sallywasinagreengrocersbuyingfruitandvegetables.Afterpaying,shemadesurethatnoonewaslooking,andstoleallthechangefromacharitycollectionthatwasondisplayatthecounter.

Additionaloutcomeinformation:1.Jennywassittinginherlivingroomwhenshereceivednewsthatshehadwonanewcarinasweepstakeshehadentered.2.Jennywasinaterriblecaraccidentthatleftherinhospitalinseriouscondition.

Additionaloutcomeinformation:1.Onherwayoutofthegreengrocers,Sallywasapproachedbyamanwhoofferedherfreesamplesofthefoodonofferinhisdelicatessenacrosstheroad.2.Onherwayoutofthegreengrocers,Sallywasapproachedbyamanwhomuggedher,stealingherbagcontainingherpurseandphoneamongotherpossessions.

SelectiveExposure46

Study6

Theshortstory,instructions,andtheadditionaloutcomeinformationusedinStudy6.

BadpersonAweekago,GeoffwaswalkingalongtheRiverWyewhenhespottedapuppyalongthebankoftheriver.Withnoregardforitslife,Geoffpickedupthepuppyandthrewitintheriver.Instructions:Onthenextpageyouwillseeabusyarrayofdifferentshapes-includingthetwobelow.Thetwoshapesbelowrepresentdifferentpiecesofinformationyoucanreceivelaterinthissurveyaboutthenarrativeabove(pleasereaditagain,justtobesure).PleaseREMEMBERandthen,onthefollowingpage,searchforandCLICKontheshapewhichrepresentstheinformationyouwanttoreceiveadditionaldetailsaboutlaterinthesurvey.Whichpieceofadditionalinformationwouldyouliketoreadmoreaboutconcerningtheabovestory?Notethatbothshapescanactuallybefoundinthevisualarray,soit'simportantthatyousearchforandclickontheshapeyouwant.Pleaseclickthenextbuttontosearchforandselecttheshapeassociatedwiththeadditionalinformationyouwanttoreviewaboutthisstory.Additionaloutcomeinformation(associatedeitherwithaneasy-orhard-to-findshape):1.Geoffwassittinginhislivingroomwhenhereceivednewsthathehadwonanewcarinasweepstakehehadentered.2.Geoffwasinaterriblecaraccidentthatlefthiminhospitalinaseriouscondition.

SelectiveExposure47

Study7

TheshortstoriesandadditionaloutcomeinformationusedinStudy7.Goodoutcomeinformationisnumbered1andbadoutcomeinformationisnumbered2.

Deservingnessnotaffirmed DeservingnessaffirmedLastweek,Sallywasinacornerstorebuyingbreadandmilk.Afterpaying,shemadesurethatnoonewaslooking,andstoleallthechangefromacharitycollectionthatwasondisplayatthecounter.

Lastweek,Sallywasinacornerstorebuyingbreadandmilk.Afterpaying,shemadesurethatnoonewaslooking,andstoleallthechangefromacharitycollectionthatwasondisplayatthecounter.

Onherwayoutofthestore,Sallywasapproachedbyamanwhomuggedandassaultedher,smashingherfacetothegroundandstealingherpursecontaininghercellphoneandothervaluablepossessions.

Additionaloutcomeinformation:

1. Sallywonaluxurycruisetripinanonlinesweepstakessheentered2. Sallycamedownwithaseriousillness

Deservingnessnotaffirmed DeservingnessaffirmedStevewasridingonthesubway.Whileatastop,afrailoldladyenteredthesamecarriageoccupiedbySteve.Insteadofgettingupandofferinghisseat,Stevescowledattheoldlady,calledhernames,andrefusedtogiveuphisseat.

Stevewasridingonthesubway.Whileatastop,afrailoldladyenteredthesamecarriageoccupiedbySteve.Insteadofgettingupandofferinghisseat,Stevescowledattheoldlady,calledhernames,andrefusedtogiveuphisseat.

Stevewascrossingthestreetafterleavingthesubwaywhenhewasstruckbyataxirunningaredlight.Stevesurvivedtheincidentbutlosttheuseofhislegs.

Additionaloutcomeinformation:

1. Stevewon$10,000playingascratch-cardlotteryticket2. Steve’sgroundfloorapartmentwasdestroyedbyflooding