Beyond Gains and Losses

JournalofPersonalityandSocialPsychology?2012AmericanPsychologicalAssociation0022-3514/12/$12.00DOI:10.1037/a0027855

BeyondGainsandLosses:

TheEffectofNeedonRiskyChoiceinFramedDecisions

UniversityofGuelph

SandeepMishra

LakeheadUniversity,Orillia

LaurenceFiddick

Substantialevidencesuggestspeoplearerisk-aversewhenmakingdecisionsdescribedintermsofgainsandrisk-pronewhenmakingdecisionsdescribedintermsoflosses,aphenomenonknownastheframingeffect.Littleresearch,however,hasexaminedwhetherframingeffectsareaproductofnormativerisk-sensitivecognitiveprocesses.In5experiments,itisdemonstratedthatframingeffectsintheAsiandiseaseproblemcanbeexplainedbyrisk-sensitivitytheory,whichpredictsthatdecisionmakersadjustriskacceptanceonthebasisofminimalacceptablethresholds,orneed.Bothexplicitandself-determinedneedrequirementseliminatedframingeffectsandaffectedriskacceptanceconsistentwithrisk-sensitivitytheory.Furthermore,negativelanguagechoiceinlossframesconferredtheperceptionofhighneedandledtotheconstructionofhigherminimalacceptablethresholds.Theresultsofthisstudysuggestthatrisk-sensitivitytheoryprovidesanormativerationaleforframingeffectsbasedonsensitivitytominimalacceptablethresholds,orneeds.

Keywords:framingeffects,risk-sensitivitytheory,prospecttheory,need,decisionmaking,minimalrequirements,aspirationlevels,Asiandiseaseproblem

Expectedutilitytheoryhasbeenthedominantframeworkforunderstandingdecisionmakinginthebehavioralsciencessincethemid-20thcentury(Friedman&Savage,1948,1952;VonNeumann&Morgenstern,1944).Thistheoryispartofabroadcategoryofnormative(orfunctional)modelsofdecisionmaking,allofwhichareconcernedwithidentifyingthemostoptimal,or“rational,”decisioninagivensituation.Inthesemodels,rationalityisdefinedbydecisionmakingmotivatedtomaximizeacurrencyofinterest.Expectedutilitytheory,forexample,positsthatpeopleseektomaximizeutilityinalldecisions,whereutilityisbroadlydefinedasameasureofhappiness,gratification,orsatisfactionderivedfromabehavior(Friedman&Savage,1952).

Alargebodyofevidence,however,suggeststhatpeoplemake“irrational”decisionsundervariouspredictableconditions(e.g.,Allais,1953;Ellsberg,1961;Kahneman&Tversky,1979;re-viewedinAktipis&Kurzban,2004;Barrett&Fiddick,1999;Mishra,2012;Rode&Wang,2000),themostfamousofwhichistheframingeffect(Tversky&Kahneman,1981).TverskyandKahnemandemonstratedthatpeopleshiftriskpreferencebetweenoptionswithequalexpectedoutcomesinidenticalproblemsthataredifferentiallyframedintermsoflossesorgains.Specifically,peoplearelargelyrisk-pronewhenfacedwithadecisionframedasalossandrisk-aversewhenfacedwithadecisionframedasagain.ConsidertheclassicAsiandiseaseproblem(Tversky&Kahne-man,1981,p.453):

ImaginethattheU.S.ispreparingfortheoutbreakofanunusualAsiandisease,whichisexpectedtokill600people.Twoalternativeprogramstocombatthediseasehavebeenproposed.Assumethattheexactscientificestimatesoftheconsequencesoftheprogramsareasfollows.

SandeepMishra,DepartmentofPsychology,UniversityofGuelph,Ontario,Canada;LaurenceFiddick,DepartmentofPsychology,LakeheadUniversity,Orillia,Ontario,Canada.

ThewritingofthismanuscriptwaspartiallysupportedbytheOntarioProblemGamblingResearchCentreandtheNaturalSciencesandEngi-neeringResearchCouncilofCanada(postdoctoralfellowshipsawardedtoSandeepMishra).AportionofdataforExperiments1,3,and4werecollectedwhileLaurenceFiddickwasapostdoctoralresearcherattheCenterforAdaptiveBehaviorandCognition,MaxPlanckInstituteforHumanDevelopment,Berlin,whowethankfortheirsupport.Theremain-derofdataforExperiments1,3,4anddataforExperiments2and5werecollectedwhileSandeepMishrawasattheUniversityofLethbridge.LaurenceFiddickpresentedportionsofExperiments1,3,and4attheannualmeetingoftheSocietyofJudgmentandDecisionMaking,Month2000,NewOrleans,LA.ThankstoAnnabreeFairweather,T.AndrewHurly,MartinLalumie`re,ChristineMichell,andKellySuschinskyfortheircommentsonthemanuscript.

CorrespondenceconcerningthisarticleshouldbeaddressedtoLaurenceFiddick,LakeheadUniversity,Orillia,500UniversityAvenue,Orillia,OntarioL3V0B9,Canada.E-mail:fiddick@gmail.com

1

InthestandardversionoftheAsiandiseaseproblem,participantsareprovidedwithadecisionscenarioinvolvingachoicebetweentwooptions,presentedineitherapositiveoranegativeframe.Thepositiveframestates:

1.IfProgramAisadopted,200peoplewillbesaved.

2.IfProgramBisadopted,thereisaonethirdprobabilitythat600peoplewillbesavedandatwothirdsprobabilitythatnopeoplewillbesaved.

Inthispositiveframe,bothoptionsarephrasedintermsofthepossibilityofsavingpeople,andthusintermsofgains.Whenpresentedwiththispositivelyframedscenario,TverskyandKah-neman(1981)foundthat72%ofparticipantspreferredthecertainoption(ProgramA),and28%preferredtheriskyoption(Program

2

MISHRAANDFIDDICK

B).ProgramBistheriskierofthetwooptionsbecauseitinvolveshigheroutcomevariance.Thenegativeframestates:

1.IfProgramCisadopted,400peoplewilldie.

2.IfProgramDisadopted,thereisaonethirdprobabilitythatnobodywilldieandatwothirdsprobabilitythat600peoplewilldie.

Here,bothoptionsarenegativelyframedintermsofthenumberofpossibledeaths,andthusintermsoflosses.TverskyandKahne-man(1981)foundthatwhenpresentedwiththislossscenario,participants’riskpreferencesreversed:22%ofparticipantspre-ferredthecertainoption(ProgramC),and78%preferredtheriskyoption(ProgramD).Thisfindingofriskreversalinmathemati-callyidenticaldecisionsmadeinlossandgainframeshasreceivedsubstantialempiricalsupport(reviewedinLevin,Schneider,&Gaeth,1998;Levin,Gaeth,Schreiber,&Lauriola,2002;forameta-analysis,seeKu¨hberger,1998).

ProspectTheory

Whydopeopleexhibitvastlydifferentlevelsofriskacceptanceinmathematicallyidenticalproblems?Prospecttheorywasdevel-opedinparttoaccountforframingeffectsindecisionmaking(Kahneman&Tversky,1979;Tversky&Kahneman,1981).Pros-pecttheorysuggeststhatpeopleexhibitframingeffectsbecausetherateofincreaseinutilityderivedfromgainsissteeplydimin-ishing:Forexample,gaining$100issubjectivelymorevaluableifonestartswith$0thanifonestartswith$10,000.Inthedomainoflosses,however,utilitydiminishesmorerapidly,andsoriskybehaviormaybeengagedintopreventanyfurtherexperienceofloss.Gainsandlossesaredeterminedaroundareferencepoint,whereoutcomesbelowthereferencepointrepresentlossesandoutcomesabovethereferencepointrepresentgains.Prospectthe-oryisadescriptivetheoryofdecisionmaking,however,anddoesnotofferanormativerationaleofwhyreferencepointsareprivi-legedindecisionmaking.Althoughprospecttheorystatesthatreferencepointsarederivedfromanindividual’spresentstate—andcanchangeonthebasisofexpectationsorbiasesofdecisionmakers(Tversky&Kahneman,1981)—itdoesnotofferanexpla-nationofwhyreferencepointsareconstructed.

Risk-SensitivityTheory

Risk-sensitivitytheoryprovidesanormativeaccountofwhyminimalacceptablerequirements,orneed,shouldcontributetotheconstructionofreferencepoints,andthusinfluenceriskychoice(Stephens,1981;Stephens&Krebs,1986).Risk-sensitivitytheorywasdevelopedbybehavioralecologiststoexplainfoodacquisitiondecisionsinanimals(e.g.,Caraco,Martindale,&Whittam,1980,reviewedinKacelnik&Bateson,1996,1997).Insteadoffocusingonthegoalofmaximizingutility,risk-sensitivitytheorywasoriginallyconceivedtoexplainriskydecisionsinthecontextofforaging.Specifically,decisionmakingwascharacterizedasre-sponsesoforganismsdesignedtomaximizeforagingreturnsinstochasticenvironments,withtheultimategoalofmaximizingbiologicalreproductivesuccessorfitness(Mishra,2012;Mishra&Lalumie`re,2008;Stephens,1981;Stephens&Krebs,1986;We-ber,Shafir,&Blais,2004).

Accordingtorisk-sensitivitytheory,decisionmakersshould

preferhigh-riskoptionsinsituationsofhighneed,whenlow-riskoptionsareunlikelytomeetthoseneeds.Here,needreferstodisparitybetweenanindividual’spresentstateandagoalordesiredstate(Mishra&Lalumie`re,2010).Forexample,someonewithapressing$5,000debtshouldprefera10%chanceofwinning$5,000(oranevenlowerprobabilisticchanceofearnings;Thaler&Johnson,1990)overreceiving$500withcertainty.Inthisscenario,eventhoughbothoptionshavethesameexpectedvalue,onlytheriskyoptionoffersthedecisionmakerachancetomeethisorherneeds.Moregenerally,risk-sensitivitytheorypositsthatdecisionmakersdonotseektomaximizecertainoutcomes(e.g.,maximizingdollarsofincome),butratherseektominimizetheprobabilityofexperiencingoutcomesthatfailtomeethisorherneeds(Mishra&Lalumie`re,2010;Rode,Cosmides,Hell,&Tooby,1999;Stephens,1981).

Althoughmostevidenceinsupportofrisk-sensitivitytheoryhasbeenquantifiedintheanimalbehaviorliterature,asmallbutgrowingbodyofevidencesuggeststhatpeoplemakedecisionsconformingtothepredictionsofrisk-sensitivitytheory.Itisdif-ficulttodirectlymanipulateenergyneedsinhumanparticipants,butothercurrencies,includingmoney,offeranaloguesthroughwhichrisk-sensitivedecisionmakingcanbeinvestigatedinalaboratorysetting.Somestudiesofhumanshavedemonstratedshiftsfromrisk-aversiontorisk-pronenessunderconditionsofneedusingmonetaryrewards.Thesestudiesusedarbitraryrequire-mentsforcreatingsituationsoflowneedandhighneed,suchaspointtotalsthathadtobemetbeforeanymoneywasearned(Deditius-Island,Szalda-Petree,&Kucera,2007;Ermer,Cos-mides,&Tooby,2008;Pietras&Hackenberg,2001;Pietras,Locey,&Hackenberg,2003;Rodeetal.,1999;Searcy&Pietras,2011).Otherstudieshavemanipulatedperceivedneedinotherdomains,suchassocialstatus(Ermeretal.,2008),hypotheticalsurvivaldecisions(Mishra,Gregson,&Lalumie`re,2012;Mishra&Lalumie`re,2010;Wang,2002),hypotheticalfinancialportfolioscenarios(Mishraetal.,2012),ordrugaccess(Bickel,Giordano,&Badger,2004)withsimilarresults.MishraandLalumie`re(2010),forexample,showedinanex-perimentalstudythatpeoplemadesignificantlymorerisky,high-variancechoicesunderinducedconditionsofhighneed.Underinducedconditionsoflowneed,participantsdidnotexposethem-selvestotheunnecessarypossibilityofnotmeetingtheirneeds.Rather,theypreferredlow-riskoptionswithlowvarianceinpay-off.Theseresultswereobtainedbothwhenpeoplewereexplicitlyawareoftheparametersofdecisionoptions(i.e.,meanandvari-anceinadecision-making-from-descriptiontask)andwhentheyimplicitlylearnedtheparametersofdecisionoptionsthroughex-perience(usingadecision-making-from-experiencetask;Mishra&Lalumie`re,2010).Theseresultswerealsoobtainedbothwithandwithoutcontrollingforindividualdifferencesinpersonalityrele-vanttorisktaking(impulsivity,sensationseeking,andself-control),suggestingthatthispatternofdecisionmakingconform-ingtorisk-sensitivitytheoryrepresentsarobust,adaptivepatternofbehaviorformostindividuals.Together,thesestudiesaddtoagrowingbodyofevidenceindicatingthatrisk-sensitivitytheorycaneffectivelyaccountforpatternsofrisktakingunderconditionsofneed.

FRAMINGEFFECTSANDNEED

3

LanguagePragmatics

Risk-sensitivitytheorysuggeststhatparticipantsmakedecisionsaroundminimalrequirements,orneeds,whichshouldcontributetotheconstructionofreferencepointsinprospecttheory.TheAsiandiseaseproblem,forexample,isrelativelynovelasadecision-makingscenario,however,inthatitdoesnotofferanyexplicitindicationofthebaserateofpotentialoutcomesoracceptableminimalrequirements.Furthermore,participantshavelittleback-groundknowledgeorexpertiseabouttheproblem.Inunfamiliarproblemswithlittlebaserateinformation,referencepointsmaybederivedfromaspeaker’schoiceofframe.Wordchoiceindecisionscenariosmayprovideimplicitcuesaboutexpectedoutcomes,andthuscontributetothedefinitionofareferencepoint.Forexample,afullglassofwaterthatsubsequentlyhashalfitsvolumepouredoutismorelikelytobedescribedashalfemptyratherthanhalffull,becausethereferencepointisafullglass(Sher&McKenzie,2006).Consistentwiththishypothesis,McKenzieandNelson(2003)demonstratedthatperceptionsofcertaindecisionsvaryonthebasisofthevalenceofworduse.Similarly,Levin,Schnittjer,andThee(1988)demonstratedthatthebase-rateperceptionoftheincidenceofcheatingamonghypotheticalcollegestudentswashigheramongparticipantswhoreceivedthestatement“65%ofthestudentshadcheatedduringtheircollegecareer,”comparedwiththestatement“35%ofthestudentshadnevercheated.”Theseresultssuggestthatbase-rateoutcomesmaybeinferredfromwordchoiceinframedscenarios.

FramedscenariosinvolvingtheAsiandiseaseproblemprovidepeoplewithinformationaboutthenumberoflivesthatcanbepossiblysavedorlostwithdifferentoptions,andtheprobabilitiesassociatedwitheachchoice,butnoexplicitinformationispro-videdaboutwhatconstitutesasatisfactoryoutcome.IntheAsiandiseaseproblem,participantsmayimplicitlyinferwhatconstitutesasatisfactoryoutcomefromthevalenceandthewordingofdeci-sionoptions.Wang(2002)foundthatwhenaskedtoprovideaminimumacceptablenumberofpeopletosave,peopleindicatedasignificantlyhigheracceptableminimumaftermakingadecisioninanegativelyframedscenariocomparedwithapositivelyframedscenario.Thisfindingsuggeststhatthewordingofdecisionout-comesplaysanimportantroleindeterminingtheperceivedrefer-encepointofadecision.Specifically,negativelyframedscenariosmaymotivatepeopletosethigherminimalacceptablethresholds,orneedlevels,andthusadjusttheirreferencepoints.

Overview

Risk-sensitivitytheorymaybeusedtoaugmentprospecttheorybyprovidinganormativerationaleforwhypeopleexhibitframingeffectsindecisionscenarios.Specifically,risk-sensitivitytheoryprovidesanexplanationforhowreferencepointsinprospecttheoryareconstructed.IntheAsiandiseaseproblem,peopleappeartousewordingandvalencecuestodetermineaminimalacceptablethreshold,orneed,intheabsenceofmoreexplicitcues.Theseminimalthresholds,orneeds,shouldinturnfacilitatethedevelopmentofananalogousreferencepoint,asinprospecttheory(Kahneman&Tversky,1979;Tversky&Kahneman,1981).Innegativelyframedsituations,peopleappeartosetahigherneedrequirement(Wang,2002).Thishigherneedrequirementmoti-vatesriskychoice:Risk-sensitivitytheorypredictsthatbothhu-

manandnonhumananimalspreferriskyoptionsinsituationsofhighneed.Thus,theclassicfindingthatpeoplearerisk-averseinthefaceofgainsandrisk-proneinthefaceoflossesintheAsiandiseaseproblemmaybeaby-productofpeopletryingtominimizethepossibilityofexperiencinganegativeoutcomethatdoesnotmeettheirneeds.

Afewpreviousstudieshaveinvestigatedtheeffectofneedonriskychoiceinframeddecision-makingtasks(e.g.,Mishraetal.,2012;Wang,1996a,1996b,2002;Xie&Wang,2003).Thesestudieshaveshownthatpeoplewillsometimesviolatetheexpec-tationsofprospecttheoryincertainconditions.Forexample,Wang(2002)showedthatself-reportedminimumrequirementswereshowntobeassociatedwithriskychoiceinframeddecisionscenariosinvolvinghypotheticalfamilymembers.However,thisstudysufferedfromseveralproblems.First,minimumrequire-mentsweredeterminedbyself-reportafterparticipantshadalreadymadeadecisionand,asaresult,mayhavebeensubjecttoconsistencyeffects.Second,anonstandardsurvivalscenariowasusedinvolvingthepossibilityofsavingimmediatefamilymem-bers.Third,thestudydidnotexaminewhetherexternallyimposedminimumrequirements,orneeds,affecteddecisionmakinginframedscenarios.Similarly,Mishraetal.(2012)demonstratedthatimposedminimalrequirements,orneeds,motivatedriskychoiceinframeddecisionscenarios(involvingbothbehavioraltasksandtheoreticalfinancialscenarios).However,thisstudydidnotexam-inedecisionmakinginthecanonicalAsiandiseaseproblem,makingresultsdifficulttocomparewithmuchoftheexistingbodyofresearchonframingeffects.Furthermore,thisstudydidnotexaminewhethernegativeframesconfertheperceptionofhighneedorhigherperceivedminimalrequirements.

Thepresentinvestigationconsistsoffiveexperimentsinwhichweexaminedwhetherrisk-sensitivitytheorycanbeusedtocom-plementprospecttheorytobetterexplaindecisionmakinginframedscenariosinvolvingtheAsiandiseaseproblem.Specifi-cally,wesuggestthatrisk-sensitivitytheoryprovidesanormativerationalefortheconstructionofreferencepoints.AsubstantialbodyofresearchonframingeffectshasfocusedonexaminingdecisionmakingintheAsiandiseaseproblem;asaconsequence,wechosetousethesameproblemasthebasisforourexperimentssoastoallowclearercomparisonsacrosstheexistingliterature.InExperiments1and2,wedeterminedwhethertheimpositionofanexternalneedaffectedriskpreferencesindecisionmaking.InExperiments3and4,weinvestigatedwhetherself-generatedneedrequirementspredictedriskydecisionmaking.Finally,inExperi-ment5,wedeterminedwhethernegativelyframedscenariosconfertheperceptionofhighneed.

Experiment1

Peopleexhibitriskacceptancewhenfacedwithaneedoraminimalrequirementthatcannotbemetwithalow-riskoption(Deditius-Islandetal.,2007;Mishraetal.,2012;Mishra&Lalu-mie`re,2010;Pietras&Hackenberg,2001;Pietrasetal.,2003;Rodeetal.,1999;Wang2002).Wang(2002)demonstratedthatparticipantsindicatedahigherminimalrequirementoflivessavedandweremorerisk-pronefollowingexposuretoanegativeframeoftheAsiandiseaseproblem.TheseresultsmayberelevanttoframingeffectsintheAsiandiseaseproblem:Shiftsinriskpref-erencemayresultfrompeopleusinglanguagecuestoinferbase-

4

MISHRAANDFIDDICK

linerequirements(andpotentialbase-rateoutcomes)andthenchoosingbehavioraloptionsthatmeetorexceedtheserequire-ments.Consequently,ifaninferredneed—perhapsderivedfromscenariowording—issupplantedwithanexplicitlystatedneed,framingeffectsshouldnotbeobserved.Inthisexperiment,wepresentedparticipantswiththeAsiandiseaseproblem,butstipu-latedanexplicitneedrequirement.Wepredictedthatparticipantswouldchooseoptionsthatweremostlikelytomeettheirneeds,regardlessofpositiveornegativeframing.Morespecifically,wepredictedthatparticipantswouldprefertheriskyoptioninsitua-tionsofhighneedandthecertainoptioninsituationsoflowneed.

Method

Participants.Ninety-fourstudentsandstaffatauniversityparticipatedinthisstudy.Twoparticipantswereeliminatedforfailingtoproperlycompletethetask,leaving33menand59women(age:M?22.5,SD?2.9,range?19–33).Participantsreceivedachocolatebarfortheirparticipation.

Measuresandprocedure.EachparticipantreceivedoneoffourversionsoftheAsiandiseaseproblem.Twoinvolvedpositiveframing,andtwoinvolvednegativeframing.Withineachtypeofframing,oneversionspecifiedaneedthatexceededtheexpectedvalueofthetwooptions(ahigh-needcondition),andtheotherspecifiedaneedthatdidnotexceedtheexpectedvalueofthetwooptions(alow-needcondition):Positiveframe/highneed(n?24),positiveframe/lowneed(n?24),negativeframe/highneed(n?24),andnegativeframe/lowneed(n?20).Theexpectedvalueoflivessavedwasequivalentoverallconditionsandchoices.

TheAsiandiseaseproblemwasmodifiedfromTverskyandKahneman’s(1981)versiontoremovereferencetotheUnitedStates(toaccommodatenon-Americanparticipants)andtoelimi-nateanypotentialeffectsoffamiliarity,backgroundknowledge,orbiasthatmayhavearisenfromcallingthehealthcrisisan“Asiandisease.”Thescenariopresentedwasasfollows:

Imaginethatyouarethehealthministerinasmallcountry.Aspartofyourjob,youneedtodecidehowtorespondtohealthcrises.Supposethat600peopleareinfectedbyafataldisease.Twoalternativemedicalplanstotreatthediseasehavebeenproposed.Assumethattheexactscientificestimatesoftheconsequencesoftheplansareasfollows.

Thetwooptionsinthepositiveframingconditionswere:

1.IfPlanAisadopted,200peoplewillbesaved.

2.IfPlanBisadopted,thereisaonethirdprobabilitythatallthepeoplewillbesavedandatwothirdsprobabilitythatnoneofthemwillbesaved.

Thetwooptionsinthenegativeframingconditionswere:

1.IfPlanAisadopted,400peoplewilldie.

2.IfPlanBisadopted,thereisaonethirdprobabilitythatnoneofthemwilldieandatwothirdsprobabilitythatallthepeoplewilldie.

Participantsinthepositiveframe/high-needconditionwereasked:“Assumingthatyoumustsaveatleast300people,whichofthetwoplanswouldyouchoose?”Inthepositiveframe/low-needcondition,participantswereasked:“Assumingthatyoumustsave

atleast100people,whichofthetwoplanswouldyouchoose?”P(pán)articipantsinthenegativeframe/high-needconditionwereasked:“Assumingthatyoumustpreventmorethan300peoplefromdying,whichofthetwoplanswouldyouchoose?”Finally,inthenegativeframe/low-needcondition,participantswereasked:“As-sumingthatyoumustpreventmorethan100peoplefromdying,whichofthetwoplanswouldyouchoose?”

ResultsandDiscussion

Theimpositionofneedsignificantlyaffectedparticipants’pref-erencefortheriskyoption.Collapsedacrossframing,participantschosetheriskyoptionmorefrequentlyunderconditionsofhighneedcomparedwithconditionsoflowneed,?2(1,N?92)?17.27,p?.001;proportionchoosingriskyoption:highneed?.77;lowneed?.34.Withineachframingcondition,thesamepatternofresultswasobserved.Inthenegativeframecondition,participantschosetheriskyoptionsignificantlymorewhenunderaconditionofhighneed,?2(1,N?44)?5.53,p?.03;proportionchoosingriskyoption:highneed?.75,lowneed?.40.Similarly,inthepositiveframecondition,participantschosetheriskyoptionmorefrequentlyunderaconditionofhighneed,?2(1,N?48)?12.08,p?.001;proportionchoosingriskyoption:highneed?.79,lowneed?.29.TheseresultsaresummarizedinFigure1.

TheseresultssupportthepredictionthatanexternallyimposedneedwouldinfluenceriskychoiceintheAsiandiseaseproblem.Participantspreferredtheriskyoptionwhenunderaconditionofhighneed,wherethecertainoptionfellbelowtherequirednumberoflivessaved(andthereforedidnotofferachanceofmeetingparticipants’needs).Theseresultswereobtainedbothacrossandwithinthepositiveandnegativeframingconditions,suggestingthatneedisanimportantmotivatorofriskychoice.

Aspredicted,noframingeffectswereobserved.Collapsedoverneedcondition,participantsdidnotvaryintheirchoiceoftheriskyoption,?2(1,N?92)?0.23,p?.68;proportionchoosing

FRAMINGEFFECTSANDNEED

5

option:negativeframe?.59,positiveframe?.54.Framingeffectswerenotobservedwithineitherneedcondition(both?2s?.57,ps?.53).InthestandardversionoftheAsiandiseaseproblem,participantsmaymakedecisionsonthebasisofthewordingofdecisionoptions(Wang,2002).Theeliminationofframingeffectsthroughtheimpositionofexplicitneedrequire-mentssuggeststhattheserequirementsmaysupplantanyimplicitcuesofbaselinerequirementsderivedfromwordingvalenceintheAsiandiseaseproblem.

Thisexperimentdemonstratedthatparticipantsmakerisk-sensitivedecisionsthataremotivatedbyexternallyimposedbase-linerequirements,orneed.Thisresultisparticularlyimportantinlightofresearchsuggestingthatpeopleoftenmake“irrational”automaticorimplicitdecisions(asopposedto“rational”deliber-ativeorprobabilisticdecisions)inmanydifferentcircumstances(cf.Ariely,2008).Thatparticipantsmademathematically“ratio-nal”decisionsinthisexperimentindicatesthatparticipantsaresensitivetoexplicitneedconstraintsondecisionmakinganddonotexhibitunqualifiedlossaversionorframingeffectsintheAsiandiseaseproblem.

Still,animportantlimitationinthisexperimentconstrainsthescopeofourconclusions.Itispossiblethatframingeffectswereeliminatedinthisexperimentbecauseparticipantswereacutelysensitivetothemathematicsofthescenario,andnotbecauseofanypsychologicaleffects.Specifically,inthehigh-needcondition,onlytheriskyoptionallowedparticipantsachancetomeettheirneeds.Asaresult,theobservedeliminationoftheframingeffectmayhavesimplybeenaconsequenceofstrongdemandcharac-teristics.WeaddressthisshortcominginExperiment2.

Experiment2

Experiment1providedsomeevidencesuggestingthatpartici-pantsaresensitivetoexternallyimposedneed.However,Experi-ment1sufferedfromanimportantlimitation:Inthehigh-needcondition,theonlyoptionthatallowedparticipantstomeettheirneedwastheriskyoption,settingupapotentialdemandcharac-teristic.InExperiment2,wesoughttoexaminewhetherexternallyimposedsituationsofneedfacilitateriskpreferenceconsistentwithrisk-sensitivitytheory,butinawaythatdidnotfacilitatesignificantdemandcharacteristics.Inthisexperiment,theneedrequirementsanddecisionoptionswereconstructedsuchthatbothdecisionoptionsofferedsomechanceofmeetingone’sneedsinthehigh-needandlow-needconditions.Wepredictedthatpartic-ipantswouldexhibitapreferenceforoptionsthatweremoststatisticallylikelytomeettheirneeds(e.g.,Mishraetal.,2012;Mishra&Lalumie`re,2010).Morespecifically,wepredictedthatparticipantswouldpreferthehigh-riskoptioninsituationsofhighneedandthelow-riskoptioninsituationsoflowneed.

Method

Participants.Seventy-twostudents(30men,42women)atauniversityparticipatedinthisstudy(age:M?19.1,SD?1.1,range?18–22).Participantsreceivedcoursecreditfortheirparticipation.

Materialsandprocedure.Eachparticipantwasrandomlyassignedtoeitherapositiveframeconditionoranegativeframecondition.Withineachcondition,participantsreceivedthreede-

cisionscenariossimilartotheAsiandiseaseproblem.Thefirstscenariospecifiedaneedthatexceededtheexpectedvalueofthetwooptions(ahigh-needcondition),thesecondspecifiedaneedthatdidnotexceedtheexpectedvalueofthetwooptions(alow-needcondition),andthethirddidnotspecifyaneedconditionatall(thecontrolcondition).Theorderofpresentationofthethreedecisionscenarioswasrandomlydeterminedforeachparticipant.Thedecisionscenariopresentedwasasfollows:

Imaginethatyouarethehealthministerofasmallcountry.Aspartofyourjob,youneedtodecidehowtorespondtohealthcrises.Supposethat1,500peopleareinfectedbyafataldisease.Twoalternativemedicalplanstotreatthediseasehavebeenproposed.Assumethattheexactscientificestimatesoftheconsequencesoftheplansareasfollows.

Thetwooptionsinthepositiveframingconditionswere:

1.IfPlanAisadopted,thereisan8/10chancethat500peoplewillbesavedanda2/10chancethat1,000peoplewillbesaved.2.IfPlanBisadopted,thereisa4/10chancethatallthepeoplewillbesavedanda6/10chancethatnoneofthemwillbesaved.

Thetwooptionsinthenegativeframingconditionswere:

1.IfPlanAisadopted,thereisa2/10chancethat500peoplewilldieandan8/10chancethat1,000peoplewilldie.

2.IfPlanBisadopted,thereisa6/10chancethatallofthepeoplewilldieanda4/10chancethatnoneofthepeoplewilldie.

Participantsinthepositiveframe/high-needconditionwereasked:“Assumingthatyoumustsaveatleast900people,whichofthetwoplanswouldyouchoose?”Inthepositiveframe/low-needcondition,participantswereasked:“Assumingthatyoumustsaveatleast300people,whichofthetwoplanswouldyouchoose?”P(pán)articipantsinthenegativeframe/high-needconditionwereasked:“Assumingthatyoumustpreventmorethan900peoplefromdying,whichofthetwoplanswouldyouchoose?”Finally,inthenegativeframe/low-needcondition,participantswereasked:“As-sumingthatyoumustpreventmorethan300peoplefromdying,whichofthetwoplanswouldyouchoose?”P(pán)articipantsinthecontrolconditionwereasked:“Whichofthetwoplanswouldyouchoose?”Thehigh-needandlow-needconditionsimposedmini-mumrequirementsof300aboveand300belowtheexpectedvalueoflivessavedacrossalldecisionoptions(600),respectively.

ResultsandDiscussion

Collapsedacrossframing,participantsexhibitedasignificantlyhigherpreferencefortheriskyoptioninthehigh-needconditioncomparedwithboththelow-needcondition(WilcoxonZ?3.71,p?.001)andthecontrolcondition(Z?3.46,p?.001);proportionchoosingriskyoption:highneed?.64,lowneed?.40,control?.47.Thesamepatternofresultswasobtainedbothwithinthepositiveframeandthenegativeframe.Withinthepositiveframe,participantsexhibitedhigherriskpreferenceinthehigh-needconditioncomparedwiththelow-needcondition(Z?2.31,p?.02)andthecontrolcondition(Z?2.65,p?.008);proportionchoosingriskyoption:highneed?.56,lowneed?.33,control?.36.Withinthenegativeframe,participantsexhib-

6

MISHRAANDFIDDICK

itedhigherriskpreferenceinthehigh-needconditioncomparedwiththelow-needcondition(Z?3.00,p?.003)andthecontrolcondition(Z?2.24,p?.03);proportionchoosingriskyoption:highneed?.72,lowneed?.47,control?.58.TheseresultsaresummarizedinFigure2.

Collapsedacrossneedconditions,anoverallframingeffectwasobserved,?2(1,N?216)?6.69,p?.01;proportionchoosingriskyoption:negativeframe?.59,positiveframe?.42.Asignificantframingeffectwasobservedinthecontrolcondition,?2(1,N?72)?3.57,p?.05(one-tailed);proportionchoosingriskyoption:negativeframe?.58,positiveframe?.36.How-ever,significantframingeffectswerenotobservedwithineitherthehigh-needcondition,?2(1,N?72)?2.17,p?.11,orthelow-needcondition,?2(1,N?72)?1.44,p?.17.

TheresultsofExperiment2replicatethoseofExperiment1withouttheproblemsofpotentialdemandcharacteristics,giventhatboththeriskyandnonriskyoptionsofferedparticipantsachancetomeettheirminimumrequirementoflivessavedineitherneedcondition.Furthermore,Experiment2addressesthepossibil-itythattheimpositionofexternalminimumrequirementssimplyuncoveredananchoringandadjustmenteffect.InExperiment1,theanchorinthepositiveframingwas200,andtheanchorinthenegativeframingwas400.Asaconsequence,itispossiblethattheseasymmetricalanchorsaffectedriskychoice.InExperiment2,however,identicalanchorswereusedinboththepositiveframeandthenegativeframe(i.e.,500and1000),andtheanchornumberswerepresentedinthesameorderinbothframes.Asaconsequence,anchoringeffectscannotexplaindifferentialriskpreferenceamongthehigh-need,low-need,andcontrolconditionsinExperiment2.

TheresultsofExperiment2providefurtherevidenceindicatingthatexternallyimposedconditionsofneedinfluenceriskychoiceconsistentwithrisk-sensitivitytheory.AsinExperiment1,partic-ipantspreferredtheriskyoptionwhenunderaconditionofhighneed,wherethecertainoptionfellbelowtherequirednumberoflivessaved(andthereforedidnotofferachanceofmeetingparticipants’needs).Framingeffectswereobservedfordecisionscollapsedacrossallconditions,aswellaswithinthecontrol

Figure2.Proportionofparticipantschoosingtheriskyoption(M,SE)inExperiment2.

condition.However,withinconditionsofhighneedandlowneed,framingeffectswerenotobserved.Imposingexplicitneedrequire-mentsmayhaveeliminatedanyeffectsoflanguagevalenceonriskychoicethatnormallycontributetoframingeffects(e.g.,McKenzie,2003;McKenzie&Nelson,2003).

Experiment3

Experiments1and2demonstratedthatparticipantsshiftriskpreferenceasafunctionofanexternallyimposedneed.ThestandardversionoftheAsiandiseaseproblemcontainsnoexplicitmentionofanybaselinerequirementsoranybase-rateinformation.Therefore,ifperceivedneedinfluencesperformanceinstandardversionsofthetask,itmustbeself-generated.Toourknowledge,onlyonepreviousstudyhasinvestigatedtheroleofinternallydeterminedneedondecision-makingintheAsiandiseaseprob-lem.

Wang(2002)foundthatpeople’sself-generatedminimumre-quirementsofnumberoflivessavedpredictedchoicesintheAsiandiseaseproblem:Peoplewhoreportedaminimumrequirementabovetheexpectedvalueofthecertainoptionshowedincreasedriskacceptance,aspredictedbyrisk-sensitivitytheory.Framingofthetaskmayhaveinfluencedpeople’sminimalreportedneedrequirements,andindeed,Wang(2002)foundthatpeople’smin-imalreportedneedrequirementswerehigherinthenegativefram-ingconditioncomparedwiththepositiveframingcondition.Wang’sstudyinvolvedaskingparticipantsabouttheirminimalrequirementsaftertheyhadalreadymadeadecision,however,andsoparticipantsmayhaveindicatedminimalrequirementsthatwereconsistentwiththechoicestheymade.

Inthisexperiment,weinvestigatedwhetherself-generatedeval-uationsofneedpredictedsubsequentdecisionmakingintheAsiandiseaseproblem.Wepredictedthatpeople’sself-generatedneedrequirementswouldpredictsubsequentriskychoiceconsistentwithrisk-sensitivitytheory.Specifically,wepredictedthatindi-vidualswithhigh-needrequirementswouldpreferriskyoptionsmorethanthosewithlow-needrequirements.Giventhefindingthatpeopleappeartoperceivehigherneedrequirementsinnega-tivelyframedscenarios(Wang,2002),wepredictedthatwewouldobserveanoverallframingeffectacrossneedgroups.However,wealsopredictedthatframingeffectswouldnotbeobservedwithinneedgroups,withtheeffectofneedsupersedingthatofframing.

Method

Participants.Seventy-ninestudentsandstaffatauniversityparticipatedinthisstudy.Threeparticipantswereremovedfromthestudyforfailingtocompletethetask,leaving76participants(30men,46women;age:M?22.6,SD?3.5,range?18–36).Participantsreceivedachocolatebarorcoursecreditfortheirparticipation.

Materialsandprocedure.EachparticipantreceivedtheAsiandiseaseproblemdescribedinExperiment1,eitherpositively(n?40)ornegatively(n?36)framed.Beforemakingadecisionbetweenthecertainandriskyoptions,participantswereasked(a)howmanypeopletheywouldliketosaveand(b)whattheirminimumrequirementwasfornumberofpeoplesaved.Thesetwoquestionswereaskedinordertodistinguishbetweentwo

potential

FRAMINGEFFECTSANDNEED

7

referencepoints:(a)participants’aspirationleveland(b)theirminimalrequirementsorneed(or“survivalpoint”;March&Shapira,1992).

Inthepositiveframe,thetwoquestionswerephrasedasfollows:(a)“Outofthe600peopleinfectedwiththedisease,howmanypeoplewouldyouliketosave?”(b)“Whatisyourbottomline:Howmanypeopleoutof600mustlive?”Inthenegativeframingcondition,thetwoquestionswerephrasedasfollows:(a)“Outof600peopleinfectedwiththedisease,howmanypeoplewouldyouliketopreventfromdying?”(b)“Whatisyourbottomline:Howmanypeopleoutof600candie?”Followingthesetwoquestions,participantsindicatedwhichofthetwoplanstheywouldprefertoimplement.Participantssawthefulldescriptionofeitherthenegativelyframedorpositivelyframedplanbeforeansweringthequestionsandmakingadecisionbetweentheplans.Questionswerepresentedseparately,althoughitwaspossibleforparticipantstogobackandchangetheiranswersforconsistency.Wenote,however,thatfewparticipantsappearedtohavechangedtheirresponses,eventhoughtheywerefreetodoso.1

ResultsandDiscussion

Everyparticipantreportedthattheywouldliketosave(orpreventfromdying)all600people,andsoreportedaspirationlevelscouldnotexplainparticipants’choices,norweretheyin-fluencedbytheframingoftheoptions.Toinvestigatetheroleofminimumrequirementsondecisionmaking,weseparatedpartic-ipantsintotwogroups:thosewhoreportedminimalneedrequire-mentsabove(n?51)andbelow(n?25)theexpectedvalueofthecertainoption(saving200people/preventing200peoplefromdying).Thesetwogroupsrepresentahigh-needgroupandalow-needgroup,respectively.Collapsedacrossframing,partici-pantswhoreportedaself-generatedneedrequirementabovetheexpectedvalueofbothoptionsweresignificantlymorelikelytoselecttheriskyoption,?2(1,N?76)?11.13,p?.001;propor-tionchoosingriskyoption:aboveexpectedvalue?.65,belowexpectedvalue?.24.

Thesamedirectionalpatternofriskychoicewasobservedwithineachframe.Inthepositiveframe,participantsmadesignif-icantlymoreriskychoicesiftheirminimalneedrequirementexceededtheexpectedvalueoftheoptions,?2(1,N?40)?5.02,p?.05;proportionchoosingriskyoption:aboveexpectedvalue?.54,belowexpectedvalue?.19.Asimilarpatternofresultswasobtainedforthenegativeframe,?2(1,N?36)?4.86,p?.05;proportionchoosingriskyoption:aboveexpectedvalue?.74,belowexpectedvalue?.33.TheseresultsaresummarizedinFigure3.

Participantsprovidedhigherminimalneedrequirementsforlivessavedinthenegativeframecomparedwiththepositiveframe(Z??1.63,p?.05,one-tailed);meanminimumneedrequire-ment:negativeframe?394.39,positiveframe?326.25.ThisfindingreplicatestheresultsobtainedbyWang(2002),demon-stratingthatparticipantsindicatedhigherminimalneedrequire-mentswhenexposedtoanegativelyframedcondition.

Asignificantframingeffectwasnotobservedinthehigh-needgroup,?2(1,N?51)?2.21,p?.16;proportionchoosingriskyoption:positiveframe?.54,negativeframe?.74.Similarly,noframingeffectwasobservedinthelow-needgroup,?2(1,N?25)?0.672,p?.63;proportionchoosingriskyoption:positive

Figure3.Proportionofparticipantschoosingtheriskyoption(M,SE)basedonself-reportedminimalacceptableneedrequirements(Experiment3).

frame?.19,negativeframe?.33.Thatframingeffectswerenotobservedwithinneedconditionconformstoourpredictions;anyeffectofperceivedneedderivedfromframingshouldhavebeensupersededbyself-reportofaminimalrequirement.Thisresultprovidessupportforthepredictionthatframingeffectswouldbeeliminatedunderconditionsofself-generatedneed,withpartici-pantschoosingoptionsthatwerelikelytomeettheirneedrequire-ments.

Acrossallparticipants,however,asignificantframingeffectwasobserved,?2(1,N?76)?4.33,p?.04;proportionchoosingriskyoption:positiveframe?.40,negativeframe?.64.Thisresultwasaspredicted;accordingtoWang(2002),negativeframesshouldfacilitateaperceptionofhighneed,andpositiveframesaperceptionoflowneed.ThiseffectisfurtherinvestigatedinExperiment5.Together,theresultsofExperiments1,2,and3indicatethatpeoplemakeriskydecisionscontingentonneed,regardlessofwhetherthatneedisdeterminedexternally(Experi-ments1and2)orinternally(Experiment3).

Experiment4

Experiments1,2,and3demonstratedthatpeopleexhibitele-vatedriskacceptanceinsituationsofhighneed,regardlessofwhetherthatneedwasdeterminedeitherexternally,throughanimposedrequirementofnumberoflivessaved,orinternally,throughself-reportofaminimumacceptablenumberoflivessaved.Theseexperiments,however,assumedthatneedsaredis-cretetargets.Minimalrequirementsneednotdisplayastep–functionrelation,withallvaluesbelowathresholdvaluebeingequallyimportantandallvaluesinexcessofthethresholdbeingequallyunimportant.Experiment4allowedparticipantstoexpress

1

Sixofthe53participantsappearedtohavechangedtheiranswers.Analysesconductedwiththeseparticipantseliminatedprovidethesamepatternof

results.

8

MISHRAANDFIDDICK

theirneedsoverarangeofvaluesratherthanspecifyingasingleminimalrequirement.

Weinvestigatedwhetherparticipants’riskpreferencespredictedtheself-ratedimportancetheyattachedtosavingdifferentnumbersofpeople.Ifparticipants’goalsarediscrete,thenonewouldexpecttoseeathresholdinreportedimportanceofsavingacertainnumberofpeople.Forexample,ifanindividualplaceshighvalueonsavingnofewerthanall600lives,thentheymayalwaysprefertheriskyoptionregardlessoftheexpectedvalueofthecertainoutcome.Wepredictedthatratingsoftheimportanceofsavingacertainnumberoflivesacrossthedistributionofallpossibleoutcomes(rangingfrom100to600livessaved)wouldbeassoci-atedwithvariationinriskpreferenceacrossthedistribution,withrisk-averseparticipantsdisplayingadifferentpatternofimpor-tanceratingsthanrisk-proneparticipants.

Method

Participants.Fifty-threestudentsparticipatedinthisstudy.Fourparticipantswereremovedfromthestudyforfailingtocompletethetask,leaving49participants(19men,30women;age:M?24.4,SD?3.6,range?19–35).Participantsreceivedachocolatebarascompensation.

Materialsandprocedure.ThesameAsiandiseaseproblemdecisionscenarioasinExperiment1wasused.Eachparticipantreceivedasinglescenarioframedeitherpositively(n?24)ornegatively(n?25).Participantsthenansweredsixquestionstoassesshowimportanttheyfeltitwastosavethelivesofvariousnumbersofpeople.Inthepositiveframingcondition,thequestionreadasfollows:“Assumingthatthesearetheonlyoptionsavail-able,howimportantisitthatatleast100[200,300,400,500,600]peoplelive?”Inthenegativeframingcondition,thequestionreadasfollows:“Assumingthatthesearetheonlyoptionsavailable,howimportantisitthatatleast100[200,300,400,500,600]peoplearepreventedfromdying?”P(pán)articipantsansweredona5-pointLikertscale(1?notimportant,5?veryimportant).Theorderofthequestionswasrandomized.Participantswereaskedwhichofthetwoplanstheywouldimplementafterallotherquestions.

ResultsandDiscussion

Participantsweresplitintotwogroupsonthebasisofwhethertheychosethecertainoption(n?16)ortheriskyoption(n?33).Weconductedarepeatedmeasuresanalysisofvariance(ANOVA)onparticipants’importancejudgments,withthenumberoflivessaved(100,200,300,400,500,600)asthewithin-subjectsfactorandoptionchoice(certain,risky)andframe(positive,negative)asthebetween-subjectsfactors.

Amaineffectoftheimportancejudgmentswasobserved,indi-catingthatsavingasmallerminimumnumberofliveswasratedasmoreimportantthansavingalargerminimumnumberoflives,F(5,205)?16.36,p?.001,?2?.29;adjustedmeanimportancerating:M100?4.54,M200?4.41,effectM300of??risky3.77,.19,choiceM400indicatingwas?3.59,M500?3.52,M600?3.56.Amainalsoobtained,F(1,41)?9.69,p?.003,?2thatparticipantswhochosetheriskyoptionratedsavinglivesasmoreimportant(collapsedoverallsixdecisions);meanimportancerating:riskychoice?4.38,certainchoice?3.42.Nomaineffect

offrameonimportancejudgmentswasobtained,F(1,41)?0.001,p?.98,?2?.01;meanimportancerating:positiveframe?3.90,negativeframe?3.89.TheseresultsaresummarizedinFigure4.Aspredicted,weobtainedasignificantnumberoflivessavedbyoptionchoiceinteractionforparticipants’importancejudgments,F(5,205)?22.03,p?.001,?2?.35.Participantswhochosethecertainoptionjudgedsavingfewerpeople(100,200)tobemoreimportantthansavingalargernumberofpeople(?300)(seeFigure4).Follow-upplannedsimpleeffectscomparisonssupportthisclaim.Amongparticipantswhochosethecertainoption,importancejudgmentsofsaving100and200peopleweresignif-icantlyhigherthanallotherpossiblenumbersoflivessaved(allps?.007),suggestingthatpeoplewhochosethecertainoptionshowathresholdofimportanceattheexpectedvalueoftheavailableoptions.Comparedwithparticipantswhochosethecer-tainoption,participantswhochosetheriskyoptionindicatedthatsavinganumberofpeoplegreaterthantheexpectedvalueofthecertainoption(greaterthan200)wasmoreimportant(allps?.006).Therewasnosignificantdifferenceinthereportedimpor-tanceofsaving100or200peopleamongparticipantswhochosetheriskyandcertainoptions(bothps?.17).Participantswhochosetheriskyoptiondidnotexhibitvariationinimportancejudgmentsacrosstherangeofpossiblelivessaved(allps?.09),suggestingthatsuchparticipantsexhibitednothresholdofimpor-tance.

Noframingeffectwasobserved,?2(1,N?49)?0.01,p?.99;proportionchoosingriskyoption:positiveframe?.67,negativeframe?.68.Byratinghowimportantitistosavedifferentnumbersofpeople,participantsmayhavecreatedaninternalminimumacceptablethresholdofneedeventhoughtheywerenotexplicitlyinstructedtodoso.Thatparticipantswhochosethecertainoptionexhibitedasteepdropintheratedimportanceofsavingmorethantheexpectedvalueof200peoplesupportsthishypothesis.Analternativeexplanationforthelackofaframingeffectinthisstudy,however,maybethataskingparticipantstoprovideminimalrequirementsacrosstherangeofoptionsmayhavechangedthedecisionscenariosignificantlyandthus

affected

Figure4.Importanceratingsforsavingatleast100,200,300,400,500,or600people(M,SE)asafunctionofriskychoice(Experiment4).

FRAMINGEFFECTSANDNEED

9

participants’riskychoice.Asaconsequence,theseresultsmustbeinterpretedwithsomecaution.Wedonote,however,thattheresultsofbothExperiments3and4provideconvergentevidenceindicatingthatself-reportedminimalrequirements(althoughmea-suredindifferentways)facilitateriskychoiceconsistentwithrisk-sensitivitytheory.

TheresultsofExperiment4suggestthatpeoplewhochoosethecertainoptionandpeoplewhochoosetheriskyoptionintheAsiandiseaseproblemhavedifferentvaluationsoftheimportanceofsavingalargenumberoflives.Participantswhochosethecertainoptiontendedtodevaluetheimportanceofsavinganumberoflivesgreaterthantheexpectedvalueofavailableoptions.Partic-ipantswhochosetheriskyoptionemphasizedtheimportanceofsavinganumberoflivesgreaterthantheexpectedvalueoftheavailableoptions.TheseresultsareconsistentwithresultsobtainedfromExperiment3,inthatpeopleappeartomakedecisionsonthebasisofself-constructedminimalrequirements.

Experiment5

Experiments3and4werepredicatedontheassumptionthatnegativelyframeddecisionscenariosconfertheperceptionofhigherneed.ThisassumptionwassupportedinExperiment3:Participantsmakingdecisionsinnegativeframesindicatedsignif-icantlyhigherminimalrequirementsforlivessavedcomparedwiththosemakingdecisionsinpositiveframes.TheseresultsreplicateWang’s(2002)findings,indicatingthesamepatternofhigherreportedminimalrequirementsinnegativeframes.However,thesefindingsmaybetheproductofartificialconsistencyeffects.InExperiment3,participantsindicatedaminimalrequirementoflivessavedandthenmadeadecision.Similarly,inExperiment4,participantsprovidedimportanceratingspriortomakingadeci-sion.Itispossiblethatparticipants’decisionsinExperiments3and4weremadetoappeartobeconsistentwiththeirreportedminimalrequirementoflivessaved.InWang(2002),participantsindicatedaminimalrequirementaftertheyhadalreadymadeadecision,andsoparticipantsmayhaveindicatedaminimalrequirementthatwasconsistentwiththechoicetheyhadmade.InExperiment5,weexaminedwhethernegativeframesconfertheperceptionofhighneed.

Method

Participants.Fifty-onestudents(21men,30women)atauniversityparticipatedinthisstudy(age:M?20.6,SD?3.6,range?18–28).Participantsreceivedcoursecreditfortheirparticipation.

Materialsandprocedure.EachparticipantwasrandomlyassignedtoreceiveasingleAsiandiseasescenario(asinExperi-ment1)framedeitherpositively(n?22)ornegatively(n?29).Followingthedecisionscenario,participantsansweredtwoques-tions.Inthepositiveframe,thetwoquestionswerephrasedasfollows:(a)“outofthe600peopleinfectedwiththedisease,howmanypeoplewouldyouliketosave?”(b)“Whatisyourbottomline:Howmanypeopleoutof600mustlive?”Inthenegativeframingcondition,thetwoquestionswerephrasedasfollows:(a)“Outof600peopleinfectedwiththedisease,howmanypeoplewouldyouliketopreventfromdying?”(b)“Whatisyourbottomline:Howmanypeopleoutof600candie?”P(pán)articipantsdidnotchooseaplanafteransweringthesequestions.

ResultsandDiscussion

AsinExperiment3,allparticipantsindicatedthattheywouldliketosaveall600people.Participantsinthenegativeframeconditionindicatedthattheirminimalrequirementfornumberoflivessavedwassignificantlyhigherthandidparticipantsinthepositiveframecondition(Mann–WhitneyU?213.00,p?.02,one-tailed);meanminimalrequirement:positiveframe?292.50,negativeframe?410.17.Thatparticipantsinthenegativeframeindicatedasignificantlyhigherminimalrequirementoflivessavedindependentofanydecisionsuggeststhatnegativelyframedsce-nariosconfertheperceptionofhighneedandthatthisperceptionisnotaby-productofanyconsistencyeffects.

Whydonegativeframesconfertheperceptionofhigherneed?Itispossiblethatparticipantsdonotexplicitlyconstructasinglethresholdthatdescribesthenumberoflivestheyfeelisaminimalrequirementtosave,butratherarangeofacceptableoutcomes.Positiveandnegativeframesmaythereforeleadtoestimatesthatare,respectively,atthelowerandhigherboundsofanacceptablerangeofcasualties(e.g.,Yaniv&Schul,1997,2000).TheresultsofExperiment4providesomeindirectevidenceforthishypoth-esis:Gainandlossframingdidnotfacilitatesignificantlydifferentmeanratingsoftheimportanceofsavinglivesacrossapossibledistribution(althoughasignificantthresholdeffectwasobservedamongparticipantswhochosethecertainoption).

Analternativeexplanationfordifferentperceivedminimalre-quirementsingainandlossframesinvolvesthedifferentialuseandperceptionoflanguage.Intheabsenceofbackgroundknowl-edgeoranyexplicitinformationaboutminimalacceptablethresh-olds(e.g.,asinExperiments1and2),decisionmakersmayrelyonimplicitcues(suchaslanguagevalence;Druckman,2001)toinferhowwelltheyareperformingrelativetoperceivedexpectations.Theseinferencesmaysubsequentlyinformpeople’sdecisions.Negativelyframedoptionsmightsignalthatoneisfartherawayfromadesiredgoalorneed(facilitatingriskpreference),whereaspositivelyframedoptionsmightsignalthatoneisclosertoadesiredgoalorneedthanexpected(facilitatingriskaversion;McKenzie,2003;McKenzie&Nelson,2003).

Furtherresearchisnecessarytoexaminetheproximatemecha-nismsunderlyingtheconstructionoflowerminimalrequirementsingainframescomparedwithlossframes.Regardlessoftheunderlyingmechanism,however,elevatedriskacceptanceinre-sponsetohigherreportedminimalthresholdsinnegativeframesinExperiments3and5maybeinterpretedastheproductoftheperceptionofhighneedfacilitatingincreasedriskychoice,consis-tentwithrisk-sensitivitytheory.

GeneralDiscussion

Theresultsoffiveexperimentsindicatedthatpeoplemakedecisionsthatconformtothepredictionsofrisk-sensitivitytheoryintheAsiandiseaseproblem.Peoplepreferredriskyoptionsinsituationsinwhichtheirneedrequirementswereunlikelytohavebeenmetbylow-riskalternatives.Theseresultswereobtainedregardlessofwhetherneedrequirementsweredeterminedexter-nally,throughtheimpositionofa“target”numberoflivestosave(Experiments1and2),orinternally,throughself-reportofadiscreteminimumacceptablethresholdoflivestosave(Experi-ment3).PeoplewhooptedforthecertainoptionintheAsian

10

MISHRAANDFIDDICK

diseaseproblemplacedlessvalueonachievingoutcomesinexcessoftheexpectedvalueoftheavailableoptions(i.e.,self-reportingthemselvesasbeinginaconditionoflowneed),whereaspeopleoptingfortheriskyoptionplacedmoreofavalueonachievingoutcomesinexcessoftheexpectedvalue(i.e.,self-reportingaconditionofhighneed;seeExperiment4).Importantly,needwasinpartdeterminedbyframing,withnegativeframesconferringtheperceptionofhighneed;participantsindicatedahigherminimalacceptablethresholdoflivessavedinnegativeframes(Experi-ments3and5).

Theresultsofthisstudysuggestthatrisk-sensitivitytheorycanexplicatemechanismsofchoicethatfacilitateframingeffectsbyprovidinganormativerationaleforriskychoicebasedaroundminimumrequirements,orneed.KahnemanandTversky(1979)suggestedthatframingeffectsariseonthebasisofdecisionsmadearoundareferencepoint,whereoutcomesbelowareferencepointrepresentlossesandoutcomesaboveareferencepointrepresentgains.Theresultsofthisstudysuggestthatminimumrequire-ments,orneeds,maydeterminetheconstructionofreferencepointsaroundwhichdecisionsaremade.Participantswhosemin-imalacceptableoutcomethresholdfellbelowtheexpectedvalueofthedecisionoptionstendedtopreferthecertainoption.Fortheseparticipants,theirminimumacceptableoutcomethresholds(andthusreferencepoint)waslikelydefinedbytheexpectedvalueofthecertainoption.Participantswhochosetheriskyoptiondidsoinsituationsinwhichtheirneedrequirementwashigher,andso,fortheseparticipants,referencepointswerelikelyhigher.

ReferencePointsandNeeds

Risk-sensitivitytheorypositsthatdecisionmakersseektomin-imizetheprobabilityofexperiencingoutcomesthatfailtomeettheirneeds(Stephens,1981).Risk-sensitivitytheoryfurtherpositsthatminimumrequirements,orneed,areimportantlyprivilegedindecisionmakingandserveasanormativemotivatorofriskychoice(Mishra&Lalumie`re,2010;Rodeetal.,1999;Stephens&Krebs,1986).Thepredictionsofrisk-sensitivitytheoryaresimilartothoseofprospecttheory,wheredecisionmakersengageinrisktakingtomaximizegain,butnotatthecostofloss(Tversky&Kahneman,1981).TverskyandKahnemansuggestedthatrefer-encepoints(aroundwhichgainsandlossesaredefined)arede-rivedfromanindividual’spresentstate,butcanchangeonthebasisofexpectationsorbiasesofdecisionmakers.Asaconse-quence,thereferencepointinprospecttheorymaybedeterminedfromneedthresholds.Belowthereferencepoint,peoplemayperceivethemselvestobeinasituationofhighneed(i.e.,farfromadesiredstate),andabovethereferencepoint,peoplemayper-ceivethemselvestobeinasituationoflowneed(i.e.,closetoadesiredstate,orhavingexceededthethresholdofadesiredstate).TheresultsofExperiments3and5suggestthatpeoplehavetwopotentialreferencepoints:aspirationlevels(orgoals;Heath,Lar-rick,&Wu,1999)orminimalacceptablethresholds,orneeds(orsurvivalpoints;March&Shapira,1992).InExperiments3and5,allparticipantsreportedanaspirationtosaveall600people.Ifparticipantsconsideredtheirreferencepointtobe600people,thenanydecisioninvolvingsavingfewerthan600peoplewouldhaveconstitutedaloss,andthusprospecttheorywouldpredictthatparticipantsshouldhavebeenrisk-proneunderthesecircum-stances.Contrarytothisprediction,participantsdidnotmake

decisionsonthebasisoftheirreportedaspirationlevel,butrathermadedecisionsonthebasisoftheirreportedminimumacceptablethresholds(Experiment3).TheresultsofExperiments3and5indicatedthatnegativeframesconferhigherminimumacceptablerequirementscomparedwithpositiveframesandthatthesemini-mumrequirementsmotivateriskychoice.Thisfindingsuggeststhatminimalrequirements,orneeds,eitherinducethecreationofareferencepointorsignificantlycontributetothedeterminationofareferencepoint.Becauseminimalrequirementsappeartopri-marilymotivateriskychoice,minimalrequirementsmaybepriv-ilegedindecisionmaking.Together,ourfindingssupportthepredictionsofrisk-sensitivitytheory,whichpositsthatminimalrequirementsarekeymotivatorsofdecisionmakingunderrisk.

LanguagePragmaticsandAffect

Theresultsofpreviousstudieshaveshownthatlanguageuseinframeddecisionscenariosmayimplicitlyprovidebaserateinfor-mation,andthussubtlyinfluencechoice(Levinetal.,1998;McKenzie&Nelson,2003;Sher&McKenzie,2006).Theresultsofourexperimentsprovidefurtherevidenceforthiseffectandsuggestapotentialmechanismthatmayinpartexplainshiftsinbase-rateperception.Experiments3and5demonstratedthatneg-ativeframesconferredtheperceptionofhighneed,inthatpartic-ipantssethigherminimalacceptablerequirementsforthemselves,andExperiment3demonstratedthatthisperceptionofhighneedleadstoincreasedriskychoice.

Theuseofloss-orientedlanguagemayserveasasubtleprimeofrisk-acceptingattitudes,leadingtoshiftsinrisk-acceptingbehavior(e.g.,Erb,Bioy,&Hilton,2002).Otherresearchhasshownthatlossframingleadstonegativeaffect(Nygren,1998),whichinturnhasbeenassociatedwithincreasedrisk-acceptingbehavior,espe-ciallyinnegativeframes(Leith&Baumeister,1996;Mittal&Ross,1998).Loewenstein,Weber,Hsee,andWelch(2001)pro-posedthe“riskasfeelings”hypothesis,whichpositedthatemo-tionssometimesaffectcognitiveevaluationofchoiceoptions,suchthattheperceptionofdecisionoptionschangesbasedonmoodstate.Peopleexposedtonegativeframesmayengageinreasoningthatisimplicitlymotivatedbysensitivitytoaffect(i.e.,peoplemay“feel”thatriskydecisionsareabetteroptioniftheyareinanegativemood).Thus,itispossiblethatnegativeaffectmaymediatethelinkbetweenloss-framedlanguage,theperceptionofhighneed,andsubsequentriskacceptance.Ourresultsprovideevidencesuggestingthatnegativeframesleadtotheperceptionofhighneed,butfurtherresearchisrequiredtoexaminewhethertheconstructionofhigherminimalrequirementsinnegativeframesisinpartfacilitatedbynegativeaffect.

IndividualDifferences

Althoughpreviousresearchhasindicatedthatindividualdiffer-encescannotexplainvariationinriskychoiceunderconditionsofexplicitlydefinedneed(Mishra&Lalumie`re,2010),littleresearchhasbeenconductedonwhetherindividualdifferencescontributetotheconstructionofdifferentacceptableminimalrequirements,orneeds.TheresultsofExperiment1,forexample,suggestthatpeoplesetdifferentminimalrequirementsevenwhenexplicitminimalrequirementsarestated:34%ofparticipantsselectedthe

FRAMINGEFFECTSANDNEED

11

riskyoption,eventhoughtheyweretoldthattheyonlyneededtosave100people(anoutcomeguaranteedbythecertainoption).Someparticipantsmighthaveerredintheirreasoning,butitisalsopossiblethatmanyoftheseparticipantssethigherminimalthresholdsforthemselves.Insupportofthishypothesis,theresultsofExperiments3and5demonstratedthatpeoplereportedhighlyvariableminimalacceptableneedrequirementsandthattheseneedrequirementsmotivateddifferentpatternsofriskychoice.Thisresultsuggeststhattherearesignificantindividualdifferencesintheconstructionofminimalneedrequirementsandthatthesevariableneedrequirementsmotivatedifferentpatternsofriskacceptance.Furtherresearchisnecessarytoexaminetheinfluenceofindividualdifferencesonminimalneedrequirementpercep-tions.

Limitations

Despitethisstudy’sresults,somelimitationsmustbenoted.TheAsiandiseaseproblemorsomemodificationoftheAsiandiseaseproblemwasusedinallexperimentsinthepresentstudy.AlthoughtheAsiandiseaseproblemhasbeenstudiedextensively,ithasbeenarguedtobeanoutlierinthatitismorelikelytoproduceframingeffectscomparedwithotherdecisionscenarios(Sickar&High-house,1998).Thegeneralizabilityofourresultsisthereforeanopenquestion.Wedonote,however,thatsomeotherrecentevidencesuggeststhatrisk-sensitivitytheorycanexplainriskychoiceinframescenariosbeyondtheAsiandiseaseproblem.Mishraetal.(2012),forexample,showedthatthepredictionsofrisk-sensitivitytheoryholdinframedscenariosinvolvingbe-havioralrisk-takingtasksinbothecologicalandfinancialdo-mains.Althoughitappearsthatagrowingbodyofevidencesuggeststhatrisk-sensitivitytheorycanexplainrisk-sensitivechoiceinframedscenarios,furtherresearchexaminingthegeneralizabilityofthesefindingsisrequired.

Inthisstudy,wedidnotexhaustivelyaddresstheprocessthroughwhichminimalrequirementsorneedthresholdsarecon-structed.Someothershavesuggestedpossiblemodelsforthegenerationofreferencepointsorminimalthresholdsfordecisionmaking.KoszegiandRabin(2006),forexample,suggestthatreferencepointsaredeterminedendogenouslyfromexpectationsaboutpastoutcomes.LopesandOden(1999)proposethatsecuritypotential/aspirationtheorycanexplainriskychoicebyaccountingforsecuritypotential(whichaccountsforoutcomesandprobabil-ities)andaspiration(theprobabilityofobtainingapreviouslyspecifiedoutcome).Thesemodelsofferpossibleproximatemech-anismsthatmayaccountforriskychoiceunderconditionsofneedinframedscenarios,andbothsharesomeimportantcomponentsofrisk-sensitivitytheory.Itwouldbeinformativetocomparethepredictiveabilityofrisk-sensitivitytheorywiththeseothermodels.Othermechanismsmayalsoplayanimportantproximateroleinfacilitatingrisk-sensitivedecisionmakinginframedscenarios.Inparticular,littleisknownaboutwhatproximatecuesareusedtoconstructneedrequirements,oracceptablethresholds,ineverydaydecisionmaking.Inthepresentarticle,wesuggestafewpossibil-ities,includinglanguagechoice(e.g.,Levinetal.,1998;McKen-zie,2003;McKenzie&Nelson,2003),negativeemotionalstates(e.g.,Loewensteinetal.,2001),andindividualdifferences.Otherresearchsuggeststhatperceptionsofone’sownaccountabilityinagivendecisionscenarioaffectaspectsofrisksensitivity,including

ambiguityaversionandlossaversion(Lerner&Tetlock,1999).TheAsiandiseaseprobleminparticularinvolveshypotheticaldecisionsaboutthewell-beingofothers,andthusdecisionmakinginthisscenariomaybeinpartinfluencedbyperceptionsofaccountability.Furtherresearchisnecessarytoshedlightonproximatemechanismsinvolvedinrisk-sensitivedecisionmakinginframedscenarios,andrisk-sensitivedecisionmakingmoregenerally.

Conclusions

Theresultsofthisstudysuggestthatrisk-sensitivitytheoryelucidatesanormativemechanismofchoicethatfacilitatesfram-ingeffectsonthebasisofminimalacceptablethresholds,orneed.Framingeffectsareoftendescribedasan“irrational”violationofexpectedutilitytheory.However,framingeffectsmayactuallybeby-productsofadaptivecognitivemechanismsthatinfertheeffi-cacyofoutcomesthroughdeterminationofaneed,facilitatingdecisionsthatminimizethepossibilityofloss(i.e.,experiencinganoutcomethatdoesnotmeetone’sneeds;Barrett&Fiddick,1999;Rode&Wang,2000).Mostdecisionsaremadeundersomeconsiderationofneed.Asaconsequence,risk-sensitivitytheoryisusefulinbeingabletoexplainriskychoiceundersuchconditions.Furtherconsiderationofsuchcognitivephenomenaasframingeffectsinlightofecologicallyrelevanttheorizing,suchasrisk-sensitivitytheory,mayleadtoabetterunderstandingofhumancognitivebiasesandhowtheymanifestinsituationsthatbetterapproximateeverydaydecisionmaking.

References

Aktipis,C.A.,&Kurzban,R.(2004).IsHomoeconomicusextinct?VernonSmith,DanielKahnemanandtheevolutionaryperspective.InR.Koppl(Ed.),AdvancesinAustrianeconomics(Vol.7,pp.135–153).Amsterdam,theNetherlands:Elsevier.

Allais,M.(1953).Lecomportementdel’hommerationneldevantlerisque:Critiquedespostulatsetaxiomesdel’e′coleAme′ricaine[Theso-calledAllaisparadoxandrationaldecisionsunderuncertainty].Econometrica,21,503–546.

Allais,M.,&Hagen,O.(1979).Theso-calledAllaisparadoxandrationaldecisionsunderuncertainty.InO.H.M.Allais(Ed.),ExpectedutilityhypothesisandtheAllaisparadox(pp.437–681).Dordrecht,theNeth-erlands:Reidel.

Ariely,D.(2008).Predictablyirrational:Thehiddenforcesthatshapeourdecisions.NewYork,NY:HarperCollins.

Barrett,H.C.,&Fiddick,L.(1999).Evolutionandriskydecisions.TrendsinCognitiveSciences,4,251–252.

Bickel,W.K.,Giordano,L.A.,&Badger,G.J.(2004).Risk-sensitiveforagingtheoryelucidatesriskychoicesmadebyheroinaddicts.Addic-tion,99,855–861.

Caraco,T.,Martindale,S.,&Whittam,T.S.(1980).Anempiricaldem-onstrationofrisk-sensitiveforagingpreferences.AnimalBehavior,28,820–830.

Deditius-Island,H.K.,Szalda-Petree,A.D.,&Kucera,S.C.(2007).Sexdifferencesinrisksensitivityunderpositiveandnegativebudgetsandpredictorsofchoice.TheJournalofGeneralPsychology,134,435–452.Druckman,J.N.(2001).Usingcredibleadvicetoovercomeframingeffects.TheJournalofLaw,Economics,andOrganization,17,62–82.Ellsberg,D.(1961).Risk,ambiguity,andtheSavageaxioms.QuarterlyJournalofEconomics,75,643–669.

Erb,H.,Bioy,A.,&Hilton,D.J.(2002).Choicepreferenceswithout

12

MISHRAANDFIDDICK

inferences:Subconsciousprimingofriskattitudes.JournalofBehav-ioralDecisionMaking,15,251–262.

Ermer,E.,Cosmides,L.,&Tooby,J.(2008).Relativestatusregulatesriskydecision-makingaboutresourcesinmen:Evidencefortheco-evolutionofmotivationandcognition.EvolutionandHumanBehavior,29,106–118.

Friedman,M.,&Savage,L.J.(1948).Theutilityanalysisofchoiceinvolvingrisk.JournalofPoliticalEconomy,56,279–304.

Friedman,M.,&Savage,L.J.(1952).Theexpected-utilityhypothesisandthemeasurabilityofutility.JournalofPoliticalEconomy,60,463–474.Heath,C.,Larrick,R.P.,&Wu,G.(1999).Goalsasreferencepoints.CognitivePsychology,38,79–109.

Kacelnik,A.,&Bateson,M.(1996).Riskytheories—Theeffectsofvarianceonforagingdecisions.AmericanZoologist,36,402–434.Kacelnik,A.,&Bateson,M.(1997).Risk-sensitivity:Crossroadsfortheoriesofdecision-making.TrendsinCognitiveSciences,1,304–309.Kahneman,D.,&Tversky,A.(1979).Prospecttheory:Ananalysisofdecisionunderrisk.Econometrica,47,263–292.

Koszegi,B.,&Rabin,M.(2006).Amodelofreference-dependentpref-erences.QuarterlyJournalofEconomics,121,1133–1166.Ku¨hberger,A.(1998).Theinfluenceofframingonriskydecisions:Ameta-analysis.OrganizationalBehaviorandHumanDecisionPro-cesses,75,23–55.

Leith,K.P.,&Baumeister,R.F.(1996).Whydobadmoodsincreaseself-defeatingbehavior?Emotion,risk-taking,andself-regulation.Jour-nalofPersonalityandSocialPsychology,71,1250–1267.

Lerner,J.S.,&Tetlock,P.E.(1999).Accountingfortheeffectsofaccountability.PsychologicalBulletin,125,255–275.

Levin,I.P.,Gaeth,G.J.,Schreiber,J.,&Lauriola,M.(2002).Anewlookatframingeffects:Distributionofeffectsizes,individualdifferences,andindependenceoftypesofevents.OrganizationalBehaviorandHumanDecisionProcesses,88,411–429.

Levin,I.P.,Schneider,S.L.,&Gaeth,G.J.(1998).Allframesarenotcreatedequal:Atypologyandcriticalanalysisofframingeffects.OrganizationalBehaviorandHumanDecisionProcesses,76,149–188.Levin,I.P.,Schnittjer,S.K.,&Thee,S.L.(1988).Informationframingeffectsinsocialandpersonaldecisions.JournalofExperimentalSocialPsychology,24,520–529.

Loewenstein,G.F.,Weber,E.U.,Hsee,C.K.,&Welch,N.(2001).Riskasfeelings.PsychologicalBulletin,127,267–286.

Lopes,L.L.,&Oden,G.C.(1999).Theroleofaspirationlevelinriskychoice:AcomparisonofcumulativeprospecttheoryandSP/Atheory.JournalofMathematicalPsychology,43,286–313.

March,J.G.,&Shapira,Z.(1992).Variableriskpreferencesandthefocusofattention.PsychologicalReview,99,172–183.

McKenzie,C.R.M.(2003).Rationalmodelsastheories–notstandards–ofbehavior.TrendsinCognitiveSciences,7,403–406.

McKenzie,C.R.M.,&Nelson,J.D.(2003).Whataspeaker’schoiceofframereveals:Referencepoints,frameselection,andframingeffects.PsychonomicBulletin&Review,10,596–602.

Mishra,S.(2012).Towardsageneraltheoryofdecision-makingunderrisk:Integratinginterdisciplinaryperspectives.Manuscriptsubmittedforpublication.

Mishra,S.,Gregson,M.,&Lalumie`re,M.L.(2012).Framingeffectsandrisk-sensitivedecision-making.BritishJournalofPsychology,103,83–97.

Mishra,S.,&Lalumie`re,M.L.(2008).Risktaking,antisocialbehavior,andlifehistories.InJ.Duntley&T.K.Shackelford(Eds.),Evolutionaryforensicpsychology:Darwinianfoundationsofcrimeandlaw(pp.139–159).NewYork,NY:OxfordUniversityPress.Mishra,S.,&Lalumie`re,M.L.(2010).Youcan’talwaysgetwhatyouwant:Themotivationaleffectofneedonrisk-sensitivedecision-making.JournalofExperimentalSocialPsychology,46,605–611.

Mittal,V.,&Ross,W.T.(1998).Theimpactofpositiveandnegativeaffectandissueframingonissueinterpretationandrisktaking.Orga-nizationalBehaviorandHumanDecisionProcesses,76,298–324.Nygren,T.E.(1998).Reactingtoperceivedhigh-andlow-riskwin-loseopportunitiesinariskydecision-makingtask:Isitframingoraffectorboth?MotivationandEmotion,22,73–98.

Pietras,C.J.,&Hackenberg,T.D.(2001).Risk-sensitivechoiceinhumansasafunctionofanearningsbudget.JournaloftheExperimentalAnal-ysisofBehavior,76,1–19.

Pietras,C.J.,Locey,M.L.,&Hackenberg,T.D.(2003).Humanriskychoiceundertemporalconstraints:Testsofanenergy-budgetmodel.JournaloftheExperimentalAnalysisofBehavior,80,59–75.

Rode,C.,Cosmides,L.,Hell,W.,&Tooby,J.(1999).Whenandwhydopeopleavoidunknownprobabilitiesindecisionsunderuncertainty?Testingsomepredictionsfromoptimalforagingtheory.Cognition,72,269–304.

Rode,C.,&Wang,X.T.(2000).Risk-sensitivedecision-makingexaminedwithinanevolutionaryframework.AmericanBehavioralScientist,43,926–939.

Searcy,G.D.,&Pietras,C.J.(2011).Optimalriskychoiceinhumans:Effectsofamountofvariability.BehavioralProcesses,87,88–99.Sher,S.,&McKenzie,C.R.M.(2006).Informationleakagefromlogicallyequivalentframes.Cognition,101,467–494.

Sickar,M.J.,&Highhouse,S.(1998).Lookingcloserattheeffectsofframingonriskychoice:Anitemresponsetheoryanalysis.Organiza-tionalBehaviorandHumanDecisionProcesses,75,75–91.

Stephens,D.W.(1981).Thelogicofrisk-sensitiveforagingpreferences.AnimalBehavior,29,628–629.

Stephens,D.W.,&Krebs,J.R.(1986).Foragingtheory.Princeton,NJ:PrincetonUniversityPress.

Thaler,R.H.,&Johnson,E.J.(1990).Gamblingwiththehousemoneyandtryingtobreakeven:Theeffectsofprioroutcomesonriskychoice.ManagementScience,36,643–660.

Tversky,A.,&Kahneman,D.(1981,January30).Theframingofdeci-sionsandthepsychologyofchoice.Science,211,453–458.

VonNeumann,J.,&Morgenstern,O.(1944).Theoryofgamesandeconomicbehavior.Princeton,NJ:PrincetonUniversityPress.

Wang,X.T.(1996a).Domain-specificrationalityinhumanchoices:Vio-lationsofutilityaxiomsandsocialcontexts.Cognition,60,31–63.Wang,X.T.(1996b).Framingeffects:Dynamicsandtaskdomains.OrganizationalBehaviorandHumanDecisionProcesses,68,145–157.Wang,X.T.(2001).Boundedrationalityofeconomicman:Newfrontiersinevolutionarypsychologyandbioeconomics.JournalofBioeconomics,3,83–89.

Wang,X.T.(2002).Riskasreproductivevariance.EvolutionandHumanBehavior,23,35–57.

Weber,E.U.,Shafir,S.,&Blais,A.(2004).Predictingrisksensitivityinhumansandloweranimals:Riskasvarianceorcoefficientofvariation.PsychologicalReview,111,430–445.

Xie,X.,&Wang,X.T.(2003).Riskperceptionandriskychoice:Situa-tional,informationalanddispositionaleffects.AsianJournalofSocialPsychology,6,117–132.

Yaniv,I.,&Schul,Y.(1997).Eliminationandinclusionproceduresinjudgment.JournalofBehavioralDecisionMaking,10,211–220.

Yaniv,I.,&Schul,Y.(2000).Acceptanceandeliminationproceduresinchoice:Noncomplementarityandtheroleofimpliedstatusquo.Orga-nizationalBehaviorandHumanDecisionProcesses,82,293–313.

ReceivedNovember7,2011RevisionreceivedJanuary31,2012

AcceptedFebruary2,2012Ⅲ

久久建筑網(wǎng)i5h4u.cn提供大量：建筑圖紙、施工方案、工程書(shū)籍、建筑論文、合同表格、標(biāo)準(zhǔn)規(guī)范、CAD圖紙等內(nèi)容。