Decomposed Similarity Measures in Brand Extensions

ABSTRACT - Recent research on consumer reactions to brand extension has mainly focused on the judgmental effects of similarity between an established brand and a brand extension. The present paper extends this research by investigating the effects of decomposed similarity. We find that similarities between an original brand and its brand extensions have to be measured using several items that better cover the similarity construct. We furthermore find that brand personality impacts brand extension evaluations, when there is a high degree of congruity between an individual’s self-image and salient personality traits of the brand extension.

Citation:

Leif E. Hem and Nina M. Iversen (2002) ,"Decomposed Similarity Measures in Brand Extensions", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 199-206.

Advances in Consumer Research Volume 29, 2002     Pages 199-206

DECOMPOSED SIMILARITY MEASURES IN BRAND EXTENSIONS

Leif E. Hem, Foundation for Research in Economics and Business Administration

Nina M. Iversen, Norwegian School of Economics and Business Administration

[We acknowledge the very stimulating and constructive inputs to this research by Professor Sharon Shavitt and her colleges at the UIUC. We would also like to thank the anonymous reviewers for helpful comments on earlier drafts of this paper.]

ABSTRACT -

Recent research on consumer reactions to brand extension has mainly focused on the judgmental effects of similarity between an established brand and a brand extension. The present paper extends this research by investigating the effects of decomposed similarity. We find that similarities between an original brand and its brand extensions have to be measured using several items that better cover the similarity construct. We furthermore find that brand personality impacts brand extension evaluations, when there is a high degree of congruity between an individual’s self-image and salient personality traits of the brand extension.

INTRODUCTION

Brand leveraging remains one of the most profitable and frequently employed marketing strategies. Extending brands enables firms to better tap into their substantial brand equity investment (Court, Leiter, and Loch 1999). For instance, in the 1950s Disney signified world-class animation. Today, Disney’s businesses include film, television, publishing, software, theme parks, hotels, cruises, and even an entire town (Celebration, Florida).

Past research on brand extension has addressed determinants of success or failure of extensions. An underlying assumption in most of this research is that brand affect product category similarity play important roles in determining success of brand extensions (e.g., Boush, et al. 1987; Aaker and Keller 1990; Boush and Loken 1991; Park, Milberg, and Lawson 1991; Herr, Farquhar, and Fazio 1996; Bottomley and Doyle 1996; Jun, Mazumdar, and Raj 1999). That is, evaluation of a brand extension is a joint function of how much the original brand is liked in its original category (the favourability of its reputation) and the level of similarity between the original brand and the extension category.

In spite of the great interest in research on brand extensions one has not yet covered all relevant similarity dimensions. To date the concept of similarity has been expanded to the awareness of its antecedents. In previous studies one has particularly focused on the role of brand characteristics like: 1) brand breadth (Boush and Loken 1991), 2) brand knowledge, and 3) brand specific associations (Broniarcyzk and Alba 1994). In most studies one has used overall measures of similarity, and in the few studies where distinct measures of similarity are utilized the dimensions tend to be chosen on an ad hoc basis. Consequently, the purpose of this study is to further explore the effects of decomposed dimensions of similarity between some existing original brands and some fictive brand extensions. We will address how various decomposed measures of similarity between an original brand and its brand extensions, depending on their relevance to the extension categories, impact judgments of the extensions. We will use decomposed similarity measures on functional, experiential, symbolic and contextual brand attributes or benefits (based on a framework by Keller 1993). We will furthermore address how various levels of congruity between the personality image of a brand extension and the self-image of the consumer conducting the judgment may impact the success of the extension. This is an alternative similarity measure to the direct measures of match between the original brand and the extension. The latter similarity dimension addresses matches between characteristics of the consumer and characteristics of the brand extension. To date this variable has not been used to explain effects of similarity on judgments of brand extensions.

In the following we first give a short review of previous findings on relations between similarity and brand extensions. Then we outline three hypotheses on some new relations between decomposed similarity dimensions and brand extensions. Thereafter, in testing of the hypotheses, we present the results from a quasi-experimental field study where ratings from a substantial sample of real consumers in a Norwegian marketplace setting are collected. The field study covers three existing original brands, which are extended to a total of 11 fictive extensions. The three original brands are one American brand (Ford automobile) and two Norwegian brands (Telenor telecom and Maarud snack). This particular research setting is also a contribution of the paper as most previous studies on brand extensions are conducted in the US, on student samples, where respondents mainly evaluate fictive brands. Finally, we discuss the results, limitations, and recommendations for future research.

SIMILARITY AND BRAND EXTENSIONS

Similarity is one of the most central theoretical constructs in psycology (Medin, et al. 1993). The concept of similarity is a component of many cognitive processes and theories of categorization and concept learning, and is used also in applied fields such as consumer behavior (see e.g., Bijmolt, et al. 1998 for an overview). Recent advances in theories of similarity have opened new avenues for consumer researchers to examine a range of empirical problems (see, Zhang and Fitzgerald 1997). To date consumer research has been highly influenced by the contrast model of Tversky (1977). This model has been utilized as a basis for constructing theories of product categorization (see, Cohn and Basu 1987), consumer learning, and brand extensions (Boush, et al. 1987; Chakravarti, et al. 1990; DubT, Schmitt, and Bridges 1992). Recent theoretical advances have extended the contrast model by addressing some of its imperfections (e.g., Medin, et al. 1990). In particular, researchers have questioned the assumption that lists of features alone can represent similarity between two items and that judgements of similarity simply requires direct feature mapping between the lists. It is proposed that similarity is better characterized as a comparison of structured representations (e.g., Falkenhainer, et al. 1990; Gentner and Markman 1994; Markman and Gentner 1993). The representations are composed of features (i.e. attributes) of objects, objects, and functions, and most importantly, the relations between these representational elements. Similarity is a function of both attributes and the relations of the attributes between a pair of items. In other words, similarity is more than Tversky’s (1977) feature-bundle view of products.

Similarity between an original brand and an extension category is perhaps the most essential criterion for success in brand extensions. Earlier research by Boush, et al. (1987) found that the greater the similarity between the original brand and the extension, the greater the transfer of positive but also negative affect to the new extension. In this study similarity was measured on one item, anchored by "Extremely dissimilar" and "Extremely similar" (Boush, et al. 1987: 232). More recent studies also support this finding (cf. Farquhar, et al. 1989; Bridges 1990; Herr, et al. 1990; MacInnis and Nakamoto 1990; Aaker and Keller 1990; Dacin and Smith 1994). [The research of Smith and Park (1992) is one exception. They did not find any significant effects of similarity on the evaluation of brand extensions. They explain the lack of support by low variability in the data set.]

Aaker and Keller (1990) found an equivalent impact of similarity on brand extension evaluations as was observed by Boush, et al. (1987). Currently the article by Aaker and Keller (1990) is seen as a "state of the art" in the field of brand extensions. In spite of the status of this study it is still only exploratory in identifying effects of similarity. Aaker and Keller (1990: 30) used three sources of similarity features and this was a great improvement from the widespread use of one-item measures. The three sources of similarity were (1) complement (the extent to which consumers view two product classes as complements), (2) substitute (the extent to which consumers view two product classes as substitutes) and (3) transfer (how consumers perceive relations in product manufacturing). Aaker and Keller (1990) do not give any theoretical arguments for using these particular measures of similarity.

Beside the basic similarity assumption, Park, et al. (1991) made another contribution to research on brand extensions. When consumers evaluate brand extensions, they consider information about the product-level feature similarity, as well as the concept consistency between the original brand concept and the extension. They found that extensions of original brands associated with a functional concept should be functional; extensions of original brands associated with a symbolic concept should be symbolic, etc. In this study similarity was measured on two items.

Recently, Herr, et al. (1996) introduced the intercategory relatedness construct, which is defined as the strength of the association between the original brand category and the target extension category. The relatedness of two product categores can depend on the similarity of common features (e.g., Boush, et al. 1987; Aaker and Keller 1990; Park, et al. 1991). Still, intercategory relatedness is a more inclusive construct than similarity (see Medin, et al. 1993; Rips 1989). Herr, et al. (1996: 149) measured intercategory relatedness by giving subjects real instructions suggesting that objects were related in a number of ways including similar features, substitution in use, or common-usage situations. Subjects were instructed to consider these points of relatedness when making their judgments. Herr, et al. (1996) argue that further studies of the broader construct of similarity is needed. Therefore, in this study we address the relations between decomposed dimensions of similarity and evaluations of brand extensions.

RESEARCH HYPOTHESES

The following discussion provides a rationale for two research hypotheses pertaining to the construct of perceived similarity.

Perceived similarity between an original brand and extension categories

Recent advances in theories of similarity have extended Tversky’s contrast model (e.g., Medin, et al. 1990). Researchers have specifically proposed that similarity is best characterized as a comparison of structured representations (e.g., Falkenhainer, et al., 1990; Gentner and Markman, 1994; Markman and Gentner, 1993). These new advances in theories of similarity are not implemented in the brand extension literature.

In initial research on brand extensions one typically measured similarity with single items. For example, Boush (1993: 71) used a single item to capture similarity. In the same way Boush and Loken (1991); Meyers-Levy, Louie, and Curren (1994); Sheinin and Smith (1994); and Smith and Andrews (1995) also used one item to measure the similarity between the original brand and the extension. Moreover, in the initial research on brand extension investigators used measures of similarity that only captured the overall dimension of the similarity construct (e.g., Park, et al. 1991: 190). Keller and Aaker (1992); John, et al. (1998); Keller and Sood (2001) also used solely overall measures of the similarity construct.

Despite the strong tendency to use single item and overall measures of similarity some exceptions exist. For instance Smith and Park (1992) measured similarity as commonalties in "needs satisfied", "usage situations", "component parts", and "manufacturing skills". Moreover, Broniarczyk and Alba (1994: 221) measured similarity as "the relevance of brand-specific associations to the extension categories".

To further illuminate effects of decomposed similarity we argue that an original brand and an extension can be linked on a whole range of dimensions. One approach used to define a brand is made by Keller (1993). In his framework Keller (1993: 7) divide the brand image in several dimensions such as brand associations, functionalbenefits (needs fulfilled), experiential benefits (sensory), symbolic benefits (belonging to a social group), user imagery, usage imagery, price, and packaging. Price and packaging could be difficult to measure since it is hard to evaluate when it comes to hypothetical brand extensions where no such information is given. Beside the framework made by Keller (1993), Aaker and Keller (1990) introduced one item covering the competence required to produce the extension. Furthermore, an overall measure of perceived similarity (see Boush, et al. 1987) could be included. Altogether these similarity dimensions capture the similarity construct more completely than single items or overall measures of similarity. We therefore believe that evaluations of brand extensions are affected positively or negatively depending on different similarity measures between an original brand and a brand extension. Therefore:

H₁: A more complete measure of similarity dimensions between an original brand and an extension predict more precisely the dimensions being crucial in evaluations of brand extensions

The role of self-image in evaluations of brand extensions

Most of the research on brand extensions has used a feature match approach similar to Tversky’s (1977) contrast model. This focus has limited researchers to only investigate the relations between the original brand and the extension (product category fit). Therefore researchers largely have overlooked the relations between consumers and their preferred brands. By so doing one has neglected the effects of similarity in terms of a match (level of congruity) between individual consumers’ self-image and the symbolic expressions of different brands (based on self-congruity theory). This type of similarity is better understood as congruity between an individual’s self-image and a brand’s personality traits. In this study we will apply recent research by J. Aaker (1995; 1997; 1999) to investigate how this type of congruity impacts evaluations of brand extensions. Thus, we are interested in impacts of similarity as a match between consumers’ self-image and the brand personality of an extension.

Researchers in marketing have become increasingly aware of the strategic importance of brand image (see e.g., D. Aaker and Biel 1993; Graeff 1997). Just as people can be described in terms of their personality traits, brands can be described in terms of their personality. Aaker (1997: 347) defines brand personality as "the set of human characteristics associated with a brand". According to Aaker creating a brand personality literally involves "personification" of a brand. She argues that brands, according to their associations with specific set of personality traits, are vehicles that consumers use for self-expression. To accommodate this need for self-expression marketers try to build brands imbued with strong personality traits that match consumers’ self-image. In one current empirical study Aaker (1999) investigated how such brands are evaluated when they possess a strong personality that may or may not match the personality of a consumer. Aaker (1999) found support for the notion that individuals who identify themselves on a particular personality dimension have a greater preference for brands that are highly descriptive on that dimension. Similarly, Vitz and Johnson (1965) found a relationship between smokers’ perceptions of cigarette image and the masculinity and femininity of the smoker. In another study, Dolich (1969) investigated the relationship between self-image and brand preferences and found that favored brands were consistent to self-concept and reinforced it. Ackoff and Emsoff (1975) studied four commercials created for four brands where each commercial was produced with the specific personality of the consumer in mind. The result indicates that most consumers select the brand that "fit" their personalities. This was true even though the four brands were identical beers except for the image created by the advertising. Thus, consumers prefer brands that are associated with a set of personality traits congruent with their self-image (Sirgy 1982). Moreover, brand preferences increase with increasing self-image congruity.

Keller (1993) argues that brands with a strong personality serve a symbolic function for consumers in expressing their actual or even ideal (aspirational) self-image (Sirgy 1982). Although the self-image concept includes more than two facets [Self-image has been considered a multidimensional construct including various types of selves, such as actual self, ideal self, social self, and sex-role self (Onkvist and Shaw 1987; Kleine, Kleine, and Kernan 1993). The facets of the self-image that are most studied are the actual self (how a consumer actually sees himself) and the ideal self (how a consumer would like to see himself).], these two are shown to have the most significant effect on brand evaluations (Sirgy 1982; 1985). High congruency implies that a particular brand personality trait matches a consumer’s actual or ideal self-image and vice versa. Aaker (1991) term consumers with a high match as "schematic" and consumers with a low "match" as "aschematic". Aaker hypothesize that individuals who are schematic versus aschematic on a particular personality dimension should have a greater preference for brands that are highly descriptive on that personality dimension. Thus, previous research has shown that the greater the congruency between the personality traits that describe an individuals actual or ideal self-image, and those that describe a brand, the greater the preference for the brand (e.g., Malhotra 1988; Sirgy 1982; 1985; 1986). Our question is whether the same logic is relevant for brand extensions. In other words, we do expect that consumers have a tendency to prefer brands extensions with personalities highly congruent with their own actual and ideal self-image. Therefore:

H_2a: The positive effect of perceived similarity on evaluations of brand extensions increases with increasing similarity between the personality of the brand extension and the consumers’ actual self-image.

H_2b: The positive effect of perceived similarity on evaluations of brand extensions increases with increasing similarity between the personality of the brand extension and consumers’ ideal self-image.

In this study we also include a variable named "brand reputation". Brand reputation is the perceived overall quality or favorability of a brand, and brand reputation is found to play a major role in evaluation of brand extensions (see e.g., Aaker and Keller 1990). In situations with high similarity it is logical that brands with a favorable reputation transfer positive affect to the extension whereas brands with a negative reputation transfer negative affect to the extension. Thus, to complete a model of similarity in brand extensions the reputation of the original brand is a key component.

METHODOLOGY

In order to test the hypotheses a quasi-experimental field research design was chosen. Stimuli: The original brands were selected with regard to the criteria of being relevant to subjects, highly familiar, and not broadly extended before. Established rather than fictitious brands were chosen. The reason for this was we that wanted to measure brand personality and brand image of established brands. Five brands were subjected to a pilot study in order to assess the degree to which they were highly familiar and associated with one (or a few) product category. Based on the results from this initial study, a brand was selected from each of the following categories: snacks (Maarud), cars (Ford), and telecommunications (Telenor). Maarud is the leading snack brand in Norway where the study was conducted. Ford has been one of the most selling automobile brands in Norway for more than fifty years. Telenor is the number one telecommunication company in Norway, and until recently it was the only telecom company. These three parent brands were leveraged to 11 hypothetical brand extensions, as can be seen in Table 1. The 11 selected extensions had to be reasonable and logical, but should also provide heterogeneity on similarity measures. (Based on a pilot study, the 11 hypothetical extensions were chosen among a sample of 20 extensions. The 11 chosen extensions varied from very similar (e.g., Ford motorbike) to not similar at all (e.g., Ford lawnmower).

Sample and data collection: A questionnaire was constructed for each of the three focal brands and administrated to subjects chosen from a population of a major Norwegian city. Respondents were contacted personally at randomly in their homes. Each respondent was asked to complete only one questionnaire for one of the three original brands. The respondents participated voluntarily and without any compensation. Still, 81 % of the individuals contacted agreed to participate in the study. The participants were given a short summary of the purpose of the study. In order to make the task more realistic they were told that the purpose of the study was to estimate consumers’ reactions to a number of planned brand extensions. Respondents were told that the researchers would collect the completed questionnaire the following day. This procedure yielded a response rate of 68.5 % of the contacted population (84.6 % of those who agreed to participate). Of the 760 questionnaires collected, only 59 questionnaires were removed due to major non-responses. In total a number of 701 questionnaires were completed satisfactorily. Together with the first author four students distributed the questionnaires. The first author conducted a preliminary training session of the interviewers. To ensure that the sample procedures were followed the interviewers started out distributing questionnaires together with the first author.

TABLE 1

THE THREE PARENT BRANDS AND THEIR BRAND EXTENSIONS

Measurement items

Dependent variable: (1) Overall evaluation of brand extensions: Subjects’ reactions to a proposed brand extension can be measured in many ways. It has typically been some form of overall evaluation of the proposed extensions (see e.g., Keller and Aaker 1992; Broniarczyk and Alba 1994; Keller and Sood 2001; Muthukrishnan and Weitz 1991). In this study we followed this practice and used three ites to measure overall attitudes towards the brand extensions. These items were measured on a six-point scale (see Table 2). Factor analysis revealed that all items loaded strongly on one factor (see Table 2, or confirm with authors).

Independent variables: (2) Similarity between the parent brands and the extensions. In order test hypotheses (H₁) similarity was measured on nine items. Some of the items were selected from the brand extension literature, and some were developed for the purpose of the study (based on Keller 1993). The similarity measures supposedly should capture decomposed aspects of similarity between the original brands and the brand extensions. Responses were given on six-point scales anchored by "not at all similar" and "very similar". Factor analysis revealed that all nine items loaded strongly on one factor (see Table 2).

(3) Similarity between the consumers’ self-image and the brands’ image: In order to test hypotheses H₂, similarity between the consumers’ self-image and the extension brands’ image was measured. Self-image/brand image similarity was measured on two Likert type six-point scales anchored as "not at all similar" to "highly similar". Factor analysis revealed one factor that captured self-image similarity (see Table 2).

(4) Reputation of the original brand: This variable was measured using six Likert type six-point scales. Factor analysis revealed one factor that captured the construct (see Table 2). Table 2 shows reliability measures (Cronbach’s alpha), eigenvalues, and total explained variance for a factor analysis capturing all the constructs and the items used to measure them.

RESULTS

Table 3 reports the descriptive statistics of the study variables. It shows the means, standard deviation, and number of respondents for each brand (Maarud snack, Ford automobile, and Telenor telecom samples).

The data shows that for all variables the standard deviations are greater than 1.04. When it comes to "reputation of the original brand", the measures are on average lower in the car sample than in the snack and the telecom samples.

Test of hypotheses

As illustrated in Table 4a and 4b, we examined the relations between the research variables by use of multiple regression. The regression models incorporate the main effects of the independent variables on the dependent variable (overall evaluation of the extensions). Table 4a and 4b, show that all models are highly significant and that they explain between 34% and 41% of total variance. This indicates a reasonable model fit. Also, he standardized regression coefficients indicate significant relations between the independent and the dependent variables.

Test of H₁

H₁ states that consumers evaluate brand extensions based on some specific similarity dimensions. This hypothese indicating that the different similarity dimensions to some extend have significant impact on the brand extension evaluations. Similarity between the original brands and the brand extensions is measured using nine items. Table 4a reports the results of the regression analysis. Table 4a show that similarity has a major impact on evaluations of the brand extensions. The overall similarity measure (item a in Table 2) and the production competence measure (item i in Table 2) are significant in all cases. The usage image (item b in Table 2) is significant in all cases, except for Maarud snacks. Brand associations (item c in Table 2) are significant in all cases, except for Ford automobiles. Sensory features (item f in Table 2) are significant for the total sample and for Maarud snacks (food). The same result is found for social/reference group (item g in Table 2). However, the similarity with the social/reference group is negatively related to the brand extension evaluation. Finally, the main decision criteria (item h in Table 2) are significant only for the total sample. Altogether, the regression models in Table 4a show that several of the decomposed similarity measures are significantly explaining the evaluation of brand extensions. Thus, H₁ is supported.

Test of H₂

H₂ postulates that consumers evaluate brand extensions more favorably when they associate themselves with the brand extensions’ image. Therefore, if the brand extensions’ image is similar to an individual’s self-image (both actual and ideal self-image) the evaluations of the extensions are higher. Self-image/brand extension image similarity is measured on two items, anchored by "Not at all similar" to "Very similar". Both actual and ideal self-image loaded on one factor, which made it hard to split it in two constructs. Table 4b reports the results of the regression analysis. A significant positive main effect of self-image/brand extension image similarity (actual and ideal self-image as one factor) on evaluation of the brand extensions is seen across all the samples (Total sample: b = .19, < .01; Maarud: b = .24, < .01; Ford: b = .20, < .01; Telecom sample: b = .17, < .01. Thus, H₂ is supported.

DISCUSSION

Past research on brand extension has traditionally focused on perceived similarities between original brands and extensions and has examined the effects of this factor on evaluations of brand extensions. This approach to research on similarity and brand extensions has some limitations. In this article we extend the understanding of the role of similarity in brand extensions by including some more decomposed measures of similarity. We furthermore test a new similarity dimension, namely the similarity between an individual’s self-image and the personality image of a brand extension. To our knowledge this dimension has not been investigated before. In a quasi-experimental field setting we find that high similarity between an original brand and an extension ought to be measured along multiple items that cover a variety of decomposed similarity dimensions. Our data show that measure of similarity between an original brand and an extension category should cover a complete set of decomposed similarity dimensions that at least include functional components, symbolic components, hedonic components (Keller 1993) and contextual components (Smith and Park 1992). The relative importance of these similarity measures should vary for different types of products. Furthermore, measures of decomposed similarity should particularly include matches between consumers’ self-image and the extended brands’ personality image. Conclusively, this study shows that a whole range of similarity dimensions may contribute in determining evaluations of brand extensions. More complete measures give better explanations of which singular similarity dimension is more crucial in determining success of failure of different brand extensions.

TABLE 2

MEASURE FROM THE CONSUMER SURVEY

TABLE 3

MEANS AND STANDARD DEVIATIONS OF STUDY VARIABLES

Managerial implications

A decade of research has shown that consumers’ initial perception of similarity is a key explanatory factor that make possible or limits the acceptance of brand extensions. This study demonstrates that similarity is not a fixed property in a brand extension setting. Similarity judgments are dynamic, that is, they do change with different similarity features. It is crucial to ensure high overall similarity between an original brand and a brand extension. However, it is also crucial to ensure high similarity between consumers’ self-image (actual and ideal self-image) and the image of the brands (the parent and the extension) that are involved in a brand extension task.

Other features that should be considered are usage situations, brand associations (see Broniarczyk and Alba 1994), sensory features, social/reference group, and the competence of the company when producing brand extensions, etc. High similarity on at least one feature is an opportunity for brand managers who can emphasize this overlapping feature through advertising. It was shown earlier that repeated exposure of only one similarity feature of an original brand and an extension serve to reduce the perceived overall incongruity between the two and, hence, lead to more positive evaluations (Lane 2000).

Theoretical implications

The self-image similarity approach applied to brand extensions cn complement the traditional categorization-based approach. The categorization approach may very well account for extension typicality effects. It is perhaps less suited to explain the effects of multiple constructs of similarity (see Keller and Sood 2001) or alternative points of intercategory relatedness. A decomposed similarity approach is undoubtedly more suitable to address these more complex extension applications.

Further research and conclusions

The study findings suggest several issues that warrant further inquiry. First, other ways to measured decomposed similarity between an original brand and brand extensions should be investigated. The psychology literature suggests that similarity is best characterized as a comparison of structured representations. Therefore some multiplicative constructions of the similarity construct may be the best representations. This could be done as suggested in, for example Gentner and Markman (1994) using transformations of the similarity distance between two objects (Exponential decay and Gaussian transformations; City block and Euclidean metrics). Some interaction effects should be investigated between some of the variables discussed in this study, but also as an influence of alternative variables. For example, consumer perceived relevance of an extension category (product category involvement) will most likely increase the perceived self-image/brand image similarity and then again influence evaluations of brand extensions. The direct effect of self-image similarity should also be studied in more depth. One hypothesis is that transfer of some salient brand personality traits from the original brand to the extension particularly should increase with increasing self-image/brand image similarity. Using functional theory (from attitude literature) could be a theoretical option to predict the responses related to the self-image questions. Attitudes towards a brand extension could serve a social identity function particularly for highly visible (conspicuous) products and/or product categories that are consumed in public and have a high "signal" value.

Another issue relevant for future research is the effect of different brand personality dimensions on the evaluation of brand extensions. It would be interesting to see how dimensions of brand personality affect evaluations of brand extensions using the measurement scale developed by Aaker (1997). It would furthermore be interesting to examine the brand personality of for instance "the Marlboro brand" and extend it to product categories that do or do not match the Marlboro brand’s personality (see Fournier 1998).

In conclusion, this investigation of similarity effects on brand extension evaluations contributes to the literature in some significant ways. We find that similarities between the original brand and the brand extensions should be measured using several items that better cover the similarity construct. Brand personality is also found to impact evaluations of brand extensions, especially if there is high similarity between an individual’s self-image and the brand extensions’ personality image. Moreover, since some product categories are more suitable as extensions of a given brand than others, a careful consideration of the match between the parent brand’s image and the new product category is vital.

TABLE 4A

REGRESSION OF DIFFERENT SIMILARITY MEASURES ON EVALUATION OF BRAND EXTENSIONS

TABLE 4B

REGRESSION OF SIMILARITY VARIABLES ON EVALUATION OF BRAND EXTENSIONS

REFERENCES

Aaker, D.A. (1991), "Managing Brand Equity: Capitalizing on the Value of a Brand Name", New York: The Free Press, 224 p.

Aaker, D.A., and Biel, A.L. (1993), "Brand Equity and Advertising. Advertising’s Role in Building Strong Brands", Lawrence Erlbaum Associates, Publishers, Hillsdale, New Jersey.

Aaker, D.A., and Keller, K.L. (1990), "Consumer Evaluations of Brand Extensions", Journal of Marketing, 54 (January), 27-41.

Aaker, J.L. (1995), "Measuring the Human Characteristics of a Brand: A Brand Personality Hierarchy", Advances in Consumer Research, 22, 393-395.

Aaker, J.L. (1997), "Dimensions of Brand Personality", Journal of Marketing Research, 34 (August), 347-356.

Aaker, J.L. (1999), "The Malleable Self: The Role of Self-Expression in Persuasion", Journal of Marketing Research, 36 (February), 45-57.

Ackoff, R.L., and Emsoff, J.R. (1975), "Advertising and Anheuser-Busch", Sloan Management Review, (Spring), 1-15.

Bijmolt, T.H.A., Wedel, M., Pieters, R.G.M., and DeSarbo, W.S. (1998), "Judgements of Brand Similarity", International Journal of Research in Marketing, 15, 249-268.

Bottomley, P.A., and Doyle, J.R. (1996), "The Formation of Attitudes towards Brand Extensions: Testing and Generalising Aaker and Keller’s Model", International Journal of Research in Marketing, 13, 365-377.

Boush, D.M. (1993), "How Advertising Slogans Can Prime Evaluations of Brand Extensions", Psychology & Marketing, 10 (January/February), 67-78.

Boush, D.M., and Loken, B. (1991), "A Process-Tracing Study of Brand Extension Evaluation", Journal of Marketing Research, 28 (February), 16-28.

Boush, D.M., Shipp, S., Loken, B., Gencturk, E., Crockett, S., Kennedy, E., Minshall, B., Misurell, D., Rochford, L., and Strobel, J. (1987), "Affect generalization to similar and dissimilar brand extensions", Psychology & Marketing, 4, 225-237.

Bridges, S. (1990), "A Schema Unification Model of Brand Extensions", Unpublished dissertation. Graduate School of Business, Stanford University, Stanford, CA 94305.

Broniarczyk S.M., and Alba, J.W. (1994), "The Importance of the Brand in Brand Extension", Journal of Marketing Research, 31, 214-228.

Chakravarti, D., MacInnis, D.J., and Nakamoto, K. (1990), "Product Category Perceptions, Elaborative Processing and Brand Name Extension Strategies". Advances in Consumer Research, 17, 910-916.

Cohen, J.B., and Basu, K. (1987), "Alternative Models of Categorization: Toward a Contingent Processing Framework", Journal of Consumer Research, 13 (March), 455-472.

Court, D.C., Leiter, M.G., and Loch, M.A. (1999), "Brand Leverage", The McKinsey Quarterly, No. 2, 101.110.

Dacin, P.A., and Smith, D.C. (1994), "The Effect of Brand Portfolio Characteristics on Consumer Evaluations of Brand Extensions", Journal of Marketing Research, 31 (May), 229-242.

Dolich, I.J. (1969), "Congruence Relationships between Self-images and Product Brands", Journal of Marketing Research, 6, 80-84.

DubT, L., Schmitt, B.H., and Bridges, S. (1992), "Categorization Research and Brand Extensions", Advances in Consumer Research, 19, 255-259.

Falkenhainer, B., Forbus, K.D., and Gentner, D. (1990), "The Structure-Mapping Engine: Algorithm and Examples", Artificial Intelligence, 41, 1-63.

Farquhar, P.H., Herr, P., and Fazio, R.H. (1989), "Extending Brand Equity to New Product Categories", Working Paper. Center for Product Research, Carnegie Mellon University.

Fournier, S. (1998), "Consumers and Their Brands: Developing Relationship Theory in Consumer Research", Journal of Consumer Research, 24 (March), 343-372.

Gentner, D., and Markman, A.B. (1994), "Structural Alignment in Comparison: No Difference without Similarity", Psychological Science, 5, No. 3 (May): 152-158.

Graeff, T.R. (1997), "Consumption Situations and the Effects of Brand Image on Consumers’ Brand Evaliations", Psychology & Marketing, 14 (January), 49-70.

Herr, P.M., Farquhar, P.H., and Fazio, R.H. (1990), "Extending Brand Equity to New Categories", Working Paper, Graduate School of Business. Indiana University.

Herr, P.M., Farquhar, P.H., and Fazio, R.H. (1996), "Impact of Dominance and Relatedness on Brand Extensions", Journal of Consumer Psychology, 5 (2), 135-159.

John, D.R., Loken, B., and Joiner, C. (1998), "The Negative Impact of Extensions: Can Flagship Prducts Be Diluted?", Journal of Marketing, 62 (January), 19-32.

Jun, S.Y., Mazumdar, T., and Raj, S.P. (1999), "Effects of Technological Hierarchy on Brand Extension Evaluations", Journal of Business Research, 46, 31-43.

Keller, K.L. (1993), "Conceptualizing, Measuring, and Managing Customer-Based Brand Equity", Journal of Marketing, 57 (January), 1-22.

Keller, K.L., and Aaker, D.A. (1992), "The Effect of Sequential Introduction of Brand Extensions", Journal of Marketing Research, 29 (February), 35-50.

Keller K.L., and Sood, S. (2001), "The Effects of Branding Strategies and Product Experience on Brand Evaluations" Forthcoming in Journal of Marketing (2001).

Kleine, R.E., Kleine, S.S., and Kernan, J.B. (1993), "Mundane Consumption and the Self: A Social-Identity Perspective", Journal of Consumer Psychology, 2 (3), 209-235.

Loken, B., and John, D.R. (1993), "Diluting Brand Beliefs: When Do Brand Extensions Have a Negative Impact?", Journal of Marketing, 57 (July) 71-84.

MacInnis, D.J., and Nakamoto, K. (1990), "Cognitive Associations and Product Category Comparisons: The Role of Knowledge Structure and Context", Working Paper, School of Business. University of Arizona.

Malhotra, N.K. (1988), "Self Concept and Product Choice: An Integrated Perspective", Journal of Economical Psychology, 9, 1-28.

Markman, A.B., and Gentner, D. (1993), "Structural Alignment during Similarity Comparisons", Cognitive Psychology, 25, 431-467.

Medin, D.L., Goldstone, R.L., and Gentner, D. (1990), "Similarity Involving Attributes and Relations: Judgments of Similarity and Difference are not Inverses", Psychological Science, 1, 64-69.

Medin, D.L., Goldstone, R.L., and Gentner, D. (1993), "Respects for Similarity", Psychological Review, 100 (2), 254-278.

Meyers-Levy, J., Louie, T.A., and Curren, M.T. (1994), "How Does the Congruity of Brand Names Affect Evaluations of Brand Name Extensions?", Journal of Applied Psychology, 79 (No. 1), 46-53.

Muthukrishnan, A.V., and Weitz, B.A. (1991), "Role of Product Knowledge in Evaluation of Brand Extension", In Holman, R. H. and Solomon, M. R. (Eds.). Advances in Consumer Research, 18, 407-413.

Onkvist, S., and Shaw, J. (1987), "Self-Concept and Image Congruence: Some Research and Managerial Implications", Journal of Consumer Marketing, 4 (1), 13-23.

Park, C.W., Milberg, S., and Lawson, R. (1991), "Evaluation of Brand Extensions: The Role of Product Feature Similarity and Brad Concept Consistency", Journal of Consumer Research, 18 (September), 185-193.

Sheinin, D.A., and Schmitt, B.H. (1994), "Extending Brands with New Product Concepts: The Role of Category Attribute Congruity, Brand Affect, and Brand Breadth", Journal of Business Research, 31, 1-10.

Sirgy, M.J. (1982), "Self-Concept in Consumer Behavior: A Critical Review", Journal of Consumer Research, 9, 287-300.

Sirgy, M.J. (1985), "Using Self-Congruity and Ideal Congruity to Predict Purchase Motivation", Journal of Business Research, 13, 195-206.

Sirgy, M.J. (1986), "Self Congruity", New York: Praeger.

Smith, D.C., and Andrews, J. (1995), "Rethinking the Effect of Perceived Fit on Consumers’ Evaluations of New Products", Journal of the Academy of Marketing Science, 23 (1), 4-14.

Smith D.C., and Park, C.W. (1992), "The Effects of Brand Extensions on Market Share and Advertising Efficiency", Journal of Marketing Research, 29, 296-313.

Tversky A. (1977), "Features of Similarity", Psychological Review, 84, 327-352.

Vitz, P.C., and Johnson, D. (1965), "Masculinity of Smokers and the Masculinity of Cigarette Images", Journal of Applied Psychology, 49, 155-159.

Zhang, S., and Fitzgerald, K.A. (1997), "New Approaches to the Role of Similarity in Consumer Research", Advances in Consumer Research, 24, 15-16.

----------------------------------------