consumer evaluations of brand extensions

David A. Aaker & Kevin Lane Keller

Consumer Evaluationsof Brand Extensions

Two studies were conducted to obtain insights on how consumers form attitudes toward brand exten-sions, (i.e., use of an established brand name to enter a new product category). In one study, reactionsto 20 brand extension concepts involving six well-known brand names were examined. Attitude towardthe extension was higher when (1) there was both a perception of "fit" between the two product classesalong one of three dimensions and a perception of high quality for the original brand or (2) the extensionwas not regarded as too easy to make. A second study examined the effectiveness of different positioningstrategies for extensions. The experimental findings show that potentially negative associations can beneutralized more effectively by elaborating on the attributes of the brand extension than by remindingconsumers of the positive associations with the original brand.

THE financial risk of entering new markets has be-come formidable for many consumer product

manufacturers. The cost of introducing a new brandin some consumer markets has been estimated to rangefrom $50 million to more than $100 million (Brown1985), with a total cost estimated to run to $150 mil-lion (Tauber 1988). The price tag is much larger thanin the 1970s in part because of the dramatic increasein media costs, the more extensive and aggressive useof promotions by established firms, and the cost anddifficulty of obtaining distribution. As a result, firmsare using established brand names to facilitate enter-ing new markets.

One such approach is line extension, whereby acurrent brand name is used to enter a new market seg-ment in its product class (e.g.. Diet Coke and LiquidTide). Another approach is brand extension, whereby

David A. Aaker is J. Gary Shansby Professor of Marketing Strategy,School of Business Administration, University of California, Berkeley.Kevin Lane Keller is Assistant Professor of Marketing, Graduate Schoolof Business, Stanford University. The authors acknowledge financial as-sistance from the Institute of Business Economic Research and thankDipankar Chakravarti, Peter Farquhar, David Reibstein, Itamar Simon-son, Allan Shocker, Douglas Stayman, Sheri Bridges, and the anony-mous JM reviewers for helpful comments.

a current brand name is used to enter a completelydifferent product class (e.g., Jello frozen pudding pops,Clorox laundry detergent. Ivory shampoo, or NCRphotocopiers). The strategy of introducing new prod-ucts as extensions has become widespread. From 1977to 1984, approximately 40% of the 120 to 175 newbrands that were introduced into supermarkets an-nually were extensions (Nielsen 1985). In 1986, morethan $15 billion in retail sales and more than 34% ofapparel and accessory sales comprised products thatwere licenses or trademarks of brand names (Kesler1987).

Brand extensions, the focus of our research, areattractive to firms that face the reality of high newproduct failure rates because they provide a way totake advantage of brand name recognition and imageto enter new markets. The leverage of a strong brandname can substantially reduce the risk of introducinga product in a new market by providing consumersthe familiarity of and knowledge about an establishedbrand. Moreover, brand extensions can decrease thecosts of gaining distribution and/or increase the ef-ficiency of promotional expenditures (Morein 1975).

The brand extension decision is strategically crit-ical to an organization. Though an extension is a wayto exploit perhaps the most important asset owned bya business, it also risks decreasing the value of that

Journal of MarketingVol. 54 (January 1990), 27-41 Consumer Evaluations of Brand Extensions / 2 7

asset. The wrong extension could create damaging as-sociations that may be expensive, or even impossible,to change (Ries and Trout 1981). Further, the decisionusually involves an important strategic growth thrust.If the judgment is wrong, substantial time and re-sources are lost and other market opportunities maybe missed.

The success of a brand extension often depends oncertain assumptions about consumer behavior, such as(1) consumers hold positive beliefs and favorable at-titudes toward the original brand in memory, (2) thesepositive associations facilitate the formation of posi-tive beliefs and favorable attitudes toward the brandextension, and (3) negative associations are neithertransferred to nor created by the brand extension. Al-most no research, however, has provided guidanceabout considerations affecting the likelihood that theseassumptions hold.

Our exploratory research consisted of two studies.In study 1, the extension reaction study, we obtainedreactions to 20 brand extensions involving six well-known brand names. Respondents provided a set ofopen-ended associations with the brand name and eachof the 20 extensions in addition to scaled measures ofattitude toward the original brand and the extensions,three measures of fit between the two product classesinvolved, and the perceived difficulty of making theextension. In study 2, the extension positioning study,more information was provided about the extension,such as cues to the positive attributes of the brand orelaborations designed to counter possible negativeperceptions of the extension, and consumer reactionsagain were assessed.

Research Questions

The purpose of our research was to explore how con-sumers evaluate brand extensions. Our goal was togain useful insights into why some brand extensionsfail and others succeed. In particular, the followingresearch questions were addressed.

1. Can useful qualitative insights into consumer evalua-tions of brand extensions be gained by exploring re-actions to the six brands and 20 extensions? What kindsof beliefs about the original brand will consumers as-sociate with the brand extension and in what ways willthose beliefs affect the extension attitude?

2. How will consumers' perceptions of the overall qualityof the original brand affect their evaluations of an ex-tension? Under what circumstances will quality per-ceptions have the largest effects?

3. What is the role of consumers' perceptions of the "fit"between the original and the new product class? Willthey affect the transfer of the quality perception of thebrand to the extension? How should fit be conceptual-ized and measured?

4. Will other aspects of the extension context, such as howdifficult the extension is to make, affect consumerevaluations?

5. How are consumer evaluations affected when differenttypes of infonnation are provided in the extension con-text?

The first four questions were explored in study 1and the fifth question was addressed in study 2. In thenext section, we discuss the study 1 research ques-tions in more detail and introduce the relevant con-structs. The methods and findings of the two empir-ical studies then are described. Finally, unansweredquestions and future research directions are presented.

Study 1. Extension Reaction Study

Research Issues

Study 1 explored how an attitude toward a brand ex-tension is formed. In this section, we introduce con-structs relevant to the process: brand attribute asso-ciations, the perceived quality of the brand, the fitbetween the two products, and the difficulty of mak-ing the brand extension. The potential relationship ofthese constructs to the attitude toward the extensionis discussed.

Brand attribute associations. A great variety ofassociations with the brand could potentially be trans-ferred to the extensions. Perhaps the most-used brandpositioning device is product attributes or character-istics. Associations have been created in the minds ofmany consumers between Viva towels and durability,BMW cars and performance, and Apple and user-friendliness. A brand also can have an association witha use situation, a type of product user, a place, or aproduct class (Aaker 1982). For example, Lowenbraubeer is associated with relaxing with good friends,Mercedes with wealthy, discriminating people, andToyota with Japan. Budweiser, Chevrolet, Levi's, andBank of America all undoubtedly have strong productclass associations. The product class of the originalbrand, especially a familiar one like beer or auto-mobiles, can itself have a set of rather strong asso-ciations that can attach to the extension.

For most brand extensions, a motivating rationaleis that the original brand has associations that will behelpful to the extension. The impact of a brand as-sociation, however, can be harmful to the extension.For example, the Betty Crocker attribute associationmight be viewed as negative if the name were usedon a fashion product designed to appeal to youngwomen. Zeithaml (1988) provides evidence suggest-ing that thickness is related to high quality in tomato-based juices but not in children's fruit-flavored drinksand pulp is related to high quality in orange juice butto low quality in apple juice. Thus, the impact of abrand belief or association highly valued in the orig-inal product class may not be positive in the contextof the new product class.

28 / Journal of Marketing, January 1990

In terms of an associative model of memory(Anderson 1983; Wyer and Smll 1986), there are linksbetween a brand such as Budweiser and a set of ele-ments such as user types and the product class. Theseelements themselves have links to other elements (e.g.,from the beer product class to beer taste). As a result,the brand has a set of associations that vary in strength.The extent to which each of these associations will"transfer" to a new product context is a central issue.It is likely to depend not only on the strength of as-sociation, but also on other factors such as the appro-priateness of the association and whether cues arepresent to activate an association.

In the first study, open-ended associations for 20hypothetical brand extensions were examined to ex-plore qualitatively what types of associations appearand what impact these associations seem to have orpotentially could have on evaluations of the exten-sion. In particular, are negative as well as positiveassociations transferred or created? These qualitativeinsights should provide useful hypotheses for futureresearch.

Attitude toward the original brand. In addition tospecific brand attributes, an important brand associ-ation is the overall brand attitude. Brand attitude isbased on certain attributes such as durability, inci-dence of defects, serviceability, features, perfor-mance, or "fit and finish" (Garvin 1984). However,it may also contain affect that is not reflected in themeasured attributes, even when a large set of attri-butes is included. Researchers building multiattributemodels of consumer preference have included a gen-eral component of attitude toward the brand that is notexplained by the brand attribute values (e.g., Srinivasan1979). The overall brand attitude may be stored andretrieved in memory separately from the underlyingattribute information (Anderson and Hubert 1963;Carlston 1980; Lingle and Ostrom 1979; Riskey 1979).

Attitude is conceptualized here in terms of theconsumer's perception of the overall quality of thebrand, termed QUALITY. The perceived quality con-struct has received considerable attention in the mar-keting literature (Holbrook and Corfman 1985; Jacobsonand Aaker 1987; Olshavsky 1985). Zeithaml (1988)defines perceived quality as a global assessment of aconsumer's judgment about the superiority or excel-lence of a product. She concludes, after reviewing aset of articles, that perceived quality is at a higherlevel of abstraction than a specific attribute of a prod-uct.

The impact of perceived quality on the attitude to-ward the extension should be unambiguously positive.If the brand is associated with high quality, the ex-tension should benefit; if it is associated with inferiorquality, the extension should be harmed. The rela-

tionship between perceived quality and the attitude to-ward the extension was explored quantitatively in thefirst study by relating a measure of perceived qualityto a measure of brand extension attitudes over thesample of respondents and extensions. The hypothesisis:

H,: Higher quality perceptions toward the original brand(i.e., higher QUALITY) are associated with more fa-vorable attitudes toward the extension.

Fit between the original and extension productclasses. Prior brand extension research has empha-sized primarily the role of "fit" or similarity betweenthe two involved product classes in the formation ofbrand extension evaluations. For example, early fam-ily branding research modeled generalized preferencesfor a family brand entry as a function of the similarityof brand competition and price levels in the particularproduct classes (Fry 1967) or similarity in shelf ar-rangements (Neuhaus and Taylor 1972). More re-cently, Tauber (1988) studied 276 actual extensionsand concluded that perceptual fit (i.e., whether a"consumer perceives the new item to be consistent withthe parent brand") is a key element in predicting brandextension success. Another study (University of Min-nesota Consumer Behavior Seminar 1987) providedempirical support for the notion that greater perceivedsimilarity between the current and new products leadsto a greater transfer of positive or negative affect tothe new product.

Why should fit be important to an extension? Onereason is that the transfer of the perceived quality ofa brand will be enhanced when the two product classesin some way fit together. Several theoretical perspec-tives are compatible with such a view: cognitive con-sistency (Heider 1958; Osgood and Tannenbaum 1955),stimulus generalization (Bierley, McSweeney, andVannieuwkerk 1985; McSweeney and Bierley 1984),affect transfer (Wright 1975), and categorization the-ory (Cohen and Basu 1987; Fiske 1982; Fiske andPavelchak 1986; Sujan 1985).

For example, categorization theory suggests that aconsumer would evaluate a brand extension in one oftwo ways: (1) by piecemeal processing, whereby anextension evaluation is a function of inferred brandattribute beliefs and their evaluative importance, or(2) by category-based processing, whereby an exten-sion evaluation is a function of some overall attitudetoward the original brand. Specifically, if consumersperceive a similarity or "fit" between the original andextension product classes, with category-based pro-cessing they would transfer quality perceptions to thenew brand extension. In fact, categorization research-ers have demonstrated that general affect can be trans-ferred fi'om one object to another (Gilovich 1981; Read1983). Hence, a second hypothesis is that:

Consumer Evaluations of Brand Extensions / 2 9

H2: The transfer of a brand's perceived quality is en-hanced when the two product classes in some way fittogether. When the fit is weak, the transfer is inhib-ited.

A second reason why fit should be important tothe extension is that a poor fit may not only detractfrom the transfer of positive associations, but may ac-tually stimulate undesirable beliefs and associations.When the fit is low, as in Betty Crocker bicycles, forexample, the consumer may question the ability of afood firm to make good bicycles. If the fit is incon-gruous, the extension may be regarded as humorousor ridiculous. Hence, another hypothesis is:

H3: The fit between the two involved product classes hasa direct positive association with the attitude towardthe extension.

Dimensions of fit. Prior research focused on fit orsimilarity but did not consider the various bases of fitbetween two product classes. Product pairs can beperceived to fit in many ways, however, and we de-veloped three such measures. Two measures take ademand-side perspective to consider the economic no-tions of substitutes and complements in product use.The third measure takes a supply-side view to con-sider aspects of the firm's manufacturing abilities.

The first fit measure, COMPLEMENT, indicatesthe extent to which consumers view two product classesas complements. Products are considered comple-ments if both are consumed jointly to satisfy someparticular need (Henderson and Quandt 1980). Thesecond fit measure, SUBSTITUTE, is the extent towhich consumers view two product classes as substi-tutes. Substitute products tend to have a common ap-plication and use context such that one product canreplace the other in usage and satisfy the same needs.

Consider Rossignol, which makes downhill skis.A complementary extension might be Rossignol skiclothing; a substitute extension might be Rossignolcross-country skis or ice skates. In both cases, be-cause fit is present, the transfer of positive associa-tions should not be inhibited. When fit is high, con-sumers are hypothesized to accept the extension conceptand not activate thought processes challenging thequality and characteristics of the extension.

The other fit measure, TRANSFER, pertains notto how consumers view relationships in product usage,but how consumers view relationships in productmanufacturing. Specifically, TRANSFER reflects theperceived ability of any firm operating in the firstproduct class to make a product in the second productclass. Do consumers feel that the people, facilities,and skills a firm uses to make the original productwould "transfer" and be employed effectively in de-signing and making the product extension? If not, theperceived quality of the brand or beliefs about the brand

in the original product class may not transfer to theextension. In fact, if a firm appears to be stretchingexcessively beyond its area of competence, negativereactions such as skepticism or even laughter mightbe stimulated and lead to negative associations.

A question of both theoretical and practical im-portance is the relative role of each of the fit measuresin terms of (1) their direct impact on the attitude to-ward the extension and (2) their moderating impacton the relationship between the perceived quality ofthe original brand and the attitude toward the exten-sion, as expressed in H2 and H3.

Perceived difficulty of making the extension. Var-ious perceptions of the new product class also mayaffect consumer evaluations of a brand extension. Weconsider one such factor, the perceived difficulty indesigning or making the extension product, termedDIFFICULT.

When consumers perceive the extended productclass to be "trivial" or very easy to make (i.e., DIF-FICULT is low), a potential incongruity occurs. Theconsumers may view the combination of a quality brandand a trivial product class as inconsistent or even ex-ploitative. The incongruity itself may trigger a rejec-tion or it might lead to a judgment that the qualityname will add a price higher than is justified and nec-essary for such a product. This logic supports the hy-pothesis that:

H4: The relationship between the difficulty of making theproduct class of the extension, DIFFICULT, and theattitude toward the extension is positive.

Summary. Study 1, the extension reaction study,was an exploratory study motivated by the hypothesesthat brand extension attitudes are influenced by theperceived quality of the brand name and the fit be-tween the two product classes. Further, an interactionwas hypothesized between the two as perceived qual-ity will be more helpful when there is a reasonablefit. Characteristics of the extension product class alsocan affect extension attitudes. In particular, productclasses that are perceived as either too trivial or tooeasy to make were hypothesized to receive lower ex-tension evaluations. In addition, attribute associationswere expected to influence extension attitudes bothnegatively and positively. The open-ended questionsin study 1 were intended to explore the nature of thatinfluence.

Method

Perceptions and evaluations of a set of six actual brandsand 20 hypothetical brand extensions were gatheredfrom 107 undergraduate business students who par-ticipated in the study as part of a course requirement.A survey cover letter said that the interest was in their


opinions, as consumers, about different brands andproducts.

Stimuli. The original brands were selected on thecriteria of being relevant to subjects, generally per-ceived as high quality, able to elicit relatively specificassociations, and not broadly extended previously. The20 extensions selected had to be reasonable and notillogical, but had to provide heterogeneity on the threefit measures. High quality brands were chosen be-cause the use of low quality brands would have tendedto generate extensions that would be less realistic. Ac-tual selection of the brand names and product classextensions involved analyzing the responses of twofocus groups and a survey of almost 100 subjects sim-ilar to those in the actual data collection. The follow-ing brands and extensions satisfied the criteria.

Original Brand Product Class ExtensionHeineken BeerVuamet Sunglasses

Haagen-Dazs Ice Cream

Vidal Sassoon Shampoo

Crest Toothpaste

McDonald's Meal

Light beer, wine, popcomSkis, wallets, sportswear,

watchesPopcom, cottage cheese,

candy barSkin cream, suntan lotion,

perfume, sportswearMouthwash, chewing gum,

shaving creamFrozen fries, theme park,

photo processing

Measures. Open-ended associations were obtainedfirst for the original brand and then for the set of ex-tensions. Respondents were asked to take roughly 30seconds to write down the associations or thoughts thatcame to mind when they considered the idea of pur-chasing each brand name product or extension. Theset of 26 open-ended association tasks was split intothree parts separated by sets of scaling tasks.

The open-ended responses to the thought-listingquestions were classified by two coders blind to thepurpose of the study. The coding was done qualita-tively by having coders group associations into clus-ters according to perceived similarity (see Tables 1and 3 for examples of the associations). The codersagreed on 82% of the associations (i.e., placed theassociation into the same type of cluster). Because thedifferences in coding were believed to be a functionof subjective interpretations in judgments and not sys-tematic deviations, the results reported are the aver-ages of the two judges' codings.

Three fit measures were used, with 7-point Likertscales (1 = strongly disagree, 7 = strongly agree).Subjects assessed the extent to which the products weresubstitutes that they would select between in certainusage situations (SUBSTITUTE) and complements thatthey would be likely to use together in certain usagesituations (COMPLEMENT). The third scale mea-

sured the perceived ability of a competent manufac-turer in the original product class to make the productextension (TRANSFER). Would the people, facili-ties, and skills used in developing, refining, and mak-ing the original product be helpful if the manufacturerwere to make the product extension (1 = not at allhelpful, 7 = very helpful)? Note that the TRANSFERmeasure was not linked to any specific brand in theproduct class.

A 7-point scale measured the difficulty in design-ing and making the product, DIFFICULT (1 = not atall difficult, 7 = very difficult). Would specializedpeople, facilities, and skills be needed to make theextended product class? A 7-point scale assessed theoverall quality of each original brand, QUALITY (1= inferior, 7 = superior). Finally, the attitude towardthe extension was operationalized by two differentmeasures: the perceived overall quality of the exten-sion (1 = inferior, 7 = superior) and the likelihoodof trying the extension assuming a purchase wasplanned in the product class (1 = not at all likely, 7= very likely).

To avoid confounding the reactions to the exten-sions, the only information subjects had was the brandname. Consumers' reaction to extension concepts sopresented should be relevant to managers who mustmake decisions about introductory marketing cam-paigns for brand extensions. It provides informationon what baseline reactions occur and the associationsthat a marketing program must counter or upon whichit can build.

Qualitative Results

The objective of the qualitative phase of the researchwas to see what types of associations would emergefrom a thought-listing about the original brands andthe extensions and thus gain insights about why eval-uations were more favorable toward some of the ex-tensions than toward others.

Original brand associations. Table 1 summarizesthe coded open-ended associations and average qual-ity ratings for the original brands. Four of the brandsreceived extremely high ratings (Heineken, Vuamet,Haagen-Dazs, and Crest), whereas the other two re-ceived above average or average quality ratings (VidalSassoon and McDonald's). These quality assessmentsare reflected in the stated associations, with manysubjects noting overall brand quality. Perceived pricewas mentioned frequently, with most subjects notingthat four of the brands were characterized by highprices. Many associations were more specific, refer-ring to particular product attributes (Haagen-Dazs—smooth/creamy), packaging (Vidal Sassoon—darkbrown bottle), facts about the manufacturer (Vuamet—French), user characteristics (Crest—family), or usage


TABLE 1Summary of Coded Brand Associations for Original Brands: Number of Respondents Mentioning Item"

Heineken Beer (5.57)ExpensiveHigh qualityGreen bottle/labelImportedEuropeanGood beer

Haagen Dazs Ice Cream (5.85)ExpensiveGood tastingGreat flavorsGood ice creamHigh qualityRichCreamyHigh calories

Crest Toothpaste (5.48)Cavity fighterTraditionalTastes goodBrand loyal

442427232315

5528201816171713

36241716

Vuarnet Sunglasses (5.87)ExpensiveSkiingQualityStylish/fashionableTrendyUV protection

Vidal Sassoon Shampoo (4.33)ExpensiveGood scentBrown bottleHigh qualityFrench/French hair designerFashionableUsed in salons

McDonald's Meal (3.33)Fast foodCheapTastes badGood friesGreasyGolden Arches/Ronald McDonald

683328271312

24151414131211

464523201913

"Numbers in parentheses are the average quality ratings. Associations and ratings are based on a sample of 107 undergraduatebusiness students.

situations (Vuamet—skiing). Thus, analysis of thestated associations for the original brands is consistentwith the notion that those brands are associated witha variety of specific attributes, as well as overall at-titudes.

Brand extension associations. Table 2 reports theaverage ratings of the 20 brand extensions and Table3 summarizes the accompanying open-ended associ-ations. An examination of the most frequently men-tioned associations for brand extensions receiving lowevaluations is illuminating. Three primary problemsemerge: (1) the fit between the original and extensionproduct classes was perceived as low, (2) the exten-sion was perceived as too easy to make, and (3) theoriginal brand carried damaging attribute character-istics to the extensions.

One problem with low rated extensions was a lackof perceived fit or similarity between the original andextension product classes. The firm was seen as lack-ing the ability to make a product that would be su-perior to competitors in the extension product class.These fit concerns were expressed in several ways.For example, some subjects reacted to the concept ofMcDonald's photo processing by stating either thatMcDonald's should stick to food and had no credi-bility as a photo processor (n = 33) or that they wouldprefer to stay with established photo processors (n =16). In evaluating Heineken popcorn, some subjects

felt that popcorn and beer do not mix (n = 26). Sim-ilarly, some subjects felt that a beer and wine pro-ducer is a bad combination (n = 32) or that Heinekenlacked the technical experience to make wine (n =17).

A second problem was that the product class wasconsidered too "easy to make." In reacting to Hei-neken popcorn, some subjects maintained that all pop-corn is the same (n = 20). Because all brands areabout the same, the extension was presumably per-ceived as being unlikely to be superior to existingproducts. This lack of perceived difference in qualityin the product class is also evident for Vidal Sassoonperfume (n = 14), Crest shaving cream (n = 20), andHaagen-Dazs cottage cheese (n = 11).

A third problem with less successful extensions wasthat the brand name carried damaging product classassociations or beliefs to the extension. For example,some subjects commented on how Heineken popcornwould probably either taste specifically like beer (n =15) or, more generally, would have an unappetizingtaste (n = 18). Vidal Sassoon perfume was associatedby some subjects with an undesirably strong sham-poo-like scent (n = 17). Crest chewing gum wasthought by some either to taste specifically like tooth-paste or, more generally, to taste bad or unappealing(n = 25).

An interesting illustration of the transfer of attri-butes is the contrast of Crest chewing gum to Crest


TABLE 2Study 1 Means"

1. McDonald's photo processing2. Heineken popcorn3. Heineken wine4. Haagen Dazs cottage cheese5. Vidal Sassoon perfume6. Crest shaving cream7. Haagen-Dazs popcorn8. McDonald's frozen french

fries9. Crest chewing gum

10. Vidal Sassoon sportswear11. McDonald's theme park12. Vidal Sassoon skin cream13. Vaurnet wallets14. Vaurnet skiis15. Vidal Sassoon suntan lotion16. Vaurnet watches17. Heineken light beer18. Haagen Dazs candy bar19. Crest mouthwash20. Vaurnet sportswearAverageStandard deviation

AttitudeToward

Extension2.032.302.943.133.243.263.28

3.373.433.483.563.633.783.913.984.074.764.814.865.153.601.65

QUALITY

3.335.575.575.854.335.485.85

3.335.484.333.334.335.875.874.335.875.575.855.485.875.081.53

TRANSFER

1.431.984.363.673.583.632.39

5.844.061.912.314.662.712.794.332.876.714.376.083.253.652.02

SUBSTITUTE

1.501.924.352.852.412.123.95

4.223.751.711.992.792.262.112.422.755.565.304.432.883.062.08

COMPLEMENT

1.695.355.262.084.274.473.19

3.403.803.193.985.094.656.093.924.865.494.265.645.754.322.03

DIFFICULT

4.751.815.192.444.963.081.81

2.103.584.295.464.123.585.933.875.434.512.883.504.293.881.86

"Study means are averages of 107 subjects' perceptions of the following measures: attitude toward the extension, the average ofperceived overall quality of the extension (1 = inferior, 7 = superior) and likelihood of trying the extension (1 = not at all likely,7 = very likely); QUALITY, the overall quality of the original brand (1 = inferior, 7 = superior); TRANSFER, the usefulness ofmanufacturing skills and resources in the original product class for making the extension product (1 = not at all helpful, 7 = veryhelpful); SUBSTITUTE, substitutabillty of the original and extension products in use (1 = low, 7 = high); COMPLEMENT, com-plementarity of the original and extension product classes in use (1 = low, 7 = high); and DIFFICULT, perceived difficulty indesigning and making the extension (1 = not at all difficult, 7 = difficult).

mouthwash. The mouthwash extension has a muchhigher quality rating than the gum extension, despitethe fact that in both cases there are many "good forteeth" associations. However, a larger incidence ofdamaging "bad taste" or "taste like toothpaste" as-sociations is found for the gum (n = 25) than for themouthwash (n = 10). The Crest taste thus seemed tobe acceptable, or perhaps even an asset, in the mouth-wash context but a liability in the gum context.

Another interesting illustration of the transfer ofattributes to an extension involves Heineken popcornand Vuamet skis. These two brands both received veryfavorable perceived quality ratings for their originalproducts and both extensions had very high comple-mentarity ratings but low marks on substitutability andtransfer. Yet, Vuamet skis received much more fa-vorable attitude ratings than Heineken popcorn. Thefact that some subjects perceived popcorn to be easyto make contributed to the negative evaluation ofHeineken popcorn, but it also could have been causedby the negative association of beer taste mentioned bysome subjects. The Vuamet name, in contrast, had arather remarkable ability to be exported to other prod-

uct classes. In this case, complementarity may haveled to an inference that the extension would have the"stylish" attribute associated with the Vuamet name,and this attribute was valued in the different extensioncontexts.

Two additional points should be made about thequalitative analysis of the associations. First, somebrands received relatively few associations (e.g., Crestshaving cream, Haagen-Dazs popcom, Vidal Sassoonsuntan lotion, and Vuamet sportswear). For thesebrands, consumers may simply have had difficultyvisualizing what the extension should be like, perhapsbecause the brand name associations were not veryrelevant to the new product class. Second, some ex-tensions received rather pronounced categorization thatallowed subjects' responses to be divided into clearlydefined segments. For example, some subjects asso-ciated Vidal Sassoon sportswear with trendy, fashion-able, or high status designer labels (though perhapspartly because of the jeans line with a similar name).Other subjects felt that Sassoon should stick to haircare and found the idea of Sassoon sportswear tacky,unoriginal, and unappealing.


TABLE 3Summary of Coded Brand Associations for Brand Extensions:

Number of Respondents Mentioning Item"

1. McDonald's Photo Processing (2.03)Stick to food/no credibilityLow qualityFastWould not useCheap

3. Heineken Wine (2.91)Beer and wine are bad comboLow or bad qualityNot much experience/stick to beerGood quality/good nameExpensive

5. Vidal Sassoon Perfume (3.24)Smells like shampooGood quality/smells goodBad scentAll are alikeLow/medium quality

7. Haagen Dazs Popcorn (3.28)Bad mix/stick to ice creamExpensiveFlavors/sweet/richAll popcorn is the sameHigh quality

9. Crest Chewing Gum (3.43)Prevents cavitiesGood for teethBad tasteSugarlessGood taste

11. McDonald's Theme Park (3.56)For kids/familiesMcDonald's charactersFunStupid/silly/awfulFood

13. Vuarnet Wallets (3.78)ExpensiveHigh qualityFashionable/stylishSportyNot leather/velcro

15. Vidal Sassoon Sun Lotion (3.98)ExpensiveHigh qualityNo technical knowledgeFashionable/glamorousLow quality

3329201615

322417128

1717171413

3123191613

3726252317

5629141411

3321241515

19151088

2. Heineken Popcorn (2.30)Popcorn and beer don't mixAll popcorn is the sameUnappetizing/bad ideaTastes like beerGoes with beer

4. Haagen Dazs Cottage Cheese (3.13)High qualityGood tasteBad associationsStick to ice cream/inconsistentAll the same

6. Crest Shaving Cream (3.26)Same as othersStick with toothpasteGood productGood qualityReasonably priced

8. McDonald's Frozen French Fries (3.37)Good qualityNot as good as real thingBad/grossGreasyConvenient

10. Vidal Sassoon Sportswear (3.48)Stylish/trendy/fashionableNo appeal/would not buyLow qualityExpensiveStick to hair care

12. Vidal Sassoon Skin Cream (3.63)Trust because of reputationHigh qualityExpensiveScentedStick to hair care

14. Vuarnet Skis (3.91)High qualityExpensiveTrendy/fashionableNo technical knowledgeStick to established manufacturer

16. Vuarnet Watches (3.78)Trendy/fashionableExpensiveLike SwatchSportyHigh quality

2620181514

1715151211

202015109

292519118

2119291612

262120127

3332251211

4035351715

l\/lodeling Consumer Evaluationsof Brand Extensions

The qualitative analysis provides support for the im-portance of some of the constructs thought to affectconsumers' evaluations of brand extensions (e.g.,perceived product class fit and perceived difficulty ofmaking the product extension). To address these ef-fects more formally and explore the role of perceived

quality, we estimated a regression model motivatedby the four hypotheses.

The dependent variable was attitude toward theextension, operationalized by the average of the per-ceived quality of the extension and the likelihood oftrying the extension measures. The use of two indi-cators provided a more reliable measure of the attitudeconstruct, as the correlation between the two was .67,


TABLE 3 (continued)Summary of Coded Brand Associations for Brand Extensions:

Number of Respondents Mentioning Item"

17. Heineken Light Beer (4.76)Good tasteHigh qualityFewer caloriesExpensiveEuropean/imported

19. Crest IVIouthwash (4.86)Good like toothpasteFights cavitiesHigh quality/works wellTastes goodTastes bad

3432242216

3523181110

18. Haagen Dazs Candy Bar (4.81)Tastes goodExpensiveWorth tryingChocolateGood quality

20. Vuarnet Sportswear (5.15)ExpensiveStylish/trendy/fashionableHigh qualitySportyStatus symbol

3223171614

4844321912

"Numbers in parentheses are the extension attitude ratings. Associations and ratings are based on a sample of 107 undergraduatebusiness students.

TABLE 4A Regression Model of the Attitude Toward the Extension

Independent Variable

StandardizedRegressionCoefficient

RegressionCoefficient t-vaiue

QUALITY (perceived quality of original brand)TRANSFER (transfer of skills/assets from original to extension

product class)COMPLEMENT (degree to which the two product classes are

complements)SUBSTITUTE (degree to which the two product classes are

substitutes)QUALITY X TRANSFERQUALITY X COMPLEMENTQUALITY X SUBSTITUTEDIFFICULT (difficulty of making extension)Sample size = 2140Adjusted ^ = .26

-.01

.15

.02

- .01

.12

-.02

- .1

2.0

- . 4

.08

.12

.25

.18

.12

-.06.02.03.02.12

-1.01.43.22.16.2

suggesting a reliability of .79. Further, when separateregressions were run for each, none of the regressioncoefficients were significantly different.

The independent variables follow the four hy-potheses and are listed in Table 4. The first variableis the perceived quality of the original brand, QUAL-ITY, from H,. Next are the three fit variables,TRANSFER, COMPLEMENT, and SUBSTITUTE,from H3. The following three terms reflect the inter-actions of the three fit variables with the perceivedquality variable, from H2. The final variable is the

'Because the analysis is conducted across subjects and across ex-tensions, a confounding scaling effect may be present. If some sub-jects tended to use the upper part of the 7-point scale on both theindependent and dependent variables and others tended to use the lowpart of the scale, even if there were no relationship for an individual,a pooled, aggregate analysis would show a significant effect betweenthe variables. The source of this effect, however, would be spurious.To check this possibility, a main-effects-only model was run with datastandardized within each subject (i.e., each subject's mean was sub-tracted for each specific value for the dependent and independent vari-ables, and this difference was divided by the standard deviation). The

perceived difficulty of making the extension, DIFFI-CULT, from H4. The regression was run over the 107subjects and the 20 extensions, making a sample sizeof 2140.' The resulting standardized regression or betacoefficients, regression coefficients, and t-statistics arereported in Table A?

Perceived brand quality. The beta coefficient forthe QUALITY variable is essentially zero, indicating

resulting regression coefficients are essentially identical to those foundfor the raw data, ruling out the possiblity that scaling effects confoundthe fmdings.^ h e correlation matrix for the Table 2 variables is based on the re-sponses of 107 subjects for 20 extensions.

QUAL- TRANS- COMPLE- SUBSTI- DIFFI-ITY FER MENT TUTE CULT

Extensionattitudes

QUALITYTRANSFERCOMPLEMENTSUBSTITUTE

.29 .34.08

.31

.19

.27

.25

.14

.44

.21

.13

.00- . 0 7

.16- . 0 9

Consumer Evaluations of Brand Extensions / 35

that, in contrast to Hj, there is no direct link fromperceived quality of the brand to the attitude towardthe extension.

Perceived product class fit. The beta coefficientsfor two of the fit variables, COMPLEMENT andSUBSTITUTE, are not significant. However, the betacoefficient for TRANSFER, the extent to which theassets and skills of a firm making the product classassociated with the original brand name could be ap-plied to making the extensions, is substantial (.15) andsignificant (p < .05). Hence, the direct association offit with the attitude toward an extension, H3, is foundfor only one of the three fit variables, TRANSFER.

Model interactions. The beta coefficients for theinteractions of QUALITY with COMPLEMENT andSUBSTITUTE are both substantial (.25 and .18, re-spectively) and significant (p < .01 and p < .05, re-spectively), as in H2. The beta coefficient for the in-teraction of QUALITY with TRANSFER (.12),however, is not significant (p > .15). The model re-sults suggest that high perceived quality for the orig-inal brand name is related to acceptance of the brandextension only when fit based on complementarity orsubstitutability is present. Thus, picking an extensionthat is complementary or a substitute may not over-come low perceived brand quality.

The ability of a firm in the original product classto make the extension by applying current skills orassets, TRANSFER, has primarily a direct relation-ship. Hence, it may detract from the attractiveness ofan extension even in the presence of high levels ofperceived quality for the core brand. Perhaps thistransfer of skills and assets establishes credibility inmaking the extension that is necessary independent ofopinions of the original's brand quality.

Which fit variable? Of the three fit variables pro-posed and explored in this study, which are the mostuseful? The results suggest that the TRANSFER andCOMPLEMENT fit variables are more important inexplaining variance in extension attitudes than theSUBSTITUTE variable. In a main-effects-only model,where the interactive effects are omitted so that bothdirect and indirect effects of the three fit variables canbe summarized, the standardized regression coeffi-cients for TRANSFER, COMPLEMENT, and SUB-STITUTE are respectively .24, .17, and .08 (all dif-ferences significant at the .01 level). Moreover, if thetwo beta weights involving TRANSFER and COM-PLEMENT are added, they total .27 for each. In con-trast, though SUBSTITUTE has an interactive effectwith perceived quality, because of the negative coef-ficient for the main effect, the two beta weights in-volving SUBSTITUTE only total . 10. In this setting,COMPLEMENT and TRANSFER appear to be themost important fit variables.

COMPLEMENT and TRANSEER—are both need-ed? If complementarity and the ability of a firm'smanufacturing skills and assets to transfer from oneproduct context to another are the fit variables moststrongly related to extension attitudes, what is the ef-fect of their combination? When a COMPLEMENT/TRANSFER interaction term is added to the model,the beta is - .17 and statistically significant (t = -2 .8 ,p < .01) and the other coefficients remain basicallythe same. This negative relationship suggests that fiton one of the two variables is adequate—little is gainedby having a fit on both dimensions.

Perceived difficulty of making the extension. Thedifficulty of making the extension, DIFFICULT, hasa significant beta of .12, which is significantly higherthan the SUBSTITUTE variable. Hence, H4, that anextremely easy-to-make extension is on average lesslikely to be accepted than other extensions, is sup-ported. Two explanations seem reasonable. First,consumers may feel it is incongruous to introduce aquality brand name in a trivial product class. Second,the association of a quality name with an easy-to-makeproduct class may suggest to consumers the likelihoodof an overpriced product.

Study 2. ExtensionPositioning Study

Research Issues

The stimulus presentation for the brand extensions instudy 1 was very terse; the only information providedwas the brand name and product class. The reactionstherefore were to extension concepts prior to any in-troductory marketing campaign. During the actuallaunch of a new extension, a firm can influence theperception of an extension by providing informationcues through advertising and other marketing mix ac-tivities. The purpose of study 2, the extension posi-tioning study, was to pursue the fifth research ques-tion: How are consumer evaluations affected whendifferent types of information are provided in the ex-tension context?

The study 1 findings that favorable brand qualityperceptions are related to favorable extension evalu-ations and that low extension evaluations can be causedby transferred attribute associations suggest two gen-eral strategies. One approach is to provide a cue toconsumers on the general quality of the original brand,so that positive aspects of the brand will be more sa-lient and negative elements less salient in the newcontext. For example, if consumers are cued and re-minded that Crest chewing gum is from a leader in


the control of cavities and tartar, positive elements ofthe extension may be more salient (e.g., that the gumwill help in dental hygiene). A second approach is toelaborate on a brand extension attribute to inhibit thedevelopment of any potentially negative beliefs thatconsumers may infer. For example, stating that Crestchewing gum is available in spearmint and pepper-mint flavors should reduce the probability that con-sumers will think the gum has a taste like toothpaste.Study 2 examined both of these approaches.

Method

The method in terms of measures, procedures, andsubjects was essentially the same as that used in study1, though the 121 students who participated came froma different semester's course offering. Only four low-rated brand extensions, however, were used as stim-uli: McDonald's photo processing, Heineken pop-com. Crest chewing gum, and Vidal Sassoon per-fume.

The design was a repeated measures, 2 x 2 fac-torial design with two between-subject factors, eachwith two levels, (1) original brand quality cue (presentor absent) and (2) brand extension attribute elabora-tion (present or absent). The cues referred to the highquality and leadership status of the original brand; theelaborations were brief, neutral descriptions of an at-tribute about which subjects in study 1 had expressedsome uncertainty and concern (see Table 5 for thespecific cues and elaborations used). Subjects wereassigned to one of four groups, each of which saw thesame type of information for all four extensions.

Group 1. No quality cues or attribute elaborations

Group 2. Quality cues

Group 3. Attribute elaborations

TABLE 5Description of Cues and Elaborations

Original Brand Quality Cue ConditionMcDonald's photo processing: From the providers

of fast, inexpensive, and convenient service.Crest chewing gum: From a ieader in the control of

cavities and tartar.Heineken popcorn: From the makers of a high

quality premium beer.Vidal Sassoon perfume: From the makers of high

quality personai care products.

Brand Extension Attribute Elaboration ConditioniVIcDonaid's photo processing: Physicaiiy separated

from the food service and using a weil-estabiished camera retaiier to process the fiim.

Crest chewing gum: in spearmint and peppermintfiavors.

iHeineken popcorn: in reguiar and cheddar cheesefiavors.

Vidai Sassoon perfume: With a subtle, but sensualfragrance.

Group 4. Quality cues and attribute elaborations

The fourth group provided a test of interactive ef-fects between cues and elaborations (i.e., will the twotypes of information have an additive impact in com-bination?). In additon to evaluating the four test ex-tensions, each group also considered two filler exten-sions, Vuarnet skis and Haagen-Dazs candy bar. Thetwo main dependent measures were perceived qualityand likelihood of trying the extension, which againwere averaged; the main independent variables mea-sured were the perceived quality of the original brandand the fit measures.

Results

The mean values for the attitude toward the extension,as operationalized by the average of the perceivedquality and intent to try scales, are summarized in Ta-ble 6. Though the general pattern across the four ex-tensions is similar to the corresponding ratings fromstudy 1, the mean values for study 2 are somewhathigher. This difference may reflect a contrast effectof the different contexts: study 1 included other morefavorably rated extensions, providing a more positivecontext, whereas the stimuli in study 2 were weightedheavily toward extensions that were not well received.

An analysis of variance was conducted with theextension attitude measure to test the main and inter-action effects of the cue and elaboration factors. Thoughthe main effect of elaboration is significant (F = 4.86,p < .03), the main effect of cue and the interactioneffect of the cue and elaboration are not significant (F< 1).

Table 6 indicates rather clearly that the elaborationwas very effective in affecting the attitude toward theextension whereas the use of a quality cue was not.The elaboration impact was strong for the aggregateand all of the test brands with the exception of VidalSassoon perfume. One explanation for the VidalSassoon case is that the elaboration ("a subtle, butsensual fragrance"), which was developed by drawingon words used in other perfume ads, was not adequateto expunge the shampoo odor association, especiallywithout strong visual presentation. The overall con-clusion, though, is that to overcome the potentialtransfer of negative associations for a proposed ex-tension, the concept should be elaborated in such away that it appears inconsistent with the potentiallydamaging attribute.

To gain more insight into how the manipulationsaffected brand evaluations, two main effects regres-sion analyses were conducted with QUALITY, COM-PLEMENT, SUBSTITUTE, and TRANSFER as in-dependent variables. The first analysis compared theregression coefficients for the no-elaboration groups

Consumer Evaluations of Brand Extensions / 37

TABLE 6Extension Attitude Means for Study

Brand Extension

McDonald's photo processingCrest chewing gumHeineken popcornVidal Sassoon perfunneAverageStandard deviationSample size

No Cue orElaboration

2.674.653.363.813.621.58

31

CueOnly

2.624.423.243.633.481.58

31

ElaborationOnly

3.104.833.873.923.931.59

29

Cue andElaboration

3.124.973.603.783.871.54

30

'Extension attitudes are the average of perceived extension quality (1 = inferior, 7 = superior) and likelihood of trying the extension(1 = not at all likely, 7 = very likely).

(1 and 2) and the elaboration groups (3 and 4),The standardized beta for TRANSFER is signifi-

cantly higher (p < .01) in the no-elaboration groups,.32, than in the elaboration groups, .11. The presenceof a brand extension attribute elaboration appears toreduce the importance of the fit issue, inhibiting theemergence of negative beliefs. If a potentially dam-aging brand attribute is present, a lack of fit accen-tuates its effect. If this attribute is neutralized, how-ever, fit is less an issue in the sense that the consumeris less likely to base evaluations on a perceived lackof fit.

The second analysis compared regression coeffi-cients for the no-cue groups (1 and 3) and the cuegroups (2 and 4). The standardized beta is not sig-nificantly higher for QUALITY in the cue groups, norare the betas significantly different for the other vari-ables (p > .20). Combined with the ANOVA results,the lack of a cue effect suggests that favorable qualityperceptions must have been salient to no-cue subjects.This favorable image, however, could not offset thefact that the relationship with the original brand'sproduct class cued undesirable associations such astoothpaste taste, greasy food, beer taste, or shampooscent in the extension context. Thus, in this researchsetting, a more effective introductory strategy for thebrand extension is to protect it from negative associ-ations, rather than attempting to reinforce positive as-sociations about the original brand.

Summary of Main FindingsOur studies not only shed light on how consumersevaluate brand extensions, but also are relevant to thebroader research objective of understanding how brandnames and brand associations are used by consumersin their purchase decisions. The findings are based ona limited set of brands and hence generalization be-yond that set should be made with caution. Further,as study 1 provides only correlational data, the strengthof its implications is limited. With these qualifyingstatements in mind, we offer the following five ob-

servations about consumer evaluations of brand ex-tensions based on the research findings from the twostudies.

1. Inferred attribute beliefs both enhanced and harmed theevaluations of a brand extension. For example, thequalitative associations suggest that the Crest taste wasan asset for a mouthwash extension but a liability fora gum extension. Inferred beliefs associated with un-favorably evaluated extensions often were for concreteproduct class attributes, such as the taste of toothpasteor beer. Inferred beliefs associated with favorably eval-uated extensions often were for abstract brand attri-butes, such as style.

2. Subjects' perceptions of the quality of the original brand,QUALITY, and the relationship or "fit" between theoriginal and extension product classes had an interac-tive effect on evaluation of an extension. The relation-ship of a positive quality image for the original brandwith the evaluation of a brand extension was strongonly when there was a basis of fit between the twoproduct classes.

3. The three dimensions of fit between the original andextension product classes were the perceived applica-bility of the skills and assets of a competent manufac-turer in the original product class for making the prod-uct extension (TRANSFER), the perceived product classcomplementEirity (COMPLEMENT), and the perceivedproduct class substitutability (SUBSTITUTE). TheCOMPLEMENT and SUBSTITUTE fit measures in-teracted with the perceived quality of the original brandto predict brand extension evaluations, but TRANS-FER had primarily a direct impact on the evaluations.Overall, TRANSFER and COMPLEMENT were moreimportant as predictors than SUBSTITUTE, and therewas evidence of a negative interaction between thosetwo fit variables. Thus, a fit on either TRANSFER orCOMPLEMENT may be adequate; a good fit on bothis not necessary.

4. Subjects' perceptions of the difficulty of making theextension (DIFFICULT) had a positive relationship withevaluations of an extension, supporting the hypKJthesisthat an extremely easy-to-make extension, on average,is less likely to be accepted. Consumers may attributethe act of placing a quality brand into what is viewedas a trivially easy-to-make product class as a blatanteffort to capitalize on a brand name image to commandhigher than justified prices or they may feel it is in-congruous to introduce a quality brand name in a trivialproduct class.


5. In study 2, cueing subjects about positive qualities ofthe original brand did not affect evaluations for exten-sions that had low evaluations in study 1. Providing abrief elaboration of an extension attribute about whichsubjects may have been uncertain and which had thepotential to damage the extension, however, led to morefavorable extension evaluations. Because generally well-known and well-liked brands were used, reminders ofquality evidently were unnecessary. The elaboration,by clarifying the nature of an important attribute, ap-peared to be effective in inhibiting the transfer of neg-ative associations. It also appeared to reduce the sa-lience of perceived credibility of a finn in the originalproduct class in making the extension.

Implications and FutureResearch Directions

Predicting Attribute Transfer to ExtensionsWhat types of brand attributes and contexts will resultin the "export" of an attribute to an extension? Thedistinction between concrete attributes, those definedby physical, tangible product characteristics, and ab-stract attributes, those involving intangible productperceptions (e.g., Johnson and Fomell 1987; Myersand Shocker 1981; Olson and Reynolds 1983; ZeithamI1988), may be helpful. One could hypothesize that anabstract attribute, such as style associated with Vuamet,might be transferred to a broader set of product classesthan concrete attributes, which usually are associatedwith specific product classes.

The Process Generating the Fit andAttitude Judgments

Does a perception of poor fit between two productclasses contribute to the transfer of negative attributesto the extension or low extension evaluations, or doesthe presence of a negative evaluation or negative as-sociation inferred about the extension lead to a per-ception of poor fit? What is the critical problem foran extension, the perception of poor fit or the asso-ciation with a negative attribute? Understanding themediating role of fit is an important research priority.

Dimensions of Fit

The conclusions about the roles of the three fit vari-ables should be explored with other stimuli and in othercontexts. There may be conditions under which theirrelative roles are affected. For example, for abstractattributes, complementarity may have a more impor-tant role in fit judgments (e.g., as with Vuamet). Forconcrete attributes, however, substitutability andmanufacturing credibility may be the more relevant fitmeasures. Other fit conceptualizations also may war-rant exploration, for example, shared attributes or fea-tures such as in the contrast model (Johnson 1986;Tversky 1977; Tversky and Gati 1982).

Extending a Brand to a Relatively TriviaiProduct Class

Our findings suggest that consumers may not alwaysaccept the extension of a high quality brand to a prod-uct class that is by comparison trivial or very easy tomake, even if fit is good. We advanced two hy-potheses: (1) the extension is perceived to be pricedtoo high and (2) the combination of a high qualitybrand with an easy-to-make product is considered in-congruent or exploitative. Future research could pro-ductively explore the conditions under which each hy-pothesis might hold.

Positioning Strategies for Extensions

In study 2, elaborating attributes of extensions is foundto be a more effective way to neutralize undesirablebrand associations than using original brand qualitycues. In addition to replicating and testing this find-ing, future research should explore other strategies.Can any other cues possibly enhance extension eval-uations? For example, perhaps fit perceptions can beinfiuenced by references to manufacturing capabilityor complementarity. Can cases be found and studiedin which such positioning strategies affected the suc-cess of the extension?

Prototypicality of the Brand

Will a brand that is considered prototypical (a goodexample) of a product class be particularly difficult toextend because of its strong product class association?Tauber (1981) discusses the risk of extending brandnames such as Kleenex, Scotch tape, or Band-Aid thatare closely associated with a product. However, anassociation between brand prototypicality and per-ceived quality was observed in our study and others(e.g., Nedungadi and Hutchinson 1985; Ward andLoken 1988). In study 1, a measure of prototypicality(how good an example of the product category is thebrand, on a 7-point scale?) had a correlation of .34with perceived quality (QUALITY).

Branding Strategies Across MultipleProduct Classes

Research is needed that explores more complexbranding strategies. How will consumers respond toextensions of brands already associated with multipleproduct classes, such as Heinz or General Electric?How do various combinations of product class asso-ciations affect brand associations? When does an um-brella brand strategy rather than distinct brand namesmake sense?

The Role of Involvement in Extending Brands

What is the effect of involvement on brand exten-sions? When the motivation or ability to process in-


formation is low, consumers might be expected to relyon perceived brand quality and brand familiarity as aperipheral cue in their brand evaluations (Baker et al.1986; Petty and Cacioppo 1986). However, high in-volvement contexts may involve higher risk and thusconsumers may need the reassurance of an establishedbrand name. MacKenzie and Lutz' (1989) model ofthe antecedents of and relationship between attitudetoward the ad (AAd) and attitude toward the brand (AB)may have some theoretical relevance to understandingthis issue.

Reciprocal Impact of Brand Extension

A critically important strategic issue is the impact ofthe extension on the original brand. It can be positive.For example, promotions for Sunkist extensions werethought to have enhanced both name recognition andassociations with good health and vitality (Kesler 1987).

An extension may, however, cannibalize the sales ofthe original brand or damage its image by creatingnew associations or by confusing the current ones (Riesand Trout 1981). For example. Miller Lite may havedamaged Miller High Life because of the light beerassociations (Tauber 1981).

This reciprocal impact is especially important invertical extensions in which an upscale or downscaleversion of the brand is introduced. Several questionsarise. What impact will downscale extensions have onthe original brand? Can positive effects be stimulatedor negative effects be minimized? Conversely, whatimpact will an upscale extension have on the originalbrand? How readily will consumers accept such anextension? Park, Jaworski, and Maclnnis (1986) pro-vide a framework for managing brand image over timethat may be useful in addressing these questions.

REFERENCESAaker, David A. (1982), "Positioning Your Product," Busi-

ness Horizons, 25 (May/June), 56-62.Anderson, John R. (1983), The Architecture of Cognition.

Cambridge, MA: Harvard University Press.Anderson, Norman H. and Stephen Hubert (1963), "Effects

of Concomitant Verbal Recall on Order Effects in Person-ality Impression Formation," Journal of Verbal Learningand Verbal Behavior, 2 (5-6), 379-91.

Baker, William, J. Wesley Hutchinson, Danny Moore, andPrakash Nedungadi (1986), "Brand Familiarity and Adver-tising: Effects on the Evoked Set and Brand Preference,"in Advances in Consumer Research, Vol. 13, Richard J.Lutz, ed. Provo, UT: Association for Consumer Research,637-42.

Bierley, Calvin, Frances K. McSweeney, and Renee Van-nieuwkerk (1985), "Classical Conditoning of Preferencesfor Stimuli," Journal of Consumer Research, 12 (Decem-ber), 316-23.

Brown, Paul (1985), "New? Improved?" Business Week (Oc-tober 21), 108-12.

Carlston, Donald E. (1980), "The Recall and Use of Traitsand Events in Social Inference Processes," Journal of Ex-perimental Social Psychology, 16 (July), 303-28.

Cohen, Joel B. and Kunal Basu (1987), "Alternative Modelsof Categorization: Toward a Contingent Processing Frame-work," Journal of Consumer Research, 13 (March), 455—72.

Fiske, Susan T. (1982), "Schema-Triggered Affect: Applica-tions to Social Perception," in Affect and Cognition: The17th Annual Carnegie Symposium on Cognition, MargaretS. Clark and Susan T. Fiske, eds. Hillsdale, NJ: LawrenceErlbaum Associates, 55-78.

and Mark A. Pavelchak (1986), "Category-BasedVersus Piecemeal-Based Affective Responses: Develop-ments in Schema-Triggered Affect," in The Handbook ofMotivation and Cognition: Foundations of Social Behavior,Richard M. Sorrentiono and E. Tory Higgins, eds. NewYork: Guilford Press.

Fry, Joseph, N. (1967), "Family Branding and Consumer BrandChoice," Journal of Marketing Research, 4 (August), 237-47.

Garvin, David A. (1984), "Product Quality: An ImportantStrategic Weapon,"BM.sine5s//orizon5, 27 (May-June), 40-3.

Gilovich, Thomas (1981), "Seeing the Past in the Present: TheEffect of Associations to Familiar Events on Judgments andDecisions," Journal of Personality and Social Psychology,40 (May), 797-808.

Heider, Fritz (1958), The Psychology of Interpersonal Rela-tions. New York: John Wiley & Sons, Inc.

Henderson, James M. and Richard E. Quandt (1980), Micro-economic Theory: A Mathematical Approach. New York:McGraw-Hill Book Company.

Holbrook, Morris B. and Kim P. Corfman (1985), "Qualityand Value in the Consumption Experience: Phaedrus RidesAgain," in Perceived Quality, Jacob Jacoby and Jerry Ol-son, eds. Lexington, MA: Lexington Books, 31-57.

Jacobson, Robert and David A. Aaker (1987), "The StrategicRole of Product Quality," Journal of Marketing, 51 (Oc-tober), 31-44.

Johnson, Michael D. (1986), "Consumer Similarity Judg-ments: A Test of the Contrast Model," Psychology andMarketing, 3 (Spring), 47-60.

and Claes Fomell (1987), "The Nature and Meth-odological Implications of the Cognitive Representation ofProducts," Journal of Consumer Research, 14 (Septem-ber), 214-28.

Kesler, Lori (1987), "Extensions Leave Brand in New Area,"Advertising Age (June 1), SI.

Lingle, John H. and Thomas M. Ostrom (1979), "RetrievalSelectivity in Memory-Based Impression Judgments,"Journal of Experimental Social Psychology, 23 (March),93-118.

MacKenzie, Scott B. and Richard J. Lutz (1989), "An Em-pirical Examination of the Structural Antecedents of Atti-tude Toward the Ad in an Advertising Pretesting Context,"Journal of Marketing, 53 (April), 48—65.


McSweeney, Frances K. and Calvin Bierley (1984), "RecentDevelopments in Classical Conditioning," Journal of Con-sumer Research, 11 (September), 619-31.

Morein, J. (1975), "Shift From Brand to Product Line Mar-keting," Harvard Business Review, 53 (September-Octo-ber), 56-64.

Myers, James H. and Allan D. Shocker (1981), "The Natureof Product-Related Attributes," in Research in Marketing,Vol. 5, Jagdish Sheth, ed. Greenwich, CT: JAI Press, Inc.,211-36.

Nedungadi, Prakash, and J. Wesley Hutchinson (1985), "ThePrototypicality of Brands: Relationships With BrandAwareness, Preference, and Usage," in Advances in Con-sumer Research, Vol. 12, Elizabeth C. Hirschman andMorris B. Holbrook, eds. Provo, UT: Association for Con-sumer Research, 498-503.

Neilsen, A. C. (1985), Testing Techniques, Vol. 1, No. 1.Neuhaus, Colin F. and James R. Taylor (1972), "Variables

Affecting Sales of Family-Branded Products," Journal ofMarketing Research, 14 (November), 419-22.

Olshavsky, Richard W. (1985), "Perceived Quality in Con-sumer Decision Making: An Integrated Theoretical Per-spective, in Perceived Quality, Jacob Jacoby and Jerry Ol-son, eds. Lexington, MA: Lexington Books, 3-29.

Olson, Jerry C. and Thomas J. Reynolds (1983), "Understand-ing Consumers' Cognitive Structures: Implications for Ad-vertising Strategy," in Advertising and Consumer Psychol-ogy, L. Percy and A. Woodside, eds. Lexington, MA:Lexington Books.

Osgood, C. E. and P. H. Tannenbaum (1955), "The Principleof Congruity in the Prediction of Attitude Change," Psy-chological Review, 62 (1), 42-55.

Park, C. Whan, Bernard J. Jaworski, and Deborah J. Maclnnis(1986), "Strategic Brand Concept-Image Management,"Journal of Marketing, 50 (October), 135-45.

Petty, Richard E. and John T. Cacioppo (1986), Communi-cation and Persuasion. New York: Springer-Verlag.

Read, Stephen J. (1983), "Once is Enough: Causal ReasoningFrom a Single Instance," Journal of Personality and SocialPsychology, 45 (August), 323-34.

Ries, Al and Jack Trout (1981), Positioning: The Battle forYour Mind. New York: McGraw-Hill, Inc.

Riskey, Dwight R. (1979), "Verbal Memory Processes inImpression Formation," Journal of Experimental Psychol-ogy: Human Learning and Memory, 5 (3), 271-81.

Srinivasan, V. (1979), "Network Models for Estimating Brand-Specific Effects in Multi-Attribute Marketing Models,"Management Science, 25 (January), 11—21.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Eval-uation Strategies Mediating Consumer Judgments," Jour-nal of Consumer Research, 12 (June), 31—46.

Tauber, Edward M. (1981), "Brand Franchise Extensions: NewProducts Benefit from Existing Brand Names," BusinessHorizons, 24 (2), 36-41.

(1988), "Brand Leverage: Strategy for Growth in aCost-Controlled World," Journal of Advertising Research,28 (August/September), 26-30.

Tversky, Amos (1977), "Features of Similarity," Psycholog-ical Review, 84 (4), 327-52.

and Itamar Gati (1982), "Similarity, Separability,and the Triangle Inequality," Psychological Review, 89 (2),123-54.

University of Minnesota Consumer Behavior Seminar (1987),"Affect Generalization to Similar and Dissimilar Brand Ex-tensions," Psychology and Marketing, 4 (Fall), 225-37.

Ward, James and Barbara Loken (1988), "The Generality ofTypicality Effects on Persuasion and Comparison: An Ex-ploratory Test," in Advances in Consumer Research, Vol.15, Michael J. Houston, ed. Provo UT: Association forConsumer Research.

Wright, Peter L. (1975), "Consumer Choice Strategies: Sim-plifying vs. Optimizing," Journal of Marketing Research,11 (February), 60-7.

Wyer, Robert S., Jr. and Thomas K. SruU (1986), "HumanCognition in Its Social Context," Psychological Review, 93(July), 322-59.

Zeithaml, Valarie A. (1988), "Consumer Perceptions of Price,Quality, and Value: A Means-End Model and Synthesis ofEvidence," Journal of Marketing, 52 (July), 2-22.

Reprint No. JMS41101


consumer evaluations of brand extensions

Documents