A Field Test of Subliminal Self-Help Audiotapes: The Power of Expectancies Author(s): Eric R. Spangenberg, Carl Obermiller and Anthony G. Greenwald Source: Journal of Public Policy & Marketing, Vol. 11, No. 2 (Fall, 1992), pp. 26-36 Published by: American Marketing Association Stable URL: http://www.jstor.org/stable/30000271 Accessed: 18-12-2015 13:37 UTC Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Public Policy & Marketing. http://www.jstor.org This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions A Field Test The Power of of Subliminal Self-Help Expectancies Audiotapes: Eric R. Spangenberg,CarlObermiller,and AnthonyG. Greenwald In theface of evidence to the contrary,consumerscontinue to believe in the efficacy of subliminalmessage self-help (SMSH)audiotapes. The authors contend that the value consumersattributeto these tapes stemsfrom the illusion of efficacy, the result of an expectancyeffect. Two double-blindstudies tested subliminal message self-help audiotapes claimed to improvememory,increase self-esteem, or induce weight loss. Actual content and labeled content were independentlyvaried; some subjects believed they were using memory tapes but were actually using weight loss or self-esteem tapes, and so on. After a monthof use (per manufacturers'recommendations)by volunteersubjects, there were no effects of subliminalcontent in the claimed direction on any of the three dimensions.In contrast, there were changes or the perception of changes correspondingto the tape labels. Theseplacebolike effects are hypothesizedto resultfrom consumerexpectancies.If consumerswhoforsake more conventionaltreatmentsfor SMSHaudiotapes are misled, they suffer both the loss of the price of the tapes and, more importantly,opportunitycosts. Thus, the potentially damaging consequencesare a major concernfrom a public policy perspective. Subliminal messageself-help(SMSH)audiotapesostensibly use subliminalmessages to treat a wide rangeof listeners'problems.Theyrepresenta large andgrowingmarket,annualsalesfiguresfor SMSHaudiobut estimatesrangefrom$50 miltapesareunobtainable, lion to morethan$1 billiona year[Lofflin1988;TheSun studieshavedemonstrated 1990].Thougha few laboratory subliminal effects replicable [e.g., CheesmanandMerikle 1985, 1986; Greenwald,Klinger, and Liu 1989; Janiszewski 1988; and Kunst-Wilsonand Zajonc 1980], virtu- ally all testsof subliminalprocesseshavefailedto findany significanteffectson complexbehaviors(fora review,see Pratkanis andGreenwald[1988]).Thelackof scientificevidence notwithstanding,there is widespreadacceptance amongconsumersthatsubliminalprocessingis aneffective aid. Severalpointscanbe madeaboutthe substantial demand for SMSHtapesandthe needfor additionalresearch.First, manyconsumersbuynotone,butmultipleSMSHtapes,so we presumethey experiencesomebenefit.Second,the belief in thevalueof SMSHtapesis notlimitedto specificsegments;it is widespread[Block and VandenBergh 1985; Zanot,Pincus,andLamp1983].Furthermore, thoughadmitis Assistant Professorof Marketing,College ERICR. SPANGENBERG of BusinessandEconomics,WashingtonStateUniversity.CARL is Associate Professorof Marketing,Albers School of OBERMILLER G. GREENWALD is Professor Business, Seattle University. ANTHONY of Psychology,Department of Psychology,Universityof Washington. The researchwas supportedin part by ResearchGrant MH41328fromthe NationalInstituteof MentalHealthandthe EdnaG. BensonFellowshipFundat theUniversityof Washington Graduate Schoolof Business. 26 tingthelackof evidenceof subliminalmessageeffects,one frompastremightraisequestionsaboutthegeneralizability search(mostlylaboratorystudiesin unnatural usage situations)to SMSHaudiotapeuse. Ourresearchintentionwas to give the producta fairtest in the field. Further,we propose an alternativeexplanationof the apparenteffects of the tapesthatis basedon consumerexpectancies.Though ourresearchis specificallya test of one product,we argue thattheeffectof expectanciesmayinfluenceconsumerevaluationsof manyproducts.SMSHaudiotapes arerepresentative of productswith ambiguousor subjectivebenefits,for which consumerexpectanciesmay bias evaluationprocesses. SMSH audiotapesrepresenta class of productswhose principalor distinguishingattributesare illusory.The notionof a productwhosecentralattribute(anddistinguishing feature)is actuallynonexistentmayraiseskepticism,butit is consistentwith marketing'sconceptof the "totalproduct" [Kotler1980].Productsconsistof morethanthephysical object;in particular, theyareenlargedby an imagecreatedby promotionandsupportedby consumerbeliefs.Exstudiesprovide resultsin sensorydiscrimination perimental have evidencethatconsumers'beliefsin productattributes Alon of effects powerful perceptions productperformance. thatbrand lison andUhl [1964],forexample,demonstrated was thebasisfordisimageratherthanobjectiveingredients have criminatingamongbrandsof beer.Otherresearchers foundsimilartastediscrimination biases due to labels for soft drinksand cigarettes[Husbandand Godfrey 1934; PronkoandBowles 1949].Hence,thereis ampleprecedent in marketing researchto supthoughtandproductperception we defineas of which the port concept illusoryattributes, thatareclaimedto exist butdo not haveany characteristics Journal of Public Policy & Marketing This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions Vol.11 (1) Fall 1992, 26-36 Journalof PublicPolicy& Marketing objectivebasis andwhosenaturecan be neitherconfirmed nordisconfirmed by the typicalconsumer. We proposethatconsumers'evaluationof SMSHaudiotapesis subjectto the samebiasesthatresultin preference discrimination differamongbrandsthathaveimperceptible ences withouttheirbrandlabels. Promotionalclaims and wordof mouthcreateexpectanciesaboutproductperformancejust as theydo for branddifferences.The expectancies, in turn,influenceevaluationsof the productsin use. Thus,the beliefthatsubliminalprocessingcanassistlearning maycreatean expectationfor learningto occur. Boththe specificclaimsof SMSHtapesandthe general role of expectanciesin productevaluationsarerelevantissues for publicpolicy consideration.Regulatoryagencies and state attorneysgeneralhave indicatedsome concern aboutthe advertisingof SMSHtapes,butwhatactionwill be takenis unknown.If consumersaremisledandexperience nothingor only an illusoryeffect,theysufferboththe loss of the priceof the tapesand,moreimportantly, opportunitycosts. Therearemorethan400 SMSHaudiotapetitles, includingtapesdesignedto treathigh bloodpressure, cancer,substanceabuse,andlearningproblems.Toforsake moreconventional treatments forSMSHaudiotapes mayresult in damagingconsequences,andone canhopethatcurrentclaimswouldbe disallowed.The purposesof ourrefieldtestsof theefsearchwere(1) to conductdouble-blind fectivenessof SMSHaudiotapesand(2) to assessthe role evaluations.Next of consumerexpectanciesin performance we reviewliterature relevantto ourtestsof subliminalmessage effectsandexpectancies. Literature 27 ing the propositionthatsubliminalmessagesinfluencemotivationor complexbehavior. SMSHaudiotapesarethe majorformin whichsubliminal techniqueshave appearedin the marketplace;they makeclaimsfordozensof therapeutic effects,includingimprovedself-esteem,cessationof smokingor druguse, and weightloss. SMSHaudiotapes typicallyhaveblandaudible contentsuchas classicalmusicor soundsof the oceanand are marketedfor self administration; usersare awarethat are a subliminal they receiving message.Thelayperson,of course,has no meansof evaluatingthe subliminalcontent of SMSHaudiotapes; consumershavefaiththatthe imperexists andthatit will be effective. ceptiblecentralattribute In fact,consumers'faithin subliminalmessagesraisesa of muchlaboratory eviquestionaboutthe generalizability dence againstsubliminaleffects. In the typicallaboratory study,subjectsare unawarethatsubliminalprocessesare being studied;thereforetheyhaveno expectationthatsubliminalcues will operateeffectively.Purchasers of SMSH audiotapes,in contrast,arguablyexpectthe tapesto work for them.If consumerfaithin the existenceandefficacyof subliminalmessagesis a necessarycondition,the question of SMSHaudiotapeeffectivenessrequiresa field test with subjectswho arefully awareof the tapes'subliminalmessages.This knowledge,of course,suggeststhe hypothesis thatit is nota subliminalmessagebutthebeliefin a subliminalmessagethatcausesanyobservedeffects.If SMSHaudiotapeshavebenefits,theymaybe dueto a responseto expectanciesratherthanany link betweensubliminalmessagesandcomplexbehaviors. Expectancies:Self-FulfillingProphecies,Placebo Effects,and Self-Deception SubliminalPersuasion Studiesdoneunderthe rubricof "subliminaltechniques" are diverse in methodand topics studied.Pratkanisand Greenwald [1988]describea sequenceof eventscharacterizing the progressionof researchon subliminalpersuasion fromtheearlieststudy[Dunlap1900]to thepresent:"A researcher reportsa subliminaleffect;theeffectcannotbe replicated;nevertheless,the moresensationalaspectsof the in textbooksandthemass findingreceivepopulartreatment media."Themediaoftenprovideprominent (andusuallyuncritical)coverageof subliminalpersuasion[e.g., Lofflin 1988] andseveralsurveyshavedemonstrated fairlywidespreadpublicacceptanceof the concept[e.g., Block and VandenBergh1985;Zanot,Pincus,andLamp1983]. Thesize of theSMSHaudiotape marketandpublicinterest in the producthavemotivatedseveralteststhatpurportedly supportthe effectivenessof theproduct.Oftenserving moreas advertisingclaimsthanscientificevidence,many of the tests supportingthe effectivenessof SMSHaudiotapes have been producedby researcherson payrollsof SMSHaudiotapemanufacturers [e.g.,Taylor1988;Vandennone of the reportedstudies Boogert1984].Furthermore, has useda double-blind designor beenpublishedin a credible scientificjournal.In separatereviewsof the literature on unconsciousprocessing,VokeyandRead[1985],Moore andGreenwald[1988]all draw [1982;1988],andPratkanis the conclusionthatthereis no scientificevidencesupport- In this section,we establishthatexpectancyeffectsprovide a useful paradigmfor explainingthe illusionsof efficacy for SMSHaudiotapesandrelatedproducts;ourdiscussion is not intendedto be an exhaustivereviewof the hundreds of studiesrelatedto theconceptsof interest[seeDarleyand Fazio 1980;Ross andOlson1981].Threeconceptsmaybe relevantto the effect of expectancieson productperformance:self-fulfillingprophecies,placeboeffects,andselfdeception. The notionthata person'sexpectanciesaboutanother often serveas self-fulfillingprophecieswas introducedby sociologist RobertMerton[1948, 1957], who observed [1957, p. 423]: "The self-fulfillingprophecyis, in the beginning, afalse definitionof the situationevoking a new behavior which makes the originally false conception come true. The specious validity of the self-fulfilling prophecyperpetuates a reign of error. For the prophet will cite the actual course of events as proof thathe was rightfrom the very beginning." Probablythe best-knownexample of self-fulfilling prophecy was demonstratedin the work of Rosenthal and Jacobson [1968]. This effect, labeled the Pygmalion effect, is manifestedwhen a personchanges in responseto another person's expectations (as students respond to teachers' expectations). Thus, a self-fulfilling prophecy occurs when expectationresultsin a changein the situationthatcreates objective expectancy confirmation. This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions 28 Field 'Ist of SubliminalAudiotapes A relatedprocessis the placeboeffect. Whena person has expectationsaboutan object,the objectcannotchange, but the person'sresponseto the objectcan be affectedby on placeboeffectsin his or herexpectancies.Theliterature is extensive(for a retreatment and medicine psychiatric and Schwartz see White,Thrsky, view, [1985]);the power in hundredsof studies.For of expectanciesis demonstrated example,Jospe [1978] reportedKissel and Barrucand's [1964] presentationof instancesin whichplaceboswere andrheuusedto treatcardiovascular, endocrine, pulmonary, maticdiseases,amongotherproblems,as well as in surgical recovery.Theseauthorscite results(adaptedfromHaas, Fink,andHartfelder [1959])summarizing placeboeffectiveness (versuscontrolsubjectswhobelievedtheyreceivedno treatment) rangingfrom28 to 58%.A placeboeffectoccurs whenexpectationchangesthe actor'sresponseto a situation,resultingin objectiveexpectancyconfirmation. in contrast,is alsoevidencedin situations Self-deception, of stiminterpretation involvingperceptionandsubsequent to depend uli. Evaluationof stimulihas beendemonstrated stronglyon priorexpectationsheld by the perceiver[e.g., Brunerand Minturn1955; Sujan, Bettman,and Sujan involvesno changein thesituationor 1986].Self-deception the actor'sbehavior,merelychangesin the actor'sperceptionof the situation,resultingin subjectiveexpectancyconfirmation. Consumerresearchershave investigatedthe biasingeffects of expectancies createdby advertising.Deighton is effectiveif it canarouse [1984]proposedthatadvertising confirmwith thatconsumerscan subsequently expectations on thepofocused productexperience.HochandHa [1986] in this tential for bias process;they conceptualizedadas inducedexpectancies "hypotheses"thatconsumerstest whenprodwith productexperience.In theirexperiments, uct performanceevidence was unambiguous,ad effects werereplacedby experienceeffects;however,whenprodevidencewas ambiguous,advertisinghad uct performance significanteffects on productevaluations.Subjectsdisplayeda bias in theirprocessingof informationfromthe use of the product,confirmingthe expectationscreatedby the advertising. HochandHa [1986] identifiedone factorthatmediates the role of expectancies-the ambiguityof performance data.Whenanoutcomeis clearlyunsatisfactory, priorexpectationsof success cannothold, but if resultsare unclear bias as positive.Thisconfirmation theymaybe interpreted was identifiedand demonstratedby Snyderand Cantor [1979] andSnyderandSwann[1978] in social exchanges andhasbeenshownto affectbothencodingof information andretrievalfrommemory.A secondmediatingfactormay be the extentto whichconsumerperformanceinfluences If the consumer'sbehaviorcan indeproductperformance. determine thesuccessof a product, priorexpectanpendently cies can be confirmedregardlessof productperformance. Such a self-fulfillingprophecyis illustratedby consumers whobuy "mileage-enhancing gasolineadditives,"thenconfoundtheiractionsby takingothermeasuresto improve in mileageto mileagebut attributeresultingimprovement the effectivenessof the product. Hypotheses Thoughthe thrustof ourresearchis the investigationof an alternativeexplanation,a necessaryconditionis the test of theclaimedsubliminalmessageeffects.Ouraimwas to designtheresearchto makethetestas strongas possible.Dedesign,the field setting,self-selected spitethe double-blind subjects,andthenaturaluse situation,thedearthof anysupportingscientificevidencein favorof subliminalpersuasion andthe lack of anycompellingtheoreticalframework led us to expectno effectsof subliminalmessagecontent. hasnomain subliminal H1: SMSHaudiotape messagecontent measures. effectonanydependent A failureto rejectH1 wouldlead to the conclusionthat subliminalmessagesare not responsiblefor any actualor perceivedchangesin dependentmeasuresthatmightbe exto periencedby subjects.Specialdesignissues appropriate the test of the nullhypothesisarediscussedsubsequently. The alternativeexplanationfor SMSHaudiotapeeffects of expectancyeffects. framework is basedon thetheoretical Productclaimsandgeneralpublicacceptanceof producteffectivenessstimulateexpectanciesheldby consumersabout productefficacy; these beliefs influenceproductperforForboth manceor the perceptionof productperformance. sublimeffect of of the absence assume H2 andH3, we any inalmessagecontent.H2 positsthestrongereffectof expectancies;labelsarehypothesizedto createexpectanciesthat bias subsequentconsumerbehaviorsin ways thatresultin benefits.H3 positsthe weakereffect objectiveperformance to createexpectanlabelsarehypothesized of expectancies; cies thatbias subsequentencodingor retrievalof information in bias-confirming ways.H2 is rootedin the concepts of self-fulfillingprophecyandplaceboeffects;if H2 holds, peoplereallyget whatthey expectto get. H3 is basedon if H3 holds,peopleonly thinktheyget what self-deception; theyexpectto get. H2: Perceived tapecontent(i.e.,tapelabels)causesobjective on relevantdimensions. (actual)improvement H3: Perceivedtapecontent(i.e., tapelabels)causessubjective on relevantdimensions. (perceived)improvement We areawareof no workthathasbeendoneon the disof prodtinctionbetweenobjectiveandsubjectiveappraisals uct performancethat might suggest whetherSMSH auconformsmoreto H2 or H3. Thenecdiotapeperformance both hypotheseshold: data about for conditions essary weightloss (not the scale reading,but perceptionsof personal appearanceand attitudestowardweight loss), selfesteem, and memory are likely to be complex and ambiguous. For all three issues, many consumer behaviors in ad- ditionto use of SMSHaudiotapesmayhaveeffects. Study 1 Design exIn study1, we testedH1 andH2 andmadea preliminary plorationof H3, usinga threeactualproduct(weightloss, self-esteemimprovement,memoryenhancement)by threela- membeledproduct(weightloss, self-esteemimprovement, were commerProducts factorial design. ory enhancement) cially preparedsubliminaltapes claimed to provide the re- This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions Journalof PublicPolicy& Marketing spective treatments.We preparedthe labels, which containedinformationleadingsubjectsto believethe sublimito the identification andmatched nal contentcorresponded other(mis)identifyinginformationgiven to subjects.Subjects receivedeithercorrectlylabeledor mislabeledtapes. Subjectsand Procedure Subjectswererecruitedfor a month-longstudyas unpaid volunteersat studentcentersignuppointsandthroughcomThe study munityandcampusnewspaperadvertisements. was describedas one in whichparticipation couldimprove problemareascommonto adults(weight,self-esteem,memory) throughthe use of SMSH audiotapes.Personswere qualifiedto participatein the studyif they wereinterested in tryingto achievetwo or moreof thethreeself-helpgoals. Becausepeopleto whomgeneralization wouldbe appropriate(i.e.,purchasing users)areself-selected,paymentof subjects would have been inappropriate;participantswere resultsof the studyanda freetapeupon given preliminary studycompletion.Subjectswereblindto thestudy'shypothwereblindto the respective eses just as the experimenters conditionassignmentsof individualsubjects.Of 112 subjects who completedthe pretest,83%fulfilledall studyrequirements.No differencesbetweensubjectscompleting the studyandthosedroppingout wereevidenton the basis of a comparisonof demographiccharacteristics including age, sex, andmaritalstatus. Subjectswere scheduledby telephoneor personalcontactfor pretestsessionsconductedin groupsof fromone to eightsubjects.Uponarrival,subjectsreadandsignedan informedconsentform,rankingon the formtheirpreference amongthe self-improvementareas(weight, self-esteem, one of two memory).All subjectswerethenadministered parallelformsof a batteryof pretestmeasures(counterbalancedbetweenpretestand posttest).Commerciallyproducedtapeswith subliminalcontentwereusedfor all conconditions.Whetherthe ditions;only labelingdifferentiated was or labeled wasdetermined rantape correctly incorrectly domly by choosingfromtapeslabeledby a personother thanthe experimenter. Criticalto bothstudieswas separation of the labelinganddistributiontasks,thusensuringa double-blindassignmentprocedurein whichboththe subadministrator wereblindto the subject andtheexperiment conditions. jects' assigned Rarelyimplementedin behavioralresearch,the double-blind ensuredagainsta procedure demandeffect(i.e., subjects'responding to consciousorunconsciousexperimenter aboutthehypotheses). expectations Eachsubjectwasgivena tapelabeledwithhis orherfirst choiceof treatment domain.Thispartialnonrandom assignmentprocedureservedto assign subjectsto conditionsin whichtheirdesireto experiencepurported tapebenefitswas relativelygreater,thus theirexpectanciesof tape efficacy werelikelyto be strongerthantheywouldbe withrandom assignments.Subjectswerethengivena packetof information including(1) tape scriptcorrespondingto subject's thelabelmanipulation andproviding tapelabel(reinforcing furtherbasis for subjectexpectancies),(2) listening frequencylog in whichsubjectsrecordeddetailsof eachlistening session, (3) tape manufacturer publicationespousing the virtuesof SMSHtapeuse, (4) copyof studydescription 29 andconsentform,and(5) directionsfor reportingtheirlisto reporttheirlisteningfrequency(subjectswereinstructed a to tening frequency telephone answering machine). Ninetypercentof the subjectsreturneda completedlistening frequencylog. Participantsreportedlistening to the tapesan averageof 42 timesoveran averageparticipation periodof 37 days.Afterlisteningto the tapesfor at least fourweeks,subjectsreturned to thelabto completeposttest procedures. Materialsand Measures Audiotapes CorrectlylabeledSMSHaudiotapesweredonatedby three establishedmail-ordersuppliers.Audibleverbalcontent andtapecontentwas carefully identifyingthe manufacturer erasedfromtapeintroductions audiolab. by anindependent labelsand affixingstudycondiRemovingmanufacturers' tion labelswas done conditionby condition,one tapeat a of tapes. time,to preventconfusionor misidentification MemoryAbility To measurepre-andposttestmemoryability,two counterbalancedparallelversionsof subtasksof theWechslerMemory Scale [Stone,Gardner,and Albrecht1946;Wechsler in a grouplabora1945]weremodifiedfor administration tory setting.Wechslermemorysubtaskswere (1) immediaterecallof briefstories,(2) digitrecall(forwardandbackward),(3) visual reproductionof drawings,and (4) word pairassociatelearningexercises. Self-esteem The threeself-administered scalesusedto test for an effect on subjects' self-esteem were previouslyvalidatedand widely acceptedindicatorsof self-esteem.The scales used were the 10-itemRosenberg[1965] self-esteemmeasure, the 16-itemTexasSocial BehaviorInventory:VersionsA andB (TSBI)[Helmreich andStapp1974],andtheBeckDepressionInventory(BDI)[Becket al. 1961],whichwas reversescoredto makeit consistentwiththe others.Correlationsamongall pairsof measureswerehigh(r> .4 forboth studies1 and2). WeightLoss An averageof readingsof threescaleswas the measureof bodyweight.Also, a specificmeasureof expectanciesfor personalweight (weight locus of controlscale; WLOC) was administered [Saltzer1982].WLOCprovideda measureof changein subjects'attitudestowardweightloss that has beenused as a predictorof futurebehaviorsrelatedto weightreduction[Weiss1977]. WLOCwas includedbecause the 30-day periodof the studymay have not been enoughtimefor a personto lose weight. Perceived Efficacy Priorto learningtheiractualtapecontentat the conclusion of the study,subjectswere askedwhethertheirlong-term memoryhadimproved(yes or no) duringthe tapelistening subperiod.Thisquestion,designedto exploreH3,measured This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions 30 Field Testof SubliminalAudiotapes jects' perceptionsof changein memoryabilityas opposed to objectivechangesmeasuredby the Wechslersubtasks. Results arepresentedin a consistent Resultsof tapemanipulations form for all treatments.An a priori combinationof the Wechslerscale componentswas summedandstandardized (mean=O,S.D.=1) to test for effects of manipulationson memory.Thethreeself-esteemmeasureswerestandardized separatelyand then averagedto form one overall selfesteem score. All dependentmeasurepre- and posttests werestandardized by usingtheirrespectivecomponentpredeviations.Thismethodof posttest testmeansandstandard scorestandardization preservedscaleunitsandthedirection of posttestchangesrelativeto pretestscores.Thougha single measure(averageof threescale readings)was used to test for pre- and posttestweight differences,we use the formatto presenttheseresultsandthose samestandardized for the single-scoreWLOCscale.Perceivedefficacymeasas pertheseresultsarepresented ureswerenotstandardized; centages.Measureswereanalyzedby meansof a 3 X 3 factorialANCOVAprocedurewith posttestscoreas criterion andcorresponding pretestscoreas covariate.In general,the controlgroupfor comparisonsconsistedof the two groups not specificallytargetedfor the dependentmeasurein each Forexample,the statisticaltestforthe specificcomparison. effect of actualself-esteemsubliminalcontentcontrasted theaverageself-esteemdataforthesubjectswhoreceivedactual self-esteemtapeswith those for the subjectswho receivedactualmemoryandweighttapes. Self-esteem Figure1 presentsthe self-esteemdata.ConsistentwithHI, thereis no maineffectof actualsubliminalcontenton subjects' self-esteem.Immediatelyapparentis the maineffect of time; posttestlevels of self-esteem(andmemory)are higherthanpretestlevels in all conditions.This effect, whichmay be due to practiceprovidedby the pretests,is not relevantto any of the hypotheses.Usingthe ANCOVA procedure,we found that adjustedposttest self-esteem scoreswerenot significantlyhigherwhensubjectsreceived a trueself-esteemtape thanwhen they receiveda mislabeled self-esteem(truememoryor weightloss) tape.The of actualmemoryandactualweightcondition combination selfsubjectsservingas a controlgrouphada standardized esteemscoreat pretestof .06, whichincreasedto a posttest scoreof .47; the actualself-esteemconditiongroup'sscore at pretest was -.11, which increased at posttest to .19 (F(1,82)=2.8,p=.098,MSE=.23).In thetestof H2, subjects listeningto tapesidentifiedas self-esteemdidexperiencean increasein self-esteemin comparisonwith subjectslistening to tapesidentifiedas memoryor weight;theeffect,however,is not statisticallysignificant(combinedlabeledmemory& weightpretest/.18to posttest/.44versuslabeledselfesteempretest/-.35to posttest/.25,F(1,82)=1.46,p=.231, MSE=.23). Memory Ability WechslerscalesubFindingsforthe averaged,standardized tasks are also shown in Figure 1. Again,consistentwith Figure 1. DependentMeasuresby Actualand Labeled Conditions(Study 1) 0.6 0.5 S T D Z D S E S C A L E 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 ACTUAL ACTUALMEMORY LABELED LABELEDMEMORY SELF-ESTEEM & WEIGHT SELF-ESTEEM & WEIGHT MEMORY 0.6 0.5 0.4 0.3 0.2 0.1 S T D Z D M E M 0 S C -0.1 A L -0.2 E -0.3 -0.4 WEIGHT LABELED LABELED WEIGHT ACTUAL ACTUAL MEMORY & SELF-ESTEEM MEMORY & SELF-ESTEEM WEIGHT 0.6 0.5 S T D Z D W G H T 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 ACTUAL WEIGHT ACTUAL MEMORY & SELF-ESTEEM PRETEST LABELED WEIGHT LABELED MEMORY & SELF-ESTEEM POSTTEST H1, no main effect of actual subliminaltape content is foundfor subjects'memoryability.Adjustedposttestmemorymeasurescoreswereno betterwhensubjectsreceiveda truememoryimprovement tapethanwhenthey receiveda mislabeled memory (true self-esteem improvementor This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions Journalof PublicPolicy& Marketing weightloss) tape(combinedactualself-esteemandweight, pretest/.05to posttest/.40versusactualmemory,pretest/.10 to posttest/.36, F(1,83)=.25,p=.621,MSE=.47).A statiseffect observedon the memorymeasure tically significant as a resultof thetapelabelmanipulation, however,supports H2. Subjectslisteningto tapesidentifiedas memoryima significantincreasein memory provementdemonstrated withsubjectslisteningto tapesidentiabilityin comparison fied as self-esteemimprovement or weightloss (combined labeledself-esteemand weight,pretest/.00to posttest/.20 versus labeled memory, pretest/-.05 to posttest/.53, F(1,83)=4.07,p=.047,MSE=.47). WeightLoss Also shownin Figure1 areresultsof actualandlabeledeffects on the averageweightmeasure.ANCOVAresultsfail to supportH2, but areconsistentwithH1. In general,subjects lost no weightas a resultof truetapecontent(combinedactualself-esteemandmemory,pretest/-.07to posttest/-.08 versusactualweight, pretest/.14to posttest/.16, F(1,82)=.07,p=.790,MSE=.01).Nor did they lose weight as a resultof tapelabel(combinedlabeledself-esteemand 13 versuslabeledweight, memory,pretest/-.14to posttest/-. to pretest/.35 posttest/.34, F(1,82)=1.32, p=.254, MSE=.01). WLOC(the measureof attitudetowardweightloss) resultsmirrorthoseof averagedweight.Thereis no effectof actualtapes(combinedactualself-esteemandmemory,pretest/-.12to posttest/-.01versusactualweight,pretest/.23to posttest/.23, F(1,82)=.06, p=.812, MSE=1.04)or labels (combinedlabeledself-esteemandmemory,pretest/.02to posttest/.12versuslabeledweight,pretest/.-.05to posttest/ -.08, F(1,82)=.7,p=.406,MSE=1.04)on WLOC. Cell by Cell Contrasts Onemightarguethatthe strongesttestof Hi-the situation in whichsubliminalmessagesaremostlikelyto operatewouldbe whensubjectsreceivethe SMSHaudiotapesthey expect (i.e., the actualtape correspondsto the label).We testedthis notionby contrastingeach of the threegroups thatreceivedactualtapescorresponding to theirlabelsto apcontrolgroups.Two contrastswereconductedfor propriate each of the threeconditions:(1) corresponding actualand label versuscorresponding actualand "other"label, and actualand label versus "other"actual (2) corresponding andcorresponding label.Noneof the six comparisons were statistically significant. Thus, not even conditions with the most potential producedevidence of a subliminal message effect. 31 mentin long-termmemoryabilitywhereasfewerthan30% of subjectsin the combinedlabeledself-esteemandweight conditionindicatedincreasedlong-termmemoryability (F(1,83)=4.36,p=.04,MSE=.29). Discussion Resultsof study1 provideevidencethatsubliminalcontent of tapesdoesnot conferbenefitsto consumersas manufacturersclaim.TheevidencesupportsH1;no effectof actual subliminal tapesis foundon eitherobjectiveorperceivedefficacymeasures.Thereis, however,aneffectof labelon the H2; and,though objectivemeasureof memory,supporting the differenceis non-significant, the self-esteemmeasures differin the samedirection.Resultsof the perceivedeffiof H3; cacy(yes orno) measurejustifyfurtherinvestigation nearlytwiceas manysubjectsin thelabeledmemorycondition perceivedtheirtapesto be effective. If, as it appears,listenerbeliefsgeneratedan expectancy andaretherebyresponsiblefor tapeefficacyor theperception of efficacy,they werenot strongenoughto convince subjectsthatthey hadlost any weightor wereon the road to losingweight.The simplestexplanationis thatthe data werenot unambiguous as to weightloss. Verylittleis left to doubtwhenone stepson a scale.Indeed,manysubjects indicateduponarrivalat posttest(priorto beingweighedby the experimenter) thatthey hadlost no weight.No matter how strongthe expectancies,the objective,inescapable measureof actualweightwas evidentto subjects.Thetiming of the studymay also haveworkedto suppresseffects of the weight loss tape labels; pretestsessions began in early Octoberand posttestsessionswere completednear Thanksgiving.It is arguablethat,in general,in the weeks precedingthe holidayseasonpeopledo not lose, andperhaps tend to gain, weight. However,no effect of SMSH tapesor labelswas foundby usingtheWLOCmeasurethat was includedto detect changesin attitudethat may not havebeenmanifestedas actualweightloss butwouldpredict futureweightloss. Study1 providesconvincingevidenceaboutH1;subliminalcontentplayedno rolein theobjectivelymeasurable or subjectivelyperceivedefficacyof SMSHtapes.The effect of tapelabelon objectivecriteriaandon the perceptionof efficacyis evidenton somebutnot all measures.To replicate the findingsof H1, furtherinvestigateH2, and more fully test H3, we conducteda secondstudy. Study 2 Design PerceivedEfficacy Study2, a simplifiedreplicationof study1, implemented a Perceived efficacy measures were assessed only for memory in study 1. There is no effect of tape content on perceived increase(yes or no) in long-termmemoryability;subjects in both the actual memory and combined actual selfesteem and weight conditions reportedlong-term memory improvementof approximately38% (F<1). There is, however, a cleareffect of the label manipulationon subjects'perceptionsof memorytape efficacy. Fifty-threepercentof subjects in the labeled memorycondition reportedan improve- two actual product(self-esteem improvement,memory enhancement)by two labeled product (self-esteem improvement, memory enhancement) factorial design. Tapes and test equipment/stimuliused in the second study were identical to those used in the first study with the exception that the weight tapes were excluded. Subjects and Procedure Subjects were recruitedby the same procedureas was used This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions 32 Field Test of Subliminal Audiotapes Figure2. DependentMeasuresby Actualand Labeled Conditions(Study2) Materialsand Measures SELF-ESTEEM Memoryandself-esteemtapesandmeasuresusedin thesecond studywereidenticalto thoseof study1. A changefrom study1 was theinclusionof a set of perceivedefficacy(yes or no) questionsaskingwhethersubjectsfelt thatthe tape had increasedthe levels of severaldimensions(see Table 1); thesequestionsprovideda test of H3. 0.7 0.6 S T D Z D to theirtapesan averageof 47 timesoveran averageparticipationperiodof 35 days. 0.5 0.4 0.3 S E Results 0.2 S C A L E 0.1 Resultsof bothself-esteemandmemorytapemanipulations arepresentedin a formsimilarto thatusedfor study1; the same methodof standardization was used. Standardized measureswereanalyzedby meansof an ANCOVA(2 X 2 factorial)procedurewithposttestscoreas the criterionand corresponding pretestscoreas covariate.As for study1, resultsarepresentedfor actualandlabeledconditions. 0 -0.1 -0.2 -0.3 ACTUAL SELF-ESTEEM ACTUAL MEMORY LABELED SELF-ESTEEM LABELED MEMORY Self-esteem self-esteem Figure2 includesresultsfor the standardized scoresby actualandlabeledconditions.As in study 1, a maineffectof timeis evidentforbothself-esteemandmemory.Resultsof an ANCOVAprocedureareagainconsistent withH1; adjustedposttestself-esteemscoresarenot found to be higherwhensubjectsreceiveda self-esteemimprovementtapethanwhentheyreceiveda mislabeledself-esteem result,in fact,is thattheac(truememory)tape.A surprising tualmemoryconditionproduceda significantlyhigherselfesteemscorethanthe actualself-esteemcondition(actual self-esteem,pretest/.00to posttest/.27versusactualmem- MEMORY S T D Z D M E M S C A L E 0.7 0.6 0.5 0.4 0.3 0.2 ory, pretest/.00to posttest/.60,F(1,62)=5.79,p=.019, 0.1 0 -0.1 -0.2 -0.3 ACTUAL MEMORY ACTUAL SELF-ESTEEM PRETEST LABELED MEMORY LABELED SELF-ESTEEM MSE=.28).Inthetestof H2,thedirectionof theresultssugbutthe gestsaneffectof theself-esteemlabelmanipulation, observeddifferenceis not statisticallysignificant(labeled 15 to posttest/.40versuslabeledmemself-esteem,pretest/-. to ory, pretest/.16 posttest/.46, F(1,62)=1.24, p=.27, MSE=.28). POSTTEST MemoryAbility in the first study.Of the 88 subjectspretested,76%comdifferencebeNo demographic pletedposttestprocedures. those and tween subjectscompleting droppingout of the studywas evident,norwas therea differencebetweensubdifferencein experijects in the two studies.Oneimportant betweenthetwo studieswasthatsubjects mentalprocedure in study2 wereassignedrandomlyto the fourconditions, as opposedto beingassigneda tapelabeledwiththeirfirst choiceas in study1. Becausesubjectsin randomlyassigned conditionswouldhave weakerexpectanciesthansubjects whohadchosenthe typeof tapeto whichtheythoughtthey wouldbe listening,randomassignmentprovideda stronger astestof label-initiated expectancyeffects.A double-blind of exclusion the With was used again. signmentprocedure for sevthe left who one subject taperunningcontinuously eralhoursa day (576 sessions),subjectsreportedlistening Findingsfor memoryby labeledandactualconditionsare also shownin Figure2. The resultsagainsupportHI; no maineffect of actualsubliminaltape contentis foundfor Table1. PerceivedEfficacy:PercentMeansof "Yes" Responsesto TapeImprovingRelatedand UnrelatedDimensions(Study2) Actual Labeled Did tapeimprove...? M S M S Memory Self-esteem Concentration Self-confidence stamina Physical Tranquility Relaxation 16 38 32 44 13 66 67 34 30 45 36 15 79 65 41 12 42 18 12 73 61 10 56 35 63 16 72 71 This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions Journalof PublicPolicy& Marketing subjects'memoryability.Adjustedposttestmemorymeasure scores were not betterwhen subjectsreceiveda true tapethanwhentheyreceiveda mismemoryimprovement labeledmemory(trueself-esteem)improvement tape.In a to who listened reverse effect, subjects nearlysignificant trueself-esteemtapesimprovedslightlymoreon the memoryteststhanthosewho listenedto truememorytapes(actualmemory,pretest/-.21to posttest/-.04versusactualselfesteem, pretest/.21to posttest/.48,F(1,62)=2.97,p=.09, MSE=.38).Contraryto the predictionof H2, no effect on memoryas a resultof label is observed(labeledmemory, pretest/.01to posttest/.25versuslabeledself-esteem,pretest/ -.01 to posttest/.20,F(1,62)=.01,p=.935,MSE=.38). Discussion PerceivedEfficacy Conclusions Table 1 gives results of posttest "yes or no" questions dimensions abouttapes'effectson perceivedimprovement by actualand labeledconditions.Some of the secondary measuresarecorrelatedwith the primarydependentmeasures(self-esteemandmemory).Onemightarguethatabilshouldcorrelatewithmemoryability(obity to concentrate served correlationof improvedconcentrationwith improvedmemory:r=-.38,p=.01).Self-esteemandself-confidencearealso highlycorrelated(r=.80,p=.001). Thepatternof the datain the "actual"columnsof Table arenot 1 supportHI: perceptionsof memoryimprovement for actual self-esteem than for actual (16% memory higher of improvedconvs. 34%);the sameis trueforperceptions centration(32%for actualmemoryvs. 45%for actualselfesteem).Likewise,perceptionsof self-esteemimprovement are not higherfor actualself-esteem(30%)thanfor actual memory (38%);nor are perceptionsof self-confidence higherfor actualself-esteem(36%)thanfor actualmemory (44%).In all fourinstances,theeffectivenesswasperceived to be no betterfor the trueproductthanfor the mislabeled product(all differencesn.s.). The datain the "labeled" columnof Table1, support H3.Forty-onepercentof subjectsin thelabeledmemoryconditionreportedan improvementin memorywhereas10% of subjectsin thelabeledself-esteemconditionindicatedincreasedmemoryability(F(1,59)=8.48,p=.005,MSE=.17). on Thoughnot significant,resultsof thelabelmanipulation areconsistentwithmemoryresults;42%of laconcentration beled memorysubjectsperceivedimprovedconcentration versus35%of labeledself-esteemsubjects(F<1,n.s.).Similarly,56%of subjectsin the labeledself-esteemcondition reported a perceived improvement in self-esteem versus 12% of subjects in the labeled memory condition (F(1,59)=18.78, p=.000, MSE=.17), and 63% of subjectsin the labeled self-esteem condition reported improved selfconfidence versus only 18%of subjectsin the labeled memory condition F(1,61)=15.88, p=.000, MSE=.24). Results of the perceived improvement(yes or no) questions therefore supportH3. The remainingscales in Table 1 show no statisticallysignificant differences. As expected, only a small percentage (16% or less) of subjectsreportedan improvementin physical staminaas a result of the SMSH audiotape.In contrast, the soothing audibleportionresultedin high reportsof tranquility and relaxationfor all subjects. 33 Study2 resultsprovidefurtherevidencein favor of H1; SMSHaudiotapeusersdidnotdemonstrate tapeefficacyas a resultof subliminalcontent.Unlikethe firstexperiment, study 2 reveals no effects of labels on objectiveperformancemeasures,andthusno supportforH2. However,the resultsstronglysupportH3. The labeleffect on perceived in study tapeefficacyfoundin the firststudywas apparent 2 forboththe labeledmemoryandlabeledself-esteemconditions.Subjectsconsistentlyreportedperceivedimprovementon dimensionslogicallyrelatedto theirlabels.Furthermore,dimensionsunrelatedto subjects'assignedlabels werenot reportedas areasof perceivedimprovement. The resultsof ourtwo studiessupportthe hypothesisof no effectof the claimedsubliminalmessagesin SMSHaudiounderrealisticlisteningcontapes.Whenusedby volunteers ditionsovera lengthof timeguaranteedby manufacturers to providepositive results, SMSH audiotapesproduced associatedwith subliminalconnoneof the improvements tentin eitherstudy.Therewas,however,someevidenceof change,or theperceptionof change,on thedimensionsthat to treat.Thesechanges,realor illuthe tapesarepurported sory,werecausedby the tapelabels,not the tapecontent, and we believe they resultedfromexpectanciesthatsubjects broughtto the use of the product. Combined,the resultsof studies1 and2 (identicalin instrumentsused andexperimentaladministration) provide evenstrongerevidenceagainstanyeffectof subliminalcontent.' For memorytapes,the combinedresultsare simply nonsignificant(standardizedadjustedposttest memory scores:actualmemory/.22versusactualself-esteem/.34, F(1,155)=1.13,p=.289,MSE=.45),butfor self-esteemthe dataaresignificantin the directionoppositethe claimedeffect (standardized adjustedposttestself-esteemscores:actual self-esteem/.25 versus actual memory/.48, F(1,154)=7.94,p=.005,MSE=.27);we haveno explanation for thisunexpectedresult. Thecombinedresultsshowa nearlysignificanteffectof label on subjects'self-esteem(standardized adjustedposttest self-esteemscores:labeledself-esteem/.44versuslabeled memory/.29,F(1,154)=3.04,p=.083,MSE=.27),but effectof memorylabelon subjects'memoa nonsignificant ries (standardized adjustedposttestmemoryscores:labeled memory/.34 versus labeled self-esteem/.23, F(1,155)=1.0, p=.32, MSE=.45). The strengthof the combined evidence favoring the null hypothesis (H1) requiresconsiderationof the combined statistical power of the test(s) performed.Power of the design to detect an effect size of .35 (between Cohen's [1987] "small" and "medium" effect sizes) was calculated (1tailed, alpha=.05). The two-study combined sample size of 160 provided power of .95 for this effect size for selfesteem and power of .94 for the same effect size for memory. This level of power is higher than that in most studies in which power issues are addressedacross the disciplines of marketing,psychology, education,and speech [Peterson, Albaum, and Beltramini 1985; Sawyer and Ball 1981]. Given the high level of power, the "real world" aspect, and This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions 34 Field Testof SubliminalAudiotapes the strengthof ourdouble-blinddesign,we feel confident confirmingH1, the null hypothesis,for subliminalcontent effects. In both studies,subjectsheld the subjectiveillusionof tapeefficacy.Specifically,of the five tests of H3, fourare statisticallysignificant,showinga bias in the processingof informationresultingin subjectiveappraisalsthatconfirm expectancies.Furtherworkshouldbe doneto identifythe sourceof such self-deception,whetherin encoding(selecevidenceor rejective attentionto expectancy-confirming tion/distortion of disconfirming evidence)or retrievalof information. The supportfor H2 is weak.Of the five testsof H2, two butnotstatistically (forself-esteem)areas hypothesized significant,one (memory,study1) is statisticallysignificant, and the othertwo show no appreciabledifferences.The SMSHaudiotapesmay not be expectanciesaccompanying to true placeboeffects-objective strongenough produce changesin targeteddimensions. It is importantin the contextof ourstudyto distinguish betweenexpectancyeffects and "demand"effects. One mightarguethatwhatwe have labeled"expectancy"effects aremerelydemandeffects-increases(or the perception thereof)in self-esteemor memoryas a resultof subjects' knowingthey werebeingmeasuredon thesedimensions.The distinctionbetweenthe demandeffectsexplanation andthe hypothesisof expectancyeffectsis subtle.Accordingto a demandeffectsexplanation,subjectsshowed improvementbecausethey were being measuredandbeof them. lievedthe researchto demandsomeimprovement Thus, the observedresults could not be ascribedto the SMSHaudiotapelabels alone,nor could they be generalized beyondthe researchsetting.The expectancyhypothesis explainsthe improvementin termsof subjects'beliefs aboutthe SMSHaudiotapes,a relationshipbetweensubjects andproductsthatexistsbothwithinandoutsidetheresearchsetting.Thus,demandis a threatto boththe internal validityof the hypothesizedrelationshipandthe external of the observedeffects. generalizability The studiesincludedseveralchecksandcontrolsfor demand.Becausesubjectswere all volunteers,participation was likely motivatedby a true desire to experiencethe claimed-effective powerof SMSHaudiotapes,not a desire to confirmresearchhypotheses.The studies were conductedin the field over a periodof at least one month, which de-emphasizedthe sense of involvementin a redistinctpretestand searchstudy.Also, thecounterbalanced, posttestversionsof the dependentmeasuresshouldhavereduceddemandeffects.Furthermore, multipleobjectivemeasureswereused,diminishingsubjects'abilityto fakeeffects. To minimizethe unlikelypossibilitythatsubjectsmaypurposefullysuppresstheirmemoryabilityor self-esteematpreatposttest,doubletestto produceanapparent improvement blindassignmentto a specificlabeledconditionwas made afterpretestingin both studies.Finally,the patternof results on the perceivedefficacy scales in Table 1 argues againsta simpledemandeffectsuchas a responseset bias; only the scalesrelatedto tapelabelsshowimprovement. In general,the relationship betweenthe truthof advertising claimsandconsumers'perceptionof benefitsmightbe consideredas a two-by-twomatrix.Whenattributeclaims are trueand benefitsare perceived,thereis obviouslyno claimsarefalseandno benefitsreproblem.Whenattribute sult,the deceptionis obviousandsuch advertisingshould be disallowed.Whenattributeclaimsaretruebutno benefits resultfromtheproduct,we shouldexpectmarketforces to work;such a productshould fail. It is when attribute claimsarefalsebutbenefitsare,nonetheless,perceivedthat properpublic policy responseis unclear.In additionto SMSHtapes,severalproductsfit thiscase-baldness cures, diet plans, and wrinklecreams,for example.Regulatory agenciesareconsideringsomeof thesemarketsas well. In additionto specificregulatoryactivity,we mightapplythe samelogic to the moregeneralpracticeof marketingnearcommodityproductson the strengthof brandimagesthat imply an illusory advantageor superiorityover other brands.We urgemarketers to considerthe ethicsof strateconsumer that on capitalize gies perceptionsthatarebased on nothingmorethanthe expectationsformedby advertising claims.Wealso encourageeducationthatcautionsconsumersagainstdamagingself-deception. Limitationsand Suggestionsfor Future Research A significantstep in progressingfromourresearchwould be to clarifythe role of advertisingin creatingthe illusion of productefficacy. One mightreplicateour studieswith the inclusionof an advertisingmanipulationto determine whetherpromotionincreasesexpectancyeffects. Wealso suggestfurtherresearchwithotherproductcatefor other gories.Expectancyeffectscouldbe demonstrated of that are valued because objectivelyillusoryatproducts tributes.Onemighthypothesizedifferencesin the natureof advertisingfor those products-more image related,less One mightfurtherhypothesizethat functionalinformation. consumerswouldbe moreskepticalof such information; of consumerswho areproneto accepting the identification wouldbe bothinterestaboutillusoryattributes information ing and valuable.Is gullibilitya generalizableconsumer Is therea gulliblesegment? characteristic? studiesprovidestrongevidenceagainst Ourdouble-blind any objectiveeffect of subliminalmessagescontainedin self-helpaudiotapes.However,subjectivebenefitsappear to be associatedwith the use of SMSHtapes;consumers thinktheyareimprovingin relevantareasof self-helpwhen in actualitytheyarenot. Ourcontentionis thatthese "perceived"benefitsaretheresultof consumer-generated expectancies-consumerself-deception. Notes 1. Becauseweightlosswasomittedfromthesecondstudy,subjectswhoreceivedactualweightlosstapesorlabelswereincludedin controlgroupswiththosewhoreceivedmemory reinanalyses of theself-esteem tapesorlabels,asappropriate, sults,andwiththosewho receivedself-esteemtapesor labels, in analysesof thememoryresults(combined as appropriate, n=160). This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions Journalof PublicPolicy & Marketing References 207, 557-558. Allison,RalphI. andKennethP. Uhl (1964), "Influenceof Beer on TastePerception,"Journalof MarketBrandIdentification ing Research, 1 (August), 36-39. Beck, A.T.,C.H.Ward,M. Mendelson,J. Mock,andJ. Erbaugh forMeasuring Archivesof (1961),"AnInventory Depression," General Psychiatry,4 (June), 53-63. Block,MartinP. andBruceG. VandenBergh(1985), "CanYou Sell Subliminal Journalof AdvertisMessagesto Consumers?" ing, 14 (3), 59-62. IdenBruner,JeromeS. andA. LeighMinturn(1955),"Perceptual Journalof GeneralPsytificationandPerceptual Organization," chology,53 (July),21-28. Cheesman,Jim andPhilipM. Merikle(1985), "WordRecognition and Consciousness," in Reading Research: Advances in Theoryand Practice,Vol. 5, D. Besner,T.G.Waller,andG.E. eds. New York:AcademicPressInc.,311-352. MacKinnon, - and- 35 (1986), "DistinguishingConsciousFromUncon- scious PerceptualProcesses," CanadianJournal of Psychology, New Lofflin,John(1988),"HelpFromthe 'HiddenPersuaders."' YorkTimes(March20), F17. Merton,RobertK. (1948), "TheSelf-FulfillingProphecy,"Antioch Review,8 (Summer),193-210. - Moore,TimothyE. (1982), "SubliminalAdvertising:WhatYou See Is WhatYouGet,"Journalof Marketing, 46 (Spring),3847. - ioral Sciences.Hillsdale,NJ:LawrenceErlbaumAssociates, Inc. ConDarley,JohnM. andRussellH. Fazio(1980), "Expectancy firmationProcesses Arising in the Social InteractionSequence," AmericanPsychologist, 35 (October),867-881. of Advertising andEviDeighton,John(1984), "TheInteraction dence," Journal of ConsumerResearch, 11 (December), 763- 770. Shadowson Dunlap,Knight(1900),"TheEffectof Imperceptible the Judgmentof Distance," Psychological Review, 7 (5), 435453. Greenwald,AnthonyG., MarkR. Klinger,andT.J.Liu (1989), "UnconsciousProcessingof DichopticallyMaskedWords," Memoryand Cognition, 17 (January),35-47. Haas,H., H. Fink,andG. Hartfelder (1959), "DasPlaceboProblem," Fortschritteder Arzneimittelforschung,1, 279-454. Helmreich,RobertandJoy Stapp(1974), "ShortFormsof the TexasSocialBehaviorInventory(TSBI),An ObjectiveMeasure of Self-esteem," Bulletin of the Psychonomic Society, 4 (5a), 473-475. Ha (1986), "Consumer LearnHoch,StephenJ. andYoung-Won ing: Advertisingandthe Ambiguityof ProductExperience," Journal of ConsumerResearch, 13 (September),221-233. Husband,RichardW. andJaneGodfrey(1934), "An Experimental Studyof CigaretteIdentification," Journalof AppliedPsychology,18 (2), 220-251. Janiszewski,Chris(1988),"Preconscious ProcessingEffects:The of AttitudeFormation andConsciousThought," Independence Journal of ConsumerResearch, 15 (September),199-209. Jospe, Michael (1978), The Placebo Effect in Healing. Lexington, MA:LexingtonBooks. (1988),"TheCaseAgainstSubliminalManipulation," Psychology and Marketing,5 (4), 297-316. Peterson,RobertA., GeraldS. Albaum,andRichardF. Beltramini of EffectSizes in Consumer Behav(1985), "A Meta-Analysis ior Experiments," Journal of ConsumerResearch, 12 (June), 97-103. Pratkanis,AnthonyR. andAnthonyG. Greenwald(1988), "RecentPerspectives on UnconsciousProcessing:Still No Marketing Applications," Psychology and Marketing,5 (4), 337-353. 40 (December),343-367. Cohen, Jacob (1987), Statistical Power Analysis for the Behav- (1957), Social Theoryand Social Structure.New York:Free Pressof Glencoe. of ColaBevPronko,N.H.andJ.W.Bowles(1949),"Identification erages," Journal of Applied Psychology, 33 (6), 605-608. Rosenberg,Morris(1965), Society and the AdolescentSelf-Image. Princeton,NJ:PrincetonUniversityPress. Rosenthal,RobertandLenoreJacobson(1968),Pygmalionin the Classroom.New York:HoltRinehartandWinston,Inc. Ross, M. andJ.M. Olson (1981), "An Expectancy-Attribution Modelof the Effectsof Placebos,"PsychologicalReview,88, 408-437. Saltzer, Eleanor B. (1982), "The Weight Locus of Control (WLOC)Scale: A Specific Measurefor ObesityResearch," Journal of PersonalityAssessment,46 (6), 620-628. Sawyer,AlanG. andA. DwayneBall (1981), "StatisticalPower andEffectSize in Marketing Research,"Journalof Marketing Research,18 (August),275-290. Snyder,Markand NancyCantor(1979), "TestingHypotheses AboutOtherPeople:TheUse of HistoricalPeople,"Journalof ExperimentalSocial Psychology, 15, 330-342. - andWilliamB. Swann,Jr.(1978), "Hypothesis TestingPro- cesses in Social Interaction,"Journal of Personality and Social Psychology,36 (11), 1202-1212. andRuthAlbrecht.(1946), "An Stone,CalvinP., JohnGardner, AlternateFormof the WechslerMemoryScale," TheJournal of Psychology, 22 (2), 199-206. Sujan,Mita,JamesR. Bettman,andHarishSujan(1986),"Effects of Consumer on Information Expectations Processing SellingEncounters,"Journalof MarketingResearch,23 (November),346353. TheSun (1990), "SubliminalTapes:Help or Fraud?"(January 29) Bremerton, WA,D1. Taylor, Eldon (1988), Subliminal Learning: An Eclectic Ap- proach.SaltLakeCity,UT:JustAnotherRealityPublishing. Kissel, P. and D. Barrucand(1964), Placebos et effet placebo en medecine. Paris:Masson. Kotler,Philip(1980),Principlesof Marketing. EnglewoodCliffs, NJ:Prentice-Hall, Inc. W.R. andRobertB. Zajonc(1980),"AffectiveDisKunst-Wilson, crimination of StimuliThatCannotBe Recognized,"Science, VandenBoogert,Carol(1984), A Studyof Potentials UnlimitedSubliminal Persuasion/Self-HypnosisTapes.GrandRapids,MI: Po- tentialsUnlimited,Inc. Vokey,JohnR. andJ. Don Read(1980), "SubliminalMessages: Betweenthe Devil andthe Media,"AmericanPsychologist40 (November),1231-1239. This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions 36 Field Testof SubliminalAudiotapes David(1945),"A Standardized Wechsler, MemoryScaleforClinical Use," The Journal of Psychology, 19, 87-95. of SuccessfulWeight Weiss,ArnoldR. (1977), "Characteristics BeA BriefReviewof Predictor Reducers: Addictive Variables," haviors,2 (4), 193-201. White,Leonard,BernardTursky,andGaryE. Schwartz(1985), Placebo: Theory,Research, and Mechanisms.New York: GuilfordPress. Zanot,EricJ.,J. DavidPincus,andE. JosephLamp,(1983),"Public Perceptions of SubliminalAdvertising," JournalofAdvertising, 12 (1), 37-45. This content downloaded from 130.15.241.167 on Fri, 18 Dec 2015 13:37:02 UTC All use subject to JSTOR Terms and Conditions