You are on page 1of 18

Measuring Brand Equity: An Evaluation of a Consumer-Based Brand Equity Scale

Author(s): Judith H. Washburn and Richard E. Plank


Source: Journal of Marketing Theory and Practice, Vol. 10, No. 1 (Winter, 2002), pp. 46-62
Published by: M.E. Sharpe, Inc.
Stable URL: http://www.jstor.org/stable/40470060 .
Accessed: 03/10/2013 16:08

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.

M.E. Sharpe, Inc. is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing
Theory and Practice.

http://www.jstor.org

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
BRANDEQUITY:
MEASURING
ANEVALUATION
OF A
CONSUMER-BASED
BRANDEQUITYSCALE

JudithH. Washburn
BowlingGreenStateUniversity

RichardE. Plank
WesternMichiganUniversity

Thisresearch reportsan independent assessment ofa recently


developedsetofconsumer-based brandequitymeasures.Yoo
andDonthu(1997) developeda multidimensional, consumer-based brandequityscale comprisedoffourtheoretically
defined
constructs
anda separatemultiple-item overallbrandequitymeasure.The presentresearchemployedslightly modifieditems
ina differentcontextin an attemptto examinetherobustness oftheproposedscale. Subjects(n=272) respondedto thebrand
equityscale fordifferentbrandsandcombinations ofbrandsin a co-branding
context.The resultssuggestthat,whiletheYoo
andDonthuscale represents an adequatefirststep,further
scale developmentis needed. Nevertheless,
thisscale development
has broughtus closerto a universallyacceptedmeasureofconsumer-based brandequity.

INTRODUCTION Anotherperspectiveon brand equity emanatesfromthe


viewpointof themarketing organizationand focuseson the
The conceptof brandequityhas been thesubjectof much assetvalueofthebrandinthemarketplace.Yoo andDonthu
recentresearch.Brandequity,whichrefers totheincremental (1997) designeda scale to measurecustomer-based brand
value added to a productby virtueof its brand,has been equityas opposed to the monetaryvalue of the brand.
thoroughly conceptualized (e.g.,Aaker1991andKeller1993), Accordingto Keller (1993), a thoroughunderstanding of
buta universally acceptedbrandequitymeasurehas notbeen customer-based brandequityis essentialforsuccessfulbrand
forthcoming.A numberof different ad hoc measureshave management since "thecontentand structureof memoryfor
beenreported (e.g., Aaker 1991,Parkand Srinivasan1994, the brandwill influencethe effectiveness of futurebrand
and Simonand Sullivan1993),butYoo and Donthu(1997) strategies"
(p.2).
further addressedthemeasurement questionby creatingand
testingthepsychometric propertiesof a set of scales in an The purposeofthisresearchis to investigate
empiricallythe
attempt to measure customer-based brand equity. psychometric propertiesof Yoo and Donthu's (1997, 2002)
customer-basedbrandequityscale withinthecontextofa co-
Keller( 1993) explicatedthetermcustomer-based brandequity brandingstudy. Giventheimportance of brandequityas a
anddefined itas "thedifferential
effect
ofbrandknowledge on usefulmeasure,themannerinwhichwe attempt tomeasureit
consumer responseto themarketing of thebrand"(p.2). As mustbe thoroughly Theoretical
investigated. andoperational
such,onewaytoexaminebrandequityis fromtheperspective definitions thatare both valid and reliableare criticalin
of thecustomerand is based in thecustomer'sknowledge, advancingtheknowledgebase ofacademiciansandproviding
familiarity, and associationswith respectto the brand. guidance for practitioners (Bagozzi 1992). The present

46 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
researchrepresentsan independent administrationoftheYoo based brandequitycontributes of the
to the effectiveness
andDonthu(1997) scaleinan attempt tovalidatetheoriginal mix.
firm'smarketing
authors'efforts.First,we discussthetheoreticalstructureof
customer-based brandequity. We proceedby outlining the Yoo and Donthu(1997) adoptedfourofthefivebrandasset
development ofYoo andDonthu's (1997) brandequityscale. categoriesthat,accordingto Aaker(1991), comprisebrand
We thensummarize threeempirical onthescale (Yoo
reports equity.Aaker( 199 1) described brandloyalty, name(orbrand)
and Donthu1997,2002; Yoo, Donthu,and Lee 2000) that awareness,perceivedquality, and brand associationsas
provideassessment andvalidity
ofscalereliability fora single representing customer perceptions and reactionsto thebrand,
datacollection.We continuebyreporting scale performance dimensionsthatcan readilybe understoodby consumers.
in the presentdata collectionand comparingthe present Aakerfurther describedthefifth brandasset,otherproprietary
findingsto thosepreviouslyreported byYoo and colleagues. brandassets,as consisting ofpatents, trademarks, andchannel
Finally,we drawconclusions abouttheusefulness ofYoo and relationships. Thisdimension is notrelevanttotheconsumer-
Donthu'sscale. oriented, customer-based (or "consumer-based" perYoo and
Donthu1997) brandequitymeasureput forthby Yoo and
We used structural equationmodeling(SEM) by employing Donthu( 1997). Thus,Yoo andDonthu( 1997) focusedonthe
Proc Calis in SAS to conductthisanalysis. We detailour fourAaker(1991) dimensions thatcomprisetheconstruct of
experimental and data collectionprocedures,present our consumer-based brand equity: brand loyalty,brand awareness,
findings andsuggestfurther research.Thisresearch highlights perceivedquality,andbrandassociations.On thisbasis,Yoo
thestrengths and,concurrently, uncoverscertainlimitations of and Donthu(1997): 1) designedfivespecificbrandloyalty
Yoo and Donthu's(1997) scale. As such,itcontributes to a itemstocapturetheattitudinal effectofbeingloyaltoa brand;
betterunderstanding of brand equitymeasurement.The 2) definedbrandawarenessas brandrecognition than
(rather
underlying purposeof thisanalysisis to further evaluatethe brandrecall)(Keller 1993) and used fourbrandrecognition
usefulnessof the scale and to make suggestionsfor its measuresfrompreviousresearch(i.e., Alba and Hutchinson
improvement. 1987,Nedungadiand Hutchinson1985,Rossiterand Percy
1987,andSrull1984);3) measuredperceivedbrandqualityas
MEASURING BRAND EQUITY consumers' subjective judgment of a brand's overall
excellencefollowingZeithaml(1988) by employingseven
Aaker( 1991) definedbrandequityas "... a setofbrandassets itemsfromDodds, Monroe,and Grewal (1991); and 4)
andliabilitieslinkedtoa brand,itsnameandsymbol,thatadd articulated brandassociationsas consistingof six new items
toorsubtract fromthevalueprovidedbya productorservice based on Keller's (1993) workto measureboththequantity
toa firm and/or tothatfirm's customers" (p. 15). Aaker( 199 1) andqualityofinformation processing.Thus,Yoo andDonthu
further explainedthata brand'sassetsandliabilities contribute (1997) employeda total of 22 itemsto capturethe four
to its equity,may differdependingon context,and can be dimensions thatcompriseconsumer-based brandequity.
groupedintofivecategories:brandloyalty, nameawareness,
perceivedquality,brandassociations,and otherproprietary Yoo and Donthu(1997) ultimately developedtwo distinct
assets.Consistent withKeller( 1993),Aaker(1991) suggested brand equity scales - Overall Brand Equity and
that brand equity providesvalue to the firm(e.g., via Multidimensional BrandEquity. The OverallBrandEquity
effectiveness of marketing programs, brand loyalty,price scale (hereafterreferred to as OBE) was reducedusingfactor
premiums, favorable environment for brand extensions, etc.) analysis to a finalset of fouritems(Table 1) froman initial
and value to the customer(e.g., via enhancedinformation pool of 18 OBE indicators. This scale was developed
processing, purchasedecisionconfidence, and increaseduse primarilyto evaluate the convergentvalidity of the
satisfaction).The literature discussesseveralmethodsof Multidimensional BrandEquityscale (hereafter referredtoas
measuringthe financialvalue to the firm(see Bello and MBE) and, according to Yoo and Donthu (1997),
Holbrooke1995fora summary), butlittleattention is devoted accomplishedthiswithconsistently highcorrelations.The
tohowwe measurevaluetothecustomer.The importance of focusofYoo andDonthu's(1997) efforts was developingthe
understanding brandequityfromthecustomer's perspectiveis MBE scale (Table 1), and consistently, MBE is thefocusof
explainedbyKeller(1993): "Thoughtheeventualgoal ofany thisresearch. In a pilot study,Yoo and Donthu(1997)
marketing program is to increasesales,itis firstnecessaryto reducedtheinitial22 itemsto a totalof 17 itemsbased on
establishknowledge structuresforthebrandso thatconsumers examinationof scale reliabilityvia Cronbach's alpha
respondfavorably tomarketing activities
forthebrand"(p. 8). coefficient. Formoredetailon thisprocedure, pleasereferto
Kellerfurther statedthatwhilepositivecustomer-based brand Yoo andDonthu's(1997) publication.
equitycan lead to greaterrevenue,lowercosts and higher
profit,it has directimplicationsfor the firm'sabilityto DEVELOPMENT OF MBE
commandhigherprices,customers'willingnessto seek out
new distribution channels,the effectiveness of marketing Yoo and Donthu's (1997) studyevaluatedsix brandsof
communications, and the success of brandextensionsand athleticshoes,fourbrandsofcamerafilm,andtwobrandsof
licensing opportunities. Inotherwords,thelevelofcustomer- colortelevision setswiththreestudent samples- 633 Koreans,

47 JournalofMarketing
THEORY AND PRACTICE Winter2002 47

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 1 used discriminant
ITEMS FOR MBE AND OBE SCALES
analysis to examine the relationships
betweenOBE andbothattitude towardthebrandandpurchase
MULTIDIMENSIONAL BRAND EQUITY (MBE): intention.
Thisanalysisestablished
convergent
validity
among
Perceivedquality thethreeconstructs
withOBE positively withboth
correlated
Xis ofhighquality*. attitude
towardthebrandand purchaseintention.Further, a
The likelyqualityofX is extremely high.
The likelihoodthatX wouldbe functional is veryhigh. pairwisecomparison,followedby a Chi square analysis,
The likelihoodthatX is reliableis veryhigh*. suggestedthatOBE was distinctfrombothconstructs.
X mustbe ofverygood quality*.
X appearsto be ofverypoorquality*,(r) Yoo andDonthu( 1997) further clarified
thesefindings byalso
on
reporting a furtherreduced,ten-itemMBE scale (i.e.,three
Brand loyalty constructs oftwoitemsforperceivedquality,three
I considermyselfto be loyalto X.
consisting
X wouldbe myfirstchoice.
items for brand loyalty and five items for brand
I willnotbuyotherbrandsifX is availableat thestore. awareness/associations).Testingthismeasureacrossthethree
culturallydistinctsamples,the authorsarguedfor an etic
Brand awareness measure(i.e., one thatis freeof culturalbias). Fits were
I knowwhatX lookslike*. acceptableacrossall threesamples.
I can recognizeX amongothercompeting
brands.
I am awareofX.
Yoo, Donthuand Lee (2000) also reportedon theMBE and
OBE scales confirming the relationshipwithina structural
Brand associations
Some characteristicsofX cometo mymindquickly. equationmodelthatexaminedthe impactof marketing mix
I can quicklyrecallthesymbolor logo ofX. elementson brandequity.In additiontohypothesesregarding
I havedifficulty
in imaginingX in mymind,(r) the effectof variousmarketing mix componentson brand
equity,theauthors and
hypothesized confirmed thatthelevel
OVERALL BRAND EQUITY (OBE) of consumer-based brandequitywas positivelyrelatedto
It makessenseto buyX insteadof anyotherbrand,even iftheyarethe consumers'perceptions of brandquality,brandloyalty,and
same.
Even ifanotherbrandhas samefeatures
brand awareness/associations. Further,the researchers
as X, I wouldprefer
to buyX.
Ifthereis anotherbrandas good as X, I preferto buyX. suggestedthatperceivedqualityandbrandassociationsmight
Ifanother brandis notdifferent
fromX in anyway,itseems affectbrandequityby firstaffecting brandloyalty. This
smarterto purchaseX.
suggestionwas made based on a muchstronger relationship
betweenbrandloyaltyand brandequitythanthatbetween
*'d itemsweredeletedfromthe15-itemto the10-itemmodel. perceived quality and brand equity or between brand
associations/awareness
andbrandequity.
320 KoreanAmericansand 577 Americans.To providea
meansto evaluatethescale's discriminant thestudy
validity, CO-BRANDING
also collecteddataon respondents' attitudetowardthebrand,
purchase intention,product category involvementand This analysisexaminesconsumer-based brandequityin the
experience, andbrandpurchaseexperience.The researchers contextofco-branded products.Co-branding is a strategy
that
analyzedthedataon threelevels: an individualanalysis(to attemptsto capturethe synergism of combiningtwo well-
evaluate commonalitiesacross samples); a multi-group knownand well-likedbrandsinto a third,
unique branded
analysistoexaminefactorial invariance;anda pooledanalysis product(e.g.,Rao andRuekert1994). Somecurrent
examples
to identify culture-freeuniversaldimensionsin the pooled ofco-branding include: Edy's ice creamwithGirlScoutthin
sample.Theindividual analysisresultedindeletingtwoitems mintcookies,LexuswithCoach leatherinterior, andKellogg's
witha totaloffifteenitemsremaining (Table1),whichshowed Poptarts withSmucker'sfruitfilling.A co-branding strategy
excellentreliability.
However,an exploratory factoranalysis is oftenused to quicklygain consumeracceptanceof a new
failedtoproducetheexpectedfourdimensions ofbrandequity productbyintroducing thenewproductwithnotone,buttwo,
bybeingincapableof separating brandawarenessand brand familiar brandnames. In somecases,a companyemploystwo
associations.Further confirmatory factoranalysissupported of itsown brandswhile in othercases the takesthe
strategy
the three-dimension model consistingof brand loyalty, formof a brandingalliancebetweentwodifferent
marketing
perceivedquality,and brandawareness/brand associations. organizations(Rao andRuekert1994).
Later in this paper, we engage in a discussionof the
distinctions
betweenbrandawarenessandbrandassociations. While has becomea popularbranding in
co-branding strategy
recentyears,empiricalstudiesthatexamineits impacthave
Next,Yoo andDonthu( 1997) compared thethree-factorMBE beenfew. Park,Jun,andShocker( 1996) examined
model with the four-item combining
OBE scale. A second order existingbrand names
(constituentbrands) to create a
confirmatory factoranalysisdemonstrated a strongfitbetween CompositeBrand Extension
(CBE), or a co-brand. This
thefour-item OBE scaleandthethree-factor MBE modelwith research focusedona strong leadbrand("header"brand)paired
a highcorrelation(.78) betweenthetwo. Finally,theauthors witha secondaryor "modifier" brand.Resultssuggestedthat

48 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
complementarity betweenthe lead and secondarybrand items)usingthesamescalesas Yoo andDonthu(1997). The
contributed moreto positiveevaluationsoftheCBE thandid questionsused wereidenticalbutwereevaluatedon a seven-
towardthesecondary
a positiveattitude brand.Theirresults point(Very StronglyAgree/Very StronglyDisagree) rather
also suggestedthat a stronglead brand was not much thanthe five-point responsescale (StronglyAgree/Strongly
influenced bythesecondary brand. Disagreeas employedbyYoo and Donthu)becausewe found
through pre-testingthattheseven-point responsescaleshowed
Simoninand Ruth(1995) examinedthe effectof specific more variabilitythan the five-pointscale. We made the
productcombinationsby experimenting with within-or decisionto tradeoffgreatervariability betweenhighand low
between-brand anactivity
bundling, analogoustoco-branding. equitybrands (by addingtwo scale pointsto createa seven
Theseauthors concludedthatjoiningtwowell-likedproducts pointscale) fordirectcomparability to Yoo andDonthu'sfive
via between-brandbundling could contributeto the point scale. While this trade off compromisesdirect
development of favorableattitudestowardthe bundle,and comparisons betweenthetwoscaleadministrations, we believe
towardthenewbrand. Simoninand Ruth(1998) thevalue of greatervariability
indirectly, justifiesthiscompromise.To
also examinedconsumer attitudes
towardbrandalliances(co- minimizetheincompatibility, we maintained thesevenpoint
brands). Their further
researchsuggested consumers' attitudes scalethroughout all aspectsofouradministration andfocusour
towarda particular brandallianceinfluencedtheirsubsequent discussionon comparisons withinourdataversusbetweenour
attitudestowardsthe individualbrandsthatcomprisedthat dataand Yoo and Donthu's(1997) data. Thus,we measured
allianceand,inthereverse, consumer attitudes
towardpartner sevenseparateconstructs: fourMBE constructs, one measure
brandspriortothealliancesignificantlyaffected theirattitudes ofOBE andtwoconstructs measuring attitudetoward thebrand
towardsthealliance. and purchaseintention.These finaltwo constructs actedto
validatethebrandequityscales.
These previousstudiesestablishedimportant effectsof co-
branding onthebrandsinvolved.Nevertheless, noneofthese To carryouttheexperiment describedhere,a listofpotential
studiesaddressedtheeffects ofco-branding on brandequity, productpairswas generated overthecourseof severalweeks
theoverriding purposeofthisdatacollection.Thespecificsof by reviewingadvertisements and free-standing insertsandby
howco-branding affects
brandequityis thesubjectofanother examining products on thegrocerystoreshelves.Productsfor
reportinwhichwe analyzedifferences betweengroupsusing themainstudywereselectedbasedon theresultsofa seriesof
ANOVA as is appropriate forexperimental design. This pretestsand thejudgmentof theresearchers.Characteristics
paper'spurposeis onlyto examinethevalidityof thebrand instrumental inproductselectionincludedtheabilityto easily
equityscale in thecontextof a co-branding study.As such, trytheproductina laboratory (classroom)setting andtheneed
thefocusof thisreportis on thebrandequityscale and its totestfictional, yetrealistic, co-brands.In addition, thetested
properties, not on the effectsof co-branding.We employ productcategoriesneededto be familiar to student subjects.
confirmatory factoranalysisto evaluatethe psychometric
properties the brandequity scale in accordancewith Pretestsubjectsevaluatedthecompatibility
of or congruency of
acceptedpracticeandYoo andDonthu'stechniques. constituentproductcategoriesand generateda listof top-of-
mindbrandnamesforthesecategories.Productpairswiththe
In thepresent we
study, employed Yoo and Donthu's (1997) highestcompatibility scores(potatochipswithbarbecuesauce
brandequityscales to help understand the impactof co- flavoringand papertowelswithan antibacterial ingredient)
branding.We usedboththeMBE and OBE scales to group were ultimately selectedfroma list including: microwave
brandsintohighequityor low equityconditions ina seriesof popcornwithbutter, toasterpastrieswithfruitfilling, pretzels
pretests.Fortheactualexperiment, we usedthesamescales withcheese,cookieswithchocolate,coffeecake withpeanut
to determine theeffecton brandequityof pairingdifferent butter,and facial tissue withcold cream. Anotherpretest
brandstogether.Forexample,we combinedtwohighequity employedthe MBE and OBE scales to measurethe brand
brands,twolow equitybrandsandco-brandsofmixedbrand equity of fourbrand names in each of the fourproduct
equityto measurethe brandequitylevel of boththe co- categories.Thebrandnameswiththehighestandlowestbrand
brandedproductsand the individualbrandsthatcomprise equity scores in each productcategorywere selectedto
them. The methodand rationaleis explainedin the next represent highand low brandequitynames.
section.
A total of 272 subjectswere randomlyassignedto eight
THE STUDY treatment conditionsconsisting of thefollowingbrandequity
combinations foreach of thetwo productcategorypairings:
The researchreported herecomes froma largerstudythat high/high, high/low, low/high, and low/low. Subjectswere
examined theimpactofco-branding onbrandequity handed questionnairesand told to carefullyfollow the
strategies
(Washburn1999). An experiment usingstudents fromthree instructions. The procedures and instructionswere
different US universitiesutilizedYoo and Donthu's(1997) standardizedacross the eight differentadministrations.
MBE andOBE scales. In addition, subjectsprovidedattitude Subjectswereexposedto a fullcolorpackageprototype ofthe
towardthe brand(five items)and purchaseintention (two co-brandintherespective treatment conditions andthenasked

49 JournalofMarketing
THEORY AND PRACTICE Winter2002 49

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
torespondtothebrandequityquestionsforthatco-brandand memory-clearing taskswereemployed.Theseremediesdo not
foreachoftheconstituent brands.The initialquestionswere alleviatetheconcernsdiscussed,whicharevalidlimitations to
answeredafterexposureto the packagebutbeforeproduct thisstudyand shouldbe keptin mindwhenevaluatingthe
trial.Subjectsthenengagedina seriesofmindclearingtasks, results.
triedtheproductand,following producttrial,againanswered
thebrandequityquestionsfortheco-brandandforeachofthe ANALYSIS
constituent brands.As a result,each subjectprovidedbrand
equityevaluations byresponding to26 questions(i.e., 15 MBE The treatment groupsweredividedin multipleways forthe
items,fourOBE items,fiveattitude towardthebranditems, purposeof thisanalysis. Recall thateach individualsubject
andtwopurchaseintention items)fortheco-brand andeachof (n=272) completedthesamequestionsfordifferent brandsor
thetwoconstituent brandsbothbeforeandafterproduct trial. combinations ofbrandsa totalofsixtimes.Thus,theeffective
In effect,each subjectprovidedsix brandequityevaluations. samplesize is 1632 (272 x 6). Sample#1(n=139x6or 834)
Whilecontributed duringthesame session,theevaluations consistsofthepotatochip/barbecue sauceco-brand evaluations
wereseparatedbytheproducttrialand bythemind-clearing (a combination a
closelyresembling productcurrently in the
tasksrequiring subjects to consider andprovide information on marketplace) in all treatment conditions. #2
Sample (n=133x6
unrelatedissues. On average,each subjecttook about25 or 798) consistsof the paper towel/disinfectant co-brand
minutesto completethetask. Therewas no indicationof evaluations(a fictitious product)in all treatment conditions.
respondent fatigue,boredomor otherconditions thatwould Sample#3(n=67x6 or 402) consistsof onlythehighbrand
leadtopoorscaleresponses.In addition, thestudyemployed equityconditionsin bothproductclasses. Low brandequity
control groupsthatcompleted brandequityevaluations forone conditionswere evaluatedin Sample #4(n=67x6 or 402).
ofeightbrands(i.e.,twobrandseachofpotatochips,barbecue Sample#5(n=137x6 or 828) consistsof mixedbrandequity
sauce,papertowels,or disinfectants). These subjectswere situations whereone highequitybrandis combinedwithone
randomly assigned to a control group and completedonly a low equitybrand. Finally,Sample #6examinedall brand
brandequityevaluation.Theydidnottrytheproductnordid combinations in total(n=272x6 or 1632). Table 2 showsa
in
theyparticipate any other activities. samplekey. The differing samplesizes aresimplyan artifact
of therandomizedtreatment groups. Further, recallthatthe
Three possible concerns could arise with the method studywas an experiment thatsoughttodetermine theimpactof
employed:theuse ofstudent subjects,theuse ofa laboratory different treatment conditions onbrandequity.Forpurposesof
setting, andtheuse ofa within-subjects orrepeatedmeasures thispaper,dividingsamplesas aboveprovidesinsight intothe
design.To addressthefirst twoconcerns, Calder,Phillips, and possibleeffectsof thetreatment conditionson theresponse
Tybout(1981) providedjustification fortheuse of student patterns and,thus,thereliability andvalidityofthemeasures.
subjectsandlaboratory settings.Theseauthorsdifferentiated
between research conducted for effectsversus theory TABLE 2
In effects SAMPLE KEY
application. applicationresearch, "the specific
effects obtainedareexpectedtomirror findingsthatwouldbe Sample#1 All potatochip/barbecue sauce co-brands(n=834)
observedif data were collectedforotherpopulationsand Sample#2 All papertowel/disinfectant co-brands(n=798)
settingsin the real world" (p.197). The concern over Sample#3 High equity co-brands in both productclasses
generalizability relatesto thetheory, notthespecificeffects (n=402)
obtained. Researchdesign should be mandatedby the Sample#4 Low equity co-brands in both productclasses
(n=402)
researcher's prioritiesregarding whether "toapplythespecific Sample#5 Mixed equity co-brandsin both productclasses
effectsobservedor to applya moregeneralizedtheoretical (n=828)
understanding" (p. 197). The presentresearchis designedto Sample#6 ' Combination of all co-brands(n=1632)
be a theory application As
study. such,Calder,Phillips, and
Tybout(1981) recommended thatresearch procedures should The purposeof the followinganalysisis to providefurther
includesubjectswho are homogeneouson non-theoreticalevidenceas tothefactorstructure andthevalidity ofthebrand
variables(e.g., students)and a researchsettingthatcan be equity scales as discussed. Recall that Yoo and Donthu (1997)
isolatedfromextraneous variables(e.g.,a laboratory). These originallydeveloped a four-factor MBE scale that was
procedures allowthestrongest testsofcausalityandminimize subsequently reducedto a three-factor scale. Thisanalysisis
threatsto internalvalidity(Cook and Campbell 1975). structured to testboththefour-factor andthree-factor models,
Nevertheless, student constitutea convenience inaddition to the four-item OBE scale, on the present data set.
subjects
our to
sample,limiting ability generalize these to
findings any
otherpopulation. The final concernregardingrepeated We examinedeach ofthesix samplesin severalways. Based
measuresdesignis justifiedby Martin(1996). A within- on theoreticallydeveloped measures, we constructed
subjectdesignis appropriate to addressthe issues herein confirmatory factormodels(CFA) examiningboththeOBE
because the differences between individualsubjects are and MBE measures. We generallyfollowed Yoo and
minimized, further assuringa homogeneous sample. In order colleagues' methodology (1997, 2002) toprovideevidenceof
tominimize theriskofsubjectcontamination construct reliability andvalidity. Specifically,we first
through learning,

50 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 3
CONFIRMATORYFACTORANALYSISOVERALL
BRANDEQUITY (OBE) SCALE

SampleNumber 12 3 4 5 | 6
SampleSize = "
834 798 1Õ2 402 828 1632
AGFI .966 '"938 .784 .986 Ì985 Wl
~CFI ^96 ^992 ^973 TÕÕ ^99 "996
~"
NNFI ^989 1)75 JÏ& ^999 ^996 ^989
Prob>chi**2 .003 .000 jÕÕÕ 325 ^68 !ÕÕÕ
CHI SQ/D.F. 5.74 9.64 1845 ÌA2 2~68 9^89
Composite
Reliability .924 .800 .927 ^912 ^912 ^918
Variance
Extracted 754 7ÏÔ 76Õ 722 722 736
Validity
Convergent YES YES YES YES YES YES
Std.Residuals
>2.58 4 4 4 Õ Õ 4
<5% oftotalresiduals
(.6) NO ÑO ÑO YES YES ÑO

TABLE 4
CONFIRMATORYFACTORANALYSIS15-ITEM,FOUR-FACTORMULTIDIMENSIONALBRANDEQUITY (MBE) SCALE
~
SampleNumber 1 |2 | 3 |4 [~5 [~6
SampleSize 834 ~798 402 402 828 1632
AGFI ~9Ï0 ^ÏÔ ^883 Ì836 ^26 ^24
CFI .971 .965 .962 ^32 ^976 ^71
NNFI .963 ~956 ^52 ^Î5 ^70 !%4
Prob> chi**2 .000 ^ÕÕ [ÕÕÕ jÕÕÕ !ÕÕÕ !ÕÕÕ
CHI SQ/D.F. "OÍ 4JS5 ÍT5 423 ÍÕÕ Î96
CompositeReliability
(1) ^93 1*34 ^875 ^66 ^867 ^71
(2) .899 .878 ~89Î ^64 ^889 ^894
_(3) .940 ~936 ^944 Í9ÍÕ ^944 Í943
J4) .875 .818 .820 .801 ~877 ^854
VarianceExtracted
(1) .735 .627 .702 .688 ^86 !693
(2) .749 ~722 .732 "679 .727 738
J3) 725 719 739 .628 737 740
J4) 700 ~6ÕI J6Õ7 ^588 7Î4 ^60
Convergem Validity YES YES YES YES YES YES
> 2.58
Std.Residuals 29 3Ì 27 26 31 42
<5% oftotalresiduals
(10) ÑO ÑO ÑO ÑO ÑO ÑO

(1) Latent
factor - items
isBrandLoyalty 5,6,7;(2) Latent
factor
is BrandAwareness- items
8,9,10;
(3) Latent
factor
isPerceived
Quality - items11,12,13,14,15,16;
(4) Latentfactor - items17,18,19
isBrandAssociations

51 JournalofMarketing
THEORY AND PRACTICE Winter2002 51

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 5
CONFIRMATORYFACTORANALYSIS15 ITEM,THREE-FACTORMULTIDIMENSIONALBRANDEQUITY (MBE) SCALE

SampleNumber 12 3 4 5 6
SampleSize 834 798 402 402 828 1632
"ÃGH !%7 ^09 M9 Ì835 ' Ì9Ì9 ^07
~CF' <969 !%2 .954 .926 .972 .968
NNFI !%2 ^954 ^44 .912 .967 .962
Prob> chi**2 MO MO MO MO MO ' MO
CHI SQ/D.F. 4~99 Ï83 ' Í53 " 4.33 " 4.32 ~" 5.00 ~~
.892 .856 .876 .874 .867
Composite (1)
Reliability ' .892
"(2) Ì934 ^915 ^U .905 .929 .936
~(3) ^52 ^939 Ì943 mÕ Í944 M'
"VarianceExtracted
(1) .735 .695 .703 .709 ~686 ~tJ5
"(2) ^03 ^45 .630 .615 .686 .704
"(3) 728 J2Õ ?735 ^28 736 728
ConvergentValidity YES YES YES YES YES YES
> 2.58
Std.Residuals 27 33 26 29 32 ~Ü
~<5% oftotalresiduals
(10) NO NO NO NO "ÑO" "ÑO

(1) Latent
factor - items5,6,7;(2) Latent
is BrandLoyalty is BrandAwareness/Brand
factor - items8,9,10,17,18,19;
associations (3) Latent
factor
is Perceived
Quality-items11,12,13,14,15,16

TABLE 6
CONFIRMATORYFACTORANALYSISTEN-ITEM,THREE-FACTORMULTIDIMENSIONALBRANDEQUITY (MBE) SCALE

"sãmpleNumber 11 I2 I3 I4 I5 I6
SampleSize 834 798 402 402 828 1632
AGFI ^953 J939 ^08 ^24 ^46 ^57
"CFI ^85 !973 ^967 ^973 ^980 ^983
NNFI ^79 ^62 ^53 .963 ^72 ^75
Prob> chi**2 ~ÕÕÕ MO MO MO MO ^ÕÕ
CHI SQ/D.F. 3Ü3 Í67 3Ì46 Ï84 436 678
Composite (1)
Reliability ^93 Ü34 ^78 ^63 ^67 Ü71
(2) "!9Í8 ^98 ^888 M6 ^U ^ÏÔ
"(3) ^847 795 Ml 780 ^64 ^47
VarianceExtracted
(1) 735 ~627 703 Ifj ^86 ^93
(2) "!693 ~626 ^14 ^ÏÔ~ Hlï ^68
(3) ~735 74Õ ?735 .639 760 735
ConvergentValidity YES YES YES YES YES YES
> 2.58
Std.Residuals 10 13 15 8 14 16
<5% oftotalresiduals
(4.5) NO "ÑO "ÑO "~NO "ÑO ÑO

1 Latent
factor - items
is BrandLoyalty factor
5,6,7;(2)Latent is BrandAwareness/Brand - items
Associations (3) Latent
9,10,17,18,19; factor
is Perceived
- items12,13
Quality

52 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 7
INTENTIONSCALE
CONFIRMATORYFACTORANALYSISTWO-FACTORATTITUDE/PURCHASE

SampleNumber 12 3 4 5 6
"SampleSize 834 798 402 402 ~828 "Î632
AGFI .939 "T955 ^47 .919 .940 Js5
~CFI ^83 ^84 ^987 ^971 ^80 ^984
NNFI ^972 ^74 ^978 ^53 !%8 ^74
Prob> chi**2 MO !ÕÕÕ MO MÕ MÕ MÒ
CHI SQ/D.F. 6^84 482 Ï95 Ï35 IM §35
CompositeReliability
(1) ~9Ï2 ~865 "~881 ~888 ~899 ]$95
"(2) ^20 ^72 ^76 78Õ ^57 ^839
VarianceExtracted
(1) ^698 T73 763 Ü39 /75Ì 724
(2) ~681 .569 ^07 ITJ M6 ^635
ConvergentValidity YES YES YES YES YES YES
> 2.58
Std.Residuals 7 6 ¡ 4 5 9
<5% oftotalresiduals
(2.1) "ÑO "ÑO "^ES ~NO "NO "ÑO

factor
(l)Latent isAttitude
TowardBrand- items
22,23,24,25,26;
factor
(2)Latent isPurchase - items
Intention 20,21

TABLE 8
CORRELATIONOF BRANDEQUITY MODELS WITH ATTITUDETOWARDBRANDAND
PURCHASEINTENTION(ALL CORRELATIONSSIGNIFICANTAT P<.00n
"SampleNumber 11 | 2 I3 I4 I5 I6
A. Overall
BrandEquity(OBE)
~
Attitude
toBrand ~645 I .480 I .560 I .521 I .582 I .571
Purchase
Intention .717 1 .634 1 .692 1 .607 1 .715 I .686
B.15-item,
Four-Factor
MBE
~~
Attitude
toBrand ~756 I .690 I .737 I .670 I .752 1 .734
Purchase
Intention .707 1 .587 | .607 I .641 | .685 I .653
C.BrandLoyaltyLatentFactor15-
item, MBE
Four-Factor
Attitude
toBrand .665 I .471 I .579 I .551 1 .588 I .589
Purchase
Intention .729 I .681 I .745 I .674 I .709 I .712
D. BrandAwarenessLatent
Factor15-
item,Four-Factor
MBE
Attitude
toBrand .543 I .394 I .466 I .403 I .506 I .489
Purchase
Intention .483 I .264 | .258 1 .345 I .438 I .389
E. Perceived Latent
Quality Factor15-
item,Four-Factor
MBE
Attitude
toBrand .781 I .768 I .770 I .728 I .789 i .776
Purchase
Intention .700 | .599 | .577 | .660 1 .676 1 .652
F.BrandAssociations
Latent
Factor
15-item,
Four-Factor
MBE
Attitude
toBrand .578 I .455 I .455 I .446 I .574 I .538
Purchase
Intention .514 1 .325 I .287 I .402 | .466 I .424
G.Ten-Item,Three-Factor
MBE
Attitude
toBrand ~22 I .624 I .695 I .601 I .712 I .690
Purchase
Intention .693 1 .569 I .618 I .595 1 .668 I .636
H.Brand LoyaltyLatent
FactorTen-
Item, Three-Factor
MBE ^_
Attitude
toBrand .665 I .471 I .579 I .552 1 .588 1 .589
Purchase
Intention .729 I .681 I .745 I .674 1 .709 1 .712
I. Perceived Latent
Quality Factor
Ten-
Item,Three-Factor
MBE
Attitude
toBrand .736 I .701 I .725 I .630 1 .750 I .724
Purchase
Intention .672 | .555 1 .556 1 .582 | .645 | .616
J.Brand Awareness/Brand
Associations LatentFactorTen-
Item, Three-Factor
MBE
~" AttitudetoBrand .584 I .450 1 .480 I .452 "| .570 | .541
Purchase
Intention 1 .520 1 .318 I .299 I .396 .474 .428

53 JournalofMarketing
THEORY AND PRACTICE Winter2002 53

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 9
BEST FIT MODELS
A. BestFitFour-Factor
MBE

"Sample 11 I2 I3 I4 IS I6
SampleSize 834 798 402 402 828 1632
AGFI £79 ,955 ,945 £47 £73 .962
CFI £96 £87 £88 £86 £94 £91
NNFI £94 £81 £82 £78 £91 .984
Prob> chi»*2 .013 .OOO .000 £01 £00 .000
CHI SQ/D.F. L80 3£5 2£0 2JX) 2,06 321
CompositeReliability(1) .853 796 .834 .862 .837 .839
(2) .852 .886 .891 .819 .819 .832
(3) .836 £21 £14 .827 £13 .895
(4) .829 .800 .830 732 .838 .817
VarianceExtracted(1) 746 .668 742 .675 12a 725
(2) 742 722 732 .694 .695 713
(3) .630 746 728 .618 727 742
(4) 708 .668 709 .582 721 .690
ConvergentValidity YES YES YES YES YES YES
> 2.58
Std.Residuals 2 4 4 2 1 5
<5% oftotalresiduals I YES 1 NO I NO | YES I YES I NO
(1) Latent
factorisBrandLoyalty - items
5,6,7;(2) Latent
factor is BrandAwareness - items8,9,10;
3. Latentfactor
is Perceived - items11,12,13,14,15,16; (4) Latent
Quality factor - items17,18,19
is BrandAssociations

B. Indicators
LoadingintoFour-Factor
BestFitMBE

Sample 11 I2 I3 I4 |5 |6
BrandLoyalty 6J 5J 5J_ 5,6,7 5J) 5J_
BrandAwareness 8^10 8,9,10 8,9,10 8110 8J0 9J0
Perceived
Quality 13,15,16 11,12,14,16 11,12,13,16 12,14,16 11,12,13,16 11,12,16
"BrandAssociations I 17,18 I 18,19 I 18,19 1 17,19 1 18,19 1 17,19
C. BestFitTen-Item, MBE
Three-Factor

Sample 11 I2 I3
|4 |5 |6
SampleSize 834 798 402 402 828 1632
AGFI £68 £80 £64 £60 £72 .983
CFI £92 £95 £94 £90 £93 .995
NNFI £88 £91 £88 £81 £87 .991
Prob>chi»»2 £00 £20 £51 £26 J001 .000
CHI SQ/D.F. 2£8 2£2 .L80 L97 2£4 3.45
CompositeReliability
(1) .839 791 .836 .802 .836 .821
(2) .893 .842 .843 791 .869 .853
(3) .847 .850 .847 780 .864 .848
Variance
Extracted
(1) 725 .667 721 ^71 72£ .699
J2à all álO ^35 J63 ^90 .659
_Í3) 735 740 735 ^39 761 736
Convergent
Validity YES YES YES YES YES YES
> 2.58
Std.Residuals 1 0 1 3 2 3
"<5% oftotalresiduals 1 YES I YES I YES 1 NO 1 NO I NO
D. Indicators
LoadingintoTen-Item,
Three-Factor
BestFitMBE

Sample 1 I2 I3 I4 I5 I6
BrandLoyalty 5J 6¿ 5¿6 SJ_ 6¿ 5,6
BrandAware-ness/Associations
10,17,18,19 10,18,19 10,18,19 9,10, 19 9,10, 18 10,18,19
Perceived
Quality | 12,13 I 12,13 1 12,13 I 12,13 I 12,13 1 12,13

54 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
Table9 (con't)

E. Overall
BestFit15-Item, MBE
Three-factor
"Sample 11 I2 I3 I4 I5 I6
SampleSize 834 798 402 402 828 1632
AGFI _36 __8 _58 .957 ,983 .989
CFI _99 ,999 _92 _88 _97 .998
NNFI __8 __8 _88 __0 __5 __6
Prob> chi**2 _87 __1 £16 J010 £53 .010
CHI SQ/D.F. L29 _23 _82 L94 177 2.27
CompositeReliability(1) _66 775 _83 ,862 _37 _4]
(2) .891 J38 _352 _04 _42 .842
(3) _51 _21 _86 784 _36 _95
VarianceExtracted(1) 798 __3 717 £77 720 727
(2) 731 _34 _50 _583 £17 725
(3) £75 J05 705 _45 721 741
ConvergentValidity YES YES YES YES YES YES
> 2.58
Std.Residuals 0 0 0 0 0 0
<5% oftotalresiduals I YES I YES I YES 1 YES I YES I YES

F. Indicators
' LoadingintoOverallBestFit15-Item, MBE
Three-Factor
Sample 11 I2 I3 I4 I5 I6
BrandLoyalty _7 5_> 5_> 5,6,7 5_ 5_
' BrandAware-ness/Associations 9,17,18 9,10,19 9,10,19 8,10,19 8,10,18 8,10
Perceived
Quality I 11,13,16 I 11,12 I 11,12,13 1 11,16 I 11,13 I 11,12,16

TABLE 10
CORRELATIONOF BEST FIT MBE MODELS WITH ATTITUDETOWARDBRANDANDPURCHASEINTENTION(ALL
' CORRELATIONSSIGNIFICANTAT P<.00n
SampleNumber 11 |2 13 I4 • I5 I6
Brand
A. Four-Factor Loyalty
toBrand
Attitude .631 I .415 I .518 I .552 I .621 I .536
PurchaseIntention 706 I .601 I .682 I .674 | 735 1 .655
B. Four-Factor
BrandAwareness
Attitude
toBrand .518 I .394 I .446 I .388 I .475 I .487
PurchaseIntention .452 I .264 I .258 I .323 I .404 | .386
C. Four-Factor
Perceived
Quality
Attitude
toBrand 764 I 762 I 757 I 707 I 784 I 765
PurchaseIntention .681 I .581 I .563 I .633 I .665 1 .634
D. Four-Factor Brand
Associations
Attitude
toBrand .574 I .415 I .398 I .409 I .546 I .523
PurchaseIntention .521 I .267 1 .212 I .360 I .420 1 .410
E. Three-Factor
Ten-ItemBrand
Loyalty
Attitude
toBrand .621 I .441 I .587 I .497 I .548 I .616
PurchaseIntention .683 1 .674 I 734 I .625 1 .677 I .723
F. Three-Factor
Ten-ItemBrand
Awareness/BrandAssociations
Attitude
toBrand .579 I .427 I .432 I .412 I .539 I .515
Purchase Intention .511 1 .273 I .243 I .351 1 .450 I .389
G. Three-FactorTen-Item
PerceivedQuality
Attitude
toBrand 736 I 701 I 725 I .630 I 750 I 724
Purchase Intention .672 1 .555 I .556 I .582 | .645 I .616
H. Three-Factor15-ItemBrand
Loyalty
Attitude
toBrand .631 I .501 I .587 I .552 I .621 I .616
PurchaseIntention 706 1 .689 I 734 I .674 1 735 I 723
I. Three-Factor15-ItemBrand
Awareness/BrandAssociations ________
Attitude
toBrand .582 I .427 I .465 I .400 I .521 I .464
PurchaseIntention .530 | .269 | .258 | .340 | .432 | .362
J. Three-Factor15-Item
Perceived
Quality
Attitude
toBrand 761 I 735 I 742 I .695 I 736 I 765
PurchaseIntention | .672 1 .578 I .572 .615 _37 __4

55 JournalofMarketing
THEORY AND PRACTICE Winter2002 55

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
TABLE 11 The factorloadingwas requiredto be .40 or greaterand at
CRITERIA FOR EVALUATING MODEL FIT thanitsloadingon a different
least. 10 orgreater factorforan
Criteria AcceptableFit
itemto load. In cases wheretheseconditions werenotmet,
theitemwas deletedfromconsideration.
AdjustedGoodnessofFit(AGFI) > .9
Comparative FitIndex(CFI) > .95 The analysisextracted fiveor six factorsin each of thesix
> .9
Nonnormed FitIndex(NNFI) samples. Recall thattheoriginal26 items(Yoo and Donthu
forChi-square
Probability < .05
NormedChi-square 1.0-2.0
1997)separatedintosevendistinct factors(i.e.,brandloyalty,
CompositeReliability > .700 perceivedquality,brandawareness, brandassociations, OBE,
Varianceextracted > .500 attitudetowardthebrandandpurchaseintention).In general,
t-test
Convergent Validity t > 3.29 the exploratoryfactoranalyses did not conformto our
StandardizedResiduals>2.58 <5% oftotalresiduals notions. Betweentwoand fourindicators did
preconceived
not load on any construct in everycase. In no case did a
examinedthefour-item OBE measureusingCFA. Table 3 seven-factormodel emerge as expected. While some
showstheresultsforall six samples. We thenused CFA to indicatorsdid load as anticipated,
themajority did not. The
examinetheoriginal15-item, four-construct MBE measureas resultsoftheanalyseswerediscouraging, butnotcompletely
developedby Yoo and Donthu(1997, 2002). This analysis unexpected.
provideda baselineforthefour-factor modelusingall ofthe
Each of Tables 3-7 and Table 9 reportthesame set of ten
items.Table4 reports theresultsforall six samples. Table
5 reports an alternative three-construct MBE model, statistics forall six samplesapplyingdifferent CFA models:
15-item,
as suggestedby Yoo and Donthu(1997, 2002), forall six Table 3 reports the CFA for the OBE scale; Table 4, the15-
item, four-factor MBE scale; Table 5,the 15-item, three-factor
samples. This analysisprovidedthebaselineforthethree- MBE
factormodel,againusingall items.Table 6 reports Yoo and model;Table6, theten-item, three-factor MBE scaleas
Donthu's(1997, 2002) ten-itemMBE measureforall six finalized by Yoo and Donthu (2002); Table 7, the two-factor
attitude toward the brand/purchase intention scale; andTable
samples,replicatingYoo and Donthu's final three-factor
model.We also rana two-factor modelforattitude towardthe 9, thebest fit for the three- and four-factor MBE models. The
brandandpurchaseintention toevaluatethepredictive ten statisticswe report are: AGFI, CGI, NNFI, CHI SQ, CHI
validity foreach latentfactor,
ofthebrandequityscales. Table 7 summarizes theseresults SQ/D.F.,compositereliability variance
forall sixsamples. Theseanalyseswereperformed extracted for each latentfactor, convergent validity,number of
usingthe standardizedresiduals
CALIS procedure inSAS, Version6.12. Followinggenerally greater than 2.58, and whether the
numberof standardized residualsis less thanfivepercentof
acceptedprotocol,we used covariancesas inputto the thetotal.Foreachoftheseten
statistics,we nextexplainwhat
analysis.Werana seriesofsimplePearsoncorrelations onthe
the statisticmeasures and the standards we usedto determine
threemodels testedas well as theirlatentfactors. We
correlated fit.Table 1 1 summarizes the fitcriteria used throughout this
simplesummedscales againstattitude towardthe
Tables 8 and 10 the correlations the
brandand purchaseintention factors. Table 8 reportsthe analysis. report between
correlations andtheirsignificance. brand equity models and attitude toward the brand and
purchaseintention.
Theanalysisthenproceedstodefinethe"bestfit"CFA models
forthesix samples. Beginningwiththefull15-itemmodels The adjustedgoodness of fitindex (AGFI) is a standard
forboththe four-factor and three-factor theoreticalMBE measurereported in moststudies,althoughBender's(1990)
comparative fit
index (CFI) andBentlerandBonnett's(1980)
models,we maximizedthefitforthemodelsusingresidual
outlieranalysisandmodification indices. We completed the non-normed fitindex (NNFI) arearguedto be moreaccurate
same analysisfor the three-factor, of
ten-itemMBE model regardless sample size (Hatcher1996). We report all three
measures. For each, a statistic of .900 or better is regarded as
developedbyYoo andDonthu(1997). Table 9 providesthe
resultsforeachbest-fit modelin additionto thelistingofthe a good fit. However, Hu and Bentler (1999) have suggested
indicators foreachmodel. .950 as a moreappropriate acceptanceofmodelfitforCFI.

Theprobability forchi-square goodnessoffit(CHI SQ) is also


Finally,Table 10 providesan analysisexaminingpredictive
reported.Thismeasurerequiresa statistic
of.05 orhigher (1.0
validityvia simplePearsoncorrelationsof thevariousMBE
modelswithattitude towardthebrandandpurchaseintention. is best)forthemodelto be considereda good fit.Giventhe
Afterreporting theseresults,
we discusstheusefulnessofthe weaknessofchi-square as a goodnessoffitmeasure(Joreskog
variousbrandequityscales. and Sorbom1989),we also reportchi-squaredividedbythe
degreesof freedom, ornormedchi-square(CHI SQ/D.F.),as
RESULTS suggestedby Hair et al. (1998). A resultbetween1.00 and
2.00 indicatesa fitregardlessoftheprobability
statistic.
To assess a maximumlikelihoodfactoranalysiswith an
obliquerotation rulesofthumb. Composite reliability(Fornell and Larcker 1981) was
we usedseveralconventional
computedforeach latentfactor.Thus,forTable 3 onlyone
56 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
latentfactoris reportedand,therefore,onlyone composite providesan adequatebrandequitymeasurebutcontinuesto
reliability
computation. However,forthe15-item,four-factor pose questionsbecause of the largenumberof residualsor
MBE model(Table 4), the15-item, three-factor
MBE model noise.
(Table5),theten-item, MBE model(Table6), and
three-factor
thetwo-factor model(Table7) we report threeand
four,three, The 15-item,four-factor MBE scale also providesmixed
twolatentfactors, As such,composite
respectively. reliability results.Recall thatthisscale was Yoo and Donthu's (1997)
statistics
are computedforeach latentfactorin each model. originaltheoretical brandequitymodel. The modelfitsvery
Compositereliability is analogousto coefficient
alpha but well forall samplesexceptSample #4based on CFI and
considereda superiormeasure. Like coefficientalpha, NNFI. AGFI is mixedandchi-squareis weak. Thecomposite
compositereliabilityshouldbe .700 or higher(Hair et al. reliabilities
are strongas is thevarianceextracted,however,
1998). the numberof residuals is very high for each sample.
Examiningtheresidualsand themodification indicesshows
The varianceextractedestimateassesses the amountof thatsignificant improvement in model fit is possible by
variancecaptured bythelatentfactor inrelationtothevariance deletingoffending indicators.
becauseofmeasurement (FornellandLarcker1981)and
error
is considered a veryconservative test. A statisticof .500 or Table 5 shows thatthe 15-item,three-factor MBE model
betteris desirable. In thiscontext,convergent validityis demonstrates a fairlygood fitto thedata,similarto thefour-
defined interms oftheindividual indicators andwhether ornot factormodel. Nevertheless, the numberof standardized
are
they measuring the same latent variable(Andersonand residualsis veryhighforeach sampleindicating, onceagain,
Gerbing1988). It is assessed by reviewingthet-testsfor a verynoisyscale. The residualsand modification indices
factorloadings.Atthestatistical levelof.001, all t-testsmust suggestsubstantial room forimprovement.
generatea statistic greaterthan3.29. The argument is that
eachindicator is measuring thelatentconstruct whenthet-test As shownin Table 6, theten-item, three-factor MBE model
is significant.Finally,we reportthenumberof normalized (Yoo andDonthu'sfinalmodel)demonstrates solid
generally
residualsover2.58. Per Hair et al. (1998), an acceptable results.In all six samples,theoverallfitofthemodelto the
number ofstandardized residualsis a maximum offivepercent data is acceptable. Composite reliabilityand variance
of thetotalresiduals. We also reportwhetheror not that extractedarealso acceptable,despitepoorchi-square results.
standard is achieved. In general,thismodel fitsslightlybetterthanthe 15-item,
four-factororthe15-item, three-factor
MBE models,butagain
RecallthatYoo and Donthu(1997) beganwitha totalof 22 is subjectto residualproblems.It shouldbe notedthatthe
itemsin the OBE scale, but,in theend,retainedonlyfour Yoo and Donthu (1997) analysiswas subjectto similar
items. We usedthosefouritems,theminimum numberthat residualsproblems, themagnitude
although is notknown.An
canbesubjected toa singlefactor
confirmatoryfactor
analysis. examinationof the residualsand modification indicesalso
Therefore, we are unableto make adjustments to the OBE indicatessomeimprovement inmodelfitcan be obtained.
scale. In contrast,
theotherscalesbeingassessed(Tables4-7
and9) aremultiple latentvariablescaleswithmultiple
factors. ThepurposeofTable 7 is to examinetheconstruct of
validity
These scales can be improvedby deletingone or more bothattitude towardthebrandand purchaseintention.The
indicator(s)andstillbe re-specified
toconfirma moreaccurate two-factor CFA reported in Table 7 is includedso thatthese
fitofthedata. scales can be usedto determinethepredictive validityofthe
brandequityscales reportedin Tables 3-6 and 9. Since
Table 3 reports
theCFA resultsforthefour-itemOBE scale. purchaseintention was onlya two-item scale and,therefore,
Theseresultsprovidegenerallystablefindings.Five of the unableto be reduced,the two separateconstructs (attitude
six samplesshowverysolid fit,althoughSample#3shows towardthebrandandpurchaseintention) werecombinedinto
onlya marginalfitwitha poorAGFI. In Sample#3,all the onemodelforvalidityestimation purposes.We wouldexpect
evaluationswereforhighbrandequityproducts.In all but thathigherbrandequitywouldbe highlycorrelated withmore
Sample#4,thechi-square datais weak. positiveattitudestowardthe brand and greaterpurchase
intention.
Criticaltotheanalysisis thenumberofstandardized residuals
exceeding2.58. Samples #4 and #5 have no residuals. All six samplesgenerallyfitwell withhighAGFI, CFI and
Samples#1,#2,#3,and #6each showfourresiduals,which NNFI statistics
and strongcompositereliability
andvariance
suggestsa "noisy"scale. In summary, fiveofthesix samples extracted. However, the model produced excessive
indicategood fitstatisticallywith only Sample #3 being standardizedresiduals. Examinationof the residualsand
marginal.However,in threeof thefive"good fit"samples modificationindicesprovidedno consistent formodel
pattern
(Samples #1,#2 and #6) standardizedresidualproblems improvement acrossthesamples.
appear. A detailedexaminationof the specificindicators
underlying theresidualssuggeststhata singleindicator
is not Table 8 (A-J) providesthe bi-variatePearsoncorrelation
thecause. Therefore,deletinganyoneindicator willnotsolve statistics OBE andthethreeMBE scalesandtheir
comparing
theproblem oflargeresiduals.Thus,thefour-itemOBE scale latentcomponents withbothattitudetowardthebrandand
57 JournalofMarketing
THEORY AND PRACTICE Winter2002 57

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
purchaseintention.Recall fromthe previoustables that We were able to improvethe model by using Yoo and
construct validityproblemssurfacedin someofthesamples. Donthu's(1997) samethree-factor structure
butbyexamining
ForboththeOBE scale and theindividualunderlying latent all 15 itemsrather
thanthetenitemsultimately used by Yoo
factorsin thethree-and four-factor MBE scales,we simply andDonthu(1997) (Table 9). Thesefindings suggestthat,as
summedthe scales to computethe Pearson correlations. proposedby Yoo and Donthu(1997), a three-factor MBE
Again,thepurposeofthisanalysisis to assessthepredictive model that groups togetherbrand awarenessand brand
validity ofthefourbrandequityscales. As notedinthetables, associationitemsprovidesthemostparsimonious modelin
all Pearsoncorrelationsarestatistically atp< .001.
significant bothYoo andDonthu's ( 1997) dataandours.However,future
Brandawarenessandbrandassociations showthe
consistently researcherswould be wise to carefullyselecttheparticular
lowestcorrelations althoughthe magnitudeof correlation items that fitbest withintheirown analyses. Yoo and
variesacross the six samples. The only otherconsistent Donthu's(1997, 2002) analysisdifferedfromours in the
pattern appearingis thatcorrelationsaregenerally higherfor nature and numberof the specific items selected for
thepotatochip/barbeque sauce co-brandsthanforthepaper evaluation.
towel/disinfectantco-brands. This resultis not surprising
sincesubjectsweremorelikelyto be familiar withthepotato Our analysis also foundacceptablefits,with acceptable
chip/barbecue sauce product inasmuch as the paper compositereliability andvarianceextracted, fora four-factor
towel/disinfectantproduct was fictitious. MBE model,again starting with15 items. Whilethefour-
factorstructureis notas stronga fitas thethree-factor model,
Table 9 reportsthebestmodelfitforthefour-factor MBE itcouldcertainly be appliedin a situation wheretheneedto
model;thethree-factor, ten-item MBE model;andthethree- separatebrandawarenessfrombrandassociationis deemed
factor,15-itemMBE model. To arriveat theseresults,we critical. Thatis, in our data,the 15-item,four-factor MBE
examinedthet-statistics forindividual problematic model works,but not as well as the 15-item,three-factor
indicators,
standardized residuals,andvariousmodification indices.Fit version. We base this assertionon lower numbersof
is indicatedby strongfitstatistics and by ensuringthatall standardized residualsand slightlyhigherAGFI, CFI and
indicatorsare significant contributorsto the model. In NNFI statistics forthethree-factor modelwithslightly higher
addition, no latent factor should have fewer than two averagecompositereliability and varianceextracted forthe
indicators.The bestfitmodelsare thosewithone or more four-factor model.
offending indicators removed from the model while still
retaining a four-orthree-factor solution. The questionof whetheror notbrandawarenessand brand
associationsshouldbe collapsediscritical.BothAaker( 199 1)
All of the best-fitmodels (Table 10 A throughJ) have and Keller (1993) distinguish betweenthe two constructs.
adequate fits with all Pearson correlationsstatistically Aaker(1991) definedbrandawarenessas "... theabilityofa
significantatp< .001. Thebestfitting modelis thatdeveloped potential buyertorecognizeorrecallthata brandis a member
using three factorsand all 15 items. This is theonlymodel of a certainproductcategory"(p. 61). He further explained
withno standardized residualsacrossall six samples. It also thatbrandawarenesscan be viewedas a continuum ranging
has slightly higherfitstatisticsthanthecompetingmodels. frombrandrecognition at thelowestlevel,to brandrecallat
We also correlated thebestfitmodelswiththefour-item OBE themid-level,totop-of-mind recall(i.e.,thefirst-named brand
model forwhichYoo and Donthu(1997) reporteda .78 in unaidedrecall)and,finally, thedominant brand,whichis
correlation.Acrossthesix samples,theOBE modelandthe the only brandrecalledby the consumer. Aaker (1991)
four-factor MBE modelaverageda .661 correlation; theOBE definedbrandassociationsas "... anything linkedinmemory
andthethree-factor MBE modelaverageda .675 correlation; toa brand"(p. 109). Clearly,thetwoconstructs willbe highly
and theOBE and thefull15-itembestfitthree-factor MBE correlatedsince, accordingto Aaker's conceptualization,
modelaverageda .684correlation.All aresubstantially lower brandawarenessmustprecedebrandassociations.Thatis,a
thanthe strongcorrelation achievedby Yoo and Donthu consumer mustfirst be awareofthebrandinorderto develop
(1997), butveryconsistent withone another. a setofassociations.Nonetheless, thetwodimensions arenot
synonymous sinceonecanbe awareofa brandwithout having
DISCUSSION a strongsetofbrandassociationslinkedinmemory.

Theresultsofthepresent insomeways,support
research, and Accordingto Keller (1993), brand awarenessand brand
buildupon the workof Yoo and colleagues(1997, 2000, associationsarethedimensions thatunderlie
brandknowledge
2002), and in otherways,departfromit. The resultsof the withbrandawarenessbeingdescribedas "...relatedto the
ten-item,three-factorMBE model (Table 6) are most strengthof the brandnode or tracein memory..." (p. 3).
consistent
withYoo andDonthu's(1997) findings.We tested Kelleralso statedthat,"... brandawarenessaffects
consumer
Yoo andDonthu's(2002) finalmodelon oursix samplesand decisionmakingbyinfluencing theformationandstrength of
foundan acceptablefitand acceptablecompositereliability brand associationsin the brand image" (p. 3). Keller
andvarianceextracted.Nevertheless,thismodelwas subject conceptualizedbrand associations as falling into three
toseriousresidualproblemsprompting us to attempt
torefine categoriesof increasingscope: attributes, benefits,and
itinorderto eliminate
as manyresidualsas possible. attitudes,with attributesbeing descriptivefeaturesthat
58 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
characterizea product.ThisdiscussionshedslightonYoo and Thefundamental questioninthisanalysisis whetherornotwe
Donthu's(1997) inability to separatethetwoconstructs. The shouldbe satisfiedwitha three-factor
brandequitymodelas
seemswelldevelopedenoughto support
literature thetheory proposedby Yoo and Donthu(1997) and confirmed by our
ofbrandawarenessandbrandassociationsbeingtwoclosely a researchquestion can be
analysis. If, theoretically,
yetdistinct
related theparticular
concepts.Nevertheless, items adequatelyaddressedwithintheconstraintsofthethree-factor
selectedby Yoo andDonthu(1997) do notappearto clearly model,ouranalysissuggeststhatitprovidesthestrongestand
discriminatebetweenthetwoconstructs. The itemsselected cleanestfit. However,we encourageresearchers to further
measureonlythelowestlevel of bothbrandawareness(i.e, items(tenor 15) thatprovidethebestfit.
refinetheparticular
recognitionas perAaker1991) andthelowestlevelofbrand
onlyattributes
associations(i.e, describing versusbenefitsor In situationswherea four-factor analysisis imperative, we
attitudes
as perKeller1993). recommend thatresearchers use thefour-factorstructure,but
with caution. While the fit of this model was seen as
The confoundingresultsof the brand awareness/brand acceptable(althoughnotthebest) in ourresearch,Yoo and
associationsquestioncan be betterunderstood by examining Donthu(1997) originally discountedsucha modelas having
theseriesof correlations betweenour best fitbrandequity weak fit.Again,thismodelshouldbe used withevencloser
modelsandattitude towardthebrandandpurchaseintention attention paidto fitandresiduals.
(Table 10A through J). Thelatentfactors
ofbrandloyalty and
perceivedqualitycorrelatehighlywithbothattitudetowardthe As in all empiricalstudies,thisstudyhas limitations.The
brandand purchaseintention regardlessof themodel being primarylimitation is thatthisresearchwas not specifically
scrutinized.Thelowercorrelations appearconsistentlyinthe designed to ascertain differences in scale reliabilityand
brand awareness and brand association latent factors, validity acrossdifferentconditions.Rather, itwas designedto
with
particularly respect to intention.
purchase Theseresults testco-branding strategies.As such,the analysissuffersfrom
are not surprising.The data indicatethat,even thougha ourinability tocontrolforsomefactors andmanipulate others.
consumer mayhave a highawarenesslevel of one brand,or For instance,thepresentstudyused a seven-point response
mayhavea richsetofassociationsinconjunction withone or scale versusthe five-point scale used by Yoo and Donthu
morebrandnames,itis notgiventhattheconsumer perceives ( 1997) and,inthecontext ofco-branding, thisresearchelicited
thebrand(s)as offering highqualityor feelsa highdegreeof multipleresponsesfroma singlesubject. It is impossibleto
loyaltyto thatbrand(s). It seemsthatbothbrandawareness determinewhich variables,if any, mightaccountforthe
and brandassociationsare necessary,but not sufficient, differences betweenthis studyand the workof Yoo and
conditions forhighcustomer-based brandequity. colleagues(1997, 2000, 2002).

Table 12 examinesthe correlations conductedon our data CONCLUSION


betweenOBE, thefour-factor MBE, thethree-factor MBE and
thetwoseparateconstructs of attitudetowardthebrandand Theresultsofthisresearchprovidesomesupport forYoo and
purchaseintention.It showsthatYoo and Donthu's(1997) Donthu's(1997) OBE modeland three-factor MBE models.
proposedOBE modeldemonstrates a veryhighcorrelation Nonetheless, the support is inadequate to offer an
(andhigherthaneitherthefour-orthree-factor MBE models) unconditionalendorsement ofYoo andDonthu's ( 1997) brand
with purchaseintentionbut relativelyweak resultswith equitymeasures. On the ourresultsencouragefuture
contrary,
attitudetowardthebrand. It could be arguedthatthemore researchersto proceedto further
refinethefactorstructure of
criticalmeasureis purchaseintention.However,thesemixed thesemodels. We concludethatYoo and Donthu'sscale is
resultsarelikelya resultofnoise(thenumberofresiduals)in notpsychometrically soundfortheorytestingresearchand
theOBE scalethatis impossibleto reducebecauseofthefew needsto be improved.Ourbestfitmodelwas similarto Yoo
number of items(four)thatcomprisethescale. As reported, and Donthu's(1997) butrequiredadditionalitemsthatYoo
deletinganyone itemdoesnotimprovethescale anddeleting andDonthudeletedintheiranalysis.We concurwithYoo and
morethanone itemwoulddestroyit. Thus,we recommend Donthuthatsucha measurement scale is neededandthatscale
thatfutureresearchers use the OBE scale withcautionby development to datehasbeensuccessful.We also agreewith
remaining verycognizantofproblematic residuals. Yoo andDonthu'sconclusionthat,whilethisscalerepresents
a forwardstep in scale development, furtherrefinementis
TABLE 12
AVERAGE CORRELATIONS BETWEEN BRAND
required.
EQUITY MODELS AND ATTITUDE TOWARD
THE BRAND AND PURCHASE INTENTION Futureresearchshouldfocuson thebrandawareness/brand
15-ItemFour- Ten-ItemThree- associationsissue. Theoryhighlightsa clear distinction
OBE FactorMBE FactorMBE betweenbrand awarenessand brand associations,which
Average Average Average makesitimportant to further
refinethescales. We arguethat,
Attitudetoward .560 .723 .674
thebrand consistentwithYoo and Donthu(1997), a three-factor
scale
Purchase .675 .646 .630 worksbest given the constraintsof the itemsselectedto
Intention | | measurebrandawarenessand brandassociations.However,
ouranalysissuggeststhatdifferent
itemsmayactto improve
59 JournalofMarketing
THEORY AND PRACTICE Winter2002 59

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
fitsof themodel/data
thestatistical interface.Therefore,
we financialvalueofa brand.An equallyimportant questionto
stronglyencourage futureresearchersto reevaluate
theuse of managers is how to measure
effectively customer-basedbrand
theparticular itemsselectedforthescale and look uponthe equity.However,muchlessattention hasbeendevotedtothe
literature
to suggestmorediscriminating indicators. design of such an instrument.The realvalueofattempting to
refineYoo and Donthu's customer-based brand equity
Much brandequityresearchwill need to relyon a short, measureis in movinga stepcloserto a universally accepted
overallperceptualbrandequityevaluation. The four-item customer-based brandequityscale. Yoo and Donthumade
OBE scale developedby Yoo and Donthu(1997) is a good great stridesin accomplishingthis task. Nevertheless,
start,but seemsto sufferfromresidualproblems. In our managers mustbe confident thattheinstrument recommended
discussion,we havereferred toresidualproblemsas noise. A is, indeed,measuringwhat theyintendto measure. By
majoreffort shouldbe undertaken to furtherrefinethisscale evaluating theYoo andDonthuscaleina different we
context,
intoa four-or five-item scale thatexhibitsmoreconsistent havemovedthisresearchforward. Ourresearch supportsYoo
responsepatterns.The questionofwhether a five-or seven- andDonthu'sconceptofan OBE scale,butsuggeststheYoo
pointresponsescale is appropriate also shouldbe further and Donthuscale would benefitfromréévaluationof the
considered. specificitems.Further, ourresearchuncoversa limitation to
Yoo and Donthu'sthree-factor MBE scale thatcontradicts
Yoo andDonthu(1997) havemadean excellentstarttowards acceptedtheoryabouttheunderlying brandequityconstructs
thedevelopment ofvalidandreliableperceptual measuresof ofbrandawarenessandbrandassociations.Whilewe hesitate
brandequity.Futureresearchshouldbeginwiththisbase and to takea standon whichscale is bettergiventhelimitations
refinethescaleswiththegoal ofachievingpsychometrically associatedwiththedifferences inmethodsofthetwostudies,
validandreliablescales. In an idealworld,a validandreliable we do nothesitate torecommend thatfurther refinementofthe
scale is one that providessimilarand consistentfactor scale is calledfor. We believethisscale shouldbe testedin
structuresregardlessof thecontextin whichit is used. The different contexts,on different typesof products,and with
current scalehas shownto be reasonably consistentovertwo different of
types subjects inorder toachievegeneralizability.
very differentresearch
design contexts, butfurtherrefinement When the scale(s) are refined to the point of general
is calledfor. acceptance,marketing managerswill findthemusefulin
tracking customers'perceptions of brandequity,comparing
MANAGERIAL IMPLICATIONS oneproduct'scustomer-based brandequitywithanother'sfor
deletionand resourceallocationdecisions,and contributing
Marketingmanagershave severalmodels fromwhich to one additionalpiece of information to estimatethemonetary
choosewhenmeasuring brandequitywithrespectto the valueofa companyand itsproducts.

REFERENCES

Aaker,D.A. (1991),ManagingBrandEquity,NewYork: The and D. G. Bonnett(1980), "SignificanceTests and


FreePress. Goodness of Fit in the Analysisof Covariance
PsychologicalBulletin,88: 599-606.
Structures,"
Alba, J.W.and J. W. Hutchinson, (1987), "Dimensionsof
Consumer Expertise," Journal of Consumer Calder,B. J., L. W. Phillips,and A. M. Tybout(1981),
Research,13 (1), 411-453. "DesigningResearchFor Application,"Journalof
ConsumerResearch8 (2): 197-207.
Anderson,J.C. and D. W. Gerbing(1988), "Structural
EquationModeling in Practice: A Review and Cook,T. andD. Campbell(1975), "The DesignAndConduct
Recommended Two-StepApproach," Psychological Of Experiments And Quasi-Experiments In Field
Bulletin,103,411-423. Settings."EditedbyMartinDunnette.Handbookof
Industrialand OrganizationalResearch.Chicago:
Bagozzi, R. (1992), "Acrimonyin the Ivory Tower: RandMcNallyandCo.
StagnationorEvolution?" JournaloftheAcademyof
Marketing Science,20 (4), 355-359. Dodds, W.B., K.B. Monroe,and D. Grewal(1991), "Effects
of Price,Brandand StoreInformation on Buyers'
Bello,D. C. andM. B. Holbrooke(1995), "Does An Absence Product Evaluation," Journal of Marketing
Of Brand Equity Generalize Across Product Research,"28 (3), 307-319.
Classes?"JournalofBusinessResearch34:125-131.
Fornell,C. and D.F. Larcker(1981), "EvaluatingStructural
Bentler,P.M. (1990), "Comparative FitIndexesin Structural EquationModels withUnobservableVariablesand
Models,"PsychologicalBulletin,107: 238-246. Measurement Error," Journal of Marketing
Research,18 (February):39-50.
60 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
Hair, J. F., R.E. Anderson,R.L. Tathamand W.C. Black Simon,C.J.,and M.W. Sullivan(1993), "The Measurement
ariateData Analysis,FifthEdition,
(1998), Multiv and Determinants of Brand Equity: A Financial
UpperSaddleRiver,NJ:PrenticeHall. Approach,"Marketing Science,12 (4), 28-52.

Hatcher,L. (1996), "Using SAS PROC CALIS or Path


StructuralEquation
Analysis: An Introduction," Siminon,B. L., andJ.A. Ruth(1998), "Is A CompanyKnown
3
Modeling, (2): 176-192. By TheCompanyItKeeps? AssessingTheSpillover
EffectsOf Brand Alliances On ConsumerBrand
Hu, L. and P.M. Bentler,.(1999), "CutoffCriteriaforFit Attitudes," JournalofMarketing Research35(1): 30-
Indexes in Covariance Structure Analysis: 42.
ConventionalCriteriaVersus New Alternatives,"
EquationModeling,6(1), 1-55.
Structural and (1995), "BundlingAs A Strategy
For New
Product Introduction: EffectsOn Consumers'
K. G. andD. Sorbom(1989), LISREL 7, Chicago:
Joreskog, ReservationPrices For The Bundle, The New
SPSS, Inc. Product,And Its Tie-In," Journal of Business
Research33: 219-230.
Keller, K.L., (1993), "Conceptualizing,Measuring,and
ManagingCustomer-Based BrandEquity,"Journal Srull,T.K. (1984), "Methodological TechniquesfortheStudy
ofMarketing, 57 (1), 1-22. of PersonMemoryand Social Cognition,"in R.S.
Wyerand T.K. Srull (eds.), Handbook of Social
Martin,D. W. (1996), Doing PsychologyExperiments,4th Cognition, vol. 2, Hillsdale,NJ:LawrenceErlbaum
Edition,Brooks/Cole PublishingCompany,Pacific Associates,1-72.
Grove,CA.
J.H.(1999), An EvaluationofCo-branding:Its
Washburn,
Nedungadi, P. and J. W. Hutchinson (1985), The onBrandEquity,SEC Attribute
Effects Performance,
ofBrands:Relationships
Prototypicality withBrand and the Moderating Role of Product Trial,
Awareness, and
Preference, Usage," inAdvancesin UnpublishedDoctoralDissertation,UMI Number:
ConsumerResearch,Vol. 12, ElizabethHirshman 9942832, UMI 300 North Zeeb Road, AnnArbor,
andMorrisHolbrook,eds. Provo,UT: Association MI.
forConsumerResearch,498-503.
Yoo, B., andN.Donthu( 1997), DevelopingandValidating
Park,C.W.,S.Y. Jun,andA. D. Shocker,(1996), "Composite a Consumer-basedOverallBrandEquityScale for
Branding Alliances: An Investigation
Of Extension Americansand Koreans:An Extensionof Aaker's
And Feedback Effects,"Journal of Marketing andKeller'sConceptualizations," at
Paperpresented
Research33 (4): 453-466. 1997 AMA SummerEducatorsConference,Chicago:

Park,C.S., and V. Srinivasan,(1994), "A Survey-Based , and (2002), "Developing and Validatinga
MethodforMeasuringand Understanding Brand Multidimensional Consumer-basedBrand Equity
Equityand itsExtendibility,"
JournalofMarketing Scale,"Journal
of BusinessResearch,Forthcoming.
Research,31 (2), 271-288.
, , and S. Lee (2000), "An Examinationof
Rao, A. R., and R. W. Ruekert(1994), "BrandAlliancesas Selected MarketingMix Elements and Brand
signals of ProductQuality,"Sloan Management Equity,"Journal of the Academyof Marketing
Review,Fall,87-97. Science,28 (2), 195-211.

Rossiter,J.R., and L. Percy (1987), Advertisingand Zeithaml,V.A. (1988), "ConsumerPerceptionsof Price,


Promotion Management,New York:McGrawHill. Quality, and Value: A Means-End Model and
Synthesisof Evidence,"Journalof Marketing,
52
(3), 2-22.

61 JournalofMarketing
THEORY AND PRACTICE Winter2002 61

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions
AUTHOR BIOGRAPHY
Judith H. Washburn (Ph. D.SaintLouis is assistant
University) professorofmarketing at BowlingGreenState
in
University Bowling Green, OH. Dr. Washburnhas publishedpreviouslyin suchjournalsas JournalofConsumer
Marketing, JournalofServicesMarketing andJournalofNonprofit and PublicSectorMarketing.Herresearchinterests
relateto organizational
alliancesintheformofbrandalliancesandcauserelatedmarketing.

AUTHOR BIOGRAPHY
RichardE. Plank,(Ph. D., CityUniversity ofNew York)is professor
ofmarketingat WesternMichiganUniversity in
Kalamazoo,MI. Dr. Plankhas publishedpreviously in over15journalsincluding
theJournalofMarketing Theoryand
Practice. His researchinterests
relateto integration
ofbuyingand sellinginbusinessmarkets.He has developeda
modelofbrandequityforbusinessmarkets.

62 JournalofMarketing
THEORY AND PRACTICE

This content downloaded from 139.132.1.23 on Thu, 3 Oct 2013 16:08:57 PM


All use subject to JSTOR Terms and Conditions

You might also like