investigate

COGNITIONANDEMOTION,2001,15(2),207-230

UsingtheImplicitAssociationTesttoinvestigate

attitude-behaviourconsistencyforstigmatid

behaviour

n

UniversityofWashington,Seattle,USA

RutgersUniversity,NewJery,USA

ald

UniversityofWashington,Seattle,USA

Toconsciouslybolsterbehaviourthatisdisapprovedbyothers(i.e.,stigmatid

behaviour)peoplemayholdandreportafavourableattitudetowardthebehaviour.

However,achievingsuchbolsteringoutsideawarenessmaybemoredifficult.

Explicitattitudesweremeasuredwithlf-reportmeasures,andtheImplicit

AssociationTestwasudtoasssimplicitattitudestowardbehaviourheldby

stigmatidactors(smokers)andnonstigmatidactors(vegetariansandomni-

vores).Smokers'showedgreaterattitude-behaviourconsistencyintheirexplicit

rast,vegetarians

andomnivoresshowed

attitude-behaviour-consistencyatbothimplicitandexplicit

s'implicitnegativeattitudestowardsmokingmayreflectitsstatus

asastigmatidbehaviour,oritsaddictivenature.

Therearemanybehavioursthatpeopleengageindespiteknowingthatothers

regardthebehaviourasunwi,objectionable,

thepeoplewhoengageinsuchbehaviourscognitivelyadjusttothisstigmatid

characteroftheirownbehaviour?Smokingprovidesaninterestingbehaviourto

studybecauofitshavingchangedinrecentyearsfrombeingasocially

CorrespondenceshouldbeaddresdtoJaneSwansonorAnthonyGreenwald,Departmentof

Psychology,Box351525,UniversityofWashington,Seattle,WA98195-1525,USAortoLaurie

Rudman,PsychologyDepartment,TilletHall,LivingstonCampus,RutgersUniversity.53Ave.E.,

Piscataway,NJ08854,USA;e-mail:swansonj@,rudman@,or

agg@

ThisrearchwassupportedbyGrantsMH-41328andMH-001533fromNationalInstituteof

MentalHealthandbyGrantSBR-9710172fromNationalScienceFoundationtothethirdauthor.

02001PsychologyPressLtd

~joumals/pp/I:10.1080/0060

208SWANSON,RUDMAN,GREENWALD

ent,lawsrestrict

smokers'behaviour,smokersareviewedasunhealthy,dirty,weak-willed,and

morallybereft(Goldstein,1991;Rozin&Singh,1998),andthemajorityof

smokersareawarethattheirhabitincreastheirchancesofheartdia,lung

cancer,andprematuredeath(Shopland&Brown,1987).Becauthisknowl-

edgeandthestigmaassociatedwithsmokingareinconsistentwithknowingthat

theysmoke,smokersmayexperienceadissonance-liketension(Festinger,

1957).Thismayprompttheircreationormodificationofcognitionstosupport

theirbehaviour(ger,1957,pp.5-6).Bycontrast,peoplewhoengagein

nonstigmatidbehaviourhavenooccasiontorespondtosuchinconsistencies.

Peoplewithstigmatidoccupations(e.g.,toplessdancersandmorticians)

maydownplaythenegativeaspectsoftheirprofessions,emphasisinginsteadthe

prosocialbenefitstheyprovide(Thompson,1991;Thompson&Harred,1992).

Alongthelines,smokersperceivelesshealth-relatedconquencesofsmok-

ingthandononsmokers(Halpern,1994;Johnson,1968),eventhoughboth

groupshavethesamefactualknowledge(McMaster&Lee,1991;Miller&

Slap,1989).Further,themoresmokersacknowledgethehealthrisksofsmok-

ing,themoretheyproducerationalisationsfortheirhabit(Johnson,1968).And,

althoughsmokers'lf-reportedattitudestowardsmokingrangefromneutralto

slightlyunfavourable,theyneverthelesshavemorepositiveattitudestoward

smokingthandononsmokers(Chassin,Presson,Sherman&Edwards,1991;

Stacy,Bentler,&Flay,1994).Insum,theliteraturesuggeststhatstigmatid

actors-includingsmokers-cognitivelybolstertheiractionsinthefaceof

widespreaddisapproval.

Allpriorrearchonsmokers'cognitivebolsteringoftheirsmokinghabithas

ntrearchadditionally

marygoalofthisrearchwastodetermine

whethercognitivebolsteringofstigmatidbehaviourwouldalsobeevidenton

gwasanobviouschoiceforthestigmatidbeha-

viour,anddietarypreferenceswereudascomparisonnonstigmatidbeha-

i,therewasnoreasontosuspectthatsmokers'attitudeswouldbe

ngstatementsofcognitiveconsistency

theoriesdonotaddressadistinctionbetweenimplicitandexplicitcognitions.

BecauGreenwaldetal.(inpress;Greenwaldetal.,1999)havereported

greaterconsistencyamongimplicitthanexplicitcognitionsinotherdomains,

I

therewasactuallysomereasontoanticipatethatimplicitmeasuresmightshow

I

-

theless,whenpeopleactinwaysthatelicitfrequentnegativefeedbackfrom

others,ortofthisI

view,Greenwaldetal.(1999)foundoneexceptiontotheirgeneralobrvation

thatpeoplewholikedthemlvesandidentifiedwiththeirgroupalsoshowed

I

ysubjectswithhighlf-esteemimplicitlydisidentifiedwith

1

theiragegroup,alsoshowingstronglygreaterimplicitpreferenceforyoungthanI

plicitfindingintheageattitudedomainmayindicatetheextentto

rly,smokers'implicit

cognitionsmayindicatetheextenttowhichsmokingisstigmatid.

Thedevelopmentofimplicitmeasuresthatarensitivetoindividualdif-

ferencesprovidestheopportunitytoexamineimplicitcognitionsassociatedwith

our-relevantcognitionsincludeattitudestoward

thelfandtowardthebehaviour,andassociationoflfwiththebehaviour.

Implicitattitudesaremeasuredbyasssingtheautomaticassociationbetween

theattitudeobjectandpositiveornegativevalence(Fazio,1990;Greenwald&

Banaji,1995;Greenwald,McGhee,&Schwartz,1998-intheemotionlitera-

ture,thisisreferredtoasautomaticaffect,e.g.,Winkielman,Zajonc,&

Schwarz,1997).'Bothcognitiveandemotiontheoristsconceptualiimplicit

cognitions(e.g.,attitudesandbeliefs)assimilartoimplicitmemory,suchthat

eachisrevealedwhenpastexperienceindirectlyinfluencesrespons"ina

fashionnotintrospectivelyknownbytheactor"(Greenwald&Banaji,1995,p.

4).Bycontrast,explicitcognitionsarepresumedtorequiredeliberateretrieval

ofinformation.

Evidencefromprejudiceandstereotyperearchindicatesthatimplicitand

explicitcognitionsareonlyweaklycorrelated(e.g.,Blair,inpress;Brauer,

Wal,&Niedenthal,2000;Greenwaldetal.,1998;Rudman,Ashmore,&Gary,

2000);Rudman,Greenwald,Mellott,&Schwartz,1999).Thisisnottosuggest

thattheconstructsarecompletelyindependentorthattheirrelationshipcannot

bemoderated(Rudmanetal.,2000;ealsoWegner&Bargh,1998;fora

discussionoftheinterfacebetweenimplicitandexplicitcognitions).However,

thefindingsdosuggestthatthepsychologicalpropertiesofimplicitand

explicitcognitionscananddodiverge.

TheImplicitAssociationTest

TheImplicitAssociationTest(IAT;Greenwaldetal.,1998)isaflexible

measureofimplicitsocialcognition,includingattitudes,stereotypes,andlf-

concept(e.g.,Greenwaldetal.,inpress;Rudmanetal.,2000).Themethod

assumesthatperformingtasksthatobligepeopletosortwell-associatedcate-

goriestogetheriasierthanperformingtasksinwhichthecategoriestobe

mple,thelf-esteemIATinvolves

fourcategories:twocontrastedtargetconceptcategories(lfandother)andtwo

contrastedattributecategories(pleasantandunpleasant;eFigure1).Inthe

data-gatheringtrialblocksoftheIAT,subjectsperformtwocombinedcate-

Affectcanbeconceptualidamotionsorastheevaluationattachedtoaparticular(attitude)

object(In&Diamond,1989).Theprentpaperisconcernedwithaffectinthelattern-

specifically,attitudestowardone'slfandone'sbehaviourwhenthebehaviourisstigmatid(e.g.,

smoking)versuswhenitisnonstigmatid(e.g.,vegetarianism).

210SWANSON,RUDMAN,GREENWALD

respondleftrespondright

Task1UNPLEASANTPLEASANT

Task2SELFOTHER

-Task3SELF+UNPLEASANTOTHER+PLEASANT

IAT

Itemsin

theFour

Categories

Task4OTHER+UNPLEASANTSELF+PLEASANT

Figure1.

IllustrationoftheImplicitAssociationTest(IAT).TheIATstartsbyintroducingsubjects

example,thecategoriesareintroducedinTasks1and2.

InTask1,subjectsareaskedtorespond"left"topleasantwordsand"right"tounpleasantwords.

InTask2,subjectsrespond"left"tolfwordsand"right"measureis

obtainedbycomparingresponlatenciesinthenexttwotasks,oneinwhichlfandunpleasantare

assignedto"left"andotherandpleasantto"right",andanotherinwhichotherandunpleasantare

assignedto"left"andlfandpleasantareassignedto"right".Ifthesubjectrespondsmorerapidly

whenlfandpleasantsharearespon,thisindicatesthatthelf-pleasantassociationisstronger

thanthelf-otherassociation.

SELFOTHER

methey

mYthem

minetheir

lfother

gorisationtasksthatmapthefourcategoriesofstimuli(lf,other,pleasantand

unpleasant)ombinedtask(lf+unpleasant),

subjectsareinstructedtorapidlypressonekeyforbothlfandunpleasant

condcombinedtask(lf+pleasant),bothlfandpleasantgetonerespon

andbothotherandunpleasantgetthealternativerespon.(Orderofthetwo

combinedtasksiscounterbalancedacrosssubjects.)TheIATeffectisthedif-

subjectswithhighimplicitlf-esteem,thelf+pleasantcombinedtaskis

expectedtobeperformedsubstantiallymorerapidlythanthelf+unpleasant

combinedtask.

PLEASANTUNPLEASANT

cuddlePain

happyAwful

smileDisaster

joyGrief

EXPERIMENTI

Inaninitialstudyofsmokers'implicitattitudestowardsmoking,Experiment1

contrastedsmokingwithtwodifferenttargetconcepts(sweetsorexerci)to

createIATsthatmightdiscriminatebetweensmokersandnonsmokers,provided

smokers'iceofcontrast

categorieswasbadonthehypothesisthatnonsmokersshouldprefersweetsto

smokingasanoralgratification,whereassmokersmightshowareverpattern.

Inaddition,nonsmokersshouldpreferahealthybehaviour(exerci)toan

unhealthybehaviour(smoking),whereassmokersmightnotshowapreference.

Finally,lf-reportattitudestowardsmokingandeithersweetsorexerciwere

assdforcomparisonpurposandwereexpectedtodiscriminatebetween

smokersandnonsmokers.

Method

Subjects.

Thewere93undergraduatesattheUniversityofWashington

tswhowereex-

smokers(n=9)alsampleconsistedof

38smokersand46nonsmokers.

Materials

tscompletedameasurethatallowedustoclassify

tsalsocompletedatofeight

manticdifferentialitemsforeachtargetconcept(smokingandsweetsor

exerci).Each7-pointitemconsistedofpolar-oppositeadjectivepairs(good-

bad,healthy-unhealthy,xy-unxy,pleasant-unpleasant,harmless-harmful,

sociable-unsociable,ugly-glamorous,calming-stressful).Subjectswerein-

structedtocheckthemiddlectioniftheattributedimensionwasirrelevant

itescoresforeachtargetconcept(e.g.,smoking)

werecalculatedbyscoringthe7-ptscalefrom

-3to+3andsummingthe

rencescore

thatcorrespondedtotheIATtarget-conceptdiscriminationwascalculatedby

takingthecompositescoresforthetwotargetconceptsandsubtractingone

ca,highscoresreflectmorepositiveattitudestoward

smoking(comparedtoexerciorsweets).

Finally,subjectsindicatedonafeeling"thermometer"howfavourablethey

ermometerwaslabelledin10degree

tion,0waslabelledas"extremelycold

orunfavourable",50as"neutral",and99as"extremelywarmorfavourable".

ThermometerdifferencescoresthatcorrespondedtoeachoftheIATtarget

conceptdiscriminationswerecalculatedbytakingthethermometerscoresfor

ca,high

scoresreflectmorepositiveattitudestowardsmoking(comparedtoexercior

sweets).

tscompletedanIATmeasuringimplicitattitudes

esubjectscompletedanIATthatcontrastedsmoking

wi

otherhalfofthesubjectscompletedasimilarIATthatcontrastedsmokingwith

king(e.g.,cigarettes,ashtray),exerci(e.g.,biking,jogging),

andsweets(e.g.,candy,cookies)

pleasantandunpleasantattributeswerelectedfromBellezza,Greenwald,and

212SWANSON,RUDMAN,GREENWALD

Banaji(1986).Acompletelistofthestimuliudinthethreeexperimentsis

includedintheAppendix.

TheIATwasadministeredonIBM-compatibledesktopcomputers.2Subjects

respondedtothecategorisationtaskbypressingeitherthe"A"keywiththeleft

forefingerorthe"5"

stimuluswasprentedinblacklettersinalightgreyrectangleinthecentreof

gramrandomlylectedwithoutreplacementitemsfromthe

stimuluslistswhilenotallowingmorethanthreeitemsinarowthatwouldbe

rtrialintervalof150mswas

sideofthestimulusrectanglewerelabelstoremindsubjectsof

ubjectresponded

correctly,agreencircleappearedinasmallboxdirectlybelowthestimulusand

ubjectrespondedincorrectly,a

red"X"appearedintheboxandremainedonthescreenalongwiththesti-

mulus,untilthesubjectrespondedcorrectly.

Procedure.

Onenteringthelab,subjectswereassignedtoindividualbooths

tscompletedtheexplicitmeasuresand

wereinstructedtoplacethemdirectlyintoaboxmarked"completed

questionnaires"erimenterthenadminis-

teredtheIAT,instructingsubjectstorespondtothestimuliasquicklyand

taskconsistedofvenblocksoftrials:(1)

practiceofsinglecategorisationtaskfortheattribute(e.g.,unpleasantlpleasant);

(2)practiceofsinglecategorisationtaskforthetargetconcept(e.g.,smoking1

exerci);(3)practiceofcombinedcategorisationtask(g+unplea-

sant/exerci+pleasant);(4)criticaltrialsfortheblock3combinedcategorisa-

tiontask;(5)practiceofsinglecategorisationtaskfortheattributedimension,

butwiththeresponkeysreverdfromtheblock1assignment;(6)practiceof

combinedcategorisationtask(e.g.,smoking+pleasant/exerci+unpleasant);(7)

nwhichsubjects

performedthemixedcategorisationblocks(i.e.,blocks34and6-7)was

acticeblockhad20trialsandeachcriticalblockhad40

letionofthecomputertask,subjectsweredebriefedandthanked.

Resultsanddiscussion

rocedureswereconsistentwithGreenwaldetal.

(1998).Thefirsttwotrialsineachblockwerediscardedbecautherespon

thathadlatenciesgreaterthan3000msor

shorterthan300mswererecodedto3000msand300ms,respectivelytocontrol

hixperimentudthe2/17/97versionoftheWinIATprogramdevelopedbyShelly

Famham.

ieswerelog-transformedtomeet

distributionalassumptionsforanalysisofvariance.

bject'ssmokingIATeffectwascalculatedby

takingthelatencyforthesmoking+unpleasanttaskminusthelatencyforthe

smoking+,morepositivescoresindicatedgreaterfacilityfor

thesmoking+pleasanttaskthanthesmoking+unpleasanttaskandwereinter-

pretedasmorefavourableimplicitattitudestowardsmokingrelativetothe

contrastcategory(i.e.,sweetsorexerci).Becauthecontrastcategoriesdid

notinfluenceresults,F(l,83)=0.23,p=.633,theywerecombinedforthe

remaininganalys.

Ifsmokers'implicitattitudesareconsistentwiththeirbehaviour,theirIAT

r,smokers

andnonsmokersalikestronglypreferredthecontrastcategoryoversmoking

(Ms=-300msvs.-354ms,respectively),andtheirIATeffectsdidnotdiffer

significantly,F(1,83)=0.83,p=.rast,theexplicitmeasures

,smokerslikedsmokingrelative

tothecontrastcategorymorethandidnonsmokers,usingboththethermo-

meter,F(1,82)=18.52,p=lo-*andthemanticdifferential,F(1,82)=

10.62,p=,indingssuggestthatsmokerscognitivelyaccommodate

theirstigmatidbehaviourattheexplicit,butnotimplicit,level.

ThecorrelationsbetweentheattitudeIATandtheexplicitmeasureswere

significantwhenthethermometerwasud,r(80)=.30,p=,007,ormarginally

significantwhenthemanticdifferentialwasud,r(80)=.21,p=.

explicitattitudemeasureswerealsorelated,r(80)=.52,p=lop7.

Thefindingsthatsmokersandnonsmokershavecomparablynegative

implicitattitudestowardsmoking,whereaxplicitmeasuresdiscriminated

them,suggestthatsmokersaremoresuccessfulatbolsteringtheirsmoking

r,analternativeexplanation

smokersdissociatethemlvesfromanactivitytheydislike(alderlypeople

dissociatedfromtheiragegroup;Greenwaldetal.,inpress),theircognitions

,Experiment2wasconducted,inpart,to

testdifferencesinimplicitidentificationwithsmokingbetweensmokersand

tion,Experiment2soughttocomparethepsychological

characteristicsofstigmatidactors(smokers)andnonstigmatidactors

(vegetariansandomnivores).

EXPERIMENT2

Thelackofdifferencesinsmokers'andnonsmokers'implicitattitudesin

Experiment1suggestedthatsmokerngageinabehaviourtheydonot

r,thecontrastsudinExperiment1werepositivefor

bothsmokersandnonsmokers(sweetsandexerci).Oneobjectiveof

214SWANSON,RUDMAN,GREENWALD

Experiment2wastotestimplicitattitudestowardsmokingusinganegative

contrast(stealing).Inthisca,thecontrastcategoryivenmorestigmatid

eforeexpectedsmo-

kersandnonsmokersaliketoprefersmokingtostealing,butifimplicitattitudes

forsmokerswereconsistentwiththeirbehaviour,weexpectedsmokerstoshow

thispreferencemoresothannonsmokers.

Wealsoexaminedtheextenttowhichsmokersandnonsmokersimplicitly

identifiedwithsmokingversustealing.^Weexpectedsmokerstoidentifymore

withsmokingthanwithstealing,andtoshowthisidentificationmorethan

ersshowedgreatertendencytoidentifywithsmoking,but

nonethelesspossdimplicitattitudesthatweresimilartothoofnon-

smokers,theresultswouldsuggestthatsmokers'behaviour-relevantcognitions

areindeedinconsistentattheimplicitlevel.

Experiment2alsoexaminedimplicitandexplicitattitudestowardvegetar-

ianism,ectivewastoreplicateearlier

findingsindicatingattitude-behaviourconsistencyamongvegetariansand

omnivoreswithrespecttoeatingmeatversusothersourcesofprotein(Swanson

&Greenwald,1998).Thecontrastsudwerewhitemeatversusotherprotein.

SwansonandGreenwald(1997)showedthatwhitemeatwaatenmorefre-

quentlyandlikedmore(explicitlyandimplicitly)byomnivoresthanredmeat.

Thecategoryofotherproteincontainedsourcesofproteinthatmostlacto-ovo

vegetariansuinplaceofmeat(e.g.,tofuandnuts).Becauvegetariansand

omnivoresarenonstigmatidactors,weexpectedeachgrouptoshowconsistent

relationsbetweentheirattitudestowardthefoodstheyate,identificationwith

theirstatusasvegetariansoromnivores,andtheirbehaviour(ealsoRozin,

Markwith,&Stoess,1997).Theconsistentcognitionscouldbecharacterid

as,"IfIdoX,thenIidentifywithX,andXisgood7'(,1958).Thus,

vegetariansshouldidentifywithotherproteinsandhavemorefavourable

attitudestowardotherprotein(andlessfavourableattitudestowardmeat),

redictionswereexaminedusingimplicitand

explicitmeasures.

Method

ere113undergraduatepsychologystudentsatthe

ts

whowereex-smokerswereexcludedfromthesmokingIAT(n=7),andsubjects

whowereex-vegetarianswereexcludedfromthevegetarianIAT(n=5).Four

subjectswereexcludedfrombothIATsonthebasisoftheirlatencydata(e.g.,

duetoerrorrates>25%);inaddition,5and3subjectswereexcludedfromthe

PastrearchhasshowntheIATtobeaneffectivemeasureofimplicitlf-conceptandidentity

(e.g.,Farnham,Banaji,&Greenwald,1999;ealsoRudman,Greenwald,&McGhee,inpress).

smokingandvegetarianIATs,respectively,alsample

sizesconsistedof59nonsmokers,37smokers,66omnivores,and34

vegetarians.

Materialsandprocedure

tscompletedameasurethatallowedustoclassify

so

completedameasurethatinquiredabouttheirsmokingbehaviour,including

armeasureassdthenumberof

timesperyearthatsubjectsatewhitemeatandothersourcesofprotein.

Subjectsalsocompletedatofsixmanticdifferentialitemsforeachofthe

fourtargetconcepts(smoking,stealing,whitemeat,otherprotein).Each7-point

itemconsistedofpolar-oppositeadjectivepairs(beautiful-ugly,good-bad,

pleasant-unpleasant,honest-dishonest,nice-awful,andharmless-harmful).

Subjectswereinstructedtocheckthemiddlectioniftheattributedimension

itescoresforeachtargetconcept

(e.g.,smoking)werecalculatedbyscoringthe7-ptscalefrom-3to+3and

ence

scoresthatcorrespondtoeachoftheIATtarget-conceptdiscriminationswere

calculatedbytakingthecompositescoresforthetwotargetconceptsandsub-

ca,highscoresreflectmorepositive

attitudestowardsmoking(comparedtostealing)andtowardotherprotein

(comparedtowhitemeat).

Finally,subjectsindicatedonafeelingthermometerhowfavourabletheyfelt

lingthermometerwasidenticalin

-

meterdifferencescoresthatcorrespondtoeachoftheIATtargetconceptdis-

criminationswerecalculatedbytakingthethermometerscoresforthetwotarget

ca,highscoresreflect

morepositiveattitudestowardsmoking(comparedtostealing)andtowardother

protein(comparedtowhitemeat).

tscompletedatotaloffourIATs:twoimplicit

target-concept

discriminationsudforeachtypeofIATweresmokingversusstealingand

thewaspairedwiththeattribute

dimensionofpleasantversusunpleasanttoasssattitudes,andwiththeattri-

butedimensionoflfversusothertoasssidentification.

Thelf,other,andwhitemeatcategorieachhadthreestimuliduetothe

difficultyoffindingitemsthatweregoodexemplarsandknowntomostpeople.

Thethreelfandthreeotherstimuliconsistedofpronounsthatreferredtolf

(i.e.,me,mine,lf)orother(i.e.,they,them,other),andthathavebeenud

successfullyinpriorrearchtomeasureimplicitidentification(e.g.,Farnhamet

al.,1999;ealsoRudman,Greenwald,&McGhee,inpress).Thethreewhite

216SWANSON,RUDMAN,GREENWALD

meat(chicken,turkey,poultry)andsixotherprotein(e.g.,tofu,nuts,chee)

itemswerefromSwansonandGreenwald(1997).Thesixsmokingitems(e.g.,

smoke,cigarette)andthesixstealingstimuli(e.g.,steal,theft)weregenerated

pleasantandsixunpleasantstimuliwerelectedfrom

Greenwaldetal.(1998).Acompletelistofthestimuliudinalltheexperi-

mentsisincludedintheAppendix.

ThesameprocedurewasudasinExperiment1,withtheexceptionthat

subjectsperformedtwoIATsinsteadofone(IATorderwascounterbalanced)

andanewerversionoftheIATsoftwarewasud(Farnham,1997,version41171

97).

Resultsanddiscussion

bject'svegetarianattitude

IATeffectwascalculatedbytakingthelatencyfortheotherprotein+unpleasant

taskminusthelatencyfortheotherprotein+,morepositive

scoresindicatedfavourableimplicitattitudestowardsotherproteinrelativeto

ogousprocedurewasudtocalculatethevegetarianlf-

conceptIATsuchthatmorepositivescoresindicatedstrongeridentificationwith

otherproteinthanwhitemeat.

Itwaspredictedthatvegetarianswouldhavemorefavourableattitudes

towardotherproteinthanmeatandidentifywithotherproteinmorethanmeat.

Omnivoreswereexpectedtohavemorefavourableattitudestowardmeatthan

1reveals

thatvegetarianspreferredotherproteintomeat(M=114ms)andomnivores

preferredmeattootherprotein(M=-70ms).Omnivoresandvegetarians

implicitattitudesweresignificantlydifferent,F(1,76)=24.03,p=lop!The

effectsizeforthisdifferencewaslarge,d=reffectmerged,with

theexceptionofanuninterpretableinteractionbetweentheproceduralvariables,

IATeffect,anddiet,F(2,76)=3.17,p=.05.

Vegetariansalsoimplicitlyidentifiedmorewithotherproteinthanmeat

(M=66ms),andomnivoresimplicitlymorewithmeatthanotherprotein

(M=-46ms).Omnivores'andvegetarians'implicitidentificationwithother

proteinandmeatwassignificantlydifferent,F(1,76)=15.19,p=104,andthe

effectsizeforthisdifferencewaslarge,d=.r,thisdifferencewas

-

ferencesbetweenomnivoresandvegetariansdecreadthelaterthedietarylf-

conceptIATwasprented,F(2,76)=4.14,p=.020.

Bothexplicitmeasuresindicatedthatvegetarianspreferredotherproteinto

ect

sizesforthegroupdifferenceswerelarge(ds>2.00;eTable1).Insum,

vegetariansandomnivoresalikeshowedcognitiveconsistencybetweenlf-

218SWANSON,RUDMAN,GREENWALD

TABLE2

Correlationsamongimplicitandexplicitmeasures(Experiment2)

MeasuresI23456

eatcomparison

Implicitmeasures

rotein+PleasantaIAT

rotein+Meb

IAT

Explicitmeasures

meter(prefersother

pr~tein)~

icdifferential(prefersother

protein)"

siyreatwhitemeat

siyreatotherprotein

ngcomparison

Implicitmeasures

g+PleasantCIAT

g+MedIAT

Explicitmeasures

meter(preferssmoking)'

icdifferential(prefers

smoking)'

rettessmokedlday

Bold=p<.s=p<,eatcomparison,thelowerhalf

ofthequadrantcontainsthecorrelationsforallsubjects(Nsrangefrom101to107)andtheupper

halfcontainsthecorrelationsforvegetarians(Nsrangefrom32to34).ng

comparison,thelowerhalfofthequadrantcontainsthecorrelationsforallsubjects(Nsrangefrom98

to104)andtheupperhalfofthequadrantcontainsthecorrelationsforsmokersonly(Nsrangefrom

35to40).

aAttitudemeasuresarescoredsomorepositivescoresindicatemorefavourableattitudestoward

otherproteinrelativetowhitemeat.

IdentificationIATisscoredsomorepositivescoresindicategreaterassociationoflfwith

otherproteinthanlfwithwhitemeat.

'Attitudemeasuresarescoressomorepositivescoresindicatemorefavourableattitudestoward

smokingrelativetostealing.

d~dentificationIATisscoredsomorepositivescoresindicategreaterassociationoflfwith

smokingthanlfwithstealing.

implicitidentityandexplicitattitudemeasures(withrsrangingfrom.40toS4).

Thus,vegetariansandomnivoresshowedconvergenceamongimplicitand

onally,lf-reported

behaviour(frequencyofeatingwhitemeatandotherprotein)eachcorrelatedin

theexpecteddirectionwithimplicitattitudes,implicitidentification,andexplicit

attitudes(i.e.,negativeforwhitemeat,butpositiveforotherprotein).4The

resultsshowthatwhenbehavioursarenonstigmatid,therelationsbetween

implicitandexplicitmeasuresarerobust(Swanson&Greenwald,1998).Per-

hapsduetodiminishedpower,thecorrelationsforvegetariansalone(topmatrix)

wereintheexpecteddirection,butonlyreachedsignificancewhenmeasures

werematchedonmethod(i.e.,thetwoimplicitmeasureswererelated,aswere

veraloftheexplicitmeasures).

bject'ssmokingIATeffectwas

calculatedbytakingthelatencyforthesmoking+unpleasanttaskminusthe

latencyforthesmoking+,morepositivescoresindicated

greaterfacilityforthesmoking+pleasanttaskthanthesmoking+unpleasant

taskandwereinterpretedasmorefavourableimplicitattitudestowardsmoking

ogousprocedurewasudtocalculatethesmoking

lf-conceptIAT,suchthatmorepositivescoresindicatedstrongeridentification

withsmokingthanstealing.

Becaustealingismorestigmatidthansmoking,itwaxpectedthatboth

smokersandnonsmokerswouldhavemorefavourableattitudestowardsmoking

r,ifsmokers'implicitattitudeswereconsistentwiththeir

behaviour,1

revealsthatbothsmokersandnonsmokershadmorefavourableimplicitatti-

tudestowardsmokingrelativetostealing(Ms=173msvs.137ms),andthatthe

differenceingroupmeanswasnonsignificant,F(1,72)=2.30,p=.-

theless,smokersmightshowconsistentbehaviour-relevantcognitionsifthey

r,axpected,smokers'

identificationwithsmokingwassignificantlygreaterthannonsmokers'identi-

fication,Ms=140msvs.93ms,F(1,72)=9.61,~=,ectsizeforthis

differencewaslargerthantheattitudeeffectsize(ds=.42vs..27).Becau

smokersautomaticallyidentifiedwithabehaviourmorethannonsmokers,but

nonethelessdidnotimplicitlylikethebehaviourmorethannonsmokers,their

implicitattitudeswereinconsistentwiththeirbehaviour,

contrast,bothexplicitmeasuresindicatedthatsmokerspreferredsmokingover

stealingmoresothannonsmokers,whoattitudesshowedlittlepreferencefor

ifferencesinsmokers'andnonsmokers'explicitatti-

tudesweresignificant(eTable1).

ThelowerhalfofTable2showsthecorrelationsamongdependentmeasures

forsmokersandnonsmokers(lowermatrix)ations

betweenimplicitandexplicitattitudemeasureswererelativelyweak,compared

tothoforthevegetariansandomnivores(allrs<.15).Nonetheless,implicit

identificationcovariedwithimplicitandexplicitattitudemeasures,andwith

Subjectswe

lf-definedvegetarians(11outof34)reportedeatingwhitemeatinfrequently.

220SWANSON,RUDMAN,GREENWALD

lf-reportedbehaviour(numberofcigarettessmokedperday),suggestingthat

greaterfrequencyofsmokingwasassociatedwithstrongerlf-identityasa

smoker,licitattitudemeasures

correlatedpositivelywiththemlves,

correlationsforsmokersalone(uppermatrix)wereexaminedforevidencefor

een,the

relationswerepositivebetweenimplicitattitudesandimplicitidentification,and

r,thecorrelationsbetween

implicitandexplicitmeasureswerenotinthepredicteddirectionandhovered

esultsshowthatwhenabehaviourisstigmatid(i.e.,

smoking),theconvergencebetweenimplicitandexplicitmeasuresisrelatively

weak,comparedtowhenabehaviourisnotstigmatid(i.e.,dietarypreference).

Insum,Experiment2replicatedSwansonandGreenwald(1998),showing

thatomnivoresandvegetarianshaveconsistentimplicitandexplicitcognitions

respreferredwhitemeatto

r-

ianspreferredotherproteintowhitemeatandidentifiedmorewithotherprotein

ment2alsoshowedthatsmokersandnonsmokersdid

notdifferintheirimplicitattitudestowardsmoking,althoughsmokersdid

ertwith

Experiment

1,thefindingssuggestthatsmokers'implicitattitudesare

rast,andasin

Experiment1,smokers'explicitattitudestowardsmokingweremorepositive

thannonsmokers'attitudes,suggestingthatsmokers'explicitcognitionsare

ogether,thefindingssuggestthatsmokers'cognitivebol-

steringoftheirbehaviourmaybemorelikelyattheexplicitthanimplicitlevel.

EXPERIMENT3

stgoalwastotestimplicitcognitions

associatedwithsmoking,es

inIATtechnologyallowingpicturestimulimadeusingthecontrastcategoryof

ically,picturesweretakenofcommon

condversionwasidentical,exceptfortheabnceofthecigaretteandashtray.

Experiment3alsoexaminedimplicitidentificationforsmokersversusnon-

smokers,ntrastcategories

correspondtothebehaviouranditsopposite(king),people

whoengageinabehaviourshouldpreferittoitsopposite(e.g.,smokersshould

prefersmokingtononsmoking)andbeidentifiedwithitmorethanitsopposite

(e.g.,smokersshouldidentifymorewithsmokingthannonsmoking).Evidence

ofconsistencyamongsmokers'behaviour-relevantcognitionsrequiresshowing

thatsmokersprefersmokingovernonsmokingandshowingadifferencein

smokers'andnonsmokers'attitudesthatmatchestheirexpecteddifferencesin

identificationwithsmoking.

Thecondgoalwastotestthepossibilitythatsmokersmightachieve

ternof

consistentcognitionscanbecharacteridas"If

IdoX,andIidentifywithX,

andXisbad,thenIamalsobad".Therefore,itwasimportanttoexamine

whethersmokers'epastrearch

hasshownrobustimplicitlf-esteemforavarietyofsocialgroups(Farnhamet

al.,1999),itwashypothesidthatsmokerswouldhaveequallypositiveimplicit

ult,anyevidenceforinconsistencyamong

smokers'behaviour-relevantcognitionswouldnotbeattributabletolowered

lf-esteem.

Method

ere87undergraduatepsychologystudentsatthe

thesubjects,53werelf-reportednonsmokersand43werelf-reported

of21subjects(12nonsmokersand9smokers)wereexcluded

fromallanalysfortechnicalreasons(e.g.,higherrorrates).'Thesimilarity

betweenthesmokingandnonsmokingpictureswashigherthanwhatis

generallyfoundbetweenthetargetcontraststimuliandmayhaveledtothe

alsampleconsistedof35smokersand41

nonsmokers.

Materialsandprocedure

gbehaviourwasassdasinExperiment2.

AttitudestowardsmokingwereassdsimilarlyasinExperiment1,withthe

exceptionthatonlyasinglefeelingthermometerandasinglemantic

differentialwereud(eachwerelabelled"Smoking").Self-esteemwas

measuredusingtheRonbergSelf-EsteemScale(Ronberg,1979)anda

feelingthermometermeasure(labelled"Yourlf').

tscompletedthreeIATsthatassdattitudes

towardsmoking,identificationwithsmoking,

attitudeandidentificationIATs,thetargetconceptsweresmokingversusnon-

g

versusnonsmokingpicturesvariedonlyintheprenceversusabnceofa

tingswerecommondomesticsituationsinwhich

onemightsmoke(e.g.,readingthenewspaperatatable;eAppendix).The

Examinationofthepracticeblockdistinguishingsmokingandnonsmokingpicturesindicated

thatsmokersandnonsmokersperformedequallywell(bothintermsoflatencyanderrors)atthis

onally,allanalysreportedintheresultsctionwererepeatedwiththe

subjectsincluded,andshowednochangeinthepatternofresultsprentedherein.

222SWANSON,RUDMAN,GREENWALD

attitudeIATpairedthepictureswithwordsthatwerepleasantorunpleasantin

ntificationIATpairedthepictureswithlfversusother

f-esteemIATudthesamelfversusotherwords,pairedwith

thepleasantandunpleasantwordsudintheattitudeIAT(eAppendix).

TheprocedurewasidenticaltothatofExperiment2withthreeexceptions.

First,theIATswereadministeredusingasoftwareprogramthatallowsboth

picturesandwordstobeudasstimuli6Second,subjectsperformedthreeIATs

(IATorderwascounterbalanced).Third,theIATpracticeblocksthatfamil-

thando

singlecategorisationpracticeblocksatthestartofeachIATtask,subjectsdid

fiveinitialblockstopracticethefollowingdiscriminations(intheorderlisted):

(1)smokinglnonsmokingpicturesfrompleasantlunpleasantwords;(2)pleasant1

unpleasantwordsfromlflotherwords;(3)pleasantfromunpleasantwords;(4)

lffromotherwords;and(5)tsthen

completedthemixedcategorisationtasks(e.g.,smoking+unpleasantinon-

smoking+pleasant)forthethreeIATsasinExperiment2(onepracticeblock

andonecriticalblockpertask).

Resultsanddiscussion

Subject'sattitudeandlf-conceptIATeffectswerecalculatedasinExperiment

ca,positivescoresindicatemorefavourableattitudestoward,and

identificationwith,f-esteemTATwas

scoredsuchthatmorepositivescoresindicatemorefavourablethanunfavour-

erencesduetoproceduralvariableswere

found;therefore,theanalysreportedbelowdonotincludethem.

3showstheresultsof

Experiment3'een,smokers'

implicitattitudesrevealedapreferencefornonsmokingoversmoking

(M=-69ms),eventhoughtheyidentifiedwithsmokingmorethannonsmoking

(M=125ms).Incontrast,nonsmokers'implicitattitudesshowedastrong

preferencefornonsmokingoversmoking(M=-245ms),andtheyidentified

withnonsmokingmorethansmoking(M=-20ms).ConsistentwithExperi-

ment2,thispatternshowsmoreinconsistentimplicitcognitionsforsmokers

thannonsmokersthatisduetosmokershavingattitudesinconsistentwiththeir

3alsorevealsthatsmokers'implicit

lf-esteem(M=322ms)wasaspositiveasnonsmokers'implicitlf-esteem

(M=330ms).Thus,smokersdidnotachieveconsistencyamongtheir

behaviour-relevantcognitionsvialowlf-esteem.

heprogramwasInquisit,writtenbySeanDraine(Draine,1998).

TABLE3

Summarystatisticsforimplicitandexplicitmeasures(Experiment3)

Measure

SmokersNonsmokers

Difference

(n=35)(n=41)Cohen's

MOD)M)ddpb

Implicitmeasures

Smoking+PleasantIATc

-69.4

(244.9)-245.3

(257.8).70,008

Smoking+MeIAT~125.3

(228.5)-20.1(192.1).71

,002

Me+PleasantIATe322.2(175.9)

329.5(143.2)-.04,371

Explicitmeasures

Smokingthermometerf45.3(23.7)

16.9

(17.9)1.36lo-'

Smokingmanticdifferentialg

-7.7

(5.6)

-13.5

(3.3)

1.33lo-'

Selfthennomete?'82.7(13.1)

84.2

(13.4)-.08,630

RonbergSelf-EsteemScale'

23.5

(5.6)

24.2

(4.7)

-.I4,610

"Theeffectsizemeasure,dwascomputedbydividingmeandifferencesbytheirpooledSDs.

Conventionalsmall,medium,andlargeeffectsfordare.2,.5,and.8,respectively.

bp-valuescorrespondtot-testsofthedifferencesbetweensmokersandnonsmokers.

king.

d.H~gherscoresreflectstrongerassociationbetweenlfandsmokingthanlfandnonsmoking.

'.

t.

angesfrom0to100with

50beingneutral.

angesfrom-18to18

with0beingneutral.

angesfrom0to100with50beingneutral.

'angesfrom0to30with15beingneutral.

Table3alsoshowsthatusingpicturestimulitooperationaliacontrast

betweensmokingandnonsmokingenhancedtheabilityoftheattitudeIATto

discriminatebetweensmokersandnonsmokers,t(74)=2.73,p=.ect

sizeforthisdifferencewasmoderatelylarge(d=.70).Thisfindingsuggeststhat

nonsmokingmaybethemostappropriatecontrasttouwhenasssingimplicit

attitudestowardsmoking,comparedtocontraststhatarepositiveforbothgroups

(e.g.,exerci)ornegativeforbothgroups(e.g.,stealing).Consistentwith

Experiment2,theidentificationIATcontinuedtodiscriminatebetweenthe

groups,despitethesubstitutionofpicturestimuliforwords,t(74)=-3.17,

p=.ectsizeforthisdifferencewascomparabletothatshownin

Experiment2(d=.71).Thus,thechangeinstimulusmodeappearstohave

improvedattitudeasssmentwithoutdiminishinglf-conceptasssment.

Thethermometerandmanticdifferentialmeasurescontinuedtodis-

criminatebetweensmokersandnonsmokers,asinExperiments1and2(e

Table3).Nonetheless,whenasingleattitudeobjectwasud("smoking"),

smokers'attitudeswere,onaverage,neutraltosomewhatunfavourable,albeit

224SWANSON,RUDMAN,GREENWALD

indingsareconsistentwithprior

rearch(Chassinetal.,1991;Stacyetal.,1994)andsuggestthatsmokersmay

bolstertheirbehaviourbyviewingtheirhabitsomewhatfavourably,compared

y,smokers'andnonsmokers'explicitlf-esteemwere

comparable,asassdbyalf(feeling)thermometerandtheRonbergSelf-

EsteemScale(eTable3),againshowingthatsmokersdidnotachievecon-

sistencybyloweringtheirlf-esteem.

Table4showstherelationshipsamongExperiment3'sdependentmeasures,

forsmokersandnonsmokerscombined(lowermatrix)andforsmokersonly

(uppermatrix).ReplicatingExperiment2,thelowermatrixshowscovariation

betweentheattitudeandlf-conceptIATs,andimplicitlf-conceptwas

positivelycorrelatedwiththeexplicitattitudemeasuresandlf-reported

behaviour(numberofcigarettessmokedperday).AsinExperiment2,the

explicitattitudemeasuresalsocovariedandwereeachrelatedtolf-reported

tion,theattitudeIATwaspositivelyrelatedtoeachexplicit

r,theuppermatrixshowsthatforsmokersalone,the

een,onlythe

twoexplicitattitudemeasuresandtwoexplicitlf-esteemmeasuresreliably

y,theimplicitandexplicitlf-esteemmeasureswerenegligibly

relatedtoanyofExperiment3'sprimarydependentmeasures(attitudes,lf-

concept,andlf-reportedbehaviour).Thelackofrelationshipbetweenthe

implicitandexplicitlf-esteemmeasuresisconsistentwithpastrearch

showingthatthetwoconstructsareindependent(Farnhametal.,1999).

Insum,Experiment3providedadditionalevidencethatsmokers'implicit

mplicitlevel,smokers

hadpositivelf-esteem,identifiedmorewithsmokingthannonsmoking,but

rast,nonsmokershadpositivelf-

esteem,identifiedwithnonsmokingmorethansmoking,andpreferrednon-

atasuggestthatsmokersaremorelikelytohave

implicitattitudesthatareinconsistentwiththeirbehaviourthannonsmokers.

Additionally,Experiment3suggestedthatsmokersmayexplicitlybolstertheir

habitbyviewingtheirbehaviourmorefavourablythannonsmokersdo(i.e.,as

somewhatneutralratherthannegative).

GENERALDISCUSSION

Asperformersofastigmatidbehaviour,smokershavebeenobrvedto

consciouslyreconciletheirperformanceofthebehaviourwiththeirnegative

knowledgeconcerningit(Chassinetal.,1991;Halpern,1994;Johnson,1968).

However,becausmokersregularlyandfrequentlyconfrontlawsthatrestrict

theirbehaviour,disapprovalfromothers,andinformationcampaignsabout

smoking'sadvereffects,itispossiblethattheymaynotbeabletoresolvethis

inconsistencyattheimplicitlevel.

TABLE4

Correlationsamongimplicitandexplicitmeasures(Experiment3)

Measure

ImplicitmeasuresExplicitmeasures

I2345678

Implicitmeasures

g+PleasantIAT

g+MeIAT

+PleasantIAT

Explicitmeasures

gthermometer

gmantic

differential

ermometer

ergSelf-EsteemScale

rettessmokediday

Bold=p<.s=p<.esarescoredsomorepositivescoresindicateahigherleveloftheconstruct

erhalfofthequadrantcontainsthecorrelationsforallsubjects(Nsrangefrom70to76)andthe

upperhalfofthequadrantcontainsthecorrelationsforsmokersonly(Nsrangefrom33to35).

226SWANSON,RUDMAN,GREENWALD

ri-

ments1and2,smokers'implicitattitudestowardssmokingweresimilarto

thoofnonsmokers,andinExperiment3smokersshowedgreaterimplicit

er,inExperiments2and3,

smokersstronglyidentifiedwithabehaviourtheydidnotimplicitlylike,even

ternofimplicitinconsistencyfor

smokerscanbecharacteridas"Iamgood,andIidentifywithsmoking,but

smokingisbad".Bycontrast,thepatternofsmokers'explicitcognitionscanbe

describedas,"Iamgood,andIidentifywithsmoking,andsmokingisnotso

bad".

Toobtaincomparisondataforperformersofnonstigmatidactions,

Experiment2assdbehaviour-relevantcognitionsforvegetariansand

ultsclearlyshowedconsistentcognitionsforvegetariansand

oupidentifiedwiththeirdiet,andshowedpositiveattitudes

towardthefoodstheyate,esults

areconsistentwithviewingnonstigmatidbehavioursasonesthatdonotcreate

dissonantimplicitorexplicitstructures.

Thefindingsdonotobligeconcludingthatsmokerssuffermorefrom

cognitivedissonancethandovegetarians,omnivores,

possiblethattheexperienceofcognitivediscomfortrequiresconsciousaware-

ore,havinginconsistentimplicitcognitionsmay

notproducediscomfortunlesstheyarebroughttopeople'

rearchshouldexaminewhetherapprisingsmokersoftheirincongruent

implicitcognitionsmightfacilitatetheirabilitytoquitsmoking,throughdis-

sonancearousalandlf-regulatoryprocess(eDevine&Monteith,1993,for

areviewofsimilarrearchintheprejudicereductiondomain,andeStoneet

al.,1994,forrelevantrearchconcerning

nonperformed-but-admiredbeha-

viour).

Thebehavioursofsmokingandvegetarianismwerelectedbecauthey

tisation,however,reflectsavariety

ofdimensions(e.g.,healthiness,normativepressures,potentialforaddiction)-

anyone(ormore)ofwhichmaycautheobrveddifferencesincognitive

,whiletheaddictivenatureofsmokingmaycontributetoits

disapprovedofstatus,italsomakesitdifficulttoeliminatedissonanceby

llknownthatsmokersfinditdifficultto

quitsmoking(Hellman,Cummings,Haughey,Zielezny,&O'Shea,1991;Ro,

Chassin,Presson,&Sherman,1996).Thus,theaddictiveelementofsmoking

maybeonereasonwhysmokersmightaccommodatetheirbehaviourratherthan

r,theaddictivenatureofsmoking,themostcommon

reasongivenbysmokersforsmoking,mayalsorvetoalleviatedissonanceby

providingaconsonantcognition(Festinger,1957).Thiscognitioneffectively

dictatestothesmokerthat"It'soutofmycontrol",thusremovinganyfreewill

orintentonthepartofthesmoker.

STIGMATEEDBEHAVIOUR227

Relationshipbetweenimplicitandexplicit

measures

Themajorityofrearchusingimplicitmeasureshasfocudonasssing

mostpart,therelationshipbetweenimplicit

andexplicitmeasuresofaffectandbeliefstowardvarioussocialgroupsisweak

(Braueretal.,2000).Asuggestedinterpretationofthispartialdissociationisthat

lf-reportmeasuresaremoresubjecttocontaminationfromlf-prentation

concernsand/orthatrespondents'unconsciouscognitionsare,bydefinition,

inaccessible(Dovidio&Fazio,1992;Greenwald&Banaji,1995).Theprent

rearchfocudonbehavioursandattitudesthatarestigmatid(inthecaof

smokers)andnonstigmatid(inthecaofvegetarians'andomnivores'diets).

TheresultsofExperiments2and3showedthatsmokers'implicitandexplicit

rast,theresultsofExperiment2showed

thatvegetariansandomnivores'implicitandexplicitattitudesweremoderately

indingssuggestthattherelationshipbetweenimplicit

andexplicitmeasurescanbemoderated-intheprentrearch,bydifferences

licit-implicitlinkmaybestronger

fordietaryattitudesbecautheyarelesssubjecttotheneedforcognitive

accommodation(asisthecaforsmokers).Futurerearchshouldcontinueto

archformoderatorsofimplicitandexplicitrelations,andtoidentifythe

processbywhichconsciousandunconsciousattitudesaredrivenapartor

broughtintoconvergence(etal.,2000).

Manuscriptreceived12May1999

Revidmanuscriptreceived10November1999

REFERENCES

Belleza,F.S.,Greenwald,A.G.,&Banaji,M.R.(1986).Wordshighandlowinpleasantnessasrated

orRearchMethods,Instruments,andComputers,

18,299-303.

Blair,I.V.(inpress).itz(Ed.),Futuredirectionsin

socialcognition.

Brauer,M.,Wal,W.,&Niedenthal,P.(2000).Implicitandexplicitcomponentsofprejudice.

ReviewofGeneralPsychology,4,79-101.

Chassin,L.,Presson,C.C.,Sherman,S.J.,Edwards,D.A.(1991).Fourpathwaystoyoung-adult

smokingstatus:Adolescentsocial-psychologicalantecedentsinaMidwesterncommunity

Psychology,10,409-418.

Devine,P.G.,&Monteith,M.J.(1993).

&on(Eds.),Affect,cognition,andstereotyping:Interactiveprocess

ingroupperception@p.317-344).NewYork:AcademicPress.

Dovidio,J.F.,&Fazio,R.H.(1992).Newtechnologiesforthedirectandindirectasssmentof

(Ed.),Questionsaboutquestions:Inquiriesintothecognitivebasof

surveys(pp.204-237).NewYork,RusllSageFoundation.

Draine,S.(1998).Inquisit[Computersoftwarer].Seattle,WA:ble:

/[version251.

228SWANSON,RUDMAN,GREENWALD

Famham,S.D.(1997).FIATforWindows[Computersoftware].Seattle,WA:ble:

http:/-sfamhamilAT/[2/l7/97&4113/97].

Famham,S.D.,Banaji,M.R.,&Greenwald,A.G.(1999).&M.A.

Hogg,(Eds.),Socialidentityandsocialcognition@p.230-248).Malden,MA:Blackwell.

Fazio,R.H.(1990).Multipleprocessbywhichattitudesguidebehavior:TheMODEmodelasan

(Ed.),Advancesinexperimentalsocialpsychology(Vol.

23,pp.75-109).NewYork:AcademicPress.

Festinger,L.(1957).to,CA:StanfordUniversityPress.

Goldstein,J.(1991).Thestigmatizationofsmokers:lofDrug

Education,21,167-182.

Greenwald,A.G.,&Banaji.M.R.(1995).Implicitsocialcognition:Attitudes,lf-esteem,and

logicalReview,102,4-27.

Greenwald,A.G.,Banaji,M.R.,Rudman,L.A.,Farnham,S.D.,Nok,B.A.,&Mellott,D.S.(1999).

UntJiedtheoryofimplicitsocialcognition:Attitudes,stereotypes,ript

submittedforpublication.

Greenwald,A.G.,Banaji,M.R.,Rudman,L.A.,Famham,S.D.,Nok,B.A.,&Rosier,M.(inpress).

Prologuetoaunifiedtheoryofattitudes,stereotypes,(Ed.),

Feelingandthinking:k:Cambridge

UniversityPress.

Greenwald,A.G.,McGhee,D.E.,&Schwartz,J.L.K.(1998).Measuringindividualdifferencesin

implicitcognition:lofPersonalityandSocialPsychology,

74,1464-1480.

Halpem,M.T.(1994).Effectofsmokingcharacteristicsoncognitivedissonanceincurrentand

iveBehaviors,19,209-217.

Heider,F.(1958).k:Wiley.

Hellman,R.,Cummings,K.M.,Haughey,B.P.,Zielezny,M.A.,&O'Shea,R.M.(1991).Predictors

EducationRearch,6,77-86.

In,A.M.,&Diamond,G.A.(1989).&(Eds.),

Unintendedthought(pp.124-152).NewYork:GuilfordPress.

Johnson,R.E.(1968).lofPersonalityand

SocialPsychology,9,26G265.

McMaster,C.&Lee,C.(1991).iveBehaviors,16,

349-353.

Miller,S.K.,&Slap,G.B.(1989).Adolescentsmoking:Areviewofprevalenceandprevention.

JournalofAdolescentHealthCare,10,129-135.

Ro,J.S.,Chassin,L.,Presson,C.C.,&Sherman,S.J.(1996).Prospectivepredictorsofquit

Psychology,IS,261-268.

Ronberg,M.(1979).ConceivingthelJNewYork:BasicBooks.

Rozin,P.,Markwith,M.,&Stoess,C.(1997).Moralizationandbecomingavegetarian:The

logicalScience,

8,67-73.

Rozin,P.&Singh,L.(1998).-

lishedmanuscript,UniversityofPennsylvannia.

Rudman,L.A.,Ashmore,R.D.,&Gary,M.(2000).Unlearningautomaticbias:Themalle-

riptsubmittedforpublication.

Rudman,L.A.,Greenwald,A.G.,&McGhee,D.E.(inpress).Implicitlf-conceptandevaluative

implicitgenderstereotypes:ali@andSocial

PsychologyBulletin.

Rudman,L.A.,Greenwald,A.G.,Mellott,D.S.,&Schwartz,J.L.K.(1999).Measuringtheautomatic

componentsofprejudice:

Cognition,17,1-29.

Shopland,D.R.,&Brown,C.(1987).Towardthe1990objectivesforsmoking:Measuringthe

HealthReports,102,68-73.

Stacy,A.W.,Bentler,P.M.,&Flay,B.R.(1994).Attitudesandhealthbehaviorindiverpopula-

tions:Drunkdriving,alcoholu,bingeeating,marijuanau,Psy-

chology,13,73-85.

Stone,

J.,Aronson,E.,Crain,A.,Winslow,M.P.,etal.(1994).Inducinghypocrisyasameansof

alit).andSocialPsychologyBulletin,20,116-

128.

Swanson,J.E.,&Greenwald,A.G.(1998,May).Doimplicitattitudesandimplicitlf-concept

discriminatebetweenomnivoresandvegetarians?

prentedatmeetingsoftheMidwesternPsychologicalAssociation,Chicago,IL.

Swanson,J.E.,&Greenwald,A.G.(1997).[Measuringomnivores'implicitattitudestowarddifferent

foods].Unpublishedrawdata.

Thompson,W.E.(1991).Handlingthestigmaofhandlingthedead:Morticiansandfuneraldirectors.

DeviantBehavior,12,403-429.

Thompson,W.E.&Harred,J.L.(1992).Toplessdancers:Managingstigmainadeviantoccupation.

DeviantBehavior,13,291-311.

Wegner,D.M.,&Bargh,J.A.(1998).t,S.T.

Fiske,&y(Eds.),Thehandbookofsocialpsychologv(Vol.1,pp.446496).New

York:OxfordUniversityPress.

Winkielman,P.,Zajonc,R.B.,&Schwartz,M.(1997).Subliminalaffectprimingresistsattributional

ionandEmotion,11,433465.

APPENDIX

Targetconceptsandstimuli

Experiment1

smokingcigarettes,ashtray,tobacco,pipe,smoking,cigars,nicotine,Camels,smokers,Marlboro

exercijog,run,swim,biking,sports,tennis,diving,gymnastics,workout,aerobics

sweetscandy,cookies,cake,pie,pastry,icecream,chocolate,desrt,fudge,sugar

pleasantcaress,gold,joy,kindness,peace,success,sunri,talent,triumph,warmth

unpleasantabu,assault,brutal,junk,war,failure,filth,bad,slime,vomit

Experiment2

whitemeatchicken,turkey,poultry,chicken,turkey,poultry

otherproteinnuts,grains,tofu,chee,soybean,yogurt

smokingsmoke,cigarette,tobacco,smokers,nicotine,ligher

stealingsteal,theft,gun,mugged,robbery,thief

pleasantpeace,paradi,joy,love,cuddle,pleasure

unpleasantdisaster,divorce,crash,grief,tragedy,agony

lfme,mine,lf,me,mine,lf

otherthey,them,other,they,them,other

230SWANSON,RUDMAN,GREENWALD

Experiment3

Scenesudinsmokingandnonsmokingpictures:Besidestablewithlampandclock-radio,End-

tablewithlampandbookopen-faceddown,Kitchentablewithnewspaperspreadopenandacoffee

mug,Twoglassofwateratanoutdoortablewithchairs,Malesmokingcigaretteonbackdoor-

stoop,Bathroomsink,BackdoorstoopwithBBQandglassofjuice,Computerondesk.

pleasantcuddle,happy,smile,joy,warmth,peace,paradi,love

unpleasantpain,awful,disaster,grief,agony,brutal,tragedy,bad

lfme,mine,lf,my,me,mine,lf,my

otherthey,them,their,other,they,them,their,other