INTERPRETATIONSOFUTILITYANDTHEIR
IMPLICATIONSFORTHEVALUATIONOFHEALTH*
PaulDolanandDanielKahneman
ThetermÔutilityÕcanbeinterpretedintermsofthehedonicexperienceofanoutcome(experiencedutility)orintermsofthepreferenceordesireforthatoutcome(decisionutility).Itisthissecondinterpretationthatliesattheheartofthemethodsthateconomistshavedevelopedtovaluenon-marketgoods,suchashealth.Inthisarticle,wearguethatdecisionutilityisunlikelytogeneratemeaningfuldataontheutilityassociatedwithdifferentexperiences,andinsteadeconomistsshouldlooktowardsdevelopingmeasuresthatfocusmoredirectlyonexperiencedutility.
Theconceptofutilityiscentraltobothnormativeanddescriptivedebatesineco-nomics.ThewordÔutilityÕhastwodistinctmeanings:itcanrefereithertothehedonicexperienceofanoutcomeortothepreferenceordesireforthatoutcome.Thesehavebeenlabelledexperiencedutilityanddecisionutility,respectively(Kahnemanetal.,1997).JeremyBenthamfirstdefinedutilityinhedonicterms,asameasureofpleasureandpain(Bentham,1789/1948),andeconomistsfollowedthatusageuntilthetwen-tiethcentury.FrancisEdgeworthevenimaginedaÔhedonimeterÕ,aninstrumentthatmeasurestheutilityofmomentsofexperienceandplotsexperiencedutilityasacon-tinuousfunctionoftime(Edgeworth,1881/1967).Heproposedthattheareaunderthecurverepresentstheindividual’stotalhappinessoveragivenperiod.
Economistsabandonedexperiencedutilityearlyinthetwentiethcentury,infavourofanewinterpretation,inwhichutilityrepresentsÔwantabilityÕ(Fisher,1918).Aper-son’sdecisionutilitiesarerevealedbyherchoices.Ofcourse,thetwodefinitionshavethesameextensionifpeoplewantwhattheywilleventuallyenjoy–anassumptionthatisimplicitlyadoptedinmanyeconomicanalyses.Ifweassumethatindividualsarerational,fullyinformedandseektomaximiseutility,thenthechoicestheymakeare,bydefinition,thosethatmaximiseexpectedutility.Neoclassicalwelfareeconomicsrestsonaconceptofdecisionutilitythatiscleansedofanyreferencetohedonicexperienceandwhichassumesthatdecisionutilitycanbeinferredfromthepreferencesthatagentsrevealintheirmarketchoices.
Wheremarketsdonotexist,asinthevaluationofpublicgoodsorofstatesofpersonalhealth,economistshavedevelopedprocedurestomeasuredecisionutilitiesbyelicitinghypotheticalchoices.1Toelicittheutilitiesofstatesofhealth,economistsanddecisionanalystshavedevelopedmethodsthatrequirearespondenttostatethe
*ThisarticlewasdevelopedwhenPaulDolanwasaVisitingResearchScholarattheCenterforHealthandWell-being,PrincetonUniversity,andwethanktheCenterforprovidingsupport.WearealsogratefulforthemanyhelpfulcommentsprovidedbyparticipantsatseminarsatHarvardUniversity,BostonUniversity,UniversityofMichigan,UniversityofMadison-Wisconsin,UniversityofChicago,UniversityofNorthCarolinaandtheMedicalUniversityofSouthCarolina.1Thecontingentvaluationmethodisfrequentlyusedtomeasurethevalueofpublicgoods,inthecontextsofenvironmentallitigationorcost-benefitanalysis.Thismethodinvolvessurveysinwhichrespondentsareaskedtostatetheirwillingnesstopayforagivenbenefit,suchasahypotheticalimprovementintheirhealth,orthecontinuedexistenceofaspeciesofbirds.
[215]
216
THEECONOMICJOURNAL
[JANUARY
probabilitymixoffullhealthanddeaththatmakesthemindifferentbetweenthatgambleandthecertaintyofanintermediatehealthstate–thestandardgamble(SG)method–orelserequiresthemtostatethelengthoftimeinfullhealththattheyconsidertobeequivalenttoalongerperiodoftimeinpoorhealth–thetimetrade-off(TTO)method(Dolan,2000;Byrneetal.,2005).2Theunitofthescaleisaquality-adjustedlifeyear(QALY),whichÔassignstoeachperiodoftimeaweight,rangingfrom0to1,correspondingtothehealth-relatedquality-of-lifeduringthatperiod,whereaweightof1correspondstooptimalhealth,andaweightof0correspondstoahealthstatejudgedtobeequivalenttodeathÕ(WeinsteinandStason,1977).SomehealthsystemsarenowusingQALYstohelpdetermineprioritiesintheallocationofresources.3Themaingoalofthisarticleistoquestionthevalidityofmeasuresofdecisionutilityfortheevaluationoftheweightsassignedtodifferenthealthstates,althoughourcritiqueappliestoanyrevealedorstatedpreferencemethoddesignedtoelicitvaluesfornon-marketgoodsfromrealorhypotheticalchoices.4Themethodsusedtoelicitweightsexpresswhatrespondentswantatthetimeoftheassessmentinrelationtofutureprofilesofhealththatdifferintheirqualityoflifeandriskofdeathorlengthoflife.Assuch,theyareunlikelytomeasuretheutilitycorrespondingtohealthÔduringthatperiodÕ.However,thechoicesthatrespondentsmakeinSGorTTOquestionsshould,tosomegreatextent,beguidedbyaccurateassessmentsoftheutilitycorres-pondingtothehealthstateinquestionoverthespecifiedperiod;seeKahneman(1997)foradiscussionofthisrequirementmoregenerally.ThefocusofourcritiqueofQALYsbasedondecisionutilitiesisthattherearegoodreasonstosupposethatthisrequirementwillnotbesatisfied.
InSection1,wediscusssomeoftheevidencepertainingtotheexperiencedutilitythatflowsfromdifferenthealthstatesandotherimportantfactors,which,onthewhole,suggeststhatpeoplequicklyadapttomany,butnotall,states.BecauseQALYsareprincipallydesignedtoinformresourceallocationinhealthcare,weconsiderinSection2someofthenormativeissuessurroundingtheroleofadaptationinthecontextofallocatingpublicresources.Allelseequal,ifadaptationtooneconditionismorecompletethantoanother,thentheformerwillbegivenlessprioritythanthelatter.Whilstwemayneedtobecautiousaboutusingadaptedpreferencesinsomecontexts,theproblemsassociatedwithallowingforadaptedpreferenceshavebeenoverstated,especiallywhendecidinghowtoallocateresourcesoncethebudgetforhealthcarehasbeendetermined.
Athirdmethod,thevisualanaloguescale,whichrequiresrespondentstoratehealthstatesonascalewithdefinedendpoints,suchas0fordeadand100forfullhealth,hasalsobeenusedbutitisnotfavouredbymosteconomistsanddecisionanalystsinvolvedinmedicaldecisionmakingbecauseitdoesnotrequiretherespondenttomakeachoice.3Forexample,theNationalInstituteforHealthandClinicalExcellence(NICE),whichoffersguidanceontheuseofnewandexistingmedicinesandtreatmentswithintheNationalHealthServiceinEnglandandWales,ismuchlesslikelytorecommendthataninterventionshouldreceivegovernmentreimbursementifthecost-per-QALYofthatinterventionishighascomparedtoifitislow.Whilstnoexplicitthresholdexists,aruleofthumbhasdeveloped,whichsuggeststhatinterventionsthatcostlessthan£30,000perQALYwillberecommendedforfunding,whilstthosethatcostmorethanthiswillnotberecommendedor,ifthecost-per-QALYisexpectedtofallovertime,willberecommendedforfundingasÔexperimentaltreatmentsÕ.4Ourcritiqueisfocusedontheuseofmeasuresofdecisionutilitytoelicitvaluesofthiskind,ratherthantheirusefulnessinothercontexts,suchaspredictingbehaviour.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
22008]
INTERPRETATIONSOFUTILITYANDVALUINGHEALTH
217
ForecastsoffutureutilityshouldthereforetakedueaccountofadaptationbuttheevidencewepresentinSection3suggeststhattheintuitiveforecastsoflaypeoplegenerallyfailtodoso.InSection4,wediscusstheresultsfromhealthstatevaluationstudies,whichareconsistentwiththeideathatpeopleunderestimatetheextenttowhichtheyandotherswilladapttochangedcircumstances.Inparticular,weconsiderhowthoseaskedtoimaginewhatitwouldbeliketobeincertainhealthstates(ÔthepublicÕ)havetheirattentiondrawnawayfromthepossibilityofadaptationandinsteadtowardstransitionalchangesinthehealthdomainalone,andtowardstheirimmediateaffectiveresponsetothehealthstateinquestion.
Theseconsiderationsmightleadonetoconcludethatdecisionutilitiesshouldbeelicitedfromthosewithdirectexperienceofthehealthstatesinquestion(ÔpatientsÕ).However,decisionutilitieswillalwaysreflectthefocusoftherespondent’sattentionatthetimeoftheassessment,ratherthanwhattheywillattendtowhileexperiencingaparticularhealthstate.PatientsÕdecisionutilitiesmaybefreeofsomeofthebiasesassociatedwithpublicvaluesbuttheydonottakedueaccountofanylossesassociatedwithadaptationthatmayhavealreadytakenplace.Whilstthepublicmayoverestimatethelossesassociatedwithagivenstateofhealth,patientsmayunderestimatesuchlossesand,importantlyinapolicycontext,therelativerankingofdifferenthealthstatesmaywellvaryfromoneanother–andfromtherankingsimpliedbyexperiencedutility.Patientscouldbeaskedtoconsidertheirpreviousexperienceswhenmakinghypo-theticalchoicesaboutthefuturebutthereisalsoevidencethatpeoplearenotverymuchbetteratrememberingtheimpactofpastexperiencesthantheyareatpredictingtheimpactoffutureexperiences.
Againstthisbackground,weareratherpessimisticabouttheuseofdecisionutilitytogenerateQALYs,althoughourcritiqueofdecisionutilitymaypromptotherswhoaremoresympathetictothisinterpretationofutilitytorefineanddevelopthemethodsofpreferenceelicitation.InSection5,wearguethatmoreelaboratestudiesdesignedtoelicitdecisionutilitiescannotovercomethefundamentalproblemwithsuchutilities,whichisthattheydonotaccuratelyrepresenttheutilitystreamsassociatedwithdifferenthealthstates.Wethereforesuggestthateconomistsshouldinsteadlooktowardsdevelopingmeasuresthatfocusmoredirectlyonexperiencedutility,andweprovidesomesuggestionsabouthowthismightbedone.Ourrecommendations,whichinvolveelicitingproxyvaluesforutilityasitisexpe-riencedmoment-to-moment,willalsoprovideeconomistsinvolvedinestimatingthewelfareeffectsofnon-marketgoodsbesideshealthwithanalternativetoexistingrevealedandstatedpreferencemethods.InSection6,weprovidesomeconcludingremarks.
1.ExperiencedUtility
Muchoftheevidenceontheimpactofarangeoffactorsonutilitycanbesummedupinoneword–adaptation.Adaptation–theprocessofadjustmenttoneworchangedcircumstances–occursatdifferentlevelsandindifferentways,rangingfrommolecularchangesatthecellularlevelthatdiminishtheperceivedorexperi-encedintensityofanobjectivestimulus(suchasmovingfromlighttodark)toovertbehaviourthatreducesexposuretothestimulus.Hedonicadaptationoccurswhen
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
218
THEECONOMICJOURNAL
[JANUARY
thereisaÔreductionintheaffectiveintensityoffavourableandunfavourablecir-cumstancesÕ(FrederickandLoewenstein,1999).Suchadaptationincludesbothsensoryadaptatione.g.adjustingtoanunpleasantsmellandcognitiveadaptatione.g.changesininterests,goals,values,or–importantlyformuchofthediscussionthatfollows–redeploymentofattention.
Inawidelycitedstudy,Brickmanetal.(1978)foundthataccidentvictimswhosustainedparaplegiaorquadriplegiainthelastyearreportedhappinesslevelsthatwere,onaverage,closertothoseofacontrolgroupthanmighthavebeenexpected(2.96ona0–5scaleforparaplegicsascomparedto3.82forcontrols).SchulzandDecker(1985)foundthatreportedhappinesslevelsofagroupofmiddle-agedandelderlyparaplegicsandquadriplegicswereonlyslightlylowerthanpopulationmeansofnon-disabledpeopleofasimilarage.Similarly,WortmanandSilver(1987)foundthatquadriplegicsreportednogreaterfrequencyofnegativefeelingsthancontrols.
Tyc(1992)foundnodifferenceinqualityoflifeorpsychiatricsymptomatologyinpatientswhohadlostlimbstocancerascomparedtoacontrolgroup.Inareviewofstudiesexaminingadaptationtoburninjuries,Pattersonetal.(1993)foundhighlevelsofpsychosocialadaptationbyoneyearafteraccident.Morerecently,Wu(2001)foundthatthosewhohavehadaheartconditioninthepastarelesslikelytoreportworseself-assessedhealthandemotionalhealthduetotheonsetofnewconditionthanthosewhohavenotpreviouslyhadexposuretohearttrouble,which,itisclaimed,isÔquitesup-portiveofthetheoryofhedonicadaptationÕ.
Whilstadaptationtochangedhealthappearswidespread,itiscertainlynotuni-versal.Thereis,forinstance,evidenceofincreasedsensitisationtopain(Thompsonetal.,1973).Inadiarystudyoverfourweeksof80patientswithunexplainedpain,Petersetal.(2000)foundthatpatientswithlessthansixmonthsofpainreportedsignificantlylesspainintensity,disabilityandfatiguethanpatientswhosepainhadpersistedformorethansixmonths.Moreover,thereisalsoevidencethatcopingwithrepeatedepisodesofpainleavespatientsmorevulnerabletostressfulevents(Lennonetal.,1990).Thereissomeevidencethatpeopledonotadapttopro-gressivediseases–seeLivnehandAntonak(1994)indegenerativedisordersandAntonakandLivneh(1995)inmultiplesclerosis–butinterpretingtheseresultsisproblematicbecausethehealthstatesassociatedwithsuchconditionsareconstantlychangingand,therefore,thehedonicstatecouldbedeterioratingataslowerratethanthecondition,whichwouldstillbeconsistentwithadaptiveprocesses(FrederickandLoewenstein,1999).Thereisgenerallytheneedformorelongitudinalstudiesthatcontrolfortheeffectsofconfoundingvariables.
Itcouldalsobethatsomeoftheseresultsareexplainedbyresponseshift(SprangersandSchwartz,1999).Paraplegics,forexample,mightcomparetheirhappinesstootherparaplegics,elevatetheircurrentratingstoreflectthecontrastwiththeextremedespairimmediatelyfollowingtheonsetofdisability,oradoptlowerstandardsfortheintensityofpositiveaffect,allofwhichwouldleadtooverinterpretationofthedegreeofadaptation.Aspossibleevidenceofthis,intheBrickmanetal.(1978)study,theaccidentvictimsrememberedtheirhappinessashavingbeenmuchhigherinthepastthandidthecontrols(4.41ascomparedto3.32,respectively).Similarly,PostulartandAdang(2000)showthatkidney-pancreas
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
2008]
INTERPRETATIONSOFUTILITYANDVALUINGHEALTH
219
transplantpatientsremembertheirpre-transplantqualityoflifetobelowerthantheyreportedatthetime.5However,itisalsoentirelypossibletheparaplegicsetc.,whenaskedtoimaginetheirlifebeforeanadversechangeintheirhealth,mayfocusonthosethingsthatmadetheirlifedifferenttonowratherthanonthosemanythingsthathavebeenunaffectedbyparaplegiaetc.Andwhilstresponseshiftmakesintertemporalandinterpersonalcomparisonsofself-reportsproblematic,itcannotexplainallchangesinpreferencesthattakeplace.Forexample,thereisstrongevidenceofadaptationevenwhenphys-iologicalorbehaviouralmeasuresareused,bothofwhichshouldbelesspronetoresponseshift:Krupat(1974)foundthatthatpriorexposuretothreatreducedgalvanicskinconductance(aphysiologicalmeasureofthreat);andDaretal.(1995)foundthatwarveteranswithmoreseverepastinjuriescouldholdtheirfingerinhotwaterforlongerbeforeclassifyingitaspainfulthanveteranswithlessseverepastinjuries.
ThereisnowevidencetoshowthatindividualsÕlifesatisfactionadaptstochangesinanumberofotherfactorsandlifeevents.Thereisasmallpositiverelationshipbetweenincomeandhappinessincross-sectionalanalysesbutvirtuallynorelationshipatallintime-seriesanalyses(Easterlin,2001).6ThereisevidencethattheincomeanindividualconsiderstobeÔsufficientÕisprimarilydeterminedbyhercurrentincome(vanPraagandFerrer-i-Carbonell,2004),andthatadaptationappearstooffsetabouttwo-thirdsofthebenefitsofanyincreaseinincome(FreyandStutzer,2002).Usingdatafroma15-yearstudyofover24,000Germans,Lucasetal.(2003)showthat,onaverage,peopleexperienceanincreaseinhappinessintheyearssurroundingmarriagebutafterthesecondyearofmarriagetheyappeartoreturntotheirbaseline.Eveninthecaseofwidowhood,adaptationisclosetocompleteafterabouteightyears.Thesedataaregenerallysupportiveoftheideathatpeopleareonahedonictreadmill(BrickmanandCampbell,1971).However,itisworthnotingthattheaverageresultsmasksomeimportantindividualdifferences.Inparticular,thosewhoreactedstrongly(eitherpositivelyornegatively)werestillfarfrombaselinelevelsyearsaftertheevent.
ThesameGermandataalsohighlightthepointthatadaptationisnotfoundforallconditions.Inthecaseofunemployment,forexample,averagelifesatisfactionfallsfromaround7.2onascalefrom1–10tothe6.3inthefirstyearandisstillonly6.5inthefourthyearofunemployment(Clarketal.,2004).Thereisalsoevidencetosuggestthatpeopledonotadapttonoisethatispoorlyunderstoodorunpredictableinitstiming.7Inastudyoffirstyearcollegestudents,Weinstein(1978)foundthatannoyancewith
Tooffsetproblemswithresponseshiftpartially,respondentsarenowsometimesaskedtorateastan-dardisedvignettealongsideratingsofsatisfactionwithownhealth,sothatthelattercanberecalibratedagainsttheformer(Salomonetal.,2004).Suchanapproachservestoincreasetheagegradientinself-reportedhealth;thatis,self-reportedhealthdeclinesmorerapidlywithageonceeachindividual’shealthratinghasbeenrecalibratedagainsthis/herratingofthevignette.6Whilstthereareotherpossibleexplanationsforthisapparentparadox(i.e.somerelationshipbetweenincomeandlifesatisfactionatonepointintimeandnorelationshipovertime),itislikelythathedonicadaptationisplayingabigpart.OneotherpossibleexplanationisthattheriseinincomeovertimehasbroughtwithiteconomicÔbadsÕ(suchasgreaterincomeinequality,risingcrimeandhigherdivorcerates),whichoffsetanyincreaseinlifesatisfactionfromanincreaseinincome.7Whennoiseispredictable,hasbeenheardmanytimesbefore,changeslittlefromonetimetothenext,andiseasilyassimilatedtopeople’spriorschemas(e.g.thesoundofaclockticking),thereislikelytobeadaptation;seeWilsonandGilbert(2005),andhowthisevidenceisconsistentwiththeirAREAmodelofemotionaladaptationinwhichpeopleattendtonovelevents,reacttothoseevents,explaintheevents,andasaresultadapttothem.
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
5220
THEECONOMICJOURNAL
[JANUARY
noiseincollegeincreased.Thesameauthoralsoobservedincreasingpessimismaboutadaptationtohighwaynoise:afterfourmonths,underone-thirdspontaneouslymen-tionednoiseassomethingtheydislikedintheneighbourhood,whereasoveronehalfdidafter16months(Weinstein,1982).
2.AdaptationinaPolicyContext
Itwouldseemthatadaptationtomany–butnotall–conditionsiswidespread.Thefactthatpeople’spreferencesundergosomeformoftransformationinaprocessofadaptationwouldnot,normally,beregardedasanymoresignificantthanthechangestopreferencesovertimethatmayoccurbecauseofalltheotherexperiencesthattheyencounter.Butinthecontextofresourceallocation,adaptationraisesanormativeproblembecause,allelseequal,themoreapatientadaptstohercondition,thelesspriorityshewillreceiveinthecompetitionforresourcesthatimprovequalityoflife.Intheextremecase,wherethereiscompleteadaptationtoahealthstate(suchthatitgetsaquality-adjustmentweightof1.0),therecanbenoincreaseinutilityfromitstreat-mentorcure.8Theextenttowhichgivinglowerprioritytothosewhohaveadaptedmostisunjustwilldependlargelyonelementsoftheadaptationprocess;seeMenzeletal.(2003)foradetaileddiscussion.Sen(1992)hasdiscussedoneparticularlyregrettableelementofadaptation;namely,entrencheddeprivation.Inhiscritiqueofutilitarianism,heclaimsthatÔdesirefulfilmentÕisÔneglectfuloftheclaimsofthosewhoaretoosubduedorbrokentohavethecouragetodesiremuch...Insituationsoflong-standingdepriva-tion,thevictimsdonotgoongrievingandlamentingallthetime...Theextentofaperson’sdeprivation,then,maynotatallshowupinthemetricofdesirefulfilment...ÕHowever,Senisconcernedwithaspecialcase.PeoplewhoadapttochangedhealthstatusarenotgenerallyÔsubduedorbrokenÕ,andtheirtastesandpreferencesshouldnotbeignored.WhilstSen’sargumentforcesustopauseabouttooreadilyusingadaptedutilities,itdoesnotjustifyarejectionofvaluesshapedbyadaptation.
Theuseofadaptedvaluesraisesarealdilemmawhenpatientssuccessfullyadjusttheiractivitiesortheirgoalsinlinewiththeirchangedcircumstances(Menzeletal.,2003).Paraplegics,forexample,maytakeupaerobicwheelchairingiftheystilldesirephysicalexerciseortheymaydevelopaninterestinmusictoreplaceapreviousinterestinphys-icalactivity.Inlightofsuchlaudableeffortandachievement,itmaybeconsideredunjusttowithholdtreatmentfromparaplegicsonthegroundsthatthepotentialforgainsinexperiencedutilityarelimited.However,ifthisargumentisaccepted,thenanadvantagegetscreatedforthosewhohaveadapted.Treatmentsforsuchpeoplewillgetgreaterprioritythanarewarrantedbythesizeoftheactualutilitygainfromthem.ResourceallocationdecisionswillthenbemadeasifadaptedpatientsÕgainsinexperiencedutilitycountformorethanthegainsofpatientswhoadaptless.Thisalsoseemsunfair.
Theoppositeistrue,ofcourse,forlife-savingorlife-extendinginterventions,wherecompleteadaptationtoahealthstatewouldmeanthatthenumberofQALYsgainedfromsavingsomeone’slifeinthatstatewould,allelseequal,bethesameasthenumberofQALYsgainedfromsavingthelifeofsomeoneinfullhealth.TheincreasedprioritygiventoadaptedpatientsthatcomesfromtheuseofÔadaptedutilitiesÕinlife-extendingcontextsisoftenignoredbythosewhoargueagainsttheuseofsuchutilitiesbutitisinthecontextoflife-enhancingtreatmentsthattheissueofadaptationpresentsuswithapotentialproblem.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
82008]
INTERPRETATIONSOFUTILITYANDVALUINGHEALTH
221
HereinliesaÔvexingmoralproblemÕ(Murray,1996)thatwecannothopetoresolvehere.But,ingeneral,itseemsentirelyappropriatetogivegreaterprioritytothosestatesthatpeopledonotadapttooverthosethattheydoadaptto.Thiswouldseemtobeparticularlytruewhenallocatingresourcesamongstpatientsoncethebudgetforhealthcarehasbeendeterminedi.e.oncewehavedecidedthepriorityaffordedtopatientsinrelationtoothergroups.Giventhis,weneedtoconsiderhowwellpeoplepredictchanges–includinganyadaptation–intheirfuturepreferences.
3.PredictionsAboutExperiencedUtility
Howaccuratelydopeoplepredicttheirfutureutilityandhowwelldotheypredictadaptationthatisrelevantforresourceallocation?Theevidenceonthesequestionscomesfromvarioussourcesanditisnotveryencouraging.Forexample,Sieffetal.(1999)askedpeopletestingforHIVtoratethedegreetowhichtheyagreedordis-agreedwithanumberofitemsaboutaffecte.g.ÔIfeelangrymostofthetimeÕ.Theresponseswerestandardisedona0–100scale,where100representscompletedistress.Theanticipateddistressfromapositivescorewas95,whereastheactualdistresswas78.Theanticipateddistressfromapositivescorewas49,whereastheactualdistresswas55.Smithetal.(2006)reportdatafromcolostomypatients,halfofwhomhadtheircolostomiesreversedatsomepoint.Thecurrentqualityofliferatingona0–100scaledidnotdifferbetweenthesetwogroups(71forformerpatients,comparedto67forcurrentpatients)butthecurrentpatientspredictedthattheirqualityoflifewithnocolostomywouldbe83i.e.12pointshigherthanwhatwasactuallyreportedbytheformerpatients.Itis,however,difficulttodeterminewhetherthesedifferences,whicharestatisticallysignificant,aresignificantinaÔutilitylossÕsense,andbettermethodsofmeasuringexperiencedutilityarerequiredbeforewecouldmakethisclaim(seeSection5below).
AstudyoflaryngealcancerpatientsprovidesevidencethatthedecisionutilitiesattachedtoÔendoftherapyÕhealthstates(asmeasuredusingtheTTOmethod)aresimilarinanticipationofthosestatesandwhentheyareexperienced(Llewellyn-Tho-masetal.,1993).However,thetimeintervalbetweenevaluationswasonlyninedaysanditispossiblethatsomerespondentsrememberedtheirearliervaluations.Importantly,thereweresignificantdifferencesintheexpecteddirectionforthosereportingthemostsevereendoftherapystates.Itisnowwidelyrecognisedthatitischangesinstates,ratherthanthestatesthemselves,thatarethecarriersofutility(KahnemanandTversky,1979),andtheresultshereareconsistentwiththeideathatpeopleareabletopredicttheiremotionalresponsetosuchchangeswithafairdegreeofaccuracy,atleastovertheveryshortterm.
Instudiesoutsideofhealth,itseemsthatpeoplegenerallyfailtorecognisetheimportanceofarangeofadaptationprocesses,expectingbothgoodandbadfeelingstolastalotlongerthantheyreallydo.Forexample,Gilbertetal.(2002a)askedvotersinTexasduringthe1990electionforgovernor(whichwaswonbyGeorgeW.Bush)howtheywouldfeelonemonthaftertheelectioniftheircandidatehadlost.Respondentsexpectedtofeelmiserablebutwhenaskedhowtheyfeltonemonthlater,peoplewerejustashappywhethertheircandidatehadwonorlost.Theseandotherresultssuggestthat,whilstpeoplecangenerallypredictthevalenceandtypeofemotion
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
222
THEECONOMICJOURNAL
[JANUARY
fromeventsreasonablywell,theyoverestimatetheintensityandespeciallythedurationoftheirreactionstothoseevents(WilsonandGilbert,2003).
OneimportantreasonforthisisthatwefailtoappreciateourabilitytoÔmakesenseÕofthethingsthathappentous(WilsonandGilbert,2003).9Inrelationtoadaptationtonegativeevents,suchasadversechangesinhealth,WilsonandGilbert(2003)suggestthatwefailtoappreciatethepowerofourpsychologicalimmunesystem(PIS),whichdetectsandneutraliseseventsthatchallengeoursenseofwell-being.Ofcourse,someadverseevents,likeparaplegia,mayresultinpermanentlossesinwell-beingbutthoselosseswouldbealotworseifwedidnotpossesspsychologicaldefencesthathastenourrecoveryfromthem.AswithÔsensemakingÕmoregenerally,thePISfunctionslargelyoutsideofconsciousawarenessandismoreeffectivebyoperatingÔbehindthescenesÕ.Itislittlewonder,then,thatwefailtoanticipateit.
Inastudyoflaypredictionsofadaptation,Cohn(1999)–seeKahneman(2000)–asked362respondentstoevaluatethewell-beingoffictitiousmembersofvariouscat-egoriesofpeople,includinglotterywinnersandparaplegics.Halfoftherespondentsweretoldthattheevent(winningthelottery,becomingparaplegic)hadoccurredonemonthbefore,andtheotherhalfweretoldthateventhadoccurredoneyearbefore.Respondentswerealsoaskedtoindicatewhethertheypersonallyknewalotterywinneroraparaplegic.Thosewhodidnotknowalotterywinneroraparaplegicwerelargelyinsensitivetothetimevariable.Thosewhoknewalotterywinnerorparaplegic,how-ever,weremuchmoresensitivetothetimecourseoftheevent:thebetween-groupscomparisonshowedthattheserespondentscorrectlypredictedthatlotterywinnerswouldbecomelesshappyandparaplegicslessmiserableovertime.Thetwogroupsdidnotdiffersignificantlyintheirjudgmentsofwell-beingonemonthaftertheevent,whichsupportstheideathatpeoplearebetterabletopredictinitialchangesinutility.
4.DecisionUtilitiesinHealth
Overthelastthirtyyearsorso,utilitiesformanydifferentstatesofhealthhavebeenelicitedfrompatientsandthepublic.10Consistentwiththeevidencepresentedabove,thereisnowplentyofevidencetosuggestthatmembersofthegeneralpublicover-estimatethelosses(andunderestimatetheadaptation)associatedwitharangeofhealthstates.Inthefirstempiricalstudyonthisissue,SackettandTorrance(1978)askedthepublicandhomedialysispatientstovaluearemaininglifetimewithchronicdialysisusingtheTTOmethod.Theaverageutilityfromthegeneralpublicwas0.39,ascomparedto0.56fromdialysispatients.Ifitwerepossibletoreturndialysispatientstofullhealth,thenthegainwouldbe0.61QALYsperyear(1.0–0.39)ifweusedpublicutilitiesand0.44ifweusedpatientutilities.Inthiscase,usingpublicvalueswouldreducethecost-per-QALYoforgantransplantationbynearly40%.
FrederickandLoewenstein(1999)notethepossibilitythatbothadaptationandsense-makingcouldbespuriouslycorrelatedthroughtheircommonrelationtoathirdfactor,suchasinnatehappiness,whichcausespeopletobothadjustsuccessfullytotheirnewconditionandtomakesenseofit.However,thisislessplausibleinlightofstudiesbyWilsonetal.(2005),whichmanipulatedtheeaseofsensemakingandfoundthattheharderitwastomakesenseofagoodevent,thelongertheaffectivereactionlasted.10Thegeneralpublicis,ofcourse,aheterogeneousgroupintermsofitshealthandwillcontainmanypeoplewhosehealthisverypoorbut,onaverage,thegeneralpublicareinbetterhealthstatesthantheonestheyareaskedtoimagineexperiencinginmostvaluationstudies.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
92008]
INTERPRETATIONSOFUTILITYANDVALUINGHEALTH
223
Sincethen,whilstsomestudieshavefoundlittledifferenceinpublicandpatientvaluations,mosthaveproducedsimilarresultstothosefoundbySackettandTorrance,withsimilarlyimportantimplicationsfortheresultsfromcost-per-QALYratios.11Inareviewof39studies,includingtheirown,deWitetal.(2000)foundthat23studiesreportpatientvaluestobehigherthanpublicones,2reportpublicvaluestobehigher,11reportnodifferenceand3reportcontradictoryfindings.Itisnotclearwhytherearedifferencesacrossstudies,althoughthesmallsamplesinsomeofthestudiesmighthelptoexplainwhysomeofthemreportnodifferences.Interestingly,thetwostudieswherepatientvaluesareactuallylowerthanpubliconesareinsamplesofmenopausalwomenandwomeninchildbirth.
4.1.UtilitiesfromthePublic
Thereareatleastthreeotherfactorsthattendtoreducethepublic’sassessmentsofhealthstatesthataredifferenttotheirown,allofwhichdrawrespondentsÕattentionawayfromthepossibilityofadaptation.First,attentionisdrawntothetransitionfromonehealthstatetoanotherandthetransitorychangeinwell-beingthatwillresult.So,whenpeopleareaskedtovalueparaplegia,theywilltendtofocusonbecomingaparaplegic,whichwillinitiallybethefocusofmuchattention–theywillbeaparaplegicÔfull-timeÕ–andhencethesourceofmuchmisery.Butafterthistransitionalperiod,aparaplegicwillonlybeaparaplegicÔpart-timeÕ,astheyattendtootherthingsintheirlifethatareunaffectedbytheirparaplegia.SovaluationsarelikelytobeaffectedbyaÔPeak-StartRuleÕ(DolanandWhite,2006),whererespondentsfocusonthepeaklossandtheimmediateloss(whichinmostcasesarelikelytooccuratthesametime).Thisisalsoconsistentwiththeideathatchangesinstatesaretherealcarriersofutility(KahnemanandTversky,1979).
Second,andevenallowingforthetransitionphase,attentionisfocusedonthehealthdomainratherthanonotherdomains(suchaspersonalrelationships)whichmaybeunaffected,orevenenhanced,bychangedhealthstatus.Healthstatesaretypicallydescribedusingonlyalimitednumberofdimensionsandalwaysinwaysthatdrawstherespondent’sattentiontothosedimensionsthatwillbeadverselyaffected.Butoneofthewaysinwhichweadapttochangedcircumstancesisbyredeployingourattention.So,ifadaptationtoparaplegiatakestheformofnotthinkingaboutit(butratherthinkingaboutdomainsoflifeotherthanhealth),thiswillnotbepartofthewayanyonethinksaboutparaplegiainadvanceoftheevent.
Third,itispossiblethatSGandTTOresponsesreflectimmediateaffectivereactionstothehealthstateinquestion(Wilsonetal.,2002),whichinthecaseofsomeseverehealthstatesislikelytobeaninitialshockreactionto,orfearassociatedwith,thatstate.So,notonlymightrespondentsbechannelledtoconsideralimitednumberof(pos-siblyrelativelyunimportant)aspectsofthefuture,theymightevenbechannelledawayfromthinkingaboutthefutureatallandtowardsfocusingoncurrentfeelings.From
Thereisthepossibilitythatsomeoftheseresultscouldbeexplainedbyresponseshift(Ubeletal.,2003).However,Baronetal.(2003)foundthatmakingtheresponsescalesmoreprecisewithwell-defineddemar-cationsservedonlytoincreasethediscrepancybetweenthevaluesofpatientsandthepublic.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
11224
THEECONOMICJOURNAL
[JANUARY
thisperspective,theapparentfailuretotakeaccountofadaptationisaproductofthewayinwhichcurrentemotionsintrudeonassessmentsofthefuture,aswellasanyadditionalfailuretofullyconsiderthatfuture.12Ofcourse,policymakersmaywishtodevoteresourcestothehealthstatesthatpeoplefearthemost.Fearhasaveryrealeffectonanindividual’sutilityandonherbehaviour,includingherconsumption.Insofarasfearssurroundingparticularhealthstatescanbereducedbyexpenditureonthetreatmentandpreventionofthosestates,policymakersmaybeabletoincreaseutilitybysuchexpenditures(DolanandWhite,2006).Moreover,thepublic,throughthedemocraticprocessorpressuregroups,maydemandthatcertainhealthstates,suchasthoseassociatedwithcertainkindsofcancer,begivengreaterpriority.However,accountingforfearisaquiteseparateissuefromaccountingforthelossesinutilityfromagivenhealthstate,andSGandTTOutilitiesconflatefearsthatpeoplehaveaboutexperiencingpoorhealthwiththeirassessmentsofhowtheirliveswillbeaffectedbypoorhealth.13Inanyevent,weareunawareofanyeconomistsarguingfortheSGorTTOonthegroundsthattheypickuppeople’slegitimateaffectiveresponsestothehealthstatesinquestion.Indeed,economistshavedescribedhealthintermsofdimensionsofhealth(mobilityetc.)ratherthanintermsofconditions,likecancer,inordertoavoidintroducingtoomuchemotionintotheresponses.TheSGandTTOareassumedtotapintopeople’scognitiveassessmentsoftheutilityassociatedwiththosestates,despitethelackofanyevidencetosupportsuchanassumption.Respondentsthemselvesmaythinkthattheyaregivingaconsideredresponsetotheutilityassessmentquestionbutmayinfactbeusingtheirimmediatefearofthehealthstateasaproxyfortheirfutureassessmentofit(GilbertandWilson,2000).
FocusingrespondentsÕattentiononthetransitionfromonestatetoanother,emphasisingthenegativeconsequencesforhealth,andpickingupsomeaffectiveresponsetothechangeinhealth,maytogetherexplainwhymanystatesareconsideredtobeworsethandeathinthosestudiesthatarebeingusedtocalculateQALYsforuseinpolicysettings.14Moreover,notonlydoesitappearthatrespondentsfailtoantici-patehowtheywilleventuallyadapttomanyadversehealthstates,itseemsthattheythinkmanystateswillgetworsethelongertheylast.Forexample,BleichrodtandJohannesson(1997)foundthatSGvaluesforstateslastingtenyearswerehigherthanwhenthesamestateslastedfor30yearsandSackettandTorrancefoundthatTTOvaluesdeclinedoverdurationsofthreemonths,eightyears,andtheremaininglifeexpectancyoftherespondent.Insomecases,ahealthstatethatisconsideredtobe
Aclassicexampleofthewaycurrentfeelingsaffectdecisionsaboutthefutureisshoppingatthesupermarketwhenhungry:whenpeoplearehungry,theytendtoshopasistheyexpecttoremainperma-nentlyhungry,butshopperswhoaregivenamuffintoeatbeforeenteringtheshoparemorelikelytolimittheirpurchasestothoseitemsontheirshoppinglist(Gilbertetal.,2002b).13Foradiscussionoftheroleoffearassessmentinthecontextofeconomicevaluation,seeAdler(2004),andforarelateddiscussioninthecontextofimaginedrisks,seeSunstein(2002).14Forexample,theEQ-5D,whichisusedwidelyinevaluativestudies,generates243possiblestatesofhealth(fivedimensionsofhealth,eachwiththreelevelsofseverity)andone-thirdoftheseare,onaverage,consideredtobeworsethandeadbyaUKgeneralpopulationsamplethatvaluedthesestatesusingtheTTOmethod(Dolan,1997).
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
122008]
INTERPRETATIONSOFUTILITYANDVALUINGHEALTH
225
betterthandeadforashorterdurationisseenasbeingworsethandeadwhenitlastsforlonger(DolanandStalmeier,2003).154.2.UtilitiesfromPatients
ElicitingSGandTTOutilitiesfromthosecurrentlyexperiencingthehealthstateinquestionwillavoidsomeoftheproblemsassociatedwithelicitingutilitiesforhypo-theticalstatesfromthepublic.However,thefundamentalproblemwithutilitieselicitedinadecisioncontextremains;namely,thattheresponseswillreflectwhatevertherespondent’sattentionisdrawntoatthetimeoftheassessmentratherthanwhatitwillbedrawntoinfutureexperiences.Patientsmaywellhaveexperienceofthestatetheyarebeingaskedtovaluebuttheycanbeexpectedtoconsideronlyalimitedsub-setofpossiblefutureexperiencesintheutilityelicitationtask.
Inaddition,theSGandTTOmethodsbothrequirepatientstoconsiderhowtheirfutureexperienceswouldbedifferentweretheytobeinfullhealth.Whilstmanypatientswouldhavehadpreviousexperienceoffullhealth,theirrecollectionofthis–atleastinanevaluativesense–maybefarfromperfect(seebelow)andtheyareagainlikelytofocusononlyalimitednumberofwaysinwhichtheirliveswouldbedifferentfromnow.Differentpatientsmayfocusondifferentaspectsoftheirfuturelivesandmayhaveverydifferentconstructionsofwhatfullhealthwouldmeantothem,andthesedifferentconceptualisationscouldgosomewaytowardsexplainingthewidevariationinutilitieselicitedfromreasonablyhomogeneousgroupsofpatients;seedeWitetal.(2000).
Moreover,tobeofuseinapolicysetting,decisionutilitiesfrompatientswillneedtoreflecthowbeinginthatstateimpactsontheaveragepatient’slife,asitwillbeexperiencedinthefuturebythatpatient.Mostpatientswillexperiencesomeinitialutilitylosseveniftheyfullyadapttotheirhealthstate.Adaptedpatientsdonothavetheirattentiondrawntohowthestateimpactedupontheminthepastandso,tofullycapturethis,decisionutilitieswouldneedtobeelicitedfrompatientsateverystageofthecondition’sprogression.16So,evenifpatientswereabletoforecastaccuratelyhowtheircurrenthealthwouldaffecttheminthefuture,theirdecisionutilitieswouldnotfullyreflecthowthatstatehadaffectedtheminthepast.
Itmightbepossibletoaskcurrentpatientstogivedueconsiderationtohowthestateimpactedupontheminthepastwhenconsideringtheirdecisionutilitiesaboutthefuture.Evenifpatientsreconstructthetimecourseoftheirconditionmoreorlessaccurately,17theirdecisionutilityforhealthwillbeaffectedbyhowtheyrecallthepreviouschangeintheirhealthchangingtheirlife.Patientsarelikelytousetheir
TheconditionsoflogicalrationalityrequiredbytheQALYmodelhavenotbeendiscussedinthisarticle,asanextensiveliteratureonthisissueexistsalready(TsuchiyaandDolan,2005)butevidencesuchasthisviolatestheassumptionthatthevalueofahealthstateisunaffectedbyitsduration.16Despitethefactthatadaptationtakestime,theliteraturehaslargelybeensilentontheissueofwhenintheadaptationprocesspreferencesshouldbeelicited(Dolan,1999andSharmaetal.,2004,representnotableexceptions).17Thereissomeevidencethatpatientshaveatendencytounderestimatethetrueincidenceofprevioussomaticsymptoms(SimonandGureje,1999)andtocombineseparateepisodesintoasingleone(Meansetal.,1989).
ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
15226
THEECONOMICJOURNAL
[JANUARY
currentpreferencestorationaliseapreviouschange(Wilsonetal.,2003),sotheymaywellrememberthetransitionintotheircurrenthealthstateasbeinglessintensethanitwasfelttobeatthetime.Aspossibleevidenceofthis,Barsky(2002)reportsonanumberofstudiesthatshowthatretrospectiverecallofhealthishighlycorrelatedwithindividualsÕcurrenthealthstateandnotsowellcorrelatedwiththeirinitialstate.Moregenerally,thereisnowplentyofevidencetosuggestthatourmemoriesdonotrecallpastutilitiesandtheirdurationparticularlywell.Forexample,RedelmeierandKahneman(1996)askedpatientsundergoingacolonoscopytoreporttheirlevelofpaineverysixtysecondsthroughouttheprocedureandtosubsequentlyratetheÔtotalamountofpainexperiencedÕonasimilarscale.ThecorrelationsbetweentheÔon-lineÕandglobalratingssuggestedthatrespondentsÕmemoriesoftheexperiencewereinfluencedprimarilybythemostpainfulmomentoftheprocedureandthelevelofpainattheveryendoftheprocedure:thedurationoftheprocedurewaslargelyignored.Therefore,itappearsthatpatientsuseaÔPeak-EndRuleÕ,whichignoresthefullsetofexperiencesandhowlongtheseexperienceslast(Kahnemanetal.,1997).4.3.IllustratingtheProblemwithDecisionUtilities
Givenallofthis,itisentirelypossiblethatdecisionutilitiesfromthepublicwillleadtoadifferentrankingofhealthstatestodecisionutilitiesfrompatientsandthatbothwillmisrepresenttheutilitylossassociatedwiththeexperienceofthosestates.Toseethis,andintheabsenceofanygooddataontheexperiencedutilityassociatedwiththechangingnatureofhealthconditionsovertime,considerFigure1.Thetwosolidlinesshowtheexperiencedlosses(fromfullhealthwithU¼1)overtime(fromT¼0)fortwohealthstates,AandB.StateAstartsoffasworsethanstateBbutthereismoreadaptationtoBthantoA.Assumingthatutilityisinterpersonallycomparableandcanbeexpressedonacardinalscale,byT¼1,thetotallossinexperiencedutilityisthesameforbothstates.18Letusassumethatrespondentsfromthegeneralpublicaccuratelyassesstheinitialutilitylossassociatedwitheachstatebutthattheyfailtoforecastanyadaptationovertime.TheutilitiestheygenerateforstatesAandBareshownbythedottedlineslabelledAGPandBGP.BecauseAstartsoffasworsethanB,AGP Inevaluatingtwointerventions,onewhichpreventedstateAandonewhichpre-ventedstateB,theformerwouldappeartopreventmoreQALYlossesiftheutilitiesofthepublicareusedandthelatterwouldappeartopreventmoreQALYlossesiftheutilitiesofpatientsareused.19Infact,thelossinexperiencedutilityandthetotalutilityfrombothstatesisidenticalovertheperiodT¼0toT¼1.Itisourcontentionthat, ThestandardQALYmodelassumesthatutilitycanbecomparedacrosspeopleandtheSGandTTOmethodsbothassumethatutility(evenifitisrisk-adjustedortimediscounted)iscardinal;seeDolan(2000).19Inasimilarway,inevaluatingtwointerventionswhichwouldeitherleavepatientswhowouldotherwisedieinstateAorstateB,thelatterwouldgeneratemoreQALYSiftheutilitiesofthepublicareusedandtheformerwouldgeneratemoreQALYsiftheutilitiesofpatientsareused.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 182008] INTERPRETATIONSOFUTILITYANDVALUINGHEALTH 227 U = 1T = 0T = 1APBPBGPAGPFig.1.LossesinExperiencedUtilityfromTwoHealthStates,AandB,andDifferencesinDecisionUtilitiesfromtheGeneralPublic(GP)andPatients(P)allelseequal,thesamepriorityshouldbeaccordedtothetreatmentandpreventionofstatesAandBoverthegivenperiod.Todootherwiseistodistortprioritiesinfavourofthosewhoadaptifpublicpreferencesareusedandagainstthosewhoadaptifpatientpreferencesareused. 5.MeasuringExperiencedUtility Itmaybepossibletodevelopmoresophisticatedmeasuresofdecisionutilityandtoproviderespondentswithmoreinformationandcontextabouttheexperiencesasso-ciatedwiththestatestheyareaskedtovalue.Animportantadvanceinthisregardhasbeentheuseofvideosofrealpatientsoractorstopresentamorecompletepictureofwhatlifeinagivenhealthstatereallywouldbelike;see,forexample,Sloanetal.(1998)andLenertetal.(2005).However,decisionutilitieswillstillreflectwhattherespon-dent’sattentionisdrawntoatthetimeoftheassessment.Ubeletal.(2001)devisedanumberofnovelstudiesinwhichtheyattemptedtodrawrespondentsÕattentionawayfromthenegativeeffectsofdifferenthealthconditionsbyaskingthemtoconsiderhoweachconditionwouldaffectarangeofdifferentdomainsoflife,butthevaluationsrespondentsgavewerelargelyunchanged.20Manypeoplemayevenbeawareoftheirgeneralpropensitytomispredictfutureutility–andtheycancertainlybemadeawareinutilityassessmentstudies–buttheymaystillcontinuetomispredictitonacase-by-casebasis,inmuchthesamewayaspeoplecansimultaneouslybeawareoftheirgeneraltendencytoprocrastinateandstillprocrastinateonacase-by-casebasis(O’DonoghueandRabin,1999). Inaddition,decisionutilitieswillneedtoconformtotheconditionsoflogicalrationalityrequiredbytheQALYmodel.Theseconditions,setwithinanexpectedutilityframework,requirethattheutilityofahealthstateisunaffectedbyhowlongthe OnlyinoneoutoftenquestionnairevariantsdidrespondentsÕvaluationsofahypotheticalstate(suchasblindnessandparaplegia)increasewhentheywereaskedtothinktheimpactthestatewouldhaveonawiderangeoflifedomains. ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 20228 THEECONOMICJOURNAL [JANUARY statelastsandunaffectedbythestatesthatcomebeforeorafterit.Generally,respondentsseehealthstatesasgettingworsethelongertheylastandadditivesepa-rabilitydoesnotholdeither,althoughtherearenoconsistentpatternsintheviolationsthatwouldallowasimplealgorithmtoadjusttheQALYmodeltobetterrepresentindividualpreferencesoverfuturehealthprospects(TsuchiyaandDolan,2005).EconomistshaverespondedtosuchevidencebyderivingQALYsusinggeneralisationsofexpectedutilitytheory,suchasrank-dependentexpectedutilityandprospecttheory(Doctoretal.,2004).SuchextensionstothestandardQALYmodelareverymuchinkeepingwiththeemphasisineconomicsanddecisionscienceonlogicalcriteriaofrationality,whichcanbeassessedentirelybyreferencetoasystemofpreferencesinwhichonlyinternalcoherencematters(Sen,1993)buttheydonotaddressthemoresubstantiverationalityrequirementthatindividualsshouldcorrectlypredicttheeffectsofchangesinhealthontheirpreferences. So,ratherthan–oratleastaswellas–refiningthemethodsforelicitingdecisionutilities,wesuggestthateconomistsshouldlookformoredirectmeasuresoftheutilityassociatedwithdifferentstatesoftheworld.Economistsareshowingincreasinginterestintheuselifesatisfactionratingsasabasisforwelfareassessment(Ferrer-i-CarbonellandFrijters,2004),andtheimpactofdifferenthealthstatesisyettobefullyexploited.Insofarasthisworkconsiderssatisfactionaccordingtodomainsoflifesuchashealth,income,etc.,ithasmanyparallelstotherecommendationbyBroome(1993)thatweuseaÔdirectmethodÕ,ratherlikeavisualanaloguescalethatsimplyasksrespondentstoratetheÔgoodnessÕofahealthstate.Economists,whohavefocusedontheelicitationofdecisionutilitieselicitedusingtheSGorTTO,havelargelyignoredBroome’srecommendation.However,aswithdecisionutilities,satis-factionratingsarelikelytobebasedonwhatevertherespondent’sattentionisdrawntoatthetimeoftheassessment. Inparticular,suchratingswillbedeterminedbythecomparisonspeoplemakebetweentheirownlifeorhealthatdifferenttimes,andbetweenthemselvesandotherpeople(DolanandWhite,2006).Soakeyquestionabouttheusefulnessofsatisfactionratingsinpolicysettingsiswhetherthesecomparisonprocessesareanimportantpartofutilityintheirownrightorunhelpfuldistortionsofexperiencedutility.Theanswerwouldseemtodependontheextenttowhichtheassessmentquestiontapsintocomparisonsthattheindividualroutinelymakesinthemoment-to-momentexperienceofhislife.Putthisway,itwouldseemthatglobalassessmentsguiderespondentstomakecomparisonsthatmaynotbethefocusofattentionintheexperienceoftheirlives,oratleastnottotheextentthattheiranswertosatisfactionquestionswouldsuggest.Inaddition,satisfactionratingsarelikelytoreflectÔjudgementsthatindividualsformonthespot,basedoninformationthatischronicallyortemporarilyaccessibleatthatpointintime,resultinginpronouncedcontexteffectsÕ(SchwarzandStrack,1999).Forthesereasons,satisfactionratingsmaynotprovidethebestproxiesforthekindofexperiencedutilitywehaveinmindhere,andweinsteadneedtodevelopmeasuresof(orbetterapproximationsfor)utilityonamoment-to-momentbasis.Experiencesamplingmethods(ESM)(Stoneetal.,1999)andthedayreconstructionmethod(DRM)(Kahnemanetal.,2004)providepromisingwaysofdoingthis.ESMtypicallyinvolvesusingpalmpilotsthataskpeopleatrandomtimesduringthedaytoratedifferentfeelings(happiness,frustration,worryetc.).However,thismethodis ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 2008] INTERPRETATIONSOFUTILITYANDVALUINGHEALTH 229 invasive,itmayinterrupttheflowofanexperienceandthereareoftenmissingobservations,whichmaybenon-random(CsikszentmihalyiandHunter,2003). TheDRMhasbeendevelopedtoovercometheseproblems,andasksrespondentstodividethepreviousdayintoanumberofepisodesandthentoratedifferentelementsofaffectduringthoseactivitiesona0–6scale.Usingthismethod,Kahnemanetal.(2004)showthatoneofthebiggestdeterminantsofgoodfeelingsissleepquality,whereasmaritalstatusandincomehavemuchsmallereffects.Largesamplescanbecollectedinarelativelyshortperiodoftimeandthemethoddoesnotdisturbtheflowofexperienceasithappens.Kahnemanetal.(2004)provideevidencethattheresultsfromtheDRMprovideagoodapproximationforthosefromtheESM. Toproducedatathatallowtherelativecost-effectivenessofdifferentinterventionstobecalculatedrequiresthedatafromtheDRMtobeexpressedonacardinalscale.Atpresent,respondentsratearangeoffeelings,includinghappiness,worry,andfrus-tration,andthereisnowaytodeterminetherelativeweightsthateachrespondentattachestoeachofthese.OnesimplerulethatKahnemanandKrueger(2006)proposeistolookatthefeelingthatgetsthehighestrating:ifthisisanegativeone,scorethetimeinthatactivityasone;otherwisescoreitaszero.Itisthenpossibletocalculatetheproportionoftimethatpeoplespendinanunpleasantstate(whichKahnemanandKruegerrefertoastheÔU-IndexÕ). TheDRMcanbeadministeredtopopulationswitharangeofhealthconditionsandatvariousstagesofdiseaseprogression.Bygatheringdataonpeople’shealthexperi-ences,aswellasdataonage,sexetc.,itwillbepossibletoshowhowtheactivitiespeopleengagein,andthemoment-to-momentutilityassociatedwiththoseactivities,areaffectedbytheirhealthstate.Thegreatadvantageofdataofthiskindisthatitshowswhataffectswhatpeopledoandhowtheyfeelonamoment-to-momentbasisratherthanreflectingwhatrespondentsthinkaffectsthematthetimeadecisionutilityorlifesatisfactionratingiselicited. Bycollectingrelevantbackgroundinformation,itwillbepossibletoshowhowarangeofhealth-relatedfactors(includinganycostsassociatedwithadaptation)affectexperiencedutilitywithouttherespondenthavingtoattributetheutilitytheyexperi-enceinanyway.Indeed,themethodcouldbeusedtoshowhowarangeoffactors,includingmarketandnon-marketgoodsandservices,affectexperiencedutility.Inanyoftheseapplications,therewouldbeclearadvantagestogatheringlongitudinaldatawherepossible.Aswellasallowingtheimpactofdifferentconditionstobetracedovertime,suchstudieswouldalsoallowforissuesofcausalitytobeaddressedand,thus,facilitateabetterunderstandingofthedegreetowhichdifferentlevelsandtypesofaffectareabletopredictfuturechangesinhealth. Werecognisethatthemeasureofexperiencedutilityweproposemaynotcaptureeverythingthatindividuals(letalonepolicymakers)aretryingtoachieve.Themethod,asitcurrentlystands,doeslittletoincorporateourhigherorderpreferencesthatgiveusoursenseofidentityanddefinewhoweare,anditignoresanypreferenceswemayhaveoverparticulartypesofhedonicexperience(Kelman,2005).Wemayalsomakejudgementsaboutthepurposeandmeaninginourlivesthatmaytranscendourhedonicexperiences(Seligman,2002).Moreover,manypeoplewouldseeminglychooseaÔrealÕlife,withitsassociatedpainandsuffering,overalifewithgreaterhedonicexperienceifthatlatterlifeweretobeartificiallycreatedbyanÔexperience ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 230 THEECONOMICJOURNAL [JANUARY machineÕ(Nozick,1971).Ofcourse,therehasbeen,andwillcontinuetobe,muchdebateabouttheseissues,buttheDRMcouldbeaugmentedtoaccountforsomeoftheseconsiderations.Forexample,questionscouldbeaddedabouttheimportantgoalsthatpeoplehaveandhowcertainactivitiescontributetowardsthem,andanÔextendedU-IndexÕcouldbecalculatedastheproportionoftimethatpeoplespendinpursuitoftheirgoals. 6.ConcludingRemarks Modernwelfareeconomicsinterpretsaperson’sutilityintermsofherpreferences,andthemethodsthateconomistsanddecisionanalystshaveusedtoelicitutilityreflectthisinterpretation.Thechoicesthatrespondentsmakeinhealthstatevaluationandotherstudiesshouldbeguidedbyaccurateassessmentsoftheutilityassociatedwiththeconsequencesofthosechoices.Evenifpeopleweretoobeytheaxiomsofrationalchoiceintheirpreferencesoverhealthstates,theirpreferenceswillstillbesub-optimaliftheyaremistakenintheirforecastsoftheutilitytheywillexperienceinthefuture(Loewensteinetal.,2003).Byandlarge,itappearsthatpeopleadapttochangesintheircircumstancesbuttheyoftenfailtoappreciatethedegreetowhichtheywilladapttothosechanges.Totheextentthatourwants,ascapturedbyourdecisions,arebasedonpredictionsofwhatwewillsubsequentlyenjoy,weareoftenguiltyofÔmiswantingÕi.e.wewantthingsthatdonotmakeushappierorwedonotwantthingsthatwouldmakeushappier(GilbertandWilson,2000). Althoughmanyeconomists,aswellasaconsensuspanelconvenedbytheUSPublicHealthService(Goldetal.,1996),recommendtheuseofutilitiesfromthegeneralpublic,elicitingdecisionutilitiesfromthosecurrentlyexperiencingthehealthstateinquestionwillavoidsomeoftheproblemsassociatedwithelicitingdecisionutilitiesfromthepublic.However,apatient’sdecisionutilitywillstillreflectwhattheywantrightnowandmaystillnotreflectthefutureutilityassociatedwiththeirhealthstate.Inshort,therearegoodreasonstosupposethathealthymembersofthegeneralpublicwillfocusonthetransitionallossinutilityassociatedwithachangeinhealthandignoretheadaptationthattakesplace,andthatpatientswillfocusontheadaptedlevelsofwell-beingandignoreanytransitionalloss.21Thatdecisionutilityandexperiencedutilitydonotproducethesameresultsshouldnotcomeasagreatsurprisetoeconomists:AdamSmith(1759)arguedthatÔThegreatsourceofboththemiseryanddisordersofhumanlifeseemstoarisefromover-ratingthedifferencebetweenonepermanentsituationandanotherÕ.Thehealthstateutilitiesthatarecurrentlyusedasthequality-adjustmentweightsinQALYscertainlyresultindifferencesbetweenfullhealthanddysfunctionalstatesthatareover-rated,andmayalsodistortthedifferencesbetweendysfunctionalstates. Torepresenttheeffectofdifferenthealthstatesonpeople’swell-beingmoreaccu-rately,weproposethateconomistsinhealthandelsewhereshifttheirattentionfromthemeasurementofdecisionutilitytowardsthemeasurementofexperiencedutility. Thereisthepossibilitythatpatientsmayoverstatetransitionallossesiftheythoughtthatbydoingsotherewouldbemoreresourcesdevotedtothetreatmentoftheirconditionbutthereisnoevidencethatrespondentsthinkinthisstrategicway. ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 212008] INTERPRETATIONSOFUTILITYANDVALUINGHEALTH 231 Wehavesuggestedawayinwhichthismightbedonethatrequirespeopletostatehowtheyfeltduringvariousactivitiesonthepreviousday.Themethodisarecentdevel-opmentanditssuccessfuluseinfutureempiricalstudiesinhealthandelsewherewillrequireaninterdisciplinaryapproachinvolvingeconomistsandpsychologists,andotherswithexpertiseinparticularapplications(suchascliniciansinthecaseofhealth).Wehopethatsomeoftheideaspresentedinthisarticleprovideacatalystforthisendeavour. ImperialCollegeLondonPrincetonUniversitySubmitted:14April2005Accepted:16October2006 References Adler,M.D.(2004).ÔFearassessment:cost-benefitanalysisandthepricingoffearandanxietyÕ,Chicago-Kent LawReview,vol.79(3),pp.977–1053. Antonak,R.F.andLivneh,H.(1995).ÔPsychologicaladaptationtodisabilityanditsinvestigationamong personswithmultiplesclerosisÕ,SocialScienceandMedicine,vol.40,pp.1099–108. Baron,J.,Asch,D.A.,Fagerlin,A.,Jepson,C.,Loewenstein,G.,Riis,J.,Stineman,M.G.andUbel,P.A.(2003). ÔEffectofassessmentmethodonthediscrepancybetweenjudgmentsofhealthdisorderspeoplehaveanddonothave:awebstudyÕ,MedicalDecisionMaking,vol.23,pp.422–34. Barsky,A.(2002).ÔForgetting,fabricatingandtelescoping:theinstabilityofmedicalhistoryÕ,Archivesof InternalMedicine,vol.162,pp.981–4. Bentham,J.(1789/1948).AnIntroductiontothePrincipleofMoralsandLegislation,Oxford:Blackwell(1948 edition). Bleichrodt,H.andJohannesson,M.(1997).ÔStandardgamble,timetrade-offandratingscale:experimental resultsontherankingpropertiesofQALYsÕ,JournalofHealthEconomics,vol.16,pp.155–75. Brickman,P.andCampbell,D.(1971).ÔHedonicrelativismandplanningforthegoodsocietyÕ,in(M.H. Appleby,ed.),AdaptationLevelTheory:aSymposium,pp.207–302,NewYork:NewYorkAcademicPress,.Brickman,P.,Coates,D.andJanoff-Bulman,R.(1978).ÔLotterywinnersandaccidentvictims:ishappiness relative?Õ,JournalofPersonalityandSocialPsychology,vol.36,pp.917–27.Broome,J.(1993).ÔQalysÕ,JournalofPublicEconomics,vol.50,pp.149–67. Byrne,M.M.,O’Malley,K.andSuarez-Almazor,M.E.(2005).ÔWillingnesstopayperquality-adjustedlifeyear inastudyofkneeosteoarthritisÕ,MedicalDecisionMaking,vol.25,pp.655–66. Clark,A.,DienerE.,Georgellis,Y.andLucas,R.L.(2004).ÔLagsandleadsinlifesatisfaction:atestofthe baselinehypothesisÕ.GermanInstituteforEconomicResearch,DiscussionPaper317,Berlin. Cohn,B.(1999).ÔThelaytheoryofhappiness:illusionsandbiasesinjudgingothersÕ,unpublishedunder-graduatedissertation.PrincetonUniversity. Csikszentmihalyi,M.andHunter,J.(2003).ÔHappinessineverydaylife:theusesofexperiencesamplingÕ, JournalofHappinessStudies,vol.4,pp.185–99. Dar,R.,Ariely,D.andFrenk,H.(1995).ÔTheeffectofpastinjuryonpainthresholdandtoleranceÕ,Pain,vol. 60,pp.189–93. deWit,G.A.,Busschbach,J.andDeCharro,F.(2000).ÔSensitivityandperspectiveinthevaluationofhealth status:whosevaluescount?Õ,HealthEconomics,vol.9(2),pp.109–26. Doctor,J.N.,Bleichrodt,H.,Miyamoto,J.,Temkin,N.R.andDikmen,S.(2004).ÔAnewandmorerobusttest ofQALYsÕ,JournalofHealthEconomics,vol.23(2),pp.353–63. Dolan,P.(1997)ÔModellingvaluationsforEuroQolhealthstatesÕ,MedicalCare,vol.11,pp.1095–108.Dolan,P.(1999).ÔWhosepreferencescount?Õ,MedicalDecisionMaking,vol.19,pp.482–6. Dolan,P.(2000).ÔThemeasurementofhealth-relatedqualityoflifeforuseinresourceallocationdecisionsin healthcareÕ,in(A.J.CulverandJ.P.Newhouse,eds.),HandbookofHealthEconomics,Vol.1,pp.1724–60Amsterdam:ElsevierScienceBV. Dolan,P.andStalmeier,P.(2003).ÔThevalidityoftimetrade-offvaluesincalculatingQALYs:constant proportionaltrade-offversustheproportionalheuristicÕ,JournalofHealthEconomics,vol.22,pp.445–58.Dolan,P.andWhite,M.(2006).ÔDynamicwell-being:connectingwhatweanticipatewithwhatweexperi-enceÕ,SocialIndicatorsResearch,vol.75(2),pp.303–33. ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 232 THEECONOMICJOURNAL [JANUARY Easterlin,R.(2001).ÔIncomeandhappiness:towardsaunifiedtheoryÕ,EconomicJournal,vol.111(473), pp.465–84. Edgeworth,F.(1881/1967).MathematicalPsychics,NewYork:Kelley. Ferrer-i-Carbonell,A.andFrijters,P.(2004).ÔHowimportantismethodologyfortheestimatesofthe determinantsofhappiness?Õ,EconomicJournal,vol.114,pp.641–59. Fisher,I.(1918).ÔIsÔÔutilityÕÕthemostsuitabletermfortheconceptitisusedtodenote?Õ,AmericanEconomic Review,vol.8,p.335. Frederick,S.andLoewenstein,G.(1999).ÔHedonicadaptationÕ,in(D.Kahneman,E.DienerandN.Schwarz, eds),Well-Being:TheFoundationsofHedonicPsychology,pp.302–29,NewYork:RussellSageFoundation.Frey,B.andStutzer,A.(2002).ÔWhatcaneconomistslearnfromhappinessresearch?Õ,JournalofEconomic Literature,vol.40,pp.402–35. Gilbert,D.T.,Gill,M.J.andWilson,T.D.(2002a).ÔThefutureisnow:temporalcorrectioninaffectivefore-castingÕ,OrganizationalBehaviorandHumanDecisionProcesses,vol.88(1),pp.430–44. Gilbert,D.T.,Pinel,E.C.,Wilson,T.D.,Blumberg,S.J.andWheatley,T.P.(2002b).ÔDurabilitybiasinaffective forecastingÕ,in(T.Gilovich,D.GriffinandD.Kahneman,eds.),HeuristicsandBiases:ThePsychologyofIntuitiveJudgement,pp.292–312,Cambridge:CambridgeUniversityPress. Gilbert,D.T.andWilson,T.D.(2000).ÔMiswanting:someproblemsintheforecastingoffutureaffective statesÕ,in(J.Forgas,ed.),ThinkingandFeeling:TheRoleofAffectinSocialCognition,Cambridge:CambridgeUniversityPress. Gold,M.,Siegal,J.E.,Russell,L.B.,andWeinstein,M.C.(1996).Cost-effectivenessinHealthandMedicine, Oxford:OxfordUniversityPress. Kahneman,D.(1997).ÔNewchallengestotherationalityassumptionÕ,LegalTheory,vol.3,pp.105–24. Kahneman,D.(2000).ÔEvaluationbymoments:pastandfutureÕ,in(D.Kahneman,andA.Tversky,eds.), Choices,ValuesandFrames,ch.38,NewYork:CambridgeUniversityPressandtheRussellSageFoundation.Kahneman,D.andKrueger,A.B.(2006).ÔDevelopmentsinthemeasurementofsubjectivewell-beingÕ,Journal ofEconomicPerspectives,vol.20(1),pp.3–24. Kahneman,D.,Krueger,A.B.,Schkade,D.A.,Schwarz,N.andStone,A.A.(2004).ÔTowardnationalwell-being accountsÕ,AmericanEconomicReview,vol.94(2),pp.429–34. Kahneman,D.andTversky,A.(1979).ÔProspecttheory:ananalysisofdecisionunderriskÕ,Econometrica,vol. 47,pp.263–91. Kahneman,D.,Wakker,P.andSarin,R.(1997).ÔBacktoBentham?ExplorationsofexperiencedutilityÕ, QuarterlyJournalofEconomics,vol.112,pp.375–405. Kelman,M.(2005).ÔHedonicpsychologyandtheambiguitiesofÔÔwelfare’ÕÕ,PhilosophyandPublicAffairs,vol. 33,pp.391–412. Krupat,E.(1974).ÔContextasadeterminantofperceivedthreatÕ,JournalofPersonalityandSocialPsychology,vol. 29(6),pp.731–6. Lenert,L.,Sturley,A.P.,Rapaport,M.H.,Chavez,S.,Mohr,P.EandRupnow,M.(2005).ÔPublicpreferences forhealthstateswithschizophreniaandamappingfunctiontoestimateutilitiesfrompositiveandnegativesymptomscalescoresÕ,SchizophreniaResearch,vol.71(1),pp.155–65. Lennon,M.C.,Dohrenwend,B.P.,Zautra,A.J.andMarbach,J.J.(1990).ÔCopingandadaptationtofacialpain incontrasttootherstressfullifeeventsÕ,JournalofPersonalityandSocialPsychology,vol.59,pp.1040–50.Livneh,H.andAntonak,R.F.(1994).ÔReviewofresearchonpsychosocialadaptationtoneuromuscular disorders:I.Cerebalpalsy,musculardystrophy,andParkinson’sdiseaseÕ,JournalofSocialBehaviorandPersonality,vol.9(5),pp.201–30. Llewellyn-Thomas,H.A.,Sutherland,H.J.andThiel,E.C.(1993).ÔDopatientsÕevaluationsofafuturehealth statechangewhentheyactuallyenterthatstate?Õ,MedicalCare,vol.31(11),pp.1002–12. Loewenstein,G.,O’Donoghue,T.andRabin,M.(2003).ÔProjectionbiasinpredictingfutureutilityÕ,Quarterly JournalofEconomics,vol.118,Novemberpp.1209–48. Lucas,R.E.,Clark,A.E.,Georgellis,Y.andDiener,E.(2003).ÔRe-examiningadaptationandthesetpoint modelofhappiness:reactionstochangesinmaritalstatusÕ,JournalofPersonalityandSocialPsychology,vol.84(3),pp.527–39. Means,B.,Nigam,A.,Zarrow,M.,Loftus,E.F.andDonaldson,M.S.(1989).ÔAutobiographicalmemoryfor health-relatedeventsÕ,VitalHealthStatistics,vol.6,pp.1–37. Menzel,P.Dolan,P.,Richardson,J.andOlsen,J.A.(2003).ÔTheroleofadaptationtodisabilityand diseaseinhealthstatevaluation:apreliminarynormativeanalysisÕ,SocialScienceandMedicine,vol.55,pp.2149–58. Murray,C.J.L.(1996).ÔRethinkingDALYsÕ,in(C.J.L.MurrayandA.D.Lopez,eds.),TheGlobalBurdenof Disease,CambridgeMA:HarvardUniversityPress. Nozick,R.(1971).Anarchy,State,andUtopia,NewYork:BasicBooks. O’Donoghue,T.andRabin,M.(1999).ÔIncentivesforprocrastinatorsÕ,QuarterlyJournalofEconomics,vol.114, pp.769–816. Patterson,D.R.,Everett,J.J.,Bombardier,C.H.,Questad,K.A.,Lee,V.K.andMarvin,J.A.(1993).ÔPsycho-logicaleffectsofsevereburnsvictimsÕ,PsychologicalBulletin,vol.113,pp.362–78.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 2008] INTERPRETATIONSOFUTILITYANDVALUINGHEALTH 233 Peters,M.L.,Sorbi,M.J.,Kruise,D.A.,Kerssens,J.J.,Verhaak,P.F.andBensing,J.M.(2000).ÔElectronicdiary assessmentofpaindisabilityandpsychologicaladaptationinpatientsdifferingindurationofpainÕ,Pain,vol.84,pp.181–92. Postulart,D.andAdang,E.(2000).ÔResponseshiftandadaptationinchronicallyillpatientsÕ,MedicalDecision Making,vol.20,pp.186–93. Redelmeier,D.andKahneman,D.(1996).ÔPatient’smemoriesofpainfultreatmentmedicaltreatments:real timeandretrospectiveevaluationsoftwominimallyinvasiveproceduresÕ,Pain,vol.116,pp.3–8. Sackett,D.L.andTorrance,G.W.(1978).ÔTheutilityofdifferenthealthstatesasperceivedbythegeneral publicÕ,JournalofChronicDiseases,vol.31,pp.697–704. Salomon,J.A.,Tandon,A.andMurray,C.J.L.(2004).ÔComparabilityofselfratedhealth:crosssectionalmulti-countrysurveyusinganchoringvignettesÕ,BritishMedicalJournal,vol.328,pp.258–61. Schwarz,N.andStrack,F.(1999).ÔReportsofsubjectivewell-being:judgementalprocessesandtheirmeth-odologicalimplicationsÕ,in(D.Kahneman,E.Diener,andN.Schwarz,eds.),Well-being.Thefoundationsofhedonicpsychology,NewYork:RussellSageFoundation. Schulz,R.andDecker,S.(1985).ÔLong-termadjustmenttophysicaldisability:theroleofsocialsupport, perceivedcontrol,andself-blameÕ,JournalofPersonalityandSocialPsychology,vol.48,pp.1162–72. Seligman,M.E.P.(2002).AuthenticHappiness:UsingtheNewPositivePsychologytoRealizeYourPotentialforLasting Fulfillment,NewYork:FreePress/SimonandSchuster. Sen,A.(1992).InequalityReexamined,Cambridge,MA:HarvardUniversityPress.Sen,A.(1993).ÔConsistencyofchoiceÕ,Econometrica,vol.61,pp.495–521. Sharma,R.,Stano,M.andHaas,M.(2004).ÔAdjustingtochangesinhealth:implicationsforcost-effectiveness analysisÕ,JournalofHealthEconomics,vol.23,pp.335–51. Sieff,E.,Dawes,R.andLoewenstein,G.(1999).ÔAnticipatedversusactualresponsestoHIVtestresultsÕ, AmericanJournalofPsychology,vol.112,pp.297–311. Simon,G.E.andGureje,O.(1999).ÔStabilityofsomatizationdisorderandsomatizationsymptomsamong primarycarepatientsÕ,ArchivesofGeneralPsychiatry,vol.56,pp.90–5. Sloan,F.A.,Viscusi,W.K.,Chesson,H.W.,Conover,C.J.andWhetten-Goldstein,K.(1998).ÔAlternative approachestovaluingintangiblehealthlosses:theevidenceformultiplesclerosisÕ,JournalofHealthEconomics,vol.17,pp.475–97. Smith,A.(1759).TheTheoryofMoralSentiments,London:A.Millar(1790.Sixthedition). Smith,D.M.,Sherriff,R.L.,Damschroder,L.J.,Loewenstein,G.,Ubel,P.A.(2006).ÔMisrememberingcolos-tomies?FormerpatientsgivelowerutilityratingsforcolostomythandocurrentpatientsÕ,HealthPsy-chology,vol.25(6),pp.688–95. Sprangers,M.A.G.andSchwartz,C.E.(1999).ÔIntegratingresponseshiftintohealth-relatedqualityoflife research:atheoreticalmodelÕ,SocialScienceandMedicine,vol.48(11),pp.1507–15. Stone,A.A.,Shiffman,S.S.andDeVries,M.W.(1999).ÔEcologicalmomentaryanalysisÕ,in(D.Kahneman, E.Diener,andN.Schwarz,eds.),Well-being.TheFoundationsofHedonicPsychology,NewYork:RussellSageFoundation. Sunstein,C.(2002).RiskandReason,Cambridge:CambridgeUniversityPress. Thompson,R.F.,Groves,P.M.,Teyler,T.I.andRoemer,R.A.(1973).ÔAdual-processtheoryofhabituation: theoryandbehaviorÕ,in(H.V.S.Peeke,andM.J.Herz,eds.),Habituation,vol.1,pp.239–71,NewYork:AcademicPress. Tsuchiya,A.andDolan,P.(2005).ÔTheQALYmodelandindividualpreferences:asystematicreviewofthe literatureÕ,MedicalDecisionMaking,vol.25(4),pp.460–7. Tyc,V.L.(1992).ÔPsychosocialadaptationofchildrenandadolescentswithlimbdeficiencies:areviewÕ, ClinicalPsychologyReview,vol.12,pp.275–91. Ubel,P.A.,Loewenstein,G.,Hershey,J.,Baron,J.,Mohr,T.,Asch,D.A.andJepson,C.(2001).ÔDononpa-tientsunderestimatethequalityoflifeassociatedwithchronichealthconditionsbecauseofafocusingillusion?Õ,MedicalDecisionMaking,vol.21,pp.190–9. Ubel,P.A.,Loewenstein,G.andJepson,C.(2003).ÔWhosequalityoflife?Acommentaryexploringdis-crepanciesbetweenhealthstateevaluationsofpatientsandthegeneralpublicÕ,QualityofLifeResearch,vol.12,pp.599–607. vanPraag,B.andFerrer-i-Carbonell,A.(2004).HappinessQuantified:ASatisfactionCalculusApproach,Oxford: OxfordUniversityPress. Weinstein,M.C.,andStason,W.B.(1977).ÔFoundationsofcost-effectivenessanalysisforhealthandmedical practicesÕ,NewEnglandJournalofMedicine,vol.296,p.716. Weinstein,N.D.(1978).ÔIndividualdifferencesinreactiontonoise:alongitudinalstudyincollegedormitoryÕ, JournalofAppliedPsychology,vol.63,pp.458–66. Weinstein,N.D.(1982).ÔCommunitynoiseproblems:evidenceagainstadaptationÕ,JournalofEnvironmental Psychology,vol.2,pp.87–97. Wilson,T.D.,Centerbar,D.B.andBrekke,N.(2002).ÔMentalcontaminationandthedebiasingproblemÕ,in (T.Gilovich,D.Griffin,andD.Kahneman,eds.),HeuristicsandBiases:ThePsychologyofIntuitiveJudgement,Cambridge:CambridgeUniversityPress.ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008 234 THEECONOMICJOURNAL [JANUARY2008] Wilson,T.D.,Centerbar,D.B.,Kerner,D.A.andGilbert,D.T.(2005).ÔThepleasuresofuncertainty:pro-longingpositivemoodsinwayspeopledonotanticipateÕ,JournalofPersonalityandSocialPsychology,vol.88(1),pp.5–21. Wilson,T.D.andGilbert,D.(2005).ÔMakingsense:amodelofaffectiveadaptationÕ,workingpaper,University ofVirginiaandHarvardUniversity. Wilson,T.D.andGilbert,D.(2003).ÔAffectiveforecastingÕ,AdvancesinExperimentalSocialPsychology,vol.35, pp.345–411. Wilson,T.D.,Meyers,J.andGilbert,D.T.(2003).ÔHowhappywasI,anyway?AretrospectiveimpactbiasÕ,Social Cognition,vol.21,pp.407–32. Wortman,C.andSilver,R.(1987).ÔCopingwithirrevocablelossÕ,inCataclysms,CrisesandCatastrophres: PsychologyinAction,Masterlectureseries,Vol.6,pp.189–235,WashingtonDC:AmericanPsychologicalAssociation. Wu,S.(2001).ÔAdaptingtoheartconditions:atestofthehedonictreadmillÕ,JournalofHealthEconomics,vol. 20,pp.495–508. ÓTheAuthor(s).JournalcompilationÓRoyalEconomicSociety2008
因篇幅问题不能全部显示，请点此查看更多更全内容