I’llHaveWhatShe’sHaving:EffectsofSocialInfluence andBody Type ontheFoodChoicesofOthers
BRENTMCFERRANDARRENW.DAHLGAVANJ.FITZSIMONSANDREAC.MORALES
Thisresearchexamineshowthebodytypeofconsumersaffectsthefoodcon-sumptionofotherconsumersaroundthem.Wefindthatconsumersanchoronthequantitiesothersaroundthemselectbutthattheseportionsareadjustedaccordingtothebodytypeoftheotherconsumer.Wefindthatpeoplechoosealargerportionfollowinganotherconsumerwhofirstselectsalargequantitybutthat
thisportionissignificantlysmalleriftheotherisobesethanifsheisthin.Wealsofindthattheadjustmentismorepronouncedforconsumerswhoarelowinappearanceself-esteemandthatitisattenuatedundercognitiveload.
besityandunhealthyfoodconsumptionaremajorpub-
lichealthissues,especiallyinindustrializedcountries. Insearchingtoidentifyacausefortheepidemic,whilesome
BrentMcFerran(brent.mcferran@ubc.ca)isassistantprofessorofmarke ting,FacultyofManagement,UniversityofBritishColumbia,3333Universi tyWay,Kelowna,BC,CanadaV1V1V7.DarrenW.Dahl(darren.dahl@sauder
.ubc.ca)isFredH.SillerProfessorinAppliedMar-ketingResearch,SauderSchool ofBusiness, University ofBritish Colum-
bia,2053MainMall,Vancouver,BC,CanadaV6T1Z2.GavanJ.Fitz-
simons(gavan@duke.edu)isF.M.KirbyResearchFellowandprofessorofmark etingandpsychologyatFuquaSchoolofBusiness,DukeUniver-
sity,1TowerviewDrive,Durham,NC,USA27708.AndreaC.Morales(acmoral
es@asu.edu)isassociateprofessorofmarketingatW.P.CareySchoolofBusines
s,ArizonaStateUniversity,P.O.Box874106,Tempe,AZ,USA85287.Thisartic
leisbasedonthefirstauthor‘sdoctoraldis-sertation,supervisedbythesecondauthor,withthethirdandfourthauthorsservi ngonthecommittee.FinancialsupportfromtheSocialSciences
andHumanitiesResearchCouncilof Canada awardedtoDarrenDahl isgrate- fullyacknowledged.Thisresearchwonthe2008SocietyforConsumerPsycholog yandAssociationforConsumerResearch– Sheth(PublicPurpose)DissertationProposalcompetitions,andthefinancialsupp ortandhelpfulcommentsreceivedfrombothorganizationsisappreciated.Theaut
horsareindebtedtoYunLiforherresearchassistance,SandraBlackieandAngusSt
rathie for costume designandconstruction,and KarlAquinoandTimSilkforservingonthedissertationcommittee,aswellastoJen
niferArgo,JimBettman,JonahBerger,PierreChandon,ErinChester-McFerran,LauretteDube′,DaleGriffin,JoAndreaHoegg,NaomiMandel,RaviM
ehta,AsheshMukherjee,StewartShapiro,JulietZhu,andtheDukeconsumerbeha viorreadinggroupfortheircommentsandsuggestions. JohnDeightonservedaseditorandBabaShivservedasassociateeditorforthisa rticle.
ElectronicallypublishedAugust25,2009
915
O
authorspointtoamoresedentarylifestyle (BlairandBrod-ney1999)orgenetics(ComuzziandAllison1998),mostresearchispointingtoamarkedincreaseinconsumption(offoodanddrink)asthemaindriverofobesity(ChandonandWansink2007a;HillandPeters1998;YoungandNestle2002).However,giventhatpeopleeatmanymealsinasocialorpublicsetting,itissurprisingthatlittleresearchhasex-aminedhowourfoodchoicesareshapedbythosearoundus.Thisresearchexamineshowviewingotherconsumers‘choicesaffectsthesizeofthefoodportionsweselect.
Whilepriorresearchhasbeguntoshowthatpeople‘sfoodconsumptionchoicesareshapedbysocialandinterpersonalinfluences(e.g.,Herman,Roth,andPolivy2003),
whathasbeenlackingintheliteraturetodateisanexaminationofhowthefoodchoicesconsumersmakeareinfluencedbythebodytypesofotherspresent.Asmanyofourneighbors,friends,andcolleaguesarelikelytobeobese,doeseatingwiththemresultinyourorderinglessormorefood?Doesseeinganobesepersonorderasteakforlunchinfluenceyoutoordermoreorlessfoodyourself?Whatifyousee
athingirlorderalargechocolateparfait?Whatifinsteadofalargeportionshehasaverysmallsaladforlunch?
Weapproachthesequestionsbyfirstreviewingthelit-eratureonsocialinfluence.Weproposethatfoodchoice,likemanyotherbehaviorsinconsumptiondomains,isstronglysubjecttointerpersonalinfluences,withpeoplechoosinglarger(orsmaller)portionsafterviewinganotherconsumerdoinglikewise.Accordingtorecentresearchonreferencegroups,totheextentthatconsumersdonotwishtoemulatemembersofagivengroup,theirconsumption
。2009byJOURNALOFCONSUMERRESEARCH,Inc.●Vol.36●April2010All rights reserved. 0093-5301/2010/3606-0002$10.00. DOI: 10.1086/644611
916
choicesreflectaheighteneddesiretoadjustawayfromchoicesmadebyamemberofthatundesirablegroup.Usingamodelofanchoringandadjustment, we proposethatcon-sumersanchor on the
consumptionquantitydecisionsmadebyotherconsumersaroundthem.However,wearguethatthebodytype(thinvs.obese)ofthisotherconsumerin-teractswithhis/herquantitychoiceininfluencingthesizeoftheportionwechooseandconsumeourselves.
Resultsfromthreeexperimentsareconsistentwiththisframeworkandprovidenewinsightsintotheliteraturesonsocialinfluenceandfoodchoice.Instudy1,weproposeandtestamodelbasedonanchoringandadjustment.Weshowthatconsumersanchoronthequantitychoicesmadebyotherconsumersbutalsoadjusttheirownchoiceandconsumptionbasedonwhethertheotherpersonisamemberofan(un)desirablereferencegroup.Wefindthattheextenttowhichconsumersadjusttheirportiondownwardafterseeinganotherconsumerselectalargeportionismoderatedbythebodytypeofthisotherconsumer.Study2considersthecaseinwhichtheotherconsumersetsupalow,ratherthanhigh,consumptionanchor,anditshowsthatanupwardadjustmentbasedonbodytypecanalsooccur.Study3providesfurtherevidenceintotheprocessunderlyingtheseeffects,identifyingtwomoderators,onesocial(appearanceself-esteem)andtheothercognitive(cognitivebusyness)thataffectourfoodselections.Together,thefindings
ofthethreestudiespresentacomprehensiveexaminationofcon-sumerfoodchoicethatcontributestotheliteraturebyshow-ingwhen(andhow)peoplearelikelytousethebehaviorofothersinshapingtheirownconsumptiondecisions.
CONCEPTUALBACKGROUND
AnchoringandAdjustmentProcesses
Modelsbasedonanchoringandadjustmenthavebeenshowntoberobustinmanycontexts,evenwhenpeoplearehighlymotivatedforaccuracy(EpleyandGilovich2006;JackowitzandKahneman1995;Plous1993).Anchorsserveasreferencepointsthataredifficultforevenexpertstoignore,andtheyrepresent arelatively simple
waytomodelconsumers‘choicesofhowmuchtopurchaseorconsume,decisionsweknowarebasedonmyriadsituationalfactorsinaconsumptionenvironment.Wansink,Kent,andHoch(1998)presentamodelofpurchasequantitybasedonan-choringandadjustment.Intheirmodel,anchorssetupbyaretailerregardingmultipleunitprices,purchasequantitylimits,andsuggestivesellingcanincreasepurchasequan-tities.Forinstance,theretailersetsupananchor(e.g.,―limit12 per person‖) that consumersuseasdiagnosticininform-ing their
ownpurchasequantitydecision.Consumersadjustupwardfromasmalldefaultanchorifapricejustifiesstock-pilinganddownwardifalargeanchorwassetup(e.g.,―buy18foryourfreezer‖).Consistentwithpreviousresearchonanchoringandadjustment,consumerstendtomakeanin-sufficientadjustmentfromtheanchorandenduppurchasingquantitiesthatreflecttheefficacyoftheanchor.
Whiletheanchoringandadjustmentmodelproposedby
JOURNAL OFCONSUMER RESEARCH
Wansinketal.(1998)focusedonanchorsthatretailerscouldsetuptoinfluencepurchasequantitydecisions,weknowthatanchorscancomefromavarietyofsourcesinacon-sumptionenvironment.Weproposethatotherconsumerscan also setup normsofpurchasethatserveasanchorsthatconsumersuseindec
idinghowmuchtoconsume.
SocialInfluencesandFoodChoice
Pastresearchhasshownthatconsumptiondecisionsareinfluencedbythosewhoarephysicallypresent.Peoplearesensitivetothebehaviorofothersinaretailcontext(ArgoandMain2008;BeardenandEtzel1982),evenifsuchapersonisonlyphysicallypresentbutdoesnotengagetheconsumerinanyway(Argo,Dahl,andManchanda2005).Inthedomain of foodconsumption,studieshavefoundthatsocialinfluencecanhaveeitherafacilitatingor
attenuatingeffectoneatingbehavior,dependingonthecontext(seeHerman,Roth,andPolivy[2003]foranexcellentreview).Hermanetal.(2003)arguethatfoodchoiceisinfluencedbyadesiretoconveyacertainimpressionoradheretosocialnorms(LearyandKowalski1990;Rothetal.2001).Theyreviewexperimentsthatshowthat,whena confederate setsupanorm,otherparticipantstendtoeatmore(orless)astheconfederatedoes.Thesenormeffectsareparticularlypoignant:thosewhoarenaturallyinclinedtoeatlargepor-tionseatlessinthepresenceofothers,andthosewhowouldnormallyeatverylittleendupeatingmore.Asthegroupsizeincreases,noonewantstostandout,andpeoplein-creasinglyconformtothegroupaverage(BellandPliner2003).Thisresearchdemonstrates how an anchor setupbyfellowconsumersinfluencesothers‘consumptionquantitydecisions.Sincesocialnormsarepowerful,weexpecttofindthatpeopleanchorontheconsumptionquantitiesofothers,eatingmoreiftheotherconsumersetsupahighanchorversusalowanchor.
However,whilethislineofresearchdemonstratesanef-fectoneatingbehaviorasafunctionofsocialinfluence,itisagnosticwithrespecttowhothe―other‖consumersarethatonemightbeorderingoreatingalongside.Accordingtothisresearch,itshouldmakenodifferenceifthepeopleonemightbesharingamealwitharethinorobesesolongastheychoosethesameamount.However,researchsuggeststhatwedonotperceiveobesepeoplethesamewayaswedonormal-weightindividuals,andthuswemaynotreactinthesamemannert
otheirfoodchoices.
ObesityandConsumption
Somerecentresearchhasbeguntoexaminetheimpactofobeseothersonconsumption.Forexample,primingpeo-plewithoverweightimageshasbeenshowntoleadtoanincreaseinquantityconsumed(CampbellandMohr2008).Usingassimilation/contrastasatheoreticalframework,theseauthorsreportedthatconsumerseatmorewhenprimedwithoverweightbutnotobeseconsumers.Inaninterestingstudy,ChristakisandFowler(2007)foundthataperson‘schanceofbecomingobesesignificantlyincreasedwhenacloseother
BODYTYPESANDFOODCHOICES
(e.g.,friend,sibling,spouse)becameobese(seeCohen-ColeandFletcher[2008]forarebuttal),andotherresearchon―imitative‖obesityhasbeguntoemergeusingeconometrictechniques(Blanchflower,Oswald,
andVanLandegham2008;BurkeandHeiland2007).Thesestudiesignore
whatchoicestheotherpersonhasmade,focusingonlyontheirbodytype,andconcludethateatingwiththosewhoareoverweightwilllead to anincreaseinone‘sfoodcon-sumption;thus,peopleemulateotherstheyarecloseto.However,obesityissomethingmostpeoplewishtoavoid,andresearchhasshownthatweavoidthebehaviorsasso-ciatedwithundesirableoutgroups(includingreducingjunkfoodconsumption;seeBergerandRand2008).
Whiletheresearchoutlinedabovehasfocusedeitheron
consumers‘reactionstohowmuchotherseatorhowthebodytypeofothersaffectsconsumption,
littleworkhasexaminedtheinfluenceofthetwojointly.Weexaminethesefactorssimultaneouslyandpredictthatobservinganother consumerchoosealarge(orsmall)portionwillresultinyoudoinglikewisebutthatthiseffectismoderatedbythebodytype(thinvs.heavy)oftheotherconsumer.
Mostculturescurrentlyplaceahighvalueonthinness,andthosewhoareoverweightorobese are often victimsof
stereotypingorstigmatization(Shapiro,King,andQuinones2007).However,unlikesomestigmas,blameforbeingobeseisattri
buteddirectly
totheindividual,theassumptionbeingthatheorsheisinfullcontrolofhisorherweight(e.g.,Crandall1994;Weiner,Perry,andMagnusson1988).Evenprofessionaldietitians(incorrectly)expectthatobesepeopleunderestimateportionsizes(ChandonandWansink2007b).Consumerresearchhasbeguntoshowthattheeffectsofsocial―others‖aremoderatedbywhetherthepersonisamemberofanaspirationalordissociativegroup(BergerandHeath2007,2008;EscalasandBettman2005;WhiteandDahl2006,2007).Aspirationalgroupsarecirclesthatonewishestobeapartof;diss
ociativegroupshavetheoppositeeffect—
peoplewishtoavoidthem.WhiteandDahl(2006)showedthatmenwerelesslikelytoorderasteakwhenitwaslabeled―ladiescut‖thanwhenitwasnamedthe―chef‘scut.‖Otherresearchhasshownthatpeoplearelikelytoseekoutproductsthatareingroupfavored
butavoidproductsthat are associatedwith outgroups
(BergerandHeath2007;2008)orevenbehaviorslinkedtoan―an
noying‖other(Coo-perandJones1969).Ourresearchextendspastresultsex-aminingdissociativegroupinfluenceonconsumerchoicebyfoc
usingsolelyonhowreferencegroupsaffectthequan-tityselectionsconsumersmake.Focusingonquantityisim-portantbecauseitallowsustotestouranchoringandad-justmentmodel.Giventhelinkbetweenbothportionsizesand
obesity, and itsimpact on publichealth,webelievethis
warrantsacloserexamination.
Sincetheobeserepresentadissociativereferencegroupandresearchshowsthatweavoidthechoicesofthosewedonotwishtoemulate,weexpecttheadjustmenttotheanchorsetupbyanotherconsumertobemoderatedbythebodytypeofthisotherindividual.Iftheotherconsumersetsupanormofalargequantityoffoodchosen,wepredict
917
thataconsumerwilladjustthechoicequantitydownwardtoagreaterdegreewhentheotherpersonisobese,resultingintheconsumereatingsignificantlylesswhentheotherpersonisobeseversusthin.However,bodytypesofothersmayactivatestereotypesaboutwhatfoodstheyarelikelytoconsume;astheobeseareseentoeatpoorlyandtoindulge(Bacon,Scheltema,andRobinson2001),itmaybethecasethatthiseffectonlyexistsforfoodcategoriesthatarecongruentwiththesestereotypes(i.e.,unhealthy,fatteningfoo
ds).
PerceivedHealthinessofFoodChoices
Whiletherehavebeenseveralstudiesexaminingeatingbehavior,suchstudieshavetendedtofocusonunhealthyitems,suchascookies(Rothetal.2001),icecream(John-ston2002),andcandy(Scottetal.2008).Consumersas-sociatelosingweightwitheatingthe―right‖foodratherthanwithhavinganappropriateportionsize(AntonukandBlock2006),butampleevidencesuggeststhatitisthelatterthatmattersatleastasmuchastheformerinachievingahealthybodyweight(Wansink2006).
Therearealsotheoreticalreasonstoexamineperceivedhealthinessofthefood.Forexample,obesepeopleareper-ceivedaseating―inappropriate‖foods,suchasthosehighinfatandsugar(Weiner,Perry,andMagnusson1988).Peo-plestereotypetheobeseassupersizingtheirburgersandfries,nottheirsalads.Theassociationwithobesityisnotasstrong,therefore,withhealthyfoods. In relatedresearch,Johnston(2002)foundthatparticipantsdidnotchangetheiricecreamintakeinresponsetoobservingthequantity cho-senbyaconsumerwithalargebirthmark.Althoughthebirthmarkcreatedastigmaandmadetheotherconsumeramemberof adissociativegroup, itwasnotlinkedwithobe-sityandthereforehadnoeffectonconsumption.Thissug-geststhatconsumers‘foodselectionsshouldbe
affectedbywhattheotherpersonchooses,showingthatthepairingofthestimulus(unhealthyfood)withthetarget (an obeseper-son)isnecessarytoinfluencebehavior.Specifically,whenthefoodisunhealthy,aconsumerwouldtakemorewhentheotherpersonisthinthanwhensheisobese;however,whenthefoodchosenbytheotherpersonishealthy,theeffectofbodytypeonconsumptionwouldbeattenuated. However, the obese are a group ofconsumersthatpeoplegenerallydonotwishtoemulate.Researchinvolvingdis-sociativereference
groupswouldpredictthatthedomainofconsumptionshouldnothaveaslargeofanimpactasthereferencegroupitself.Forexample,BergerandRand(2008)foundthatwhenvideogamers(anoutgroup)werelinkedtohighjunkfoodconsumption,
participantsdecreasedtheirownjunkfoodchoiceseventhoughthereisnothingaboutvideogamesthatnecessarilycausesonetobecomeobese.Basedonthislogic,regardlessofthetypeoffoodoffered,whentheotherconsumersets upa high consumption quan-tityanchor,consumerswilladjusttheirownconsumptiondownwardtoagreaterdegreeiftheotherconsumerisobesethanifsheisthin.Study1wasdesignedbothtotestthepropositionsofananchoringandadjustmentprocessbased
918
onbodytypeandtoexaminewhetherthemodelmightbebounde
dwithinunhealthyfood.
STUDY1
ParticipantsandProcedure
Thehypothesesweretestedusinga2(confederatebodytype:thinvs.obese)#2(food:healthyvs.unhealthy)+2(controls:noconfederate,M&Msvs.granola)between-sub-jectsexperimentaldesign.Participantsincluded95under-graduatefemalesfromtheUniversityofBritishColumbia,whocompletedthestudyinexchangefor
$10remuneration.Femalesaremoresensitivetosocialcomparisonsregardingbodytype(Trampe,Stapel,andSiero2007),andgiventhatourconfederatewasfemaleandfollowingotherresearchinthisarea(e.g.,SmeestersandMandel2006),werestrictourinquirytofemalesinthisstudy.Participantswhoeitherindicatedthattheydidnotnotice what the confederatetook(np4)orwhobothtookandatemorethanthreestandarddeviationsoverthemean(np2)weredeletedfromtheanalyses.Onepersonhadfoodallergiesandelectednottoeatanything.
Participantswereinvitedindividuallyintothelabbetweenthehoursofnoonand6:00p.m.purportedlytoparticipateinastudy examining
people‘sexperiencesviewingmovies.Inalloftheconditions(exceptthecontrols),purportedly―inordertosavetime,‖participantsweretoldtheywouldberuninpairs(theotherparticipantwasalwaysatrainedconfederate).―Tomaketheexperiencemorerealistic‖theywereoffered a snack toenjoyeatingwhileviewingthefilmclip.Theconfederatetook5heapingtablespoonsofthesnackfood(approximately71gramsofgranolaor108gramsofM&Ms)inviewoftheparticipant,anamountthatwaspretestedtobealargequantityforonetotake.Thepartic-ipantwastheninvitedtotaketheamountofsnackfoodthatshewantedbeforewatchingthefilm.Neithertheconfederatenortheresearchassistantwatchedwhatamountthepartic-ipantselected.TheparticipantandtheconfederatewerethenledintoseparateroomswhereaTVwaslocated.Participantsweretoldtowatchthefilm,abenign5minuteclipfromthefilmI,Robot,andthentofilloutaquestionnaireabouttheirexperience.Theythencompletedthequestionnaire,whichcontainedanumberofdummyquestionsaboutthefilm(includingproductplacement),theroom(includingthesuitabilityofthelightingandchairs),arestrainedeatingscale,theirheightandweight,manipulationchecks,anda
suspicionprobe.
Manipulations.The
same confederatewas
usedinboththethinandoverweightconditions,andshe wasof thesameethnicityasthevastmajorityoftheparticipants.Tomanip-ulateconfederatebodytype,aprofessionallyconstructedobe-sityprosthesiswaswornbytheconfederateintheoverweightcondition(seefig.1).Thissuitwascustomdesignedfortheconfederate‘sbodybyanAcademyAward@–winningcos-tumestudio.Theconfederate‘snaturalheightwas5feet,2inches(157.5centimeters),andsheweighed105pounds
JOURNAL OFCONSUMER RESEARCH
(47.6kilograms);shehadabodymassindexlevel(BMI)of19.2(whichisonthelowendofnormalbutnotunder-weight),andsheworeasize00.Withthesuiton,sheap-pearedtohaveaweightofabout180pounds(81.8kilo-grams,aBMIofapproximately33),andsheworeasize16,makingherappearobese.Identicalclothesweretailoredinbothsmall(tofithernaturalbodytype)andlarge(overtheprosthesis)sizes,anddifferent setsof clotheswerecho-senrandomlyforeachsession. Thefoodchoiceofferedtoparticipantswasmanipulatedtobeperceivedaseitherhealthyorunhealthy.Inamanip-ulationborrowedfromWansinkandChandon(2006),gra-nolaandM&Mswereusedasthehealthyandunhealthyfoodssincetheyaresimilarincaloricdensitybutdifferstronglyinhealthinessperception.Toensuretheinternalvalidityofthismanipulationinthestudypopulation,apre-testwasconducted;itvalidatedthatgranolawasindeedperceivedtobehealthier,lesshedonic,andlesslikelytocontributetoobesitythanwereM&Ms.
Measures
DependentMeasures.Themainvariablesofinterestwereth
eweightofthesnackfoodthattheparticipanttookandateasafunctionoftheconfederate‘sbodytype.Toassesshowmuchparticipantstookandate,thebowlcon-tainingeitherM&Msorgranolawasweighedbothbeforeandafterthesession,accountingforhowmuchwasfirsttakenbytheconfederate.Becausethemovieclipwasshortinduration,notallparticipantsateallofwhattheytook.However,theywerenotpermittedtoleavetheroomwiththeirbowls,andthuswewereabletoobservetheuneatenquantitytocalculateameasureofactualconsumptionbyeachparticipant.Ourmeasuresadvancepriorresearch,asweareabletodecouplethechoiceandconsumptiondeci-sions.Inourparadigm,whiletheparticipantseeshowmuchfoodtheconfederatetakes,shedoesnotobservethecon-federate‘sactualconsumption(unlikeCongeretal.1980;Johnston2002;Polivyetal.1979).Aswell, in ourresearchthechoicedecisionofhowmuchtoputontheplateisaone-shotdecision.Unlikepastresearch,theparticipantisunableto―goforseconds‖ortoconsumemorefoodthansheputonherplateattheinitialdecisionphase.Assuch,thisrepresentsamoreconservativetest,astheparticipantcannotupdateherchoiceasaresultofviewinganotherpersoncontinuingtoconsume.
OtherMeasures.Participants‘propensityfordietingorrestr
ainedeatingwasmeasuredwitha10-itemscalefromHermanandPolivy(1980).Thisincludedsuchitemsas,―Howoftenareyoudieting?‖―Doyoueatsensibly in frontofothersandsplurgealone?‖and―Doyouhave feelings ofguiltafterovereating?‖Thereliabilityofthisscalewasap.83.Manystudieshaveshownthatrestrainedeatersbehavedifferentlythanthosewhoarenot(e.g.,Antonukand Block 2006; Scott et al. 2008),and thusweincludethisvariableasacovariateinouranalysis.Thismeasurewas
BODYTYPESANDFOODCHOICES
919
FIGURE1
THECONFEDERATEWITHOUTTHEPROSTHESISANDWITHTHEPROSTHESIS
NOTE.—Colorversionavailableasanonlineenhancement.
assessedatleast1weekinadvanceofthestudy,usinganonlinesurvey.
Attheendofthequestionnaireamanipulationcheckassessedthebodytypeoftheconfederate,measuredonthree7-pointscales(-3to+3):―Theothersubjectinthisex-perimentis...‖(veryoverweight/veryunderweight;veryobese/verythin)and―Comparedtome,theotherstudentinthisexperimentis...‖(muchheavier/muchthinner);re-liabilitywasap.76.
Inthisandsubsequentstudiesthevastmajorityofourparticipants were of normal BMIs, andsincecontrollingforBMIdoesnotaffectourresultsormoderatethem,BMIasa participant variableisnotdiscussedfurther.Resultsofthe
suspicionprobeshowedthatnoparticipantsweresuspicious
thattheconfederate‘sobesitywasnotgenuine,norwereanyawarethatshewasnotafellowparticipant.Inthisstudy,wealsorecordthetimeofdaythesessionwasrun,andwecontrolforitintheanalysis.
Results
cessful.Ananalysisofcovariance(ANCOVA)usingtheperceivedweightindexasthedependentvariable,amounttakenandconfederatebodytypeasindependentvariables,andparticipants‘restrainedeatingorientationandtimeofday ascovariates revealed onlyasignificantmaineffectfor
confederatesize(F(1,59)p52.95,p!.001).
Themean
ManipulationCheck.Themanipulationcheckwassuc-