REMEMBERING ATTACK ADS: An Experimental Investigation of Radio

To view this page ensure that Adobe Flash Player version 9.0.124 or greater is installed.

Please login or register to make a comment!

...Description...... more. less.

anindividualisexposedtoanadthatdescribespositiveornegativeinforma- tion,otherinformationassociatedwiththatmaterialmayalsobeactivated. Whenaskedtorecallthepoliticalad,theseassociatesmightwellberecalled ashavingbeenpresent,eventhoughtheywerenotinthestimulusmaterial. Weexaminedourdatatoseeifattackadsinfactproducedifferentmemory effectsthanpositiveads. 8 Ourfindingsaresimple,buttelling.Participantsin thisstudyrecalljustasmanythings(whatweshallcall cmemoryunits d)about thecontentofpositiveadsastheydothenegativeads.Hence,itappearsthat attackadsstimulatevoterrecallatthesameratesasdopositivespots 4a findingthatsuggestsattackadsdonotturnoffpeoplemorethanpositiveads. However,whatoursubjectsrecalledabouttheattackadswasfarmorelikely tobein error .Thatis,participants 9recallprotocolscontainedmoreincorrect negativememoryunitsthanincorrectpositivememoryunits.Now,wewant tobeclearaboutwhatwemeanamemoryerror.Informationthatisreported byparticipantsashavingbeenintheadwhenindeeditwasnotpresentis treatedasamemoryerror(orwhatwewillcallan cintrusion d).Itis,therefore, possiblethattheinformationrecalledmaybeaccurateingeneral,butnotas itappliestothecontentofthead.Thisdistinctionisimportantandwilldraw attentionintheconclusion. Theseresultshavepotentiallyimportantimplicationsforpoliticallearning andspeaktoissuesaboutthemeritsofattackads,topicsthatweshalladdress intheconclusion.Theremainderofthisarticleisdividedintofoursections. First,wepresenttheexperimentaldesignusedtocollectourdata.Second, wediscusshowwemeasuredaccuracyinmemoryrecallofsubjects.Third,we describeourresults.Finally,weofferafewconclusionsanddirectionsfor futurework. EXPERIMENTALDESIGN Ourexperimenthadfourconditions:attackRepublicanads,positiveRepub- licanads,attackDemocraticads,andpositiveDemocraticads.Ineachcase, theactualissuesraisedandtheirwordingwasthesameacrossallads.The spotsmentionedcrime,education,SocialSecurity,andtaxes(seeAppendix A).Weranpartyadsasopposedtocandidateadsforthreereasons.First,our datawerecollectedinthespringof2001,whenfewcandidateswereformally runningforpublicoffice.Second,sincemostpeoplehaveapartisanidentifica- tion,memoryaboutouradscouldplayoffthesewidelyheldschemas.Ifwe hadusedlocalcandidates,forexample,alargenumberofourrespondents mightnotpossessanyinformationaboutthem.Third,politicalpartiesdoin- vestalotofmoneyandtimepromotingtheirowncausesviapaidadvertising, soweareassessingaquestionofobviousimportance. Inanefforttomakesureourmanipulationwassound,weundertooka GEERANDGEER 74 numberofprecautions.First,werancopiesofouradsbyalocalpolitical consultanttoseeiftheywererealistic.Heindicatedthatthespotswereplausi- blerepresentationsofthekindofadsonemighthearrunduringanelection, whichgivesusgreaterconfidenceinthegeneralizabilityofourresults.Sec- ond,weaskedstudentsfromoneoftheauthor 9suniversitytoratehownega- tiveandhowpositivewereeachofthesimulatedads.Wewantedtomakesure thatouradscapturedaccuratelythevariablesofinterest.Theundergraduates 9 ratingsoftheadssquarednicelywithourjudgmentsofthem.Inotherwords, studentsviewedour cattack dadsasattacksandviewedour cpositive dadsas positive.Thisverificationincreasedourconfidenceinthemanipulation.Even so,wetookonefurtherstep.Ratherthanjustrunningonekindofnegative orpositivespotforeachparty,weconstructedtwoparallelformsofeachad toavoidrestingourconclusionsonasingleversionofthespot.Copiesofall adsusedinourexperimentareavailableonrequest(seeAppendixAfortwo samples). 9 Experiments,obviously,havetheadvantageofprovidinggreaterlevelsof control,whichpermitscleanertestsofkeyhypotheses.Whenstudyingthe variouseffectsofthemassmediaonthepublic,thisassetisparticularlyimpor- tantinlightofthemanyproblemsassociatedwithself-reportsofmediaexpo- sure(PriceandZaller,1993). Radioads,ourparticularmanipulation,haveanadditionaladvantageofbe- ingabletocontrolwithgreatprecisionthecontentandcharacteristicsofthe materialpresentedtooursubjects.Bypresentingtheinformationonlyver- bally,weavoidthecomplicationsassociatedwithvisualimages.Notonlywould itbetrickytocreaterealisticaudiovisualads,itwouldbedifficulttoensureall adswereequivalent.Ofcourse,ourfocusonradiolimitsthegeneralizabilityof theseresultstoothermediums,especiallyvisuallyorientedonesliketelevi- sion.Butgiventhecentralityofradioinpoliticalcampaigns(Herrnson,2000), weviewthisfocusasanassetoftheresearch. Participants Participantswere29maleand92femaleundergraduatestudentsatalarge southernstateuniversity. 10 Nearly20percentofthestudentswereminority, providingsomediversityforoursample.Thesubjectswereenrolledinpsy- chologyclassesinthespring2001semesterandwereparticipatinginresearch inordertoobtainextracreditinthoseclasses. Methods Subjectssignedupforourstudy,agreeingtocometotheresearchlabfor theexperiment.Whentheyarrived,aresearchassistantdescribedallproce- REMEMBERINGATTACKADS75 duresusedinthestudy. 11 Eachparticipantwasthenaskediftheywishedto participate.Noonedeclined.Theparticipantthensignedaninformedconsent formthatbrieflyrestatedtheprocedurestobefollowedinthestudy.After signingtheform,theparticipantwastakenintothenextroomwheredata werecollected. Eachsubjectwasrandomlyassignedtooneofthefourexperimentalcondi- tions.Therefore,approximatelyonequarteroftheparticipantsheardanattack adfromtheDemocrats,anotherquarterlistenedtoapositivespotfromthe Republicans,andsoon.Withinthatrandomization,wealsovariedtheversion ofeachtypeofad.Recallthatinordertoreducethelikelihoodthatone specificadmightproduceunusualeffects,wecreatedtwoexamplesofeach ofthefourtypesofradiospots.Thus,therewasatotalofeightconditionsin theexperiment. Afteragreeingtoparticipateandsigningtheinformedconsentform,partici- pantswereseatedinthelaboratory.Whentheyindicatedbeingready,they listenedtoahighqualitytape-recordedpoliticalad.Adcontentvariedasde- scribedpreviously.Immediatelyfollowingtheplayingoftheadtheparticipant wasgivenapassagefromauniversitycatalogtoread.Thepassagewasneutral incontentandcontainednoreferencetopoliticalmaterial.Researchassistants monitoredthereadingofthismaterialtoassureparticipant 9scompliance.This procedurelastedfor5minutes.Ourgoalwastopreventparticipantsfrom rehearsingthead 9scontentinanticipationofbeingaskedquestionsconcern- ingthead. Itisimportanttonotethateachsubjectlistenedtotheadinisolation. Thisprocedure,whiletimeconsuming,mirroredmorecloselytherealworld conditionsinwhichpeopletypicallylistentotheradio.Themostcommon waypeoplelistentoradioisintheircars,whichisusually(90%)donealone (Schulberg,1996). 12 Giventhattheaveragecommutetoschoolortowork, accordingto1990censusdata,is25minutes,asubstantialpartofpeople 9s daysarespentlisteningtotheradioinisolation. 13 Ofcourse,manypeoplealso listentoradioathome.Butagain,theevidencesuggeststhatthisactivityis alsodonealone(Schulberg,1996,p.2).Inshort,listeningtotheradioisfre- quentlydoneinisolation,strengtheningtheexternalvalidityofourdesign. 14 Laboratorysettingsdohavemanyartificialaspects.Butourmoretime-consum- ingproceduremakesthiscriticismlesstelling,especiallybycomparisonto havingsubjectslistentotheadinagroupsetting. Afterreadingthedistractingneutralmaterial,allparticipantswerein- structedtoprovideawrittenrecallofthematerialinthepoliticalad.They weretoldtowritetheiranswersasclosetoverbatimaspossible.Viaverbal instructionstheywereencouragedtowritethegistofthematerialifthey couldnotprovideaverbatimrecord.AppendixBcontainsacopyofthemem- oryquestionnairethatwasgiventoallparticipants.Whenthememoryques- GEERANDGEER 76 tionnairewascompleted,allparticipantsweregivenadditionalquestionsthat askedabouttheirpoliticalviews(AppendixB).Weincluded cfiller dquestions inthesecondquestionnaireinanefforttodisguisethenatureofthestudy. Whentheparticipantscompletedquestionnaires,eachwasaskedifthey hadanyquestions.Afteranyquestionswereanswered,theparticipantwas creditedforparticipatingandthendismissedfromtheexperiment.Thetotal timerequiredforparticipationwasapproximately30minutes. MEASURINGMEMORY UsingthecriterionestablishedbyPritchertandAnderson(1977),theads weredividedinto cideaunits. dIdeaunitsaredefinedasmeaningfuldivisions intheads.Thatis,eachadwasexaminedandthetextwasparsedintoseg- mentsthatcontainedanidentifiableunitofmeaning. 15 Sincetheattackand positiveadswereconstructedtocontainidenticalamountsofinformation,the numberofideaunitswasthesameforthetwotypesofads.Thisprocedure assuredthatdifferencesinmemorywerenottheresultofvarianceinthe numberofideaunitspresentedtothesubjects. Asinthecaseoftheoriginalads,webrokeuptherecallprotocolsintoidea units.Theseparsedrecallprotocolsweretheninspectedandeachrecallidea unitwasassignedthelabelcorrectorintrusion.Acorrectmemoryunitwas thepresenceintherecallprotocolofanideaunitthathadbeenpresentinthe ad.Anintrusionoccurswhenanideaunitintherecallprotocolisproducedby thesubjectbuthadnotbeenpresentinthead.Agistcriterionwasusedto estimatewhetherthegistoftheideaunitwaspresent(PritchertandAnder- son,1977).Inotherwords,foraunittobescoredascorrect,thesubjectdid nothavetorecallexactwordingsbutdidhavetoproducethebasicmeaning. Afterthecorrectversusintrusiondecisionwasmade,theunitwasassignedto oneofthreecontentvalues.Theunitwasidentifiedaseitherneutralincon- tent,attackincontent,orpositiveincontent.Thisyieldedsixcategoriesto whichallrecallideaunitscouldbeassigned. Tworesearchassistantsindependentlyjudgedeachparticipant 9smemory protocol.Ininstanceswheretheoriginalcodersdisagreed,athirdindepen- dentjudgereadtheprotocolandscoredit.Thethreejudgesthendiscussed theunitinquestionandselectedanagreed-uponcategory.Therewereno instancesinwhichcompleteagreementamongscorerswasnotobtained.The firsttwocodersagreed83percentofthetimeabouttheaccuracyofthemem- ory,whichisamorethanacceptablerateofinter-raterreliability.Inthelim- itedinstanceswheredisagreementinitiallytookplace,theconferencewith thethirdcoderresolvedtheissuetoallraters 9satisfaction.Followingthescor- REMEMBERINGATTACKADS77 ingofrecallprotocols,eachparticipanthadsixscoresthatreflectedthenum- berofideaunitsineachofthesixcategories. MEMORYRESULTS Weconductedapreliminaryanalysistodetermineifitwaspossibletosim- plifyourresults.RecallthatwehadseparateconditionsforDemocraticand Republicanads.Whilewesawlittlereasontoanticipatepartisandifferences, itwasessentialtocontrolforthispossibilityratherthanjustassumeitwould notbeanissue. 16 Thedatarevealedthatthepartysponsorshipoftheaddid notaffecttherecalledmemoryofoursubjects. 17 Thisfindingallowsusto simplifyouranalysesbydroppingpartysponsorshipasavariableinourexami- nationofthememorydata. 18 Animportantpayoffofthissimplificationwasan increasedsamplesizeacrossthefourcollapsedconditions,whichpermits greaterstatisticalpreciseinourtestsofsignificance.Insum,ouroverallanaly- sis,tobedescribedbelow,nowinvolvesa3-factormixed-modelANOVA. Ouranalysesfocusonthreeprinciplevariables:(a)AdType(Attackor Positive,abetweensubjectvariable),(b)MemoryType(CorrectorIntrusion, awithin-subjectsvariable),and(c)MemoryContent(Positive,Neutral,Nega- tive,awithin-subjectsvariable).Asnoted,wecreatedthenumberofmemory units,ourkeydependentvariable,throughthescoringofrecallprotocols. Thatis,eachsubject,asmentionedpreviously,receivedsixseparatescores. Thosescoresforeachparticipantwerethenumberofmemoryunitsineach ofoursixcategories. 19 Thesesixscoreswerethenidentifiedasbeingpositive, negative,orneutralincontentandassignedtothecategoryofeithercorrect orincorrect.Itisquitelikelythatthereisacorrelationbetweensomeofthese sixmeasures.We,therefore,usedaGeneralLinearModel(GLM)repeated measuresapproachtoanalyzethedata.Anadvantageofthisprocedureisthat wecantestfortheviolationoftheindependenceassumptionofrepeated measuresandprovideappropriatecorrections.Thisapproachincreasesour confidenceintheaccuracyofourstatisticalanalyses.Whensignificantviola- tionsofsphericityareidentified,wepresenttheGreenhouse-GeisserCorrec- tion(GGC).GGCadjustsstatisticalteststoaccountforthisproblem. 20 We alsopresenttheeta-squaredstatistic,whichdescribestheproportionofthe variationexplainedbyourindependentvariable. Webeginourdescriptionofthedataanalysisbyreportingthesignificant effectsintheoverallrepeatedmeasuresanalysis(seenotesforthespecific detailsonrelevantstatisticaltests).Amongthemaineffects,theonlyonethat cameclosetonormalsignificancetestswasMemoryType.Whencollapsed acrossallconditionsthereweremorecorrectideaunits( M = 1.49)thanintru- sions( M = 1.04). 21 However,therewerenosignificantdifferencesforthemain GEERANDGEER 78 effectsofAdTypeorMemoryContent. 22 ThefindingthatAdTypedidnot, overall,yieldasignificanteffectindicatesthatattackadsdonotleadsubjects torecallmoreinformationthanforpositiveads. 23 Similarly,therewereno overalldifferencesbetweenthenumberofpositive,attack,orneutralunits recalled. Thefirsttwo-wayinteractionweconsiderisbetweenAdTypeandMemory Content.Figure1containstherelevantmeans. 24 Inthisanalysis,wecollapsed thedataacrossMemoryTypetosumcorrectunitsandintrusionstogether.As onecanreadilyobserve,participants 9memoriesstrongly(andsignificantly) reflectthecontentoftheadtheyheard.Negativememoriesoccurwithattack adsandpositivememoriesoccurwithpositiveads.Neutralcontentappears unaffectedbyAdType. Whilenooneshouldbesurprisedthatnegativeinformationtriggersnega- tivememories,theseresultsserveasavaluablemanipulationcheck,verifying thatouradsapparentlycontainedthetypeofcontentwesought.So,whilewe donotwanttotoutthesubstantiveimportanceoftheseresults,theydoinstill agooddealofconfidenceinthevalueofourdata. Thenexttwo-wayinteractioninvolvesMemoryTypeandMemoryContent. Figure2presentsthemeans. 25 Thesedataarecollapsedacrossthevariableof AdType.Whenthememorycontentwaseithernegativeorneutral,there werenodifferencesbetweencorrectandincorrectmemories.However,when thememorycontentwaspositivethereweremanymorecorrectmemories FIG.1. Adtypebymemorycontent. REMEMBERINGATTACKADS79 FIG.2. Memorytypebymemorycontent. ( M = 1.92)thanintrusions( M = .56).Thisdifferenceishighlysignificant( p < .001.)Thisresultispotentiallyimportart.Weshalltakeanothercutatthis generalissuebelowandsavecommentaryonitssubstantivesignificancefor later. AdTypebyMemoryTypeisourfinaltwo-wayinteraction.Wedisplaythe relevantmeansinFigure3. 26 Forattackads,thenumberofcorrectandincor- rectmemoriesisessentiallyequal.However,forpositiveads,correctmemo- ries( M = 1.71)faroutnumbertheincorrect( M = .89).Thesedataagainshow thatattackads,whilenotincreasingthetotalnumberofmemoryunits,do resultinmanymoreincorrectrecalls. Finally,theoverallanalysis,againusingGGC,yieldedahighlysignificant triple-orderinteraction. 27 Toclarifythethree-wayinteraction,weseparately analyzedthememorydatafrompositiveadsandattackads.Theindependent variablesinthesetwo2 × 3repeatedmeasuresANOVAswereMemoryType (CorrectvsIntrusion)andMemoryContent(Positive,Neutral,andNega- tive.).Thefirstanalysisislimitedtopositiveads.Wepresentthemeansrele- vanttothisanalysisinFigure4.Therewasahighlysignificantmaineffect, whenusingGGC,forMemoryType. 28 Overall,thereweremore( p < .001) correctpositiveideaunits( M = 1.708)thanintrusivepositiveideaunits( M = .895).TheMemoryContentmaineffectvariablealsowashighlysignificant. 29 Forpositiveads,thereweremorepositiveunits( M = 2.43)thanneutralunits ( M = 1.43)andmoreneutralthannegativememoryunits( M = 0.04). 30 Of course,onewouldnotexpectnegativememoryunitsfrompositiveads.This GEERANDGEER 80 FIG.3. Adtypebymemorycontenttype. findingreinforcestheresultswedescribedinthetwo-wayinteractionsabove. However,inthisinstanceweareabletoidentifymorepreciselythefindings asoccurringwhenthespotswerepositive. Ourlastanalysismakesuseoftheideaunitsrecalledfromattackads.The independentvariablesareidenticalwiththepreviousanalysis.Intheseresults, FIG.4. Memorytypebymemorycontentforpositiveads. REMEMBERINGATTACKADS81 theonlysignificantfinding(GGCwasapplied)wasmaineffectofMemory Content. 31 InspectionoftherelevantmeansinFigure5clarifiesthemain effect.Fordatafromattackads,themeannumberofnegativecontentidea unitswas2.51,forneutralcontentthemeanwas1.14,whileforpositivecon- tentthewereessentiallynopositivememoryunits( M = 0.05).Eachofthese threepointssignificantly( p < .001)differedfromeachother.Asintheprevi- ousanalysis,thelimitationofthesefindingstodatafromattackadsreinforces theanalysesofthetwo-wayinteractions. DISCUSSION Ourfindingswarrantattention.Tostart,subjectsrecalledthesamenumber ofmemoryunitswhethertheyheardapositiveoranegativead.Thisresultis consistentwiththeclaimbysomescholarsthatattackadsdonotdepress people 9sinterestinorlearningofpolitics(FinkelandGeer,1998;Geer,2000; FreedmanandGoldstein,1999;KahnandKenney,1999).Thisfindingalso speakstothegeneralquestionofwhetherthepublicrecallsnegativeinforma- tionbetterthanpositiveinformation.Thereisasizableliteratureonthisques- tion,withmixedresults(BriansandWattenberg,1996;CappellaandJamie- son,1997;HitchonandChang,1995;ShapiroandReiger,1992).Thelackof differenceswereportincreasestheevidencethatnorelationshipexistsbe- tweenrecallandthetoneoftheadvertisement.Ourstudy,therefore,addsto thegrowingchorusthatquestionsthelongstandingassumptionthatattackads stickwithpeoplemorethandopositiveads.Withthesenullresults(see,in FIG.5. Memorytypebymemorycontentforattackads. GEERANDGEER 82 particular,LauandSigelman,2000,p.19),itmaybetimetorethinkthis assumption.Perhapsattackadsarecommonenoughintoday 9spoliticalcam- paignsthattheyarelosingtheirpunch.Wecannotbesure.Butitmightbe timetoreconsiderwhatconstitutes cconventionalwisdom dinregardstokey aspectsofattackadvertising(Lauetal.,1999). Thisisonlypartofourstory,however.Themoreimportantfindingarises whenlooking cinsidetheblackbox datthecontentofthememoryunits.Here, wediscoverthatattackadsproducemoreinaccuraterecallsthandopositive ads. 32 Atthesametime,positiveadsproducemoreaccuraterecallthando negativeads.Inotherwords,theratioofcorrecttoincorrectrecallschanges whetheroneisexaminingattackorpositiveads.Itappears,therefore,that whilethetypeofadgeneratesthesameamountofmemoryrecall,the kind ofrecallvariesinimportantways. Whilethereisnouniversallyacceptedmodelofmemoryinthefieldof psychology,scholarsgenerallyacknowledgethatinformationisstoredinpack- agesorunitsoftencalledschema(Bartlett,1932).AsSolso(2001,p.216) noted, c(p)erhapsthemostpervasiveassumptionaboutLTM(longtermmem- ory)isthatinformationinitisorganizedinsomeorderlyway. dOnewayof representingschema,anorganizingconcept,isasanetworkofassociations (CollinsandLoftus,1975).Schvaneveldt(1990)illustratestheusefulnessof thisapproachbypresentinganetworkmodelfromproximitydatausingthe algorthymnamed cPathfinder. dAnimplicationofthenetworkapproachfor thisworkisthattheactivationofoneunitofthenetworkorschemaviaa memorysearchwillresultinactivationofallassociatedunits.Itisassumed thattheeffectivenessoftheactivationisreducedasthelinksbetweenassoci- atesweakenandasthespreadextendsfurtherandfurther.Thisnotion,called spreadofactivation(CollinsandLoftus,1975),maybeusefullyappliedtothe currentwork.Ourdatasuggestthatsubjects,afterhearinganegativead,are recallingnotonlythecontentofthatadbutrelatedpiecesofinformation thatexistintheindividual 9sschemaofnegativepoliticalinformationthatwas activated.Thatis,conceptsnotpresentintheoriginalattackadareactivated andthenareaccessedinthememorysearch.Participantsdonotaccurately discriminateamongitemspresentedintheadandassociateditemsthatare linkedinthenegativeschema.Bycontrast,positiveadsdidnotproducenearly asmanyincorrectrecalls,suggestingthatthememorynetworkassociatedwith positiveinformationislesstightlylinked.Unfortunately,neitherthemodel northedatatelluswhythisshouldbethecase.Norwithinourdatasetcan wetestthissuggestion. Thesefindings,whileinteresting,arehardlydefinitive.Weneedtotease outinfutureworkwhyintrusionsinattackadsaremorecommonthanin positiveads. 33 Lau(1985)pointstotheimportanceofnegativeinformation andtodatathatsuggestthatsuchinformationhasamorepowerfulaffecton REMEMBERINGATTACKADS83 theevaluationofpoliticiansthanpositiveinformation.Itdoesnotappearthat Lau 9sworkwouldspeakspecificallytomemoryerrors,buthedoesemphasize theneedtoconsidercarefullytheinfluenceofnegativeinformation.That reinforcesourcontentionthatunderstandingthedetailsofmemoryeffects whenexaminingattackadsisanimportanttask,oneinwhichourworkoffers someinitialinsights. Thequestionofwhetherornotthesecategoriesofmemorydifferentially fadeovertimeisparticularlysalientinthecontextofworkbyLodge,Steen- bergen,andBrau(1995). 34 Theysuggestthatovertimememoryforpolitical informationfadesandwhatisleftisasummaryoftheinformation.Itisthis overallimpression,notspecificmemory,thatformsthebasisforcandidate evaluation.Tothedegreethatmemoriesplayanimportantroleinthedevel- opmentofthesummaryaffectiveevaluation,thedoorisopenedbyattackads todistorttheimpressionuponwhichvotingdecisionsaremade.Ourdatado notspeakdirectlytoLodgeandcolleagues 9suggestion.It,however,provides additionalhintsatdirectionsforfuturework. 35 Whileourfindingshaveimportforpoliticalpsychology,therearealsonota- bleimplicationsfortheongoingdebateovertheconsequencesofattackadver- tising.Atfirstblush,ourcorefindingappearsdisconcerting.Thefactthat peoplearerecallingmoreincorrectthingsfromnegativeadsthanpositiveads suggestthatattackscanbemisleading.Ifso,ourresultspaintadarkersideof attackadvertisingthathasgoneunnoticedinthespateofrecentwork.A democracyrestsonawell-informedelectorate,andifnegativeadsproduce incorrectrecallofinformation,thepublicmaybelessabletomeetthede- mandsitplacesonthem. However,thatconclusionmaybehasty.Itmaybethattherecalledinforma- tionisinaccurateforthatad,butmaynotbeinaccurateingeneral.Consider oneofoursubjects,aDemocrat,whowhenlisteningtoaDemocraticattack adrecalledthefollowing: ctheRepublicanswillmoreorlessdestroythecoun- tryandyourlife. dConsideranotherparticipant 9sreaction.Inthiscase,itwas aRepublicanlisteningtoaGOPattackad,whichgeneratedthefollowing recall: ctheadstatesthatDemocratswilltrytotakeovereverything. dInboth cases,therewasareactiontotheadthatwasnotofficiallyapartofit.Yet fromapartisanperspective,isitmisleadingorincorrectforamemberofone partytothinkthattheotherparty,ifelected,willyieldundesirableoutcomes forthenation?Thisisatoughquestionandonenotansweredbyourresults. Moreover,itmaybethatattacksonpolicyissuessparkmemoriesofother problemswiththeopposingparty.SoiftheRepublicansattacktheDemocrats forbeingbigspenders,itmayalsogenerateanassociationforthatpersonof theDemocratsfavoringtaxincreases.Fromapartisanperspective,thismay beviewedasagoodoutcomeofattackads.Andevenfromademocratic perspective,itmaybeimportantthatcampaignsremindvotersofissuesbe- GEERANDGEER 84 yondthosecontainedinanyspot.Ifattackadsactivatetherecallofrelatedbits ofmemoryonrelevantissues,itisnotatallclearthisoutcomeisworrisome. Allofthesecommentsunderscorethatthedebateovertheimpactofattack adsisfarfromover.Muchworkremains.Weneed,forexample,tocontinue toposequestionsthatadvancethefieldinnewandimportantdirections.As weseektoanswerthesequestions,itisessentialthatweconsidertheimpact ofmediumsotherthanTV.Weareamultimediasociety.Citizenswatchalot ofTV,buttheyalsolistentotheradio,scantheInternet,andreadnewspa- pers.Ifwearetoforgeamorecompleteunderstandingofhowpoliticaladver- tisingaffectselections,especiallystateandlocalcontests,wemustexpandour horizonsbeyondtelevision.Ouruseofradiospotsrepresentsonesuchat- tempt,butweonlyscratchthesurface. Theneedtopushthisresearchagendafurtherisclear.Itcomesintoeven sharperfocuswhenwerealizethatanenduringfeatureofcampaignsistheir negativity(Riker,1996).Attackswillnot,norshouldthey,disappear.Members oftheopposingpartyhavearesponsibilitytoraisedoubtsabouttheiroppo- nents.Wecertainlycannotexpectpoliticianstobeself-criticalduringacam- paign.Evenso,manyobserversfeartheadverseconsequencesofexcessive attacksinelections.Whilesomeofthosefearsaresurelymisplaced,itisessen- tialwecontinuetograpplewiththisfundamentalandenduringaspectof elections.Insodoing,wecanadvanceourunderstandingoftheelectoralpro- cess. Acknowledgments .WewouldliketothankBillFletcher,GeoffLayman,BruceOp- penheimer,andJasonHicksfortheirhelpfulfeedbackonearlierversionsofthisessay. Inaddition,weappreciatethecommentsfromthereviewersfor PoliticalBehavior and DianaMutz.JohnGeerwantstothanktheCenterfortheStudyofDemocraticPolitics atPrincetonUniversityforsupportduringthecompletionofthisproject. APPENDIXA:TRANSCRIPTSOFRADIOADVERTISEMENTS Belowarewrittencopiesoftwooftheradioadsusedinourexperiments.Tosave space,wehavenotincludedcopiesofallspotsusedinthisstudy.Onrequest,wewill provideafullsetofthetranscripts. 1.RepublicanAttack1 Nextyearinthe2000electionsyouwillbedecidingthefutureofthiscountry. Electionsarealwaysimportant.Butthesecontestswillshapeanddirectionthisnation 9s destinyasweembarkonanewcentury.Acenturythatoffershopeforallpeopleof theUnitedStates.Dowewanttopursuepoliciesthathavenotworkedinthepast? REMEMBERINGATTACKADS85 Ordowewanttochartacoursethatwillblendnewideaswithtriedandtruepolicies? Itisyourchoice. Asyouconsiderwhichpartyismostlikelytoleadthisnationtowardspeaceand prosperity,considerthefollowing: " TheDemocraticPartywillraiseyourtaxes,bluntingeconomicgrowth " TheDemocraticPartywillweakenSocialSecurity,underminingthewell-beingof ourcitizensastheyentertheirgoldenyears. " TheDemocraticPartywillpursuepoliciesthatdonotprotectthevictimsofcrime, allowingcrimeratestoriseandcriminalstogofree. " TheDemocraticPartywillnotcontrolgovernmentspending,therebyincreasing thenationaldebt. Thischoiceisanimportantone.Wewanttoapproachthecenturywithhopeand withoptimism.Byputtingtherightpeopleinchargeyouwillensurethatthiscountry prospersandenjoythefreedomsandrightsourforefathersfoughtsocourageouslyto protect. PaidforbytheRepublicanParty. 2.DemocraticPositive1 Nextyearinthe2000electionsyouwillbedecidingthefutureofthiscountry. Electionsarealwaysimportant.Butthesecontestswillshapeanddirectionthisnation 9s destinyasweembarkonanewcentury.Acenturythatoffershopeforallpeopleof theUnitedStates.Dowewanttopursuepoliciesthathavenotworkedinthepast? Ordowewanttochartacoursethatwillblendnewideaswithtriedandtruepolicies? Itisyourchoice. Asyouconsiderwhichpartyismostlikelytoleadthisnationtowardspeaceand prosperity,considerthefollowing: " TheDemocraticPartywillnotraiseyourtaxes,promotingeconomicgrowth. " TheDemocraticPartywillprotectSocialSecurity,ensuringthewell-beingofour citizensastheyentertheirgoldenyears. " TheDemocraticPartywillpursuepoliciesthatprotectthevictimsofcrime,lower- ingcrimeratestoriseandmakingsurecriminalsdonotgofree. " TheDemocraticPartywillcontrolgovernmentspending,therebypayingoffthe nationaldebt. Thischoiceisanimportantone.Wewanttoapproachthecenturywithhopeand withoptimism.Byputtingtherightpeopleinchargeyouwillensurethatthiscountry prospersandenjoythefreedomsandrightsourforefathersfoughtsocourageouslyto protect. PaidforbytheDemocraticParty GEERANDGEER 86 APPENDIXB:QUESTIONNAIRESFOREXPERIMENT FirstQuestionnaire Participantswerehandedthefollowing: Inthespacebelowwriteascompletelyandaccuratelyaspossiblethepoliticaladthat youheardafewminutesago. Onthescalebelownotethedegreetowhichyouagreeordisagreewiththecontent ofthead. 12345 StronglyAgreeNeitherDisagreeStrongly AgreeAgreeorDisagree Disagree Aftercompletingtheabovesetofquestions,participantswerethenhandedtheques- tionnairebelow: SecondQuestionnaire InformationalQuestionnaire Name Pleaseprint Pleaseanswereachofthefollowingquestions. 1.Somepeopledon 9tpaymuchattentiontopolitics.Howaboutyou?Wouldyou saythatyouareverymuchinterested,somewhatinterested,ornotmuchinter- estedinpolitics? a.Verymuchinterested b.Somewhatinterested c.Notmuchinterested d.Don 9tKnow 2.Whichpartyismorelikelytoraisetaxesorwouldn 9ttherebeanydifference amongthem? a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 3.Whichpartyismorelikelytocutsocialsecuritybenefitsorwouldn 9ttherebe anydifferenceamongthem? REMEMBERINGATTACKADS87 a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 4.Whichpartyismorelikelytoeliminatethenationaldebtorwouldn 9ttherebe anydifferenceamongthem? a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 5.WhichpartyismorelikelytocutMedicarebenefitsorwouldn 9ttherebeany differenceamongthem? a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 6.Whichpartyismorelikelytobringeconomicprosperityorwouldn 9ttherebeany differenceamongthem? a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 7.Whichpartyismorelikelytolowercrimeorwouldn 9ttherebeanydifference amongthem? a.Democrats b.Republicans c.Therewouldn 9tbeanydifference d.Don 9tKnow 8.Wehearalotoftalkthesedaysaboutliberalsandconservatives.Hereisaseven- pointscaleonwhichthepoliticalviewsthatpeoplemightholdarearrangedfrom extremelyliberaltoextremelyconservative.Wherewouldyouplaceyourselfon thisscale,orhaven 9tyouthoughtmuchaboutthis? a.Extremelyliberal b.Liberal c.Slightlyliberal d.Moderate;middleoftheroad e.Slightlyconservative f.Conservative g.Extremelyconservative h.Haven 9tthoughtmuchaboutit 9.Generallyspeaking,doyouusuallythinkofyourselfasa: a.Democrat b.Republican c.Independent d.Don 9tKnow GEERANDGEER 88 10.IfthinkofyourselfasaRepublican,wouldyoucallyourselfastrongRepublican oranotverystrongRepublican? a.Strong b.Notverystrong c.Don 9tKnow 11.IfthinkofyourselfasaDemocrat,wouldyoucallyourselfastrongDemocrator anotverystrongDemocrat? a.Strong b.Notverystrong c.Don 9tKnow 12.IfyouthinkofyourselfasaIndependent,doyouthinkofyourselfascloserto theRepublicanPartyortotheDemocraticParty? a.ClosertoRepublican b.Neither c.ClosertoDemocratic 13.HowmanyhoursdoyouwatchTVeachweek? a.Lessthan5hours b.Between5and10hours c.Morethan10hours d.IrarelywatchTV 14.Howoftendoyoulistentotheradioeachweek? a.Lessthan5hours b.Between5and10hours c.Morethan10hours d.Irarelylistentotheradio 15.Howmanymoviesdoyouwatchinthetheatreeachmonth? a.1 32 b.2 33 c.Morethan3 d.Irarelygotothemovies 16.Howoftendoyoureadanewspapereachweek? a.Everyday b.Mostdays c.Occasionally d.JustonSundays e.Irarelyreadanewspaper 17.HowoftendoyousurftheInterneteachweek? a.Everyday b.Mostdays c.Occasionally d.IdonotsurftheInternet 18.Wheredoyougetmostofyourinformationaboutpolitics? a.TV b.Newspapers c.Radio d.Friendsandfamily REMEMBERINGATTACKADS89 19.Howmanydoyoureadmagazineseachweek? a.Irarelyreadamagazine b.1 32 c.2 33 d.Morethan3 20.Pleaseindicateyourraceorethnicity. a.White/Caucasian b.African-American c.Asian d.Latino e.Other 21.Whatisyourgender? a.Male b.Female 22.Pleaseindicateyourageinyearsthespacebelow. 23.Whichfigurebelowbestrepresentsyourfamily 9sincome? a.Lessthan$15,000 b.Between$15,000 3$30,000 c.Between$30,000 3$50,000 d.Between$50,000 3$75,000 e.Greaterthan$75,000 f.Don 9tKnow 24.Whatisyourreligiousaffiliation? a.Protestant b.Catholic c.Jewish d.Other NOTES 1.Solso(2001,p.231)infactobservesthatamong cthegreatchallengesthatcognitivepsycholo- gistsfaceaswebeginthetwenty-firstcenturynoneismorevitaltoourunderstandingofwho wearethanthenatureofourbrainandhowmemoriesareorganizedtherein. d 2.Communicationscholarshavealsoshownaninterestinthestudyofmemory.Forexample, CappellaandJamieson(1997),Lang,Newhagen,andReeves(1996),NewhagenandReeves (1992),andReevesetal.(1991)examinedtheimpactofthemassmediaoncitizens 9memo- ries. 3.SeeShapiroandRieger(1992)foranexception.Theyexamineddifferentialeffectsofnegative andpositiveradiospots.Theirfocuswaslargelyonhownegativeissueadscomparedto negativeimageads.Therehasalsobeeninterestintheimpactof ctalkradio d(Barker,2002). 4.Thereiseveryreasontothinkthatcandidatescompetingingubernatorialelectionswillalso makewidespreaduseofradio.UsingLexis-Nexis,welocated363articleswritteninthepast 6months(MarchtoAugust2002)thatmentionedradioadvertisinginitscoverageofthe electionsforgovernorduringthe2002cycle. 5.Toillustratethispoint,considerthe2002midtermelections.LamarAlexander,forexample, airedadsonconservativetalkradiowiththeaimofappealingtocriticalvotersinhisbattle GEERANDGEER 90 fortheRepublicannomination(Humphrey,2002).SimilarthingshappenedinNewYork 9s battleforgovernor,whereH.CarlMcCallusedradiotoremindAfrican-Americanvotersof hisroots,bluntinganyeffortbyAndrewCuomotocourttheAfrican-Americanvote(Hum- bert,2002).GeorgePatakialsomadeuseofHispanicradioinhisefforttowinreelection (Rau,2002).Texastoo,withitsgrowingHispanicpopulation,sawitsmajorcontendersallair radiospotsonHispanicradiotosecurethosecriticalvotes(Robison,2002). 6.Whilethetopicofmemoryerrorshasnotbeenwidelyappliedtothefieldofpoliticalscience, ithasbeenthefocusofresearchinotherarenas.Wepointtotheworkontheaccuracyof eyewitnesstestimony(Loftus,1979)wherefalseidentificationandinaccuratereportsincrimi- nalinvestigationsposeaseriousthreattothecorrectapplicationofjustice.Similarly,there hasbeenmuchrecentdebate(Loftus,1993;Loftus&Read,1994)onthetopicofrepressed memoriesrecoveredinthecourseofpsychotherapy.Thedebateoftencentersonwhether ornotthememoriesareaccurateorfalse.Instancesoffalserecoveredmemorieswithserious consequencesforaccusedarewelldocumentedintheliterature. 7.Therearecriticsofschematheorywithinpoliticalscience(Kuklinski,Luskin,andBolland, 1991).Wedonotseeourarticleascommentingonthisdebate.Ourgoalissimplytousethe ideaofaschemaasausefultooltoorganizethediscussionofmemory. 8.Wedidrunasetofbasicanalysesthatlookedtoseewhetherattacksinfluencedsubjects 9 interestinpolitics,politicalattitudes,andstrengthofpartisanship.Ourexperimentalmanipu- lationdidnotproduceanysubstantiveorstatisticallysignificantfindings.Thoseresultssquare withscholarswhoarguethatattackadsdonotadverselyaffectcitizens 9commitmenttothe politicalprocess.Butnullfindingsareeasytoproduceandnotsomethingwewanttostress here.Theseresultsareavailableonrequestfromtheauthors. 9.Oneoftheauthorsservedasnarratorfortheradioads.Thisauthorwasnotamemberof thefacultywheretheexperimentwasconducted,ensuringthatthestudentscouldnotrecog- nizethenarrator 9svoice.Wealsohadagroupofstudentslistentotheadsbeforeweranthe experimenttocheckontheirquality.Theseindividualsjudgedthesespotstobeacceptable radioads. 10.Becauseoursamplewasheavilyfemale,werananalysestoseeifthesedifferencesmight underminethegeneralizabilityofstudy.Wefoundnosucheffects. 11.TheInstitutionalReviewBoard,whichreviewsresearchprotocolstoensurecompliancewith humansubjectprotection,requiresthatwedescribealltasksinastudytoparticipantsprior tobeginningtotakepartinanystudy.Thus,allparticipantsknewtheywouldbeaskedto recallthecontentofthead.Thisaspectofthestudyisunavoidable.But,atleastallpartici- pantsfacedthesamesetofinstructions,suggestingthatanydifferencesthatariseacross conditionscannotbeattributedtothisproceduralrequirement. 12.Foradditionaldataonthispoint,seehttp://www.psaresearch.com/americans.html. 13.Seehttp://www.psaresearch.com/americans.htmlfordocumentationofthisclaim. 14.Itisalsoimportanttopointoutthatoursubjects(i.e.,undergraduates)arefrequentlisteners toradio(seehttp://wargod.arbitron.com/scripts/ndb/ndbradio2.asp).Notonlydotheylistento radiointheircars,buttheyarelatenightconsumersaswell(seehttp://www.psaresearch.com/ americans.html).StudentsalsomakeextensiveuseofWalkmanandheadphonesasawayto listentotheradioinprivate.Thepointbeingthatwearenotexposingstudentstoanything newinourexperiment.Theyshouldfeelquitecomfortablelisteningtoradioinisolation. 15.Usually,anideaunitcorrespondstoasentence.Asentenceoftencontainsoneideaortheme. Ofcourse,inlonger,compoundsentences,therewouldbemorethanoneideaunit.On request,wewillprovideafullaccountingofhowwemeasuredideaunitsinthisstudy. 16.Wealsotestedforapossibleinteractionbetweenthepartisanaffiliationsofthesubjectand whichpartysponsoredthead.Democratsmightreactdifferently,forexample,toDemocratic ads.Thesetestsdidnotyieldanysignificantfindings. REMEMBERINGATTACKADS91 17.Forexample,thenumberofcorrectrecallsforsubjectsexposedtoGOPspotswas,onaver- age,3.3.AmongparticipantshearingDemocraticspots,themeanwas3.1.Thisgapyielded an F of1.3, p < .25. 18.Wealsoranteststomakesurerandomizationworked.Forexample,welookedforpossible effectsofdemographicdifferencesbetweensubjects.Weenteredsuchvariablesintotheoverall ANOVAascovariates.Innocasedidsignificantdifferences(substantiveorstatistical)arise. 19.Forpurposesofclarity,MemoryTypeandMemoryContentscoringofrecallprotocols yieldedsixcategoriesofdependentvariables:CorrectPositive,CorrectNeutral,CorrectNeg- ative,IntrusionPositive,IntrusionNeutral,IntrusionNegative. 20.Thetechnicalnameforthisassumptionis csphericity. dOurapproachtodealingwiththis statisticalproblemiscommonlyusedinpsychology.Thereare,ofcourse,otherwaystocon- trolforcorrelationsamongobservations.Theissueisnotwhetherourprocedureis cbest, d butwhetherweoffervalidstatisticalinferences.Ourapproachclearlymeetsthatgoal. 21.Thespecificstatistics,usingGGC,wereasfollows: F = 22.069, df = 1,119,eta-squared = .156, p < .001. 22.Therewere,onaverage,5.1recallsofattackadsbyrespondents.Forpositiveads,themean numberofrecallsstoodat4.9.Thisdifferencewasnotclosetostatisticalsignificance p = .66). 23.Thesenullfindingsareconsistentwithotherexperiments,suchasAnsolabehereandIyengar 9s (1995,p.49)findingthat cvoterscameawayfrompositiveandnegativeadvertisementswith aboutthesameamountofissueinformation. dLauetal. 9s(1999)summaryofpreviouswork inthisareafindsnoclearevidencethatthetoneofthespotinfluencesmemory. 24.Thespecificstatistics,usingGGC,wereasfollows: F = 331.93, df = 1.878,223.471,eta- squared = .736, p < .001. 25.Thespecificstatistics,usingGGC,wereasfollows: F = 29.735, df = 1.622,195.588,eta- squared = .200, p < .001. 26.Thespecificstatistics,usingGGC,wereasfollows: F = 15.007, df = 1,119,eta-squared = .112, p < .001. 27.Thespecificstatistics,usingGGC,wereasfollows: F = 40.987, df = 1.644,195.588,eta- squared = .256, p < .001. 28.Thespecificstatistics,usingGGC,wereasfollows: F = 29.02, df = 1,56,eta-squared = .341, p < .001. 29.Thespecificstatistics,usingGGC,wereasfollows: F = 123.56, df = 155,86.787,eta-squared = .688, p < .001. 30.Thesedifferenceswerebothhighlysignificant( p < .001). 31.Thespecificstatistics,usingGGC,wereasfollows: F = 228.85, df = 1.0725,108.863,eta- squared = .784, p < .001. 32.Theaveragenumberofintrusionswas2.4forrespondentsexposedtoanattackad.Themean shrunkto1.1forpositiveads.Thisdifferenceishighlysignificant( F = 22, p < .001). 33.Forexample,onethingthatwarrantsexaminationistheimpactofcandidate-orientedspots. Ourfocuswasonspotsabouttheparties.Itmaybethatbypersonalizingthead,theremay bedifferenteffectsonmemory. 34.SeealsoLodgeandTaber(2000)andKuklinski(2002)foradditionalinsightsconcerningthe on-lineprocessingmodel. 35.OnemightalsoturntoLodge,McGraw,andStroh 9s(1989)findingsonpositiveandnegative memoryinregardstoevaluationofcandidates.Inthatwork,theydonotreportdifferences inrecognitionindexesofmemorybetweenpositiveandnegativeinformation.They,however, reportgreateroverallrecognitionaccuracywhenparticipantsareinstructedtoprocesson- line.Fromacognitivepsychologyperspective,thissuggestsanotherreplicationofthewell- establisheddepth-of-processingeffectonmemory.Materialthatisprocessedmoreinten- GEERANDGEER 92 sivelyisrememberedmoreaccurately.Inagreementwiththegeneralthrustoftheirresearch, ourfindingsreinforcetheclaimthatmemoryisanimportanttopictoconsiderwhenexamin- ingpoliticalphenomena. REFERENCES Anderson,John.R.(1983).Aspreadingactivationtheoryofmemory. JournalofVerbal LearningandVerbalBehavior, 22:261 3295. Ansolabehere,StephenD.,andIyengar,Shanto(1995). GoingNegative .NewYork: FreePress. Ansolabehere,StephenD.,Iyengar,ShantoandSimon,Adam(1999).Replicatingex- perimentsusingaggregateandsurveydata. AmericanPoliticalScienceReview 93: 901 3910. Ansolabehere,StephenD.,Iyengar,Shanto,Simon,Adam,andValentino,Nicholas (1994).Doesattackadvertisingdemobilizetheelectorate? AmericanPoliticalSci- enceReview 88:829 338. Barker,DavidC.(2002). RushedtoJudgment? NewYork:ColumbiaUniversityPress. Bartlett,F.C.(1932). Remembering:AStudyinExperimentalandSocialPsychology . Cambridge:CambridgeUniversityPress. Brians,CraigL.,andWattenberg,Martin(1996).Campaignissueknowledgeandsa- lience. AmericanJournalofPoliticalScience 40:172 3193. Brewer,W.F.(1977).Memoryforthepragmaticimplicationofsentences. Memory andCognition 5:673 3678. Buchanan,Bruce(1996). RenewingPresidentialPolitics .NewYork:RomanandLittle- field. Buchanan,Bruce(2000).Regimesupportandcampaignreform.InLarryBartelsand LynnVavreck(eds.), CampaignReform ,pp.173 3200.AnnArbor:Universityof MichiganPress. Cappella,Joseph,andJamieson,KathleenHall(1997). TheSpiralofCynicism .New York:OxfordUniversityPress. Clinton,Joshua,andLipinski,JohnS.(2003).Negativeadsandturnout:Experimental resultsfromthe2000presidentialelections. JournalofPolitics ,forthcoming. Collins,A.M.,andLoftus,ElizabethF.(1975).Aspreadingactivationtheoryofse- manticprocessing. PsychologicalReview 82:407 3428. Druckman,JamesN.,andLupia,Arthur(2000).Preferenceformation. AnnualReview ofPoliticalScience 3:1 324. Finkel,Steven,andGeer,JohnG.(1998).Aspotcheck. AmericanJournalofPolitical Science 42:573 3595. Freedman,Paul,andGoldstein,Kenneth(1999).Measuringmediaexposureandthe effectsofnegativecampaignads. AmericanJournalofPoliticalScience 43:1189 31208. Garand,JamesC.,andGraddy,KristyL.(2001).Explainingthemediatedeffectsof negativecampaignadvertisingonvoterturnout.Paperdeliveredatthe2001Mid- westPoliticalScienceAssociation,Chicago. Geer,JohnG.(2000).Assessingattackadvertising:asilverlining.InLarryM.Bartels andLynnVavreck(eds.), CampaignReform ,pp.62 378.AnnArbor:Universityof MichiganPress. Geer,JohnG.,andLau,RichardR.(2001).Simulationandtheory:anewwaytomodel theeffectsofcampaigns.unpublishedmanuscript. REMEMBERINGATTACKADS93 Harris,R.J.(1974).Memoryandcomprehensionofimplicationsandinferences ofcomplexsentences. JournalofVerbalLearningandVerbalBehavior 13:626 3 637. Herrnson,Paul(2000). CongressionalElections ,3rdedition.Washington,DC:Con- gressionalQuarterlyPress. Hitchon,JacquelineC.,andChang,Chingching(1995).Effectsofgenderschematic processingonthereceptionofpoliticalcommercialsformenandwomencandidates. CommunicationResearch 22:430 3458. Humbert,Marc(2002).McCallstressesblackrootsinnewradioad.APwireservices, August9. Humphrey,Tom(2002).Alexanderradioadstoutconservativeside. KnoxvilleNews- Sentinel .March27,b3. Iyengar,Shanto,andMcGuire,WilliamJ.(1993). ExplorationsinPoliticalPsychology . Durham,NC:DukeUniversityPress. Jacobson,GaryC.(2000). ThePoliticsofCongressionalElections ,5thedition.New York:Longman. Jamieson,KathleenHall(2000). EverythingYouThinkYouKnowaboutPolitics... AndWhyYouareWrong. NewYork:BasicBooks. Kahn,KimF.,andKenney,PatrickJ.(1999).Donegativecampaignsmobilizeor suppressturnout? AmericanPoliticalScienceReview 93:877 3890. Kuklinski,JamesH.(2002). ThinkingaboutPoliticalPsychology. Cambridge:Cam- bridgeUniversityPress. Kuklinski,JamesH.,andHurley,NormanL.(1994).Onhearingandinterpretingpolit- icalmessages:acautionarytaleofcitizencue-taking. TheJournalofPolitics 56: 729 3751. Kuklinski,JamesH.,Luskin,RobertC.,andBolland,John(1991).Whereisthe schema?Goingbeyondthe"S dwordinpoliticalpsychology. TheAmericanPolitical ScienceReview ,85:1341 31356. Lang,A.,Newhagen,J.,andReeves,B.(1996).Negativevideoasstructure:emotion, attention,capacity,andmemory. JournalofBroadcastingandElectronicMedia 40: 460 3477. Lau,RichardR.(1985).Twoexplanationsfornegativityeffectsinpoliticalbehavior. AmericanJournalofPoliticalScience 29:119 3138. Lau,Richard,andPomper,GeraldM.(2001).Effectsofnegativecampaigningon turnoutinU.S.senateelections,1988 31998. JournalofPolitics 63:804 3819. Lau,RichardR.,andSears,DavidO.(1986). PoliticalCognition .Hillsdale,NJ:Lawer- enceErlbaumAssociates. Lau,RichardR.,andSigelman,Lee(2000).Effectivenessofnegativepoliticaladver- tisements.InJamesA.Thurber,CandiceJ.Nelson,andDavidA.Dulio(eds.), CrowdedAirWaves ,pp.17 329.Washington:Brookings. Lau,Richard,Sigelman,Lee,Heldman,Caroline,andBabbit,Paul(1999).Theeffects ofnegativepoliticaladvertisements:ametaanalysis. AmericanPoliticalScienceRe- view 93:851 3876. Lawton,L.Dale,andFreedman,Paul(2001).Beyondnegativity:advertisingeffectsin the2000Virginiasenaterace.Paperdeliveredatthe2001MidwestPoliticalScience Association,Chicago. Lodge,Milton,andHamill,Ruth(1986).Apartisanschemaforpoliticalinformation processing. AmericanPoliticalScienceReview 80:505 3519. Lodge,Milton,andMcGraw,Kathleen(1995). PoliticalJudgment .AnnArbor:Univer- sityofMichigan. GEERANDGEER 94 Lodge,Milton,McGraw,Kathleen,Stroh,andPatrick(1989).Impressivedrivenmodel ofcandidateevaluation. AmericanPoliticalScienceReview 83:399 3419. Lodge,Milton,Steenbergen,MarcoR.,andBrau,Shawn(1995).Theresponsivevoter: campaigninformationandthedynamicsofcandidateevaluation. AmericanPolitical ScienceReview 82:309 3326. Lodge,Milton,andTaber,Charles(2000).Threestepstowardatheoryofmotivated politcalreasoning.InArthurLupia,MathewMcCubbinsandSamuelPopkin,(eds.), ElementsofReasoning .Cambridge:CambridgeUniversityPress. Loftus,ElizabethF.(1993).Therealityofrepressedmemories. AmericanPsychologist 48:515 3537. Loftus,ElizabethF.(1979). EyewitnessTestimony .Cambridge,MA:HarvardUniver- sityPress. Loftus,ElizabethF.,andRead,J.D.(1994).Psychotherapyandmemoriesofchild- hoodsexualabuse:Acognitiveperspective. AppliedCognitivePsychology 8:281 3 338. Lupia,Arthur,McCubbins,Mathew,andPopkin,Samuel(2000). ElementsofReason . Cambridge:CambridgeUniversityPress. Massaro,DominicW.,andCowan,Nelson(1993).Informationprocessingmodels: microscopesofthemind.InLymanPorter,andMarkR.Rosenzweig,(eds.), The AnnualReviewofPsychologyVol.44 .PaloAlto:AnnualReviewsInc. Matlin,Margaret(2002). Cognition ,5thedition.NewYork:InternationalThomson Publishing. McGraw,Kathleen,andLodge,Milton(1996).Reviewessay:politicalinformationpro- cessing. PoliticalCommunication 13:131 3138. McGraw,Kathleen,Lodge,Milton,andJones,Jeffery(2001).Thepanderingpoliti- ciansofsuspiciousminds.Unpublishedmanuscript,OhioStateUniversity. Miller,George(1956).Humanmemoryandthestorageofinformation .IRETransac- tionsonInformationTheory IT-2,129 3137. Newhagen,J.,andReeves,B.(1992).Theevening 9sbadnews:effectsofcompelling negativetelevisionnewsimagesonmemory. JournalofCommunication 42:25 341. Price,Vincent,andZaller,John(1993).Whogetsthenews?Alternativemeasuresof newsreceptionandtheirimplicationsforresearch. PublicOpinionQuarterly 57: 133 3164. Pritchert,James.W.,andAnderson,Richard.C.(1977).Takingdifferentperspectives onastory. JournalofEducationalPsychology 69:309 3315. Rau,Jordan(2002).Firstadhitstheairways. Newsday ,May21st,a21. Reeves,B.,Newhagen,J.,Maibach,E.,Basil,M.,andKurz,K.(1991).Negativeand positivetelevisionmessages:effectsofmessagetypeandcontextonattentionand memory. AmericanBehavioralScientist 34:679 3694. Riker,William(1996). StrategyofRhetoric:CampaigningfortheAmericanConstitu- tion .NewHaven:YaleUniversityPress. Robison,Clay(2002).Cornynpromisesinadtofightforwaterrights. TheHouston Chronicle, April26th,p.42. Roediger,HenryL.(1996).Memoryillusions. JournalofMemoryandLanguage 35: 76 3100. Schulberg,Pete(1996). RadioAdvertising:TheAuthoritativeHandbook ,2ndedition. NewYork:McGrawHill. Schvaneveldt,RogerW.(1990). PathfinderAssociativeNetworks:StudiesinKnowl- edgeOrganization .Norwood,NJ:AblexPublishingCorp. REMEMBERINGATTACKADS95 Shapiro,MichaelA.,andRieger,RobertH.(1992).Comparingpositiveandnegative politicaladvertisingonradio. JournalismQuarterly 69:135 3145. Solso,RobertL.(2001). CognitivePsychology ,6thedition.NeedhamHeights,MA: Allyn&Bacon. Valentino,Nicholas(1999).Crimenewsandtheprimingofracialattitudesduring evaluationsofthepresident. PublicOpinionQuarterly 63:293 3320. West,Darrell(2001). AirWars ,3rdedition.Washington,DC:CongressionalQuarterly Press.

less