brand extension evaluation

Upload: usama

Post on 08-Apr-2018

235 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/7/2019 Brand extension evaluation

    1/14

    A Process-Tracing Study of Brand Extension EvaluationAuthor(s): David M. Boush and Barbara LokenSource: Journal of Marketing Research, Vol. 28, No. 1 (Feb., 1991), pp. 16-28Published by: American Marketing AssociationStable URL: http://www.jstor.org/stable/3172723

    Accessed: 17/12/2010 07:05

    Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

    http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

    you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

    may use content in the JSTOR archive only for your personal, non-commercial use.

    Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

    http://www.jstor.org/action/showPublisher?publisherCode=ama.

    Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

    page of such transmission.

    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 [email protected].

    American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to

    Journal of Marketing Research.

    http://www.jstor.org/action/showPublisher?publisherCode=amahttp://www.jstor.org/stable/3172723?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=amahttp://www.jstor.org/action/showPublisher?publisherCode=amahttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3172723?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ama
  • 8/7/2019 Brand extension evaluation

    2/14

  • 8/7/2019 Brand extension evaluation

    3/14

    BRANDEXTENSIONEVALUATION 17of brandcategoriesrepresentinga narrowrangeof prod-uct types?To answerthesequestions,we first examinethe structureof brandcategories.Specifically,we dis-cuss howa brandcategorycanbe describedby its breadth(i.e., by thevariationamongits products)andpositthatany currentor potentialbrandedproductcan be de-scribedby how typicalit is of the brandcategory.Nextwe describethe processesby which a brandextensioncanbe evaluated.Hypothesestest thewaybrandbreadthandbrandextensiontypicalityinfluence the evaluationof brandextensions.Our intent is to demonstratehowtheoriesof categorizationcan help marketersto antici-patetheeffects of brandextensions,and thusplanmorecoherentbrandextensionstrategies.

    BACKGROUNDCategory StructureMervisandRosch(1981, p. 89) statethat"acategoryexistswhenevertwo ormoredistinguishableobjectsaretreatedequivalently."However,it is thenonequivalenceof cateogrymembers,or "gradedstructure,"that setscategoriesapartfrom unorderedsets (Rosch, Simpson,andMiller1976;Smith,Shoben,andRips 1974).Gradedstructureanda secondcharacteristicof categorystruc-ture,brandbreadth,are relevantforunderstandinghowjudgmentsof new brandsareformed.Gradedstructure.Peopleperceivemembersof mostnaturallyoccurringcategoriesas varyingin theirdegreeof typicality,or representativeness,of thosecategories.Forexample,a robinis perceivedas moretypicalof thecategory"bird"thanis an ostrich.Furthermore,evennonmembersdifferin howtypicala nonmembertheyareof a category.Forexample,an unrelatedobjectsuch asa chairis abetternonmemberof thecategory"bird"thanis a butterfly.Thisrangein categoryrepresentativenessfromthe mostrepresentativemembersof a categorytothe nonmembersthat areleast similarto thecategoryiscalled "gradedstructure"(see, e.g., Barsalou 1985;Mervisand Rosch 1981)andhasbeendemonstratedfora varietyof consumercategories(LokenandWard1987;Wardand Loken 1986). In the contextof brandexten-sion,gradedstructureimpliesthatsomeproductsaremorerepresentativeof a brandcategorythanareothers,andpreliminarydatasupportthis assumption.2

    Brand breadth. Anotheraspectof brandknowledgestructurethatmayinfluencejudgmentsaboutabrandex-tension is brandbreadth.Brandbreadthrefers to thevariabilityamongproducttypesrepresentedby a brandname.Forexample,thecategory"Heinzproducts"wouldbe extremelynarrowif the only producttypewereket-chup;it wouldbe extremelybroadif it alsoincludedlawnmowersanddishwashers.ThecurrentHeinzproductof-feringthatincludesbabyfood, sauces,soups,and otherprocessedfoods is somewherebetween those two ex-tremesin breadth.Brandbreadthappearsto be a resultof the typicalityof brandextensions.If brandmanagersconsistentlyex-tend the brandby offeringnew productsthat are verymuchlike (i.e., typicalof) currentones, a narrowbrandresults.If brandextensionsareverydifferentfrom(i.e.,atypicalof) currentproducts,a broadbrandresults.Itseemsclear, furthermore,thatas new productsbecomeestablished,people'sbeliefs aboutwhat is "typical"ofthe categorywill be revised.The Effect of Brand Breadth on Perceived Typicality

    Now let us supposethat one of two conditionspre-vails: a brandname is associated(1) exclusivelywithsoups(a narrowbrand)or (2) with soups, condiments,and frozenvegetables(a broadbrand).Considera newkind of soupandcannedvegetablesas two potentialex-tensionsof eitherthenarrowbrandor of the broadbrand.For the narrowsoupbrand,anothersoupis anextremelytypicalmemberof the family-brandcategory.The newsoupprobablysharesnearlyall the salientfeaturesof thecurrentfamily-brandedproducts(e.g., a healthfulfood,a stand-alonemeal, wateryconsistency).The potentialbrandextensioncannedvegetables,in contrast,is not averytypicalmemberbecause it sharesonly a few char-acteristicswithsoups(e.g., it is a healthfulfood butnota stand-alonemeal). Further,soupcannotbe as typicala brandextension for the broad brandas it is for thenarrowbrandbecause it sharesfewer of the character-istics of currentproducts(e.g., it shares most charac-teristicswith soupsbut few with condimentsor frozenvegetables).Cannedvegetables,however,aremoretyp-ical of the broadbrandthan of the narrowsoup brandbecausecannedvegetablessharemanyof the featuresoffrozenvegetables,one of thecurrentfamily-brandmem-bers.As the precedingexampleillustrates,brandbreadthshouldinteractwith brandextensiontypicality;percep-tions of typicality should be more extreme for narrowbrands than for broad brands. Brand extensions that areessentiallythe same as the brand'scurrentproductsshouldbe perceived as more typical if the brand is narrowthanif the brand is broad. However, brand extensions thatare very different from (any of) the brand'scurrentprod-ucts should be perceived as less typical if the brand isnarrow than if the brandis broad.It is importantto note that the preceding propositionswould not hold if categorization of novel objects oc-

    2Toverifythatgroupsof productsrepresentedby brandnamescanhavecategorystructure,30 men and46 women18 to 75 yearsof agewere interviewedby telephonefroma sampleof 190 namesdrawnatrandomfromthe telephonedirectoriesof two largemidwesternmet-ropolitanareas.Respondentswere askedto rate variousproductsasexamples(0 = verypoorexample,10 = verygoodexample)of thebrandsSony, Jello,Kraft,and Gucci. For eachbrand,threeproductswere very good examples(e.g., Sony televisionset, Gucci shoes),threeweremoderatelygoodexamples(e.g., Sonycamera,Guccicos-metics),andthreewerevery poorexamples(e.g., Sonyshoes,Guccitelevisionset). Intersubjectagreementaboutthe relativegoodnessofexamplefor theseproductswas highly significant(p < .000) acrossall four brandsas measuredby Kendall'scoefficientof concordance(Barsalou1983).

  • 8/7/2019 Brand extension evaluation

    4/14

    18 JOURNALOFMARKETINGRESEARCH,FEBRUARY1991curredby matchingthe new instancewith any currentcategory exemplar(e.g., its "nearestneighbor").Re-viewsof recentevidence(cf. Barsalou1990;Malt1989)supporta model in which eitherexemplaror abstractprototypescan be used. Therefore,H1:Typicalityratingsforextensionsof narrowbrandsaremoreextremethantypicalityratingsfor extensionsof broadbrands.Piecemealand CategoricalEvaluationProcessesRelevantto thestudyof brandextensionis thewayinwhichthe affect associatedwith thecategorygeneralizesto the brandextension.FollowingFishbein and Ajzen(1975), we view an attitude(e.g., towarda brandex-tension)as the locationof an objector concepton anaffectivedimension.Attitudestowardbrandextensionscan be formedin at leasttwo ways. In one, an attitudeis "computed"fromspecificattributesof the extension.Such a process has been termed "analytical"(Cohen1982), "piecemeal"(Fiske 1982), or "computational"(Brooks1978).A varietyof computationalmodels(e.g.,Bettman,Capon,andLutz1975)haveprovedusefulbothforpredictingconsumerattitudesand fordiagnosingthebases of attitudes.However,the modelsdo not purportto describeconsciousevaluationprocesses.A differentgroupof models(cf. Brewer1988;Fiske1982;SrullandWyer1989)relies on categorizationpro-cesses to describeattitudeformation.If a new instance(e.g., a brandextension)is identifiedas belongingto apreviouslydefinedcategory(e.g., a brand),the attitudeassociatedwith thatcategorycan be transferredto thenew instance.Fiske, for example, arguesfromdataonpersonperceptionthat anobjectwill receive the attitudeassociatedwith a categoryor schemato the extentthatit is perceivedto fit the category. Similarly,Srull andWyer propose that people attemptto form generalimpressionsof otherpeople and will use such impres-sionsbothto assessnew informationand to make sub-sequentjudgments.Otherresearcherswho describeat-titudes as being associatedwith a categoryratherthangeneratedby inferencesaboutindividualattributeshaveexaminedaffectreferral(Wright1976)andcategory-basedprocessing(Sujan1985).Retrievalof prioraffectandcomputationof affect aretwoprocessesthatshouldnotbe assumedto be mutuallyexclusivein any given affective reaction.Forexample,retrievedaffectcould influencebeliefsabouta product,which in turn could influence the attitudetoward theproduct.Somemodels,twoof whichare discussedhere,explicitly incorporateboth retrievaland computationalprocesses.In one suchmodel,FiskeandPavelchak(1986)proposea two-stepprocessfor evaluation.The firststepinvolvesanattemptto matchthe new objectwith a cur-rentcategory.If categorizationis successful(i.e., if thereis a match),the affect associatedwith thecategorylabelis appliedto the new objectand the evaluationprocessis complete.If thereis a poormatchbetweenthe objectandcategoryknowledge,piecemealprocessesare evoked

    andaffectis computedthrougha weightedcombinationof attributes.Smith, Shoben,and Rips (1974) proposea two-stepmodelforjudgingwhethera conceptorobjectis a mem-ber of a category.In the firststep, which is rapid,onetriesto matchthe featuresof the categorywith the fea-turesof theobject.Whenthereis a clearmatchor a clearmismatch,theprocessis complete.Whensome featuresmatch and some do not match,a second, slowerstageof processingis necessary.Duringthe secondstage, amorecarefulcomparisonof definingfeaturesof thecat-egoryis madeto determinewhethertheobjectis a mem-berof the category.Both the Fiske and Pavelchakmodelandthe Smith,Shoben,and Rips model involve a two-stage process.The first stage is rapidandglobalwhereasthe secondstageis slowerandmore deliberate.However,the for-mermodeldescribesaffectiveresponsesto a newobject;the lattermodel addressescategorymembershipbutnotaffect.Inthe caseof brandextensions,we are interestedprimarilyin the way in whichpeopleevaluatethe newproductaffectively,which in turndependson the newproduct's"fit"to theoriginatingbrand.Oncepeopleareinformedthat a new extensionhas been madeby a par-ticularcompany,they mayevaluatethe new productasdesirableorundesirableon the basisof (1) whethertheylike theoriginatingbrandand(2) whetherthenewprod-uct is representativeof or similarto the brand'scurrentproducts.Thoughnot explicitly addressingthis issue of evalu-ation,the Smith, Shoben,andRipsmodel raises the in-terestingpossibilitythat when an extensionis verydis-crepantfrom the originalcategory,a rapidevaluationmayoccur. The Fiske andPavelchakmodelpredictsthatmismatches,whetherclearor ambiguous,evoke slow,piecemeal processes. Following Smith and his coau-thors,we believe thereis an alternativeto the Fiske andPavelchakmodelwherebyboth clearmatchesandclearmismatchesof theoriginalbrandcategoryareevaluatedmorerapidlythanmoderatemismatches.Thispredictionis also consistentwith the literature(e.g., Mervis andRosch1981),whichshowsthatboth(1) the moresimilaranexemplaris to a prototypeof the category,the morequicklyit will be judgedas a memberof thatcategory,and(2) the more dissimilara nonmemberof a categoryis to a prototype,the morequicklyit will be judgedtobe a nonmemberof (or"poorfit"to) thecategory.Eval-uatinga new brandextensionpresentspeoplewith a cat-egory verification task (e.g., they verify that a new con-diment is an acceptable member of the brand category"Heinz products"). Very atypical extensions are quicklyjudged to be a "poor fit" with the current brand imageand elicit fewer piecemeal processes than do moderatelytypical brandextensions. Very typical brandextensionsalso should elicit only the first stage of evaluation, with-out need for slow, piecemeal processes.In summary, the relationshipbetween brand extensiontypicality and evaluation response time is hypothesized

  • 8/7/2019 Brand extension evaluation

    5/14

    BRANDEXTENSIONEVALUATION 19to be an inverted U. That is, extremely atypical potentialmembers of a brand (e.g., Heinz floor wax) should beevaluatedrapidlybecause a clear mismatchbetween brandand brand extension elicits only the first stage of eval-uation.Moderatelyplausiblebrandextensions(e.g., Heinzinstant coffee) should elicit the second, slower stage ofevaluation because neithercategory membershipnor cat-egory nonmembershipis obvious. The consumer is likelyto judge, in a piecemeal way, whether the attributesthatmake up the abstractrepresentationassociated with thecurrent brand name can be used to develop the potentialextension. Finally, extremely typical brand extensions(e.g., a new Heinz sauce), like extremely atypical brandextensions, should require only the rapid first stage ofevaluation. Because piecemeal processes are indicatedby slow response times and more attribute-focused orinferentialcognitive responses (cf. Sujan 1985), we pre-dict:

    H2:Extremelytypicalbrandextensionsareevaluatedmorerapidlythanmoderatelytypicalbrandextensions.H3:Extremelyatypicalbrandextensions are evaluatedmore rapidlythanmoderatelytypicalbrandexten-sions.H4:Extremelytypicalbrandextensionselicit fewer at-tribute-related(piecemeal)cognitive responsesthando brandextensionsthat are moderatelytypicalofthe brand.H5:Extremelyatypicalbrandextensionselicit fewer at-tribute-related(piecemeal)cognitive responsesthando brandextensionsthatare moderatelytypicalofthe brand.Additionally, because narrow brands are expected topromote extreme perceptions of brand extension typi-cality (HI), it follows that extensions of narrow brandsshould be less likely than extensions of broad brands torequire the second stage of evaluation. Therefore:

    H6:The evaluationof anextensionof a narrowbrandismorerapidthan the evaluationof an extensionof abroadbrand.H7:The evaluationof an extension of a narrowbrandelicits fewerpiecemealcognitiveresponsesthandoesthe evaluationof an extensionof a broadbrand.Effects of Brand Extension Typicality on Evaluation

    So far we have examined the natureof the evaluationprocess rather than the outcomes of categorization pro-cesses. Suppose people have information about productsrepresentinga family brand and then reporttheir attitudetoward a potential brand extension. A spreading acti-vation model of memory (Collins and Loftus 1975) de-picts this information as nodes connected by links thatindicate relative proximity in a semantic network. Sucha model suggests that typical category members (i.e.,typical brand extensions) are linked more closely to theaffect of the category (the brand). The concept of sche-matic fit (Fiske 1982) also suggests that similar objectselicit similar affect. Therefore, a judgment of poor fit islikely to have affective consequences. Unless the orig-

    inal brand has a negative affective label, the more dis-crepantthe brand extension is, the less positively it willbe evaluated as an instance of the original brand. Intu-itively, from a brand managementperspective, this no-tion seems logical. Atypical brand extensions are lesslikely to share the advantages that the brand name hasaccrued, and the resultingevaluationwill be less positivethan the evaluation of a brand extension that is typicalof the family brandcategory(Boush et al. 1987). Hence,

    Ha:Forpositivelyevaluatedbrands,a brandextension'sdegreeof typicalityof the familybrandis relatedpositivelyto subjects'evaluationsof the extension.Thatis, brandextensionsare liked betterif theyaretypicalof the originalfamily-brandcategorythan ifthey are not.Finally, if the breadth of a brand's current productsinfluences the perceived typicality of potentialbrandex-tensions, then, by the same arguments proposed previ-ously, brand breadth should likewise influence attitudetoward the brand extensions (cf. Judd and Lusk 1984;Linville 1982; Parducci and Wedell 1986). Therefore,

    Hg: Subjects'reportedaffect towardan extensionof anarrowbrandis more extremethan that towardanextensionof a broadbrand.METHOD

    The objective of our study was to determine the effectof brand category structure, in particularthe effect ofbrand breadth and brand extension typicality, on theevaluation of potentialbrand extensions. Though the sit-uation described here for brandextensions differs fromthat described in the person perception literature-theobjectsareproductsratherthanpeople-the models shouldnevertheless be relevant to testing hypotheses about theprocesses and outcomes of categorizations of potentialnew extensions.Brandinformationwas manipulatedand subjectseval-uated a series of potential brand extensions that variedsystematically from an established brand concept. Theevaluation process was traced by using response timesand verbal protocols. Attitudes toward the potential ex-tensions were assessed by means of rating scales. Ratherthan using actual brands, for which brandbreadthcouldnot be varied systematically without introducing brandhistory confounds into the study design, we developedfictitiousbrandsthatsystematicallyvaried in brandbreadthand the typicality of brandextensions.DesignThe study was a 2 x 5 mixed design with two levelsof brand breadth (narrow or broad) and five levels ofbrand extension typicality. The latter was a repeatedmeasure. Two replicates of this experiment were con-ducted for each subject, one involving brandsassociatedwith grocery productsand the other involving brandsas-sociated with electronic products. All brand extensionsin the grocery products replicate were extensions of a

  • 8/7/2019 Brand extension evaluation

    6/14

    20 JOURNALOF MARKETINGRESEARCH,FEBRUARY1991branddescribedin a cover storyas "brandB" and allextensionsin the electronicproductsreplicatewere de-scribedas extensionsof a brandcalled "brandG."Independent Variables

    Brand extension typicality. Brand extension typicalitywas manipulatedas a within-subjectsvariable.Subjectsevaluatedpotentialbrandextensionsthatrangedfromverysimilarto the brand'scurrentproductsto verydifferentfromcurrentproducts.Ratingscales have been usedpre-viouslyas anappropriatemeasureof perceivedsimilarity(SujanandDekleva1987;Wardand Loken1986), andtypicalitymeasuresthatemploy similarityscales havebeen used frequently(e.g. Barsalou1985; Rosch andMervis1975;Tversky1977).A pretestwas conductedto selectpotentialbrandex-tensionsthatwouldprovidea rangein similarityto soup,condiments,and frozenvegetables(the "current"brandB products)andto televisionsets, calculators,anddig-ital watches(the "current"brandG products).Pretestsubjectswere also askedto thinkof extensionsthatwereextremely unlikely. On the basis of the gradedmeansimilarityto currentproducts,thepotentialextensionsofbrandB (thegroceryproducts)were selectedto be fro-zen vegetables,chunkystew, steaksauce,breakfastcer-eal, cannedfruit, toothpaste,floor wax, andpet food.The potentialextensionsof brandG (the electronicsproducts)were selectedto be televisionsets, calculators,digitalwatches,cameras,refrigerators,ballpointpens,bicycles, and garbagecans. Potentialbrandextensionsin eachreplicateincludedproductsthatwere "thesame"as those of the currentbrand,productsthat were ex-tremelydifferentfromthoseof thecurrentbrand(almostnonsensical),andproductsthat weremoderatelydiffer-ent from thoseof the currentbrand.Brand breadth.Brand breadthwas manipulatedforfictitiousbrandsas a between-subjectsfactor with twolevels. Brandswere constructedartificiallyto be com-posedeitherof very similarproductsor of a moredi-versegroupsof products.A narrowbrandin thegroceryproductsreplicatewas composedof soupsonly, condi-mentsonly, or frozenvegetables only; a broad brandwas composedof soups,condiments,and frozenvege-tables.A narrowbrandin the electronicproductsrepli-cate was composedof televisionsets only, calculatorsonly,or digitalwatchesonly;abroadonewascomposedof televisionsets, calculators,and digitalwatches.Byincludingthreenarrow-brandconditionsforeach broad-brandcondition, we attemptedto hold the type of prod-uct constant between the two breadth levels and so elim-inate confounds due to type of product. Also note that,for the broadbrands, three different potential extensionsin each replicate could be "the same" as current prod-ucts.Procedure

    The subject sat at a desk that was facing a video cam-era approximatelyseven feet away. The subjectthen went

    throughthe followingprocedurefor each of two repli-cates. Thesubjectwas askedto reada bookletcontainingthreekindsof informationabouta fictitiousbrand(e.g.,brandB) anditscompetitors(e.g., brandsA, C, andD).Thefirstkindof brandinformationconveyedsimilaritiesin size andage of the parentcompanies.This informa-tion was providedas an overviewof the companiesandwas includedto minimizeunwantedinferencesby sub-jects whentheyweretold lateraboutdifferencesin theproductssold undereach brand.The secondkindof in-formationwas intendedto manipulatebrandbreadth.Itincludeda descriptionof eachbrand'sproductmix. Thethirdkindof informationprovidedmorespecificinfor-mationabouteach brand'sproducts.It was designedtoinducea positiveattitudetowardthe targetbrand.Eachbrandwas describedby listing productsalongwith eval-uationsof thoseproductsina ConsumerReportsformat.At the end of the bookletwere instructionsfor the sub-ject to look overall the informationon the brandsuntil"you feel you have formedan impressionof the fourbrands."The rationalefor includinginformationaboutthe competitivebrandswas to providea contextfor thetargetinformation.When the subjecthad formed animpressionof each brand,he or she was instructedtoturnthe page and rateeach brand'sproductson threesemantic differentialattitudescales (describedsubse-quently).Afterevaluatingthe fourbrands,the subjectreturnedthe firstbooklet and receiveda second booklet. He orshe wasaskedto readalongastheexperimenterreadtheinstructionsaloud.Subjectsweretoldthatthe nextpartof the studypertainedto the way they would ratesomenew productsthatmightbe madeby brandB (or brandG) and thattheywouldbe askedto rateseveralpotentialbrandB (orbrandG) productsone at a time.Theywereaskedto thinkout loudwhile evaluatingeach product.Thesubjectthenturnedthe pageand readalouda ques-tion such as, "How wouldyou ratebrandB breakfastcereal?",and completedtwo semanticdifferentialatti-tuderatingscales(describedsubsequently).The subjectevaluatedallof thepotentialbrandextensionsin thisway.(Thefirstwas for practiceonly.) The timerequiredforthesubjectto evaluateeach brandextensionandthever-balizationsmadeduringeach evaluationwere recordedby meansof a video camera(PentaxPV-C55A)equippedwith a stopwatchfunction(RCACGA020).The orderin whichsubjectsevaluatedthebrandextensionswasde-terminedrandomlyandthenreversedfor halfof thesub-jects. The order of the replicates was reversed for halfof the subjects and the design was balanced for combi-nations of order effects between replicates.After subjectshad evaluated brandextensions for bothreplicates,they completedbackgroundquestionsand gavepersonal demographicinformation.Finally, similarityandrecall measures were taken to ensure that brand breadthand brand extension typicality were manipulatedas in-tended. The entire proceduretook from 20 to 30 minutesper subject.

  • 8/7/2019 Brand extension evaluation

    7/14

    BRANDEXTENSIONEVALUATION 21Dependent Variables

    Typicalitywas measuredby askingsubjectsto ratethesimilarityof eachpotentialbrandextensionto "productsthe brandcurrentlymakes"on a 7-pointbipolarscale(1= dissimilar, 7 = similar). (Remember that typicality ofthe brandextensionis includedin the studyas not onlyan independentvariablebut also a dependentvariablebecauseperceivedtypicalityis predictedto dependonwhetherthe brandis narrowor broad.)Attitudestowardthe brandextension(evaluativeout-comes)were measuredon two 7-pointsemanticdiffer-ential scales anchoredby favorable/unfavorableand de-sirable/undesirable(cf. BurkeandEdell1989).Evaluativeprocessesweremeasuredby the lengthof timetakenbya subjectto makean evaluationandby concurrentverbalprotocols.Theresponsetime usedin theseanalyseswasmeasuredfromwhensubjectsfinishedreadingthe ques-tion on the brandextension until they markedthe firstratingscale. This intervalomits both the readingtimeand the time for subjectsto read and considermultiplescales, which can varyacrosssubjects.Subjects

    Subjectswere 144 universitystudents,72 in each ofthe two breadthconditions.The narrowbrandconditionswere balancedfurtherby product.Thatis, in the elec-tronicsreplicate,24 subjectsreactedto brandsrepre-sentedby televisionsets only, 24 to brandsrepresentedby calculatorsonly, and24 to brandsrepresentedby dig-italwatchesonly. In the narrowbrandconditionsof thegroceryproductsreplicate,24 subjectsreactedto brandsrepresentedonly by frozenvegetables,condiments,andsoups, respectively.Subjectseach werepaid$5.00 and,in somecases, givenextracreditforparticipation.Sub-jectswereassignedrandomlyto experimentalconditionswithin each of the two replicates.Thosewho were as-signedto the narrowbrandexperimentalconditionsinthe firstreplicatewereassignedto the broadbrandcon-ditionin the second, and vice versa.RESULTS

    The responseswere analyzed,for the mostpart,in a2 (narrowbrand/broadbrand)x 5 (potentialbrandex-tensionsrangingfrom the same as currentproductstoverydifferentfromcurrentproducts)repeatedmeasuresanalysisof variancedesign.Hypothesizedeffects were testedby performingbothunivariate and multivariateanalyses of variance. The ho-mogeneity of variance assumptionwas not met in any ofthe analyses. That is, the observations were significantlycorrelated within subjects (Greenhouse-Geisser epsilonranged from approximately .80 to .90). Conventionalunivariateanalyses therefore would produce inflatedtypeI errors, implying an overly liberal test of the hypotheses(Geisser and Greenhouse 1958). The preferredsolutionto this problem is to perform multivariate analyses ofvariance on the within-subjects effects, which produce

    accurateestimatesof typeI errorswithoutanassumptionof anyparticularform of the variancesandcovariancesamongtherepeatedmeasures(LaTourandMiniard1983).Manipulation Checks

    Affect toward the brands. The assumptionwas that inorderto transferattitudefroma brand'scurrentproductsto new products,subjectshad to constructan attitudetowardthe brand'scurrentproducts.Providingproductratingsin a ConsumerReportsformat(as describedpre-viously)was the meansby whichwe triedtocreatethesepositiveattitudes.Evaluationson three7-pointbipolaradjectivescales anchoredby desirable/undesirable,fa-vorable/unfavorable,andhigh quality/lowquality,re-spectively,yieldedmeansrangingfrom5.3 to 6.0, allacceptablypositive. There were no significantdiffer-ences betweenbrandson thesemeasures.Brand extensiontypicality. Subjectswere asked to ratethesimilarityof eachpotentialbrandextensionto "prod-uctsthe brandcurrentlymakes"on a 7-pointbipolarscale(1 = dissimilar, 7 = similar). The results indicated thatthe potentialbrand extensionsrepresenteda range insimilarityto the brand'scurrentproductsin bothrepli-cates (see Table 1). However,amongelectronicprod-ucts, theballpointpenandbicyclewereequallydissim-lar to currentproducts.Because the bicycle did notconstitutean additionallevel of the typicalitymanipu-lation,it was droppedfrom the analyses.3Brand breadth.Subjects were asked to name theproductsthey recalledthat brandB and brandG "cur-rentlymake."Forthemanipulationto have worked(andbe recalled)perfectly,all subjectsin the narrowbrandconditionsshould have rememberedonly one kind ofproduct(e.g., frozenvegetablesonly, condimentsonly,or soup only) andall subjectsin the broadbrandcon-ditionsshouldhavenamedthreekindsof products(e.g.,frozenvegetables,condiments,and soup). The manip-ulationworkedwell, if notperfectly.Nearlyall subjects(98%)in the narrowbrandconditionsfor both brandBand brandG recalledone producttype. Most subjects(78%)in the broadbrandconditionsaccuratelyrecalledthreeproducttypes.Brand Extension Typicalityas a Dependent Measure

    H, predictsthattypicalitywouldinteractwith brandbreadth;specifically,that typicality ratingsfor exten-sions of narrowbrandswouldbe moreextremethantyp-icality ratingsfor broadbrands.As predicted,there is asignificant interaction between brand breadth and theproduct within-subjectsfactor in multivariateanalyses ofvariancefor both the grocery productsreplicate (F4,139 =4.27; p < .003) andthe electronicproductsreplicate(F3,138

    3Analyses conducted without the bicycle and without the ballpointpen were substantivelythe same. The bicycle was chosen to be droppedbecause of indications that it producedthe only significant ordereffectin the analyses.

  • 8/7/2019 Brand extension evaluation

    8/14

    22 JOURNALOF MARKETINGRESEARCH,FEBRUARY1991Table 1CELLMEANSFORDEPENDENTMEASURES

    PercentTypicality rating Attituderating Response time piecemeal responsesNarrow Broad Narrow Broad Narrow Broad Narrow Broadbrand brand Overall brand brand Overall brand brand Overall brand brand Overall

    Grocery products"Same" product 6.72 6.32 6.52 12.22 11.44 11.83 8.41 9.66 9.02 6.9 15.3 11.1Canned fruit 4.67 5.07 4.87 9.76 10.53 10.14 8.93 11.14 10.03 18.1 18.1 18.1Breakfast cereal 3.34 3.76 3.56 8.66 8.94 8.80 9.45 12.52 10.98 11.1 11.1 11.1Toothpaste 1.99 1.92 1.95 6.45 6.87 6.66 8.58 10.58 9.57 6.9 9.7 8.2Floor wax 1.38 1.32 1.35 5.59 5.79 5.69 9.50 8.41 8.97 6.9 5.6 6.3

    Electronic products"Same" product 6.83 6.56 6.69 11.66 12.22 11.94 8.04 8.26 8.15 4.2 21.4 12.7Camera 4.69 5.53 5.11 10.00 10.83 10.44 10.65 10.23 10.44 22.2 20.0 21.1Refrigerator 3.08 3.76 3.43 8.14 8.94 8.53 10.25 12.55 11.40 12.5 17.1 14.5Ballpoint pen 2.26 2.11 2.18 7.87 7.66 7.79 11.56 10.27 10.91 9.7 7.1 8.4

    = 9.74; p < .000). In both replicates, results indicatethat brand breadthinfluenced the perceived typicality ofproducts. For potential extensions that were essentiallythe same as current products, the perceived typicalityratings were higher when the brand originally repre-sented a narrowrange of producttypes than when it rep-resented a broad range of product types (mean = 6.72vs. 6.32 in the grocery replicate; 6.83 vs. 6.56 in theelectronics replicate).The opposite patternof results occurred for potentialextensionsthat were moderatelydifferentfromthe brand'scurrentproducts. Canned fruit, for example, was ratedas more typical of the current grocery products brandwhen the brand represented a broad range of productsthan when it representeda narrowrange (mean = 5.07vs. 4.67). Similarly, the camera was rated as more typ-ical of current electronic products when the brand wasbroadthan when it was narrow (mean = 5.53 vs. 4.69).Hence, narrow brands increased the perceived typi-cality of brand extensions that were about the same ascurrentproducts and decreased the perceived typicalityof products that were moderately different from currentproducts, supporting HI. However, as shown in Table1, when the potentialextensionswere very differentfromthe brand's current products, differences between nar-row and broad brandconditions were small.Effects on Response Time

    Several hypotheses were made about the speed withwhich subjects make judgments. Faster judgments areassumed to be associated with categorical processing andslowerjudgments with piecemeal processes. Preliminaryanalyses of response times revealed substantialpositiveskewness, indicating that a transformationto normalizethe data was appropriate (Kirk 1982). Therefore the

    analyses of variancereportedhere employed naturallog-arithmictransformationsof the response times.4Effects of typicalityon response time. H2predictsthatextremely typical brand extensions would be evaluatedmore rapidly than moderately typical brandextensions;H3 predictsthatextremelyatypicalbrandextensions wouldbe evaluated more rapidlythan moderately typical ones.A main effect for the product within-subjects factor issignificant in the electronic products replicate (F3,133 =14.20; p < .000). Results vary in the predicteddirectionbut are not significant in the grocery products replicate(F4,127 = 1.92; p < .111). Brand extensions that were"the same" as currentproductswere ratedmore rapidlythan any of the other brand extensions in the eletronicsreplicate, supportingH2 (see Table 1). Further,the brandextensions that were most atypical were evaluated fasterthanmore moderatelyatypicalextensions, supportingH3.Because typicality also was included in the study as adependentvariable, and hypotheses pertainto perceivedtypicality of extensions, we believe it is also useful toexamine the changes in response times as a function ofthis continuous-level variable. Therefore, for each rep-licate, response times are also shown plotted against de-gree of reportedtypicality in Figure 1, A and B. Again,overall, the predicted curvilinear relationship betweenperceivedtypicalityandspeedof judgment(responsetime)is indicated.

    Effect of brand breadthon response time. H6predictsthat extensionsof narrowbrandswould be evaluatedmore

    4Response times could be recorded fairly precisely by using slow-motion and freeze-frame controls. A subset of 30 randomly chosenresponses were timed twice. The mean difference between the timeswas just under .15 second and the maximum difference was .30 sec-ond. Errorsof this magnitudeshould not significantly affect the anal-yses.

  • 8/7/2019 Brand extension evaluation

    9/14

    BRANDEXTENSIONEVALUATION 23

    Figure 1EFFECTOF BRANDEXTENSIONTYPICALITYAND BRANDBREADTHON RESPONSETIMES,PIECEMEALCOGNITIVERESPONSES,ANDATTITUDERATINGSA. ResponseTim ByTypcaltyRating B. ResponseTmeByTypicaltyRating

    OroceryProducts lectron ProductsResponse e (snds)ReposTime(Reoneim (seconds)12 . . . . .......... .' *............. ... . ..... . .12

    3 41 2 3 4 6 6 1TykFt PyTyp mFtnP

    C. Piecemeal Responses By TypicalityRating D. Piecemeal Responses ByTypicalityRatingroceryProducts ElectronicProductsPRtPiecemeslResponses PctPiecemealResponses20 2518-..18 15..................o.. .......8 ?5

    41iI 0I I I I I I1 2 3 4 5 6 7 1 2 3 4 5 86 7TypicatyPating TypicWnyRang

    E. AttitudeRatingBy TypicalityRating F. AttitudeRatingByTypicalityRatingOroceryProducts ElectronicProductsAttitudeRating AttitudeRating14 13111010

    6I -SI I I 7 I I I I1 2 3 4 5 7 1 2 3 4 5 6 8TypicalityRating TypicalityRating

    Narrow Broad Overalli--] .... 0.. ...... ....

  • 8/7/2019 Brand extension evaluation

    10/14

    24 JOURNALOFMARKETINGRESEARCH,FEBRUARY1991rapidlythanextensionsof broadbrands.Thoughon av-eragepotentialextensionsof narrowbrandswere eval-uatedmorerapidlythanthoseof broadbrands(mean=8.97 vs. 10.46 for the groceryreplicate;mean = 10.12vs. 10.32 for the electronicsreplicate),the main effectof brandbreadthis notsignificantin eitherreplicate(F,1o30= 1.98;p < .161 in the groceryreplicate;F1,135 < 1 inthe electronicproductsreplicate).However,an interac-tion betweenbrandbreadthandproducttypicalityis sig-nificantin multivariatetestsinvolvingthegroceryprod-ucts replicate(F4,127 = 2.41; p < .05) and is close toconventionalsignificancein theelectronicproductsrep-licate(F3,133 = 2.14;p < .098).Becauseof its marginalsignificance,this interactionmustbe interpretedcautiously;however, it can be in-terpretedin two ways. First,as shownin Table 1, ex-tensionsof narrowbrandswere evaluatedmorerapidlythanthoseof broadbrands,butonly when thepotentialextensionwas notverydifferentfromthebrand'scurrentproduct(the one exceptionwas the camerain the caseof electronicproducts).Forexample,as predicted,theproductthatwas "thesame" as one that the brandal-readymadewas evaluatedmorerapidlyfornarrowbrandsthanfor broadbrands(mean= 8.41 vs. 9.66 secondsinthe groceryreplicate;mean= 8.04 vs. 8.26 secondsinthe electronicsreplicate).However,the potentialbrandextensionthatwas most atypicalof the brand'scurrentproductswas evaluatedfasterfor broadbrandsthan fornarrowones (forfloorwax, mean= 8.41 vs. 9.50 sec-onds;for ballpointpen, mean = 10.27 vs. 11.56 sec-onds).A secondinterpretationof thesedatasheds somelighton H2andH3. As can be seen in Figure1, A andB, thepredictedinverted-U-shapedrelationshipbetweentypicalityandresponsetime is more evidentfor broadthanfor narrowbrands.For narrowbrands,not only istheshapeof thecurveflatter,but alsotheresponsetimesfor the mostatypicalextensionincrease.Effects on Cognitive ResponsesProtocolswerecodedintocategoriesfor thoughtsre-latingto specificproductattributesandthoughtsdrawinginferencesaboutthe products.Thoughtsaboutproductattributesor inferencesdrawnaboutthesuitabilityof thepotentialbrandextensionsexpressed"piecemeal"pro-cesses. Subjects'thoughtswerecodedby threejudges.Onejudgecodedall the responsesand two judgeseachcodedresponsesfrom a differentgroupof 20 subjects.Disagreementswereresolvedby the firstauthorandallresponses were coded. Interjudge agreement averaged83%. Because the majority of subjects made no piece-meal responses, cognitive responses were treated as acategorical variable and the proportionof subjects ren-dering piecemeal responses when evaluating a particularproduct was compared across factors.Effects of typicalityon cognitive responses. H4and H5propose respectively that extremely typical and ex-tremely atypical brand extensions would evoke fewerpiecemeal processes than would moderatelytypical brand

    extensions.Goodness-of-fittests confirmthis resultbothforgroceryproducts(x2 = 9.58;p < .05) and forelec-tronicproducts(x2= 14.01;p < .003). The overallper-centage of subjectsrenderingpiecemealresponsesisgreaterfor the two moderatelytypicalbrandextensionsin eachreplicate(18.1%for cannedfruit;21.1%forthecamera)than for eitherthe "same"(11.1%in the gro-cery replicate;12.7%in the electronicsreplicate)or theextremelyatypicalbrandextensions(6.3%forfloorwax;8.4% for ballpintpen). H4 andH5aresupported.Fur-thermore,incontrastto responsetimeresults,whencog-nitiveresponsesareplottedby perceivedtypicality(seeFigure1, C andD), thepredictedinverted-U-shapedre-lationshipappearsto be more evidentfor narrowthanfor broadbrands.

    Effect of brand breadth on cognitive responses. H7predictsthat extensionsof a narrowbrandwould evokefewer piecemealthoughtsthan extensionsof a broadbrand.Thishypothesisis supportedonly for "thesame"productin the electronicproductsreplicate(X2= 8.1;p< .004). Brandextensionsthat were the same as thebrand'scurrentproductselicitedpiecemealprocessesmoreoften forbroad(21.4%)thanfor narrow(4.2%)brands.Resultsfor "thesame"productin the groceryproductsreplicate(15.3%vs. 6.9%for broadand narrowbrands,respectively)varyin the expecteddirectionbut arenotsignificant.A newproductthatis justlike thosethebrandalreadymakesseems morelikely to evoke thoughtful-ness if the brandconsistsof a varietyof productsthanif it consistsof only one kind of product.Thoughthisis an interestingresult,H7 is not supportedoverall.Effects on AttitudeThe correlationsbetweenthe "desirability"and "fa-vorability"scalesrangefrom.81 to .92. Consequently,attitudewas measuredby summingthetwoscales to pro-duce a single 14-pointscale (with a ratingof 8 as theneutralpoint).Effectof product typicalityon attitude. H8predictsthatthe moretypicala potentialextensionis of the brand,the morepositivelyit wouldbe evaluated.The resultssupportthathypothesis.Main effects for producttypi-cality are significantin both the groceryproductsrep-licate (F4,139 = 117.71;p < .000) and the electronicproductsreplicate(F3,140 = 64.98;p < .000). Atypicalextensionswereperceivedverynegativelyin thegroceryproductsreplicateandslightlybelowtheneutralpointinthe electronicproductsreplicate(see Table1). The cor-relation between attitudeand perceived typicality ratingsaverages .50 in the grocery replicate and .48 in the elec-tornicsreplicate.This relationshipbetween typicalityandattituderatings is demonstratedgraphically in Figure 1,E and F.Effect of brand breadth on attitude. H9 predicts thatextensions of narrowbrands would elicit more extremeattitudes than extensions of broad brands. The resultsgenerally support that hypothesis. The main effect ofbreadthis significant in the electronic productsreplicate

  • 8/7/2019 Brand extension evaluation

    11/14

    BRANDEXTENSIONEVALUATION 25(F1,142 = 5.07; p < .026);the interactionbetweenbrandbreadthandproducttypicalityis significantin the gro-ceryproductsreplicate(F4,139= 2.92;p < .023) andisclose to conventionalsignificancefor the electronicproductsreplicate(F3,14 = 2.40;p < .070).Inthegroceryproductsreplicate,aspredicted,therangeof attitudeis greaterfor narrowthanfor broadbrands.Forexample,brandextensionsthatwere the same as thebrand'scurrentproductswereperceivedas more desir-able when the brandwas narrow(mean = 12.22) thanwhenthe brandwas broad(mean= 11.44). Incontrast,brandextensionsthatwere most dissimilarto thebrand'scurrentproductswereperceivedas less desirablewhenthe brandwas narrow(mean= 5.59) than when it wasbroad(mean= 5.79). In bothreplicatesthe attituderat-ingswerehigherformoderatelytypicalbrandextensionswhenthe brandwasbroadthanwhenit wasnarrow.(Forthe moderatelytypicalextensioncannedfruit,the meanis 10.53 for the broad brandand 9.76 for the narrowbrand.Forthe camera,the meanis 10.83 for the broadbrandand 10.00 for the narrowbrand.)Apparently,ex-tensionsthataresomewhatdifferentfromthecurrentof-feringsare consideredmoreacceptableif the companyhas alreadyextendedto differentproducts.Finally,attitudeswereplottedagainstperceivedtyp-icalityratings.However,note thatbecausebrandbreadthhad the same effect on typicalityratingsas on attituderatings,theplotsinFigure1, E andF, disguisethe effectof brandbreadth.

    DISCUSSIONConsistentwith conceptualframeworksdevelopedincognitiveand social psychology,brandextensiontypi-cality and brandbreadthhad significanteffects on theevaluationprocessesandevaluationoutcomesassociatedwith a potentialbrandextension.Thesefindingsand im-plicationsfor categorizationmodelsthatexamine eval-uationprocessesare addressedin the sections that fol-low.

    BrandExtensionTypicalityBothreplicatesof our studyshow a directlinear re-lationshipbetweentypicalityand attituderatingsforpo-tentialbrandextensions. As predicted,attitudetransferfrom the currentbrandto potentialextensionsdid notoccur in anall-or-nothingway. Thenegativeattitudeto-wardextremelyunlikelyoratypicalproductsthata brandmightmakewentbeyondthefailureof apositiveattitudeto "ruboff" on the newproduct.Particularlyin thegro-cery productsreplicate,subjectsseemedto findatypicalbrandextensionsfundamentallywrongfor thebrand.Thisresult is differentfromwhat has been predictedin thepersonperceptionliteraturefor schematicfit. Specifi-cally, Fiske(1982)positedthatwhen a persondoes notfit a particularschema,the affectiveresponseis neutralor slightly positivependingidentificationof the personas a good fit withanotherschema.One possiblereasonfor this differencein predictions

    is that different classifications have different conse-quences.You shouldbe less botheredwhena librariandoesnot fit your"librarian"stereotypethanwhen a doc-tor whois aboutto operateonyoudoes not fityour"sur-geon"stereotype.Similarly,thoughto a lesserdegree,negativeconsequencesareimpliedby atypicalbrandex-tensions.Forexample,an atypicalbrandextensionmaybe regardednegativelyif it is perceivedto requireex-pertiseoutsidethe originatingcompany.Brand Breadth

    Resultsindicatethatbrandbreadthinteractedwithbrandextensiontypicality.Specifically,when the brandmadea varietyof products,an extensionthat was essentiallythe same as a currentproductwas perceivedas not astypicalas when the brandmadeonly one typeof prod-uct. However,greaterbreadthincreasedthe perceivedtypicalityof moderatelydiscrepantextensions.Thepre-cedingdiscussionmay implythat a narrowbrandsuchas Campbell'shas an advantageover a broaderbrandsuchas Heinz in offeringa new soup,but Heinz has anadvantageoverCampbell'sin offeringa moderatelydif-ferentextensionsuch as a newline of frozenvegetables.Finally,brandbreadthhadlittle effect on the perceivedtypicalityof extremelydiscrepantextensions,suggestingthat breadthand typicalityarerelative,ratherthanab-solute, magnitudes.For any particularlevel of brandbreadth,someproductsare so discrepantas to makethebrandcategoryseem relativelynarrow.Therefore,nei-thera narrowbrandsuch as Campbell'snor a broaderbrandsuchas Heinzwouldhave anadvantageinofferingan extremelydiscrepantextensionsuchas toothpaste.A narrowbrandmay not always have an advantagewhenofferingnewproductsthatareessentiallythesameas currentones. Results in the electronicproductsrep-licateindicatethattheextensionsof the broadbrandgen-erallywereperceivedmorefavorablythantheextensionsof thenarrowbrand.Inthereplicatein whichbrandnar-rowness seemed disadvantageous,the brand'scurrentproductswere calculators,television sets, and digitalwatches.Perhapsa brandthatmadeonly digitalwatchesor calculatorswas viewed as overspecialized.Recentfindings(Aakerand Keller 1989) substantiatethe im-portanceof perceivedexpertisein positiveevaluationsof brandextensions.Possiblygroupsof relatedproductsfortifya brandimage,particularlyoneinwhichtechnicalexpertisehas an importantrole.Evaluation Processes

    The evidence reportedhere is consistent with an eval-uation model that includes both categorical and piece-meal processes. As hypothesized, categorical processespredominatedwhen consumers evaluated products thatwere "the same" as the brand's current products andpiecemeal processes predominated when consumersevaluated products that were moderately different fromthe brand's current products. However, there are alsoseveral deviations from the predictedU-shaped relation-

  • 8/7/2019 Brand extension evaluation

    12/14

    26 JOURNALOF MARKETINGRESEARCH,FEBRUARY1991shipbetweenbrandextensiontypicalityand the processmeasures.Responsetimes for extensionsof narrowbrandsde-viate fromthe predictedcurvilinearpatternin that ex-tremelyatypicalextensionsof narrowbrandstooklongerto evaluatethandidmoderatelytypicalextensions,a re-sultconsistentwith the FiskeandPavelchakmodel. Per-hapssubjectswere somewhatconfusedby the prospectof a companyextendingits brandfrom,say, condimentsto toothpaste.Atypicalextensionsof broadbrandsmayhave evoked less confusionbecause the brandat leasthadalreadydemonstratedexpertisein multipleproductlines.Therelationshipbetweentypicalityandresponsetimeis muchflatterfor narrowbrandsin thegroceryproductsreplicatethan in the electronicproductsreplicate.Oneexplanationfor this differenceis the types of productstested.Subjectsmayhaveevaluatedgroceryproductsandelectronicproductsvery differently.Because electronicproductsare more technicallycomplex, the evaluativecriteriafor a brandextensionmaybe based on the per-ceivedexpertiseof themanufacturer.However,the eval-uationof a groceryproductbrandextensionmaybe basedmoreon subjectivecriteriafortheappropriatenessof us-ingthe samebrandon productswithextremelydifferentattributes.Perhapsbrandsthatmaketechnicallycomplexproducts(calculators,television sets) encouragecon-sumers to scan theirmemoriesmorethoroughlyfor away to matchthe brandextensionwith a featureof thebrand'scurrentproducts.Futureresearchmightexplorethis issue further.A notableaspectof the responsetimesandcognitiveresponsesin the electronicproductsreplicateis thattheextremelyatypicalproduct(ballpointpen) took as longto evaluateas the moderatelytypical products(cameraandrefrigerator)but did not elicit piecemealcognitiveresponsesin nearlyas manysubjects.This findingsug-geststhatthe twoprocessindicatorsmayhave measureddifferentaspectsof piecemealprocessing.Theresponsetimes seem to have measuredthe difficultyof findingpointsof similarityor differencebetweenthe brandex-tensionandcurrentproducts,whereasthe cognitivere-sponsesmeasuredthefrequencywithwhichrelevantpointsof similarityordifferencewerefound.Apparently,sub-jects scannedtheirmemoriesfor pointsof similarityordifferencebetweencurrentproductsandthepotentialbrandextension,and verbalizedanythat seemedrelevant.Re-sponsetime is arguablythe best measureof the extentof the evaluative processing because it includes the silentsearchof memory. In this context, verbal protocols maybe more useful to characterize the natureof processingthan to measure its extent.Limitations

    The strategyin our study was to manipulatestructuralvariables that seemed likely to influence evaluation. Aswith all laboratory experiments, interpretationof the re-sults must be temperedby an understandingof how well

    the experimentaltaskrepresentedthe real issues inves-tigated. The primarydifferences between the experi-mentaltaskandanevaluationof a true brandextensionare thatexperimentalevaluations(1) did notoccur in astore as they often would in consumerdecision situa-tions, (2) werebasedon vicariousexperienceratherthanpersonalexperiencewith productsin use, and(3) werebasedon informationpresentedin exclusivelyverbalformratherthan as a reactionto othertactile andvisualcuessuch as wouldoccurin mostproductevaluationsettings.In addition,there is a strongpossibilitythatthe experi-mentaltaskprimedthetypicalitygradientrepresentedbypotentialbrandextensions(it was probablyevident tosubjectsthatpotentialbrandextensionsvariedin simi-larityto currentbrandedproducts).A secondlimitationof thestudyis thatwhenresponsetimes andverbalprotocolsare collectedconcurrentlytheymaybecomeconfounded.Thatis, theact of verbalizingprobablylengthensresponsetime.Thoughtverbalizationmay either accentuateor attenuatethe extentof piece-mealprocessingas measuredby responsetime becauseall verbalresponsestake time and only some of themindicatepiecemealprocessing.Concluding Remarks

    Despitethe limitations,the resultsof ourexperimentreveal some interestingaspectsof the relationshipbe-tween categorystructureand the processof evaluatingpotentialnew membersof thecategory.Thestrengthsofthe researchmethodincludethe experimentalmanipu-lation of brandbreadthandthe comparisonof responsetimesandverbalprotocolmeasuresacrosstworeplicatesof the experiment.The results suggest that categorystructurehelpsto shapeevaluationprocessesandthatat-titudeandtypicalityareclosely linked.To the extentthat the experimentaltaskmirroredas-pects of an actualproductevaluation,the resultsalsoprovideinsightintothe way brandextensionevaluationis influencedby brandextensiontypicalityand brandcategorybreadth.Specifically,brandextensionsappearto be evaluatedmorehighlyto the extentthatthey areperceivedas similarto (ortypicalof) the brand'scurrentproductoffering.Theperceivedtypicalityof a potentialextensionis influenced,in turn,by the breadthof thebrand'scurrentproductoffering. Narrow brandcate-goriesappearto haveadvantagesfor extensionsthat arehighlysimilar,andbroadbrandcategoriesappearto haveadvantagesforextensionsthataremoderatelysimilar,tothe brand's currentproducts. In our study, broad brandcategories included a greatervarietyof productsthandidnarrowbrandcategories. This variety may be specifiedbetteras having both range (operationallydefined as thedifference between the two most different products)andvariability (operationallydefined as a summarymeasureof interproductdifferences, such as pairwise differencessquaredand summed). Futureresearchmight investigaterange and variability as separate factors of brand cate-gory breadth.

  • 8/7/2019 Brand extension evaluation

    13/14

    BRANDEXTENSIONEVALUATION 27It is significant that some potential brand extensionswere perceived as moderately typical ratherthan eitherextremely typical or extremely atypical of the brandcat-egory. These moderately typical extensions appearedtogenerate greater thought and processing time than eitherextremely typical or extremely atypical extensions. Per-haps they would be aided by promotions, and other mar-keting devices, that take into account this greater pro-

    cessing.Our findings suggest that categorization theory pro-vides a promisingframeworkfor furtherresearchon brandextension. By betterunderstandingthe structureof brandcategories and the way differences (or changes) in thatstructureaffect brandattitudes, marketerscan better an-ticipate the effects of brandextension.REFERENCES

    Aaker,DavidA. and KevinLaneKeller(1989), "ConsumerEvaluationsof BrandExtensions," Journal ofMarketing, 54(January),27-41.Barsalou,LawrenceW. (1983), "AdHoc Categories,"Mem-ory and Cognition, 11 (3), 211-27.- (1985), "Ideals,CentralTendency,andFrequencyofInstantiationas Determinantsof GradedStructurein Cate-

    gories," Journal of Experimental Psychology: Learning,Memory and Cognition, 11 (4), 629-48.(1990), "On the Indistinguishabilityof ExemplarMemoryand Abstractionin CategoryRepresentation,"inAdvances in Social Cognition, Vol. 3, T. K. Srull and R.S. Wyer, Jr., eds. Hillsdale,NJ: LawrenceErlbaumAs-sociates.Bettman,JamesR., Noel Capon,and RichardJ. Lutz(1975),"CognitiveAlgebra in MultiattributeAttitudeModels,"Journal of MarketingResearch, 12 (May), 151-64.Boush, David, ShannonShipp, BarbaraLoken,Ezra Genc-turk,SusanCrockett,EllenKennedy,BettyMinshall,Den-nis Misurell,LindaRochford,andJon Strobel(1987), "Af-fect Generalization to Similar and Dissimilar BrandExtensions," Psychology and Marketing, 4 (3), 225-37.Brewer, MarilynnB. (1988), "A Dual Process Model ofImpression Formation," in Advances in Social Cognition,Vol. 1, T. K. SrullandR. S. Wyer,Jr., eds. HillsdaleNJ:LawrenceErlbaumAssociates,1-36.Brooks, Lee (1978), "NonanalyticConceptFormationandMemory for Instances," in Cognition and Categorization,E. Rosch and B. B. Lloyd, eds. Hillsdale,NJ: LawrenceErlbaumAssociates,169-215.Burke,MarianChapmanandJulieA. Edell(1989), "TheIm-pactof Feelingson Ad-BasedAffect andCognition,"Jour-nal of MarketingResearch, 26 (February), 69-83.Cohen,JoelB. (1982), "TheRoleof Affect inCategorization:Towardsa Reconsiderationof theConceptof Attitude,"inAdvances in ConsumerResearch, Vol. 9, Andrew Mitchell,ed. Ann Arbor,MI: Associationfor ConsumerResearch,94-100.Collins,AllanM. andElizabethF. Loftus(1975), "ASpread-ingActivationTheoryof SemanticProcessing,"Psycholog-ical Review, 82 (6), 407-28.Fishbein,MartinandIcekAjzen(1975), Belief,Attitude,In-tention and Behavior. Reading, MA: Addison-Wesley Pub-lishingCompany.Fiske, SusanT. (1982), "Schema-TriggeredAffect:Applica-

    tions to Social Perception," in Affect and Cognition: The17th Annual Carnegie Symposium on Cognition, MargaretS. ClarkandSusanT. Fiske,eds. Hillsdale,NJ:LawrenceErlbaumAssociates,171-90.- andMarkA. Pavelchak(1986), "Category-BasedVer-sus Piecemeal-BasedAffectiveResponses:DevelopmentsinSchema-TriggeredAffect," in The Handbook of Motivationand Cognition: Foundation of Social Behavior, Richard W.SorrentinoandE. Tory Higgins,eds. New York:GuilfordPress, 167-203.Geisser,Seymourand SamuelW. Greenhouse(1958), "AnExtensionof Box's Resultson the Use of F DistributioninMultivariate Analysis," Annals of Mathematical Statistics,29 (April),885-91.Judd,CharlesM. andCynthiaM. Lusk(1984), "KnowledgeStructuresandEvaluativeJudgments:Effects of StructuralVariablesonJudgmentalExtremity,"JournalofPersonalityand Social Psychology, 46 (6), 1193-207.Kirk, Roger E. (1982), Experimental Design: Procedures forthe Behavioral Sciences. Monterey, CA: Brooks/Cole Pub-lishingCompany.LaTour,StevenA. andPaulW. Miniard(1983), "TheMisuseof RepeatedMeasuresAnalysis in MarketingResearch,"Journal of Marketing Research, 20 (February), 45-57.Linville,PatriciaW. (1982), "TheComplexity-ExtremityEf-fect andAge-BasedStereotyping,"Journalof Personalityand Social Psychology, 42 (2), 193-211.Loken,Barbaraand JamesWard(1987), "Measuresof Attri-buteStructureUnderlyingProductTypicality,"inAdvancesin Consumer Research, Vol. 14, Melanie Wallendorf andPaul F. Anderson,eds. Provo, UT: Associationfor Con-sumerResearch,22-8.Malt,BarbaraC. (1989), "An On-LineInvestigationof Pro-totypeandExemplarStrategiesin Classification,"Journalof Experimental Psychology: Learning, Memory and Cog-nition, 15 (4), 539-55.Mervis,CarolynB. andEleanorRosch(1981), "Categoriza-tion of NaturalObjects," Annual Review of Psychology, 32,89-115.Parducci,Allen andDouglasH. Wedell(1986), "TheCate-gory Effect With RatingScales: Number of Categories,Numberof Stimuli,andMethodof Presentation,"Journalof Experimental Psychology: Human Perception and Per-formance,12 (4), 496-516.Rosch, EleanorandCarolynB. Mervis(1975), "FamilyRe-semblances:Studiesin theInternalStructureof Categories,"Cognitive Psychology, 7 (4), 573-605.- , CarolSimpson,and R. Scott Miller(1976), "Struc-tureBasesof TypicalityEffects,"Journalof ExperimentalPsychology:HumanPerceptionand Performance,2 (4), 491-502.Smith,EdwardE., EdwardJ. Shoben,andLanceJ.Rips(1974),"Structureand Processin SemanticMemory:A FeaturalModel for SemanticDecisions,"PsychologicalReview,81

    (3), 214-41.Srull,ThomasK. and RobertS. Wyer(1989), "PersonMem-ory andJudgment,"PsychologicalReview,96 (1), 58-83.Sujan,Mita(1985), "ConsumerKnowledge:Effectson Eval-uationStrategiesMediatingConsumerJudgments,"Journalof ConsumerResearch, 12 (June), 31-46.- and ChristineDekleva(1987), "ProductCategoriza-tion andInference-Making:SomeImplicationsforCompar-ative Advertising," Journal of ConsumerResearch, 14 (De-cember, 372-8.

  • 8/7/2019 Brand extension evaluation

    14/14

    28 JOURNALOFMARKETINGRESEARCH,FEBRUARY1991Tauber,EdwardM. (1988), "BrandLeverage:StrategyforGrowthin a Cost ControlWorld,"Journalof AdvertisingResearch,31 (August/September),26-30.Tversky,Amos (1977), "Featuresof Similarity,"Psycholog-ical Review, 84 (4), 327-52.Ward,Jamesand BarbaraLoken(1986), "TheQuintessentialSnack Food: Measurementof ProductPrototypes,"in Ad-vances in Consumer Research, Vol. 13, Richard Lutz, ed.Provo,UT: Associationfor ConsumerResearch,126-31.

    Wright,PeterL. (1976), "AnAdaptiveConsumer'sView ofAttitudesandChoiceMechanismsas Viewedby anEquallyAdaptive Advertiser," in Attitude Research at Bay, WilliamD. Wells, ed. Chicago:AmericanMarketingAssociation,113-311.

    ReprintNo. JMR281101

    b lfIY1~L~Ipublicat ion a r e n o w a v a i a e r oYes!Iwouldliketo knowmoreaboutUMIArticleClearinghouse.I am interestedin electronicorderingthroughthefollowingsystem(s):D DIALOG/Dialorder D ITTDialcomO OnTyme O OCLCILLSubsystemEIOther(pleasespecify)ElIaminterestedinsendingmyorderbymail.l Pleasesendmeyourcurrentcataloganduserinstructionsforthesystem(s)I checkedabove.

    NameTitleInstitution/CompanyDepartmentAddressCity State ZipPhoneMailto:UniversityMicrofilmsInternational300 NorthZeebRoad,Box91 AnnArbor,MI48106