Modeling the brand extensions' influence on brand image

Eva Martnez Salinas

a , a,

and Jos Miguel Pina Prez

1, a,

Departamento de Economa y Direccin de Empresas, Facultad de Ciencias Econmicas y Empresariales, Gran Va, 2 50.005, Zaragoza, Spain Received 1 August 2007; revised 1 December 2007; accepted 1 January 2008. Available online 13 February 2008.

Abstract
Brand -extension strategies enhance success chances of new products, even though they expose brand image to dilution risks. The present work analyzes how brand -extension evaluation can affect the current brand image and proposes a theoretical model formed by five main factors related to brand associations, extension congruency and extension attitude. The model estimation includes structural equation analysis using data from 699 surveys developed under market conditions. The results verify that extension attitude influences brand image, whereas initial brand associations and perceived fit between the new product and either the remaining products (category fit) or the brand image (image fit) are able to strengthen consumer attitude. The study also explains the role of consumer innovativeness as a moderating factor, suggesting that the characteristics of consumer personality could be more important than expected. Keywords: Brand extensions; Brand image; Brand equity; Consumer behavior

Article Outline
1. Introduction 2. Influence of extensions on brand image 3. Hypotheses 4. Method 4.1. Pre-tests 4.2. Sample and procedure 4.3. Measures 5. Results 5.1. Scale validation 5.2. Global effects of brand extensions on brand image 5.3. Model contrasting 6. Conclusions and implications 6.1. Limitations and future research References

1. Introduction
Today, firms invest substantial resources to develop names with a favorable image. Among other advantages, a positive image facilitates business expansion through brand extensions (Keller and Aaker, 1992). Thus, product introductions with the same brand name are able to leverage the brand image, brand awareness and, on the whole, brand equity obtained in the

established markets (Milberg et al., 1997). Considering low costs and high success chances ([Milewicz and Herbig, 1994] and [Keller, 2003]), more than 80% of firms opt for brand extensions as a way of marketing goods and services (Keller, 2003). Associating an existing brand to a new entity such as a geographical name, a person or, as in this case, a product may affect the existing brand's image associations. The effect can be positive, thus increasing the memory and strengths of such associations ([Morrin, 1999] and [Aaker, 2002]), although a dilution of current beliefs often occurs (Ries and Trout, 1993; John et al., 1998). On the basis of psychology-based theories, several authors try to explain the extension- brand feedback effect through different theoretical approaches ([Grhan-Canli and Maheswaran, 1998] and [Loken and John, 1993]). Others analyze the effect of several factors, such as consumer attitude to the extended product, on brand image ([Alexander and Colgate, 2005] and [Chang, 2002]). These studies, however, use a limited number of variables that narrow the scope of analysis and make generalization difficult. With the aim of gaining insight into the process by which consumer attitude toward brand extensions transfers to brand image, the present paper extends previous research in several ways. First, the study analyzes the interrelation between the attitude toward the brand and the attitude toward the extension, trying to discover if potentially successful extensions can be damaging to the current associations. Second, the main contribution of the present study is the proposal and validation of a theoretical model that, starting from initial brand image, explains brand -extension attitude formation and the influence on brand image. Regarding previous research, mainly focused on either category fit or image congruency, the study also verifies whether the moderating role of consumer innovativeness in the extension evaluation process is different depending on the type of perceived fit. In satisfying the aforementioned goal, the study divides into five sections. Sections 2 and 3 contain a brief review of the literature to justify the theoretical model and the relations established in the hypotheses. The fourth section describes the method used to validate the model, reporting the results in the fifth section. Finally, the paper concludes with theoretical and managerial implications.

2. Influence of extensions on brand image

Extensions are a suitable strategy when they contribute to the improvement of the parent brand, reinforcing brand equity associations (Aaker, 2002) and, hence, improving brand positioning (Park et al., 1986). Nevertheless, extensions can also dilute brand -associated values, creating new associations or confusing the current ones ([Tauber, 1988] and [John et al., 1998]). Prestige brands that provide self-expression benefits run a high dilution risk, especially if they launch products in lower price-quality levels ([Kim et al., 2001] and [Aaker, 2002]). This negative effect can affect general brand associations (Martnez and de Chernatony, 2004) and the beliefs in specific attributes ([Keller and Aaker, 1992] and [Loken and John, 1993]), such as those related to the most representative product of the brand or flagship product ([John et al., 1998] and [Chang, 2002]). Diamantopoulos et al. (2005) find no evidence that extensions modify those perceptions related to brand personality. The sign of brand image variation will depend on the reciprocal transfers between the new product associations and the extended brand beliefs. According to Czellar (2003), a transfer of affection or concrete knowledge, either related or not to the products, may occur. Several models and theories allow understanding of the extension-parent brand transfer process. These models often assign a specific role to new product-extended brand perceived fit, an extremely important concept in the field of brand extensions ([Aaker and Keller, 1990] and [Vlckner and Sattler, 2006]). According to the associative network theory, brand image is a mental scheme formed by a network of concepts (nodes) interconnected by linkages or associations ([Anderson, 1983] and [Morrin, 1999]). When the company launches an extension coherent with the pre-

existing associations, the brand scheme will assimilate the new concept with no significant alterations. On the contrary, if the new example substantially differs from the scheme, the latter changes and adapts to the new associations (Park et al., 1993). Viswanathan (1997) explains that these adaptation or assimilation processes act as an anchor adjusting to the new information. The weight that consumers attach to the information will depend on its accessibility (Ahluwalia and Grhan-Canli, 2000). In high accessibility conditions, any extension category might improve or dilute the brand scheme depending on the positivity or negativity of the information. On the contrary, in a low accessibility situation, positive information will mainly benefit far extensions, whereas negative information will be more damaging to the closest extensions to the brand and its products. Researchers have proposed alternative theories to understand feedback effects of brand extensions. According to the bookkeeping model, consumer beliefs always change when faced with any kind of information, even in the case of extensions coherent with the parent brand (Weber and Crocker, 1983). On the contrary, the typicality model highlights the importance of the congruency of the information, in such a way that incongruent extensions would lead to sub-types with independent associations within the original category ([Weber and Crocker, 1983] and [Grhan-Canli and Maheswaran, 1998]). Finally, the conversion model establishes that brand schemes only change in extremely atypical examples ([Weber and Crocker, 1983] and [Grhan-Canli and Maheswaran, 1998]). Loken and John (1993) prove that two models seem to coexist in the field of extensions, a typicality one for low-fit extensions and a bookkeeping one for the rest. However, Grhan-Canli and Maheswaran (1998) find that high-motivated individuals always change their beliefs according to the bookkeeping model; low-motivated individuals, however, act according to sub-typing models, being their evaluations more extreme when perceived fit is high.

3. Hypotheses
Brand stretching into new products or services can be a fast way to remove brand associations from the consumers' mind. Most experiments on brand association dilution expose participants to information on low-fit or failure extensions ([Loken and John, 1993], [Leong et al., 1997] and [John et al., 1998]), although risk also arises when it comes to apparently successful extensions. Keller and Sood (2003) state that parent brand dilution is more likely to occur when the experience with the new category is wider (strength), when consumers consider the extension as relevant for the brand (diagnosis) and when the experience is more incongruent with the brand image (inconsistency). The present work analyzes the extension consistency issue, or fit, which is the factor that all the cognitive models of brand -extension feedback consider as more relevant ([Loken and John, 1993] and [Park et al., 1993]). The literature reveals that the attitude to an extended brand directly depends on the fit degree with the extension (Grime et al., 2002). The expansion to far categories will involve losing brand differentiation and credibility, whereas extensions to related markets will avoid potential damage (Aaker, 2002). Authors like Milberg et al. (1997) demonstrate that far extensions generate negative feedback in terms of attributes or image. Similarly, Lee and Ulgado (1993) verify that fit has a positive effect on the image of service firms, whereas Martnez and de Chernatony (2004) verify the same for tangible product extensions. According to the following hypotheses, low-fit extensions will exert a dilution effect on brand image not only when consumers hold an unfavorable attitude toward brand extensions but also when extension attitude is positive. Regardless of the extension evaluation, far extensions give rise to new associations and, hence will vary the brand schema (Park et al., 1993). On the contrary, launching close extensions to the parent brand can be a way of strengthening current associations. This positive influence probably appears when extension attitude is favorable, although unsuccessful extensions are only able to maintain the brand beliefs. In

this case, the unfavorable judgment should inhibit the positive effect of communicating the brand associations through new categories. This paper puts forward the following hypotheses. H1a The effect of high-fit extensions on brand image is positive only when the attitude to the extension is favorable. H1b The effect of low-fit extensions on brand image is negative when the attitude to the extension is both favorable and unfavorable. Next, the model in Fig. 1 aims to gain insight into how extensions affect brand image. The model includes the variables with a higher impact on extension attitude ([Aaker and Keller, 1990], [Hem et al., 2003] and [Vlckner and Sattler, 2006]; etc.). Such attitude will determine the development of brand image (Lane and Jacobson, 1997).

Full-size image (31K)

Fig. 1. Proposed model to analyze the effect of brand extensions on brand image.

The model stems from initial brand image and attempts to explain the relations and interactions that occur until a brand extension gives rise to a new brand image. The first variable to consider is thus initial brand image, which symbolizes brand knowledge prior to extension information. Regarding the new category and its relation with the parent brand, the model also considers perceived category fit and perceived brand image fit. What is more, the variables include extension attitude and consumer innovativeness, a personality trait that has some moderating effects. It is important to note that researchers have not previously tested the whole process brand -extension evaluation and feedback effects in the same model. Consequently, it is necessary to test whether the relationships found in the literature are consistent in a model context, showing the most relevant variables and the interrelation between them. The attitude toward the brand extension is the first factor that the model seeks to explain. According to the model, initial brand image, category fit and image fit will be the main determinants of the consumer assessment. Brand image is a complex construct that can be made up of different dimensions. As several authors prove, extension attitude is better for those brands related to high quality standards ([van Riel et al., 2001] and [Vlckner and Sattler, 2006]), reputation (Hem et al., 2003), prestige (Park et al., 1991) or affection (Sheinin and Schmitt, 1994), among other associations. In the corporate and services field as well, a positive image clearly generates favorable perceptions of the new products ([Brown and Dacin, 1997] and [de Ruyter and Wetzels, 2000]). In fact, the essence of extension strategies is leveraging brand image by using the same brand name. Therefore, H2 Initial brand image has a positive direct influence on brand -extension attitude.

Perceived fit is a strategic factor in the brand -extension evaluation process, bridging the gap between brand beliefs and new product assessment. Several works find image -perceived fit interaction effects ([Aaker and Keller, 1990] and [Boush et al., 1987]), thus fit may have a mediating effect between image and extension attitude. Brand image will affect fit perception, and such perception will have a direct effect on the assessment of the new product or service. In order to analyze the role of that factor properly, some authors distinguish between category fit and brand image fit ([Bhat and Reddy, 2001],[Grime et al., 2002] and [Czellar, 2003]). The former reflects the similarity between the new category and the other products of the extended brand, whereas the latter defines the degree to which the extension shares global brand concept feelings and associations, such as prestige or functionality ([Grime et al., 2002] and [Czellar, 2003]). As brand perceptions become more favorable, the extension credibility increases (de Ruyter and Wetzels, 2000), and so does parent brand -extended product congruency (Czellar, 2003). From an image approach, Park et al. (1991) claim that brand concept consistency is higher for prestige brands than for those with a lower reputation, since prestige concepts are more widely known and generalizable. Boush (1997) proves that consumers are able to perceive a higher physical similarity between two products simply because both share the same brand name. These results lead to the next hypotheses. H3 Initial brand image has a positive direct influence on perceived category fit. H4 Initial brand image has a positive direct influence on perceived image fit. Perceived fit directly affects the opinion on the extended category as well. Consumers believe that extensions to non-related categories are not very reliable, which generates a negative opinion on them (Kirmani et al., 1999). Therefore, the assessment of an extension will be more positive as perceived closeness with the brand grows ([Aaker and Keller, 1990] and [Vlckner and Sattler, 2006]), even in the case of non-prestige brands (Park et al., 1991). The favorable fit effect appears in the studies that consider fit from both the category ([Boush et al., 1987] and [Boush and Loken, 1991]) and the image (Park et al., 1991) perspective. Romeo (1991) considers category fit as the most relevant dimension, whereas Bhat and Reddy (2001) hold the opposite view. In any case, consumers who perceive higher fit will have a better opinion on brand extensions, regardless of their viewpoint. The empirical evidence supports the following hypotheses. H5 The higher the perceived category fit is, the more favorable the attitude to the extension. H6 The higher the perceived image fit is, the more favorable the attitude to the extension. The following hypotheses have final brand image as a dependant variable. Both theory and practice suggest that most brand associations will remain once the extension competes in the market, thus initial brand image will condition final brand attitude ([Lee and Ulgado, 1993] and [Martnez and Pina, 2003]). Initial brand associations can vary according to the formed extension attitude. Low quality extensions or extensions that receive negative assessments will entail a detriment of brand image ([Chang, 2002] and [Martnez and Pina, 2003]), diluting both general and specific beliefs (Martnez and de Chernatony, 2004). Other elements such as brand personality appear to be more dilution resistant (Diamantopoulos et al., 2005), although the previous literature indicates the existence of a positive relationship between extension attitude and

brand image. The subsequent hypotheses synthesize these arguments. H7 Initial brand image has a positive direct influence on final brand image. H8 Extension attitude has a positive direct influence on final brand image. The last variable of the model is consumer innovativeness, a concept that represents consumer proneness to buy new products (Roehrich, 2004). Unlike the relationships proposed above, the following and last hypotheses do not reflect direct but moderating effects. Specifically, consumer innovativeness could moderate the relationships between perceived fit factors and extension attitude. According to Rogers (1995), innovative people enjoy buying new products, do not excessively trust in social norms, and are self-confident and daring. These personality characteristics lead to a better extension attitude for both good (Vlckner and Sattler, 2006) and service ([Hem et al., 2003] and [Siu et al., 2004]) extensions. Some authors explain that innovators are also more risk prone ([Klink and Smith, 2001] and [Hem et al., 2003]) and, consequently, more receptive to new ideas and categories associated to the brand. In this sense, Klink and Smith (2001) prove that the influence of perceived fit on extension attitude is lower among innovative consumers, who are more likely to buy extensions unrelated to the current markets. Although Klink and Smith (2001) show this moderating effect only for category fit, the role of consumer innovativeness is likely to appear in image fit. H9a The effect of perceived category fit on extension attitude is weaker when consumer innovativeness is high than when innovativeness is low. H9b The effect of perceived image fit on extension attitude is weaker when consumer innovativeness is high than when innovativeness is low.

4. Method
An empirical study contrasted the hypotheses and validated the model in Fig. 1. Adopting the usual procedure, the study analyzes real brands and realistic hypothetical extensions ([Aaker and Keller, 1990], [van Riel et al., 2001] and [van Riel and Ouwersloot, 2005]) coming from three pre-tests. The following paragraphs explain the methodology in depth.

4.1. Pre-tests
A sample of undergraduates took part in the pre-tests ([Sheinin and Schmitt, 1994] and [Kim, 2003]). With the purpose of selecting the specific brands and extensions, the lack of normality in the data suggested the carrying out of Wilcoxon tests. The aim of the first pre-test, conducted with 91 undergraduates, was to choose six actual brands coming from different sectors (fast moving consumer goods, durable consumer goods and services) and having a different image perception. All the brands had to be familiar enough in order to generate specific brand associations (Low and Lamb, 2000). Consequently, individuals assessed both brand familiarity (F) and brand image (I) in two 7point Likert scales (1 = totally unfamiliar/7 = very familiar; 1 = bad image/ 7 = excellent image) for a total of 11 brands. The results gave rise to the selection ofColgate and Signal (FC = 6.4; FS = 5.5), Nike and Puma (FN = 6.6; FP = 5.6), Telefnica Movistar and Amena (FT = 6.6; FA = 6.3) as familiar brands. As required, the image is significantly different for toothpaste brands (IC = 5.7; IS = 5.0; Z = 4.62; p < 0.01), sports

brands (IN = 6.2; IP = 5.1; Z = 5.45;p < 0.01) and mobile phones (IT = 5.7; IA = 4.9; Z = 4.00; p < 0.01). The second and third pre-tests, where 98 and 81 undergraduates respectively participated, aimed at finding two extensions one for each sector with differences in perceived fit. The pre-tests included measures of both perceived category fit (CF) and brand image fit (IF) (Bhat and Reddy, 2001) in two Likert scales (1 = not at all similar/7 = very similar; 1 = non-coherent/7 = very coherent). Regarding toothpaste brands, the individuals provided suitable values for sugar-free whitening tooth decay-preventing sweets and sunglasses. The first extension shows a higher perceived fit than the second one forColgate (CF1 = 5.4; CF2 = 1.3; Z = 5.34; p < 0.01) (IF1 = 5.7; IF2 = 1.5; Z = 5.34; p < 0.01) and Signal (CF1 = 4.9; CF2 = 1.2; Z = 5.12; p < 0.01) (IF1 = 5.2; IF2 = 1.2; Z = 5.02; p < 0.01). On the other hand, as regards sports brands, skis is the close extension and DVD players the far product, both from the perspective of product category of Nike (CF1 = 3.3; CF2 = 1.3; Z = 5.12; p < 0.01) and Puma (CF1 = 3.3; CF2 = 1.1; Z = 4.91; p < 0.01) and in relation to the image of Nike (IF1 = 4.2; IF2 = 1.4; Z = 5.56; p < 0.01) and Puma (IF1 = 3.9; IF2 = 1.1; Z = 5.11; p < 0.01), respectively. Finally, the results recommended to choose telecommunication on-line courses and insurance service extensions, because of the differences in perceived fit for Telefnica Movistar(CF1 = 4.7; CF2 = 1.8; Z = 5.47; p < 0.01) (IF1 = 4.7; IF2 = 1.7; Z = 5.54; p < 0.01) and Amena (CF1 = 3.7; CF2 = 1.8; Z = 4.28; p < 0.01) (IF1 = 4.3; IF2 = 1.8;Z = 4.56; p < 0.01).

4.2. Sample and procedure

Subsequent to the pre-tests, the study included 12 questionnaires with a different brand extension combination. Each questionnaire started with questions about consumer innovativeness and some issues related to the particular brand, such as initial brand image. Shortly afterwards, respondents knew that the brand had decided to launch the potential extension and they had to assess the new product in different ways (perceived fit, etc.). The survey finished with the same questions about brand image, but considering the new product category. It is significant that the order of the questions tried to minimize order effects (Klink and Smith, 2001). The fieldwork took place in May 2005 in a foremost Spanish city, Zaragoza, which is sometimes considered as a test market for products aimed at Spain. The respondents consisted of workage individuals (16 to 65 years) that randomly answered one of the questionnaires. A team of trained interviewers approached them in different parts of the city, days and times, following a quota sampling procedure. Specifically, the sample was representative of the population in terms of gender (50.9% women and 49.1% men) and age (46.5% 2645 years, 33.3% 16 25 years, 20.2% 4664 years). The number of surveys obtained added up to 720, resulting in 699 valid questionnaires after deleting non-valid responses. Furthermore, response rates did not vary in function of sociodemographical variables (gender, age, educational level and household income). Table 1 shows the specific number of individuals who satisfactorily responded to each questionnaire. Table 1. Questionnaires used in the research

N o .

Brand

Extension (high fit)

N o .

Brand

Exte nsio n (low fit) Sungla

49 Colgate

Sugar-free whitening tooth

50 Colgate

N o .

Brand

Extension (high fit)

N o .

Brand

Exte nsio n (low fit)

sses 49 Signal 48 Signal Sungla sses 48 Nike 49 Nike DVD players 49 Puma Skis 49 Puma DVD players 74 Telefnica Telecommunication on-line 80 Telefnica Insura Movistar courses Movistar nce 79 Amena Telecommunication on-line 75 Amena Insura courses nce

decay-preventing sweets Sugar-free whitening tooth decay Skis

4.3. Measures
Several 7-point Likert questions measured the variables, with items extracted or based on the literature. Table 2 displays the scales measuring each factor. Table 2. Scales used in the questionnaires

Scale

Measured concept

Consumer Hedonist innovativeness (HINN) innovativeness Roehrich HINN1: I am more interested in buying new than (1995) known products HINN2: I like to buy new and different products HINN3: New products excite me Social innovativeness (SINN) SINN1: I am usually among the first to try new products SINN2: I try new products before my friends and neighbors SINN3: I know more than others on latest new products Brand image Functional image (FUIM) (initial/final) Martin and FUIM1i/FUIM1f: The products have a high quality Brown (1990) Aaker (1996) FUIM2i/FUIM2f: The products have better characteristics than competitors'

Scale

Measured concept

Weiss et al. FUIM3i/FUIM3f: The products of the competitors (1999) are usually cheaper Villarejo (2002) Affective image (AFIM) (initial/final) AFIM1i/AFIM1f: The brand is nice AFIM2i/AFIM2f: The brand has a personality that distinguishes itself from competitors AFIM3i/AFIM3f: It's a brand that doesn't disappoint its customers Reputation (REIM) (initial/final) REIM1i/REIM1f: It's one of the best brands in the sector REIM2i/REIM2f: The brand is very consolidated in the market Perceived fit Category fit (CAFI) Aaker and CAFI1: The extension is similar to the brand's Keller (1990) products Taylor and CAFI2: The firm's resources are helpful to make Bearden (2002) the product extension Image fit (IMFI) IMFI1: The product extension fits with the brand image IMFI2: Launching the extension is logical for the company IMFI3: Launching the extension is appropriate for the company Extension EXAT1: Favorability of the extension attitude (EXAT) Aaker and EXAT2: Perceived quality of the extension Keller (1990) Pryor and EXAT3: Likelihood of trying the extension Brodie (1998)
First, the scale of consumer innovativeness includes the items that Roehrich (1995) proposes from both a hedonistic and social perspective. Initial and final brand image scales gather items from several works ([Martin and Brown, 1990] and [Weiss et al., 1999]; etc.) which attempt to assess tangible (functional image) and intangible (affective image) attributes and benefits, as well as the global attitude to the brand (reputation). On the other hand, the scale of perceived fit considers the distinction between category fit or similarity and image fit or consistency with brand image ([Park et al., 1991], [Bhat and Reddy, 2001] and [Grime et al., 2002]). Thus, a series of items from both perspectives measure perceived fit ([Aaker and Keller, 1990] and [Taylor and Bearden, 2002]). Finally, extension attitude items come from works such as Aaker and Keller (1990) and Pryor and Brodie (1998).

5. Results
The data analysis consisted of three stages, using SPSS 13.0 and EQS 5.b software. The analysis includes examining the statistical properties of the scales in terms of unidimensionality, reliability and validity. Next, the analysis followed a series of t-test aimed at testing hypothesis H1. Finally, the study addressed the estimation of the proposed model.

5.1. Scale validation

The scale validation first involved checking the dimensionality of initial and final brand images, since the distinction between the proposed factors (functional image, affective image and reputation) could be statistically non-advisable. After deleting items with a weak item-total correlation (FUIM3i, FUIM3f), some exploratory factorial analyses showed that a two dimension-solution was more adequate to represent the brand image construct than the proposed three sub-scales. The first dimension, which puts the items of functional image (FUIM1, FUIM2) and reputation (REIM1 and REIM2) together, showed a Cronbach's alpha higher than 0.7 (Cronbach, 1951) for both initial ( = 0.85) and final images ( = 0.88). The second dimension reflects the affective component (AFIM1, AFIM2, AFIM3) and its Cronbach's alpha was also above the threshold of 0.7 for the initial ( = 0.79) and final ( = 0.81) settings. In order to ensure the external validity, the analysis continued with the comparison of two alternative models through Structural Equation Modeling (SEM) and ERLS (elliptical reweighted least squares) estimation method ([Hair et al., 1998] and [Kline, 2005]). Table 3 reveals that the first model, which considers brand image (initial or final) as unidimensional, fits worse to the data than the second model, which take into account the aforementioned dimensions. Being decisive for choosing between models with a different number of latent variables, AIC indicator suggests to select the multidimensional structure of initial (AIC = 127.41 < 154.18) and final (AIC = 158.52 < 239.56) brand image. Table 3. Indicators of the alternative models of brand image (initial and final)

Comparative indicators

Initial image

Final image

Unidim Multidim Unidim Multidim ensiona ensional ensiona ensional l l 2 RMSR (Root Mean Square Residual) ECVI (Expected Cross-Validation Index) NCP (Noncentrality Parameter) SNCP (Scaled Noncentrality Parameter) PNFI (Parsimonious 126.18 0.05 0.22 97.41a 0.04a 0.18a 84.41a 0.12a 211.56 0.05 0.34 128.52a 0.04a 0.23a 115.52a 0.17a

112.18 0.16 0.64a

197.56 0.28 0.63a

0.60

0.60

Comparative indicators

Initial image

Final image

Unidim Multidim Unidim Multidim ensiona ensional ensiona ensional l l Normed Fit Index) PGFI (Parsimonious 0.47a Goodness-of-Fit Index) AIC (Akaike 154.18 Information Criterion)
a

0.44 127.41a

0.45a 239.56

0.43 158.52a

Coefficients that are favorable to the specified model.

After verifying the multidimensional character of the brand image construct, the next step was to conduct a factor analysis for all the scales hypothesized to co-vary (Fig. 1). The methodology applied for testing the measurement model consisted of SEM methodology with ERLS estimation, yet without the moderating factor consumer innovativeness. Table 4 displays the results, which indicate that the scales present good statistical properties. Table 4. Convergent validity and reliability of the scales

Factor

Items

Convergent validity

Reliability

t (> 1.96) (> 0. CRC EVA 5) (> 0. (> 0. 6) 5) FUIMREIM (i) FUIM1i FUIM2i REIM1i REIM2i FUIM1f FUIM2f REIM1f REIM2f AFIM1i AFIM2i AFIM3i 21.23 19.53 21.99 16.49 22.94 22.62 22.69 17.71 20.44 18.92 17.29 0.80 0.76 0.82 0.66 0.84 0.83 0.84 0.70 0.79 0.74 0.69 0.85 0.58

FUIMREIM (f)

0.88

0.65

AFIM (i)

0.78

0.55

Factor

Items

Convergent validity

Reliability

t (> 1.96) (> 0. CRC EVA 5) (> 0. (> 0. 6) 5) AFIM (f) AFIM1f AFIM2f AFIM3f EXAT1 EXAT2 EXAT3 CAFI1 CAFI2 IMFI1 IMFI2 IMFI3 20.99 21.03 17.20 22.35 17.98 20.77 21.70 18.20 25.87 25.85 23.82 Increment al fit AGFI = 0.8 1 CFI = 0.96 IFI = 0.96 NFI = 0.95 NNFI = 0.9 5
Note: CRC: Composite Reliability Coefficient, EVA: Extracted Variance Analysis, GFI: Goodness-of-Fit Index, RMSEA: Root Mean Square Error of Approximation, SRMR: Standardized Root Mean Square Residual, AGFI: Adjusted Goodness-of-Fit Index, CFI: Comparative Fit Index, IFI: Incremental Fit Index, NFI: Normed Fit Index, NNFI: Nonnormed Fit Index.

EXAT

CAFI IMFI

0.80 0.80 0.69 0.83 0.71 0.79 0.84 0.73 0.91 0.90 0.86

0.81

0.59

0.82

0.61

0.76 0.92

0.62 0.79

Fit indexes Global fit

2 = 1003.46 (188) p < 0.01 GFI = 0.85 RMSEA = 0.07 SRMR = 0.04

As seen in Table 4, all the proposed items unidimensionally fit to the respective seven factors or latent variables. The values obtained in Composite Reliability Coefficients and Extracted Variance Analysis (EVA) are above 0.6 and 0.5 respectively, which guarantee the internal consistency of the scales. Given that lambda coefficients are significant (t > 1.96) with standard loadings above 0.5, the scales also show convergent validity. Finally, the verification of the

discriminant validity criterion involved checking that the variance explained (AVE) within constructs is above the variance explained across constructs (Voss et al., 2003). The main goodness-of-fit indicators for the measurement model appear at the bottom of Table 4, distinguishing between global and incremental fit indexes. On the whole, the indicators are positive and above the minimum established by researchers ([Mueller, 1996], [Hair et al., 1998] and [Kline, 2005]). Regarding global fit, GFI is above 0.8 (GFI = 0.85), which is lower than the standard level of 0.9, although acceptable in practice (Mueller, 1996). Moreover, RMSEA and SRMR error statistics do not exceed the maximum values of 0.08 (RMSEA = 0.07) and 0.1 (SRMR = 0.04). 2 The only unsuitable indicator is Chi-square test ( (188) = 1003.46; p < 0.01), which often occurs in samples of over 400 observations. On the other hand, all the incremental fit measures are above the required 0.8 (AGFI = 0.81) and 0.9 (CFI = 0.96; IFI = 0.96; NFI = 0.95; NNFI = 0.95) levels, thus corroborating the suitability of the proposed model. As a final step of the scale validation process, and with the aim of representing the model relationships adequately, second-order factors included the dimensions of initial and final brand images. These models present favorable fit indicators both for initial (GFI = 0.95; SRMR = 0.04; NFI = 0.97; IFI = 0.97; etc.) and final (GFI = 0.94; SRMR = 0.04; NFI = 0.96; IFI = 0.97; etc.) images.

5.2. Global effects of brand extensions on brand image

Prior to checking any hypotheses, the study analyzed the effect of extensions on brand image for each brand -extension combination. Given that Cronbach's alphas exceeded 0.7, the mean of all the items in the factors of initial ( = 0.89) and final ( = 0.90) images replaced the individual measurements. Table 5, which shows the initial and final image scores, reveals that brand extensions usually influence current associations. Generally speaking, these results suggest that firms should avoid entering markets far from their sector, since such extensions clearly entail brand image dilution. Table 5. Effect of brand extensions on brand image in each scenario

Brand

High-fit extension Initial image Final image 5.34 4.98 4.87 4.61 4.68 4.36

Low-fit extension Initial image 5.02 4.72 5.61 4.70 4.80 3.87 Final image 4.26 4.06 4.27 4.12 4.42 3.44

Colgate 5.30 Signal 4.77 Nike 5.00 Puma 4.69 Telefnica 4.62 Amena 4.39

Cases where there is a statistical difference with the initial image scores (p < 0.01).

According to hypothesis H1, both potentially successful and unsuccessful extensions may influence current associations, depending on perceived fit. This hypothesis was evaluated by comparing the initial and final image scores for each combination of fit (high/low) and extension attitude (favorable/unfavorable) through mean tests. The extension attitude levels resulted from calculating the mean of EXAT1 to EXAT3 ( = 0.80) and recoding each obtained value into a negative (< 4) or positive (> 4) category. Fig. 2 gathers the results.

Full-size image (16K)

Fig. 2. Influence of fit and extension attitude on brand image variation.

According to Fig. 2, high-fit extensions can improve brand image only for favorably received extensions (IMAGi = 5.0; IMAGf = 5.1; t = 3.34; p = 0.00). On the contrary, potentially unsuccessful products generate a dilution effect (IMAGi = 4.0; IMAGf = 3.8; t = 2.59; p = 0.01) despite having a high fit. This finding supports hypothesis H1a. On the other hand, low-fit extensions give rise to image dilution when extension attitude is both favorable (IMAGi = 5.3; IMAGf = 4.9; Z = 3.02; p = 0.00) and unfavorable (IMAGi = 4.5; IMAGf = 3.7; t = 10.39; p = 0.00). This finding lends support to hypothesis H1b. As well as testing hypothesis H1, Fig. 2 further shows that consumers with a better attitude toward the brand have a better attitude to the extension. In order to understand the effect of initial brand image on the extension attitude and the process of image dilution, the following part shows the result of the model estimation (Fig. 1).

5.3. Model contrasting

The estimation of the structural model underlying hypotheses H2 to H8 involves SEM methodology. Table 6 displays the results, which are favorable for all the specific relationships and the model as a whole. Table 6. Results of the structural model

Hypotheses

Standardized (t) Hypotheses validation Yes Yes Yes Yes Yes Yes Yes Incremental fit AGFI = 0.87 CFI = 0.98 IFI = 0.98 NFI = 0.97 NNFI = 0.98

H2: IMAG (i) EXAT 0.22 (6.81) H3: IMAG (i) CAFI 0.22 (4.37) H4: IMAG (i) IMFI 0.19 (4.05) H5: CAFI EXAT 0.27 (2.25) H6: IMFI EXAT 0.58 (5.06) H7: IMAG (i) IMAG (f) 0.47 (13.64) H8: EXAT IMAG (f) 0.57 (15.36) Fit indexes Global fit 2 = 654.61 (197) (p < 0.01) GFI = 0.90 RMSEA = 0.06 SRMR = 0.05
Note: See note in Table 4, significant at p 0.05.

First, initial brand image has a direct and significant influence on extension attitude (est = 0.22; t-value = 6.81), supporting hypothesis H2. Therefore, those brands with more favorable associations in the consumer mind will generate a better attitude to the purchase of new products or services. As well as having a direct influence on extension evaluation, initial brand image will affect fit perception. Thus, a better image will generate higher levels of perceived category fit (est = 0.22; t-value = 4.37), giving support to hypothesis H3. Similarly, image fit will also improve when the parent brand has a positive image (est = 0.19; t-value = 4.05), as hypothesis H4 states. The coefficients obtained suggest a slightly stronger effect for the first fit type. According to hypothesis H5, category fit seems to be a determinant of extension attitude (est = 0.27; t-value = 2.25). The relationship between image fit and attitude presents a significant coefficient as well (est = 0.58; t-value = 5.06), which leads to confirm hypothesis H6. Consumers will prefer those extensions similar to the firm's products or services, mainly if they are consistent with brand image. Given that brand image explains perceived category fit (H3) and image fit (H4), mediating effects are likely to occur. A Sobel Test corroborated the existence of the aforementioned mediating effects, both for category (Z = 1.97; p < 0.05) and image fit (Z = 3.13; p < 0.05). The last hypotheses indicate the factors that explain brand image after marketing an extension. According to hypothesis H7, final brand image has a positive (est = 0.47) and significant (t = 13.64) relation with initial brand image. Nevertheless, final associations depend more on the attitude to the new product or service (est = 0.57; t = 15.36), which proves hypothesis H8. Therefore, firms must be aware that their image, or its brands' , is unlikely to remain invariable when launching products under the same brand name. Table 6 also contains goodness-of-fit measurements, which are acceptable and better than the ones obtained with the measurement model. Again, the error coefficients (RMSEA = 0.06; SRMR = 0.05) and the GFI (0.90) show acceptable values. Similarly, all the incremental fit measurements are above the established standards (AGFI = 0.87; CFI = 0.98; IFI = 0.98; NFI = 0.97; NNFI = 0.98). The contrasting of hypotheses H9a and H9b involved a multi-sample analysis that compares the results of replicating the structural model according to consumer innovativeness (high vs. low). Consumer innovativeness levels resulted from estimating the mean of the constituent items 2 ( = 0.88), without HINN1 (R < 0.5), and considering the highest (mean + 0.25 standard deviation) and lowest values (mean 0.25 standard deviation). The Lagrange Multiplier (LM) Test and the maximum likelihood estimation method determined whether the model coefficients are significantly different (Iglesias and Vzquez, 2001). Tests of moderation effects, such as LM and Wald statistics, are usually based on estimating the Chi-square difference across groups.Table 7 displays the results. Table 7. Results of the multi-sample analysis

Hypotheses

Sample I (high)

Sample II (low)

LM test

Hypo these s valid ation

Standar Standar 2dif. dized ( dized ( (prob t) t) ab) H9a (CAFI EXAT) high/low consumer innovativeness H9b (IMFI EXAT) high/low consumer innovativeness
Significant at p 0.05;

0.08 ( 0.20) 0.90 (3.07)

0.36 (2.44) 0.52 (3.75)

2.73 Yes (0.09) 2.96 No (0.08)

significant at p 0.1.

In this case, the results lead to reject one of the moderating hypotheses (H9b) because of some discrepancies with the expectations. On the contrary, the results lend support to 2 hypothesis H9a with a 90% confidence level ( dif = 2.73; p < 0.1), proving that the effect of category fit on extension attitude decreases with highly innovative consumers ( 0.08 n.s. < 0.36). Innovative consumers prefer to assess image fit (0.90 > 0.52), in such a way that 2 Chi-square statistical value significantly improves at 90% ( dif = 2.96; p < 0.1). In consequence, moderating effects appear for the two types of fit, although the effect of image fit is contrary to the hypothesized direction.

6. Conclusions and implications

Organizations frequently follow brand -extension strategies, trying to leverage current brand associations. Given that the brand is one of the most important assets (Aaker, 1996), marketing managers must know which extensions are more suitable and less risky. An important finding of the study is that extensions that seem successful in terms of consumer acceptance can produce brand image dilution. If the extension has a low fit, there will be a negative effect on the attitude toward the brand, regardless of the attitude toward the extension. However, a high-fit extension does not guarantee the protection of brand image in cases where extension attitude is unsatisfactory. These results reveal an adaptation process of the brand association scheme (Park et al., 1993) and lend support to the existence of bookkeeping processes ([Weber and Crocker, 1983] and[Loken and John, 1993]). The new associations coming from seemingly successful extensions could thus dilute the established brand equity (Sheinin, 2000). The proposition and validation of an empirical model that explain feedback effects of brand extensions also represent a step forward in the literature. The model estimation produced positive goodness-of-fit indexes and provided support for all the specific relationships. The coefficients obtained indicate that perceived image fit, perceived category fit and initial brand image are strong determinants of extension attitude. On the other hand, consumer final beliefs about the brand are the result of the attitude to the new product and initial brand image. The estimation of the model also shows that initial brand image influences perceived category fit and image fit, which act as mediating variables in the initial image -extension attitude relation. All these results appear to be in line with those works that indicate that extension attitude mainly depends on perceived fit ([Aaker and Keller, 1990], [van Riel et al., 2001] and [Vlckner and Sattler, 2006]). Apart from perceived fit and initial associations, the findings also highlight the importance of consumer innovativeness to determine extension attitude. Klink and Smith (2001) demonstrate that perceived fit is less important for innovative consumers, although the effect of specific

dimensions is unknown to date. In the light of the results, innovative consumers do not consider category fit when evaluating brand extensions, something that never occurs in cases of low consumer innovativeness. However, those individuals prone to new ideas and products ([Roehrich, 1995] and [Hem et al., 2003]) put more stress on image fit when evaluating the extension. Therefore, an innovative consumer will not object to buying an extension that largely differs from the current products of the brand, though he will demand higher coherence with the brand image. The results of the research hold important implications for companies launching brand extensions. Despite the extensive research warning of the risks of over-extending the brand, most companies continue leveraging their brand whenever they market a new product ([Keller, 2003] and [Vlckner and Sattler, 2006]). One of the reasons behind this behavior is the wrong belief that successful brand extensions, in terms of market acceptance, cannot be harmful for the pre-existing associations. Consequently, a major recommendation for companies is not to entirely trust market success indicators, and only launch the extension provided that perceived fit is high (Mitchell and Edelman, 2003). That is the way to avoid damaging current associations. The results emphasize that the most important aspect for the extension success is coherence with the image of the extended brand. Though positive, the new product or service does not necessarily have to belong to the same category, but the firm has to be able to transmit the brand essence from one to another market (Kim, 2003). The first consumers of the product, who will probably be the most innovative ones, will especially assess this type of fit. In any case, companies aiming to mature markets should not forget category fit, given that consumers are usually less innovative than in pioneering markets and category fit reduces the level of risk (Smith and Andrews, 1995). Moreover, communicating a favorable initial image will be always positive for consumer acceptance, increasing both the appeal of the new product and fit perception.

6.1. Limitations and future research

The findings of the present research raise some questions that open a variety of future research issues. The first issue refers to the fact that the brand extensions of the study are not real products. In the future, researchers should examine real market conditions in order to verify whether the validated relationships still remain under the effect of other variables (marketing mix, competition actions, etc.). Some researchers opt for real extensions in their studies ([Swaminathan et al., 2001] and [Vlckner and Sattler, 2006]), although they focus on brand -extension evaluation. In order to analyze the effects on brand image, further research should develop the study before and after the actual introduction of the extension, which will require the collaboration of the firm. The second issue to consider is the applicability of the model to extensions of the same category or line extensions. Since line extensions are products with a higher perceived fit degree (Grime et al., 2002), the relationships should show similar coefficients. More interesting could be to study the incidence among brands that compete in the tangible product market and extend into services and vice versa. Given that the participants in the study assessed one only brand and one only extension, future research should overcome the resulting limitations. Having more than one assessment per individual would allow applying new theories to the brand -extension area, such as the matching relationship or matching law (Herrnstein, 1997). According to this theory, individuals frequently match their behavior in proportion to the reward or punishment this behavior obtains ([Herrnstein, 1997] and [Foxall and Schrezenmaier, 2003]). Hence, consumers should opt for the brands or the brand extensions that deliver a higher value or a lower cost (real or perceived).

Finally, methodological questions are susceptible to generate errors and affect the validity of the study. The methodology led to the selection of the scales for each variable, sampling method, fieldwork area and statistical techniques. In order to verify whether the methodology produces consistent results, alternative techniques deserve further attention. For instance, moderate regressions technique is an adequate methodology to analyze the moderating effects of variables (Bagozzi et al., 1992).

References
Aaker, 1996 D.A. Aaker, Measuring brand equity across products and markets, Calif Manage Rev 38 (3) (1996), pp. 102120. View Record in Scopus |Cited By in Scopus (176) Aaker, 2002 D.A. Aaker, Brand portfolio strategy, The Free Press, New York (2002). Aaker and Keller, 1990 D.A. Aaker and K.L. Keller, Consumer evaluations of brand extensions, J Mark 54 (January) (1990), pp. 2741. Full Text via CrossRef Ahluwalia and Grhan-Canli, 2000 R. Ahluwalia and Z. Grhan-Canli, The effects of extensions on the family brand name: an accessibilitydiagnosticity perspective, J Consum Res 27 (3) (2000), pp. 371381 (December). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (50) Alexander and Colgate, 2005 N. Alexander and M. Colgate, Customers' responses to retail brand extensions, J Mark Manag 21 (2005), pp. 393419. Full Text via CrossRef Anderson, 1983 J.R. Anderson, A spreading activation theory of memory, J Verbal Learn Verbal Behav 22 (3) (1983), pp. 261295. Abstract | By in Scopus (297) Bagozzi et al., 1992 R.P. Bagozzi, H. Baumgartner and Y. Yi, State versus action orientation and the theory of reasoned action: an application to coupon usage, J Consum Res 18 (1992), pp. 505518 (March). Full Text via CrossRef Bhat and Reddy, 2001 S. Bhat and S.K. Reddy, The impact of parent brand attribute associations and affect on brand extension evaluation, J Bus Res53 (2001), pp. 111 122. Article | PDF (146 K) | View Record in Scopus | Cited By in Scopus (44) Boush, 1997 D.M. Boush, Brand name effects of interproduct similarity judgments, Mark Lett 8 (4) (1997), pp. 419427. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (4) Boush and Loken, 1991 D.M. Boush and B. Loken, A process-tracing study of brand extension evaluation, J Mark Res 28 (1991), pp. 1628 February. Full Text via CrossRef Boush et al., 1987 D.M. Boush, S. Shipp, B. Loken, E. Genturck, S. Crockett and E. Kennedy et al., Affect generalization to similar and dissimilar brand extensions, Psychol Mark 4 (3) (1987), pp. 225237. Brown and Dacin, 1997 T.J. Brown and P.A. Dacin, The company and the product: corporate associations and consumer product responses, J Mark 61(1997), pp. 6884. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (371) Chang, 2002 J.W. Chang, Will a family brand image be diluted by an unfavorable brand extension? A brand trial-based approach, Adv Consum Res 29 (2002), pp. 299304. Cronbach, 1951 L.J. Cronbach, Coefficient alpha and the internal structure of test, Psichometrica 16 (1951), pp. 297334 (October). Full Text via CrossRef Czellar, 2003 S. Czellar, Consumer attitude towards brand K) | View Record in Scopus | Cited By in Scopus (49) de Ruyter and Wetzels, 2000 Ko. de Ruyter and M. Wetzels, The role of corporate image and extension similarity in service brand extensions, J Econ Psychol 21 (2000), pp. 639659 (Winter). Article | PDF (194 K) | View Record in Scopus | Cited By in Scopus (19) extensions: an integrative model PDF (566 and research propositions, Int J Res Mark 20 (1) (2003), pp. 97115. Article | PDF (2841 K) | View Record in Scopus | Cited

Diamantopoulos et al., 2005 A. Diamantopoulos, G. Smith and I. Grime, The impact of brand extensions on brand personality: experimental evidence,Eur J Mark 39 (1/2) (2005), pp. 129149. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (25) Foxall and Schrezenmaier, 2003 G.R. Foxall and T.C. Schrezenmaier, The behavioral economics of consumer brand choice: establishing a methodology,J Econ Psychol 24 (2003), pp. 675 695. Article | PDF (405 K) | View Record in Scopus | Cited By in Scopus (15) Grime et al., 2002 I. Grime, A. Diamantopoulos and G. Smith, Consumer evaluations of extensions and their effects on the core brand, Eur J Mark 36(11/12) (2002), pp. 1415 1438. Full Text via CrossRef Grhan-Canli and Maheswaran, 1998 Z. Grhan-Canli and D. Maheswaran, The effects of extensions on brand name dilution and enhancement, J Mark Res 35 (1998), pp. 464473 November. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (62) Hair et al., 1998 J.F. Hair, R.E. Anderson, R.L. Tatham and W.C. Black, Multivariate data analysis (5th Ed.), Prentice-Hall, New Jersey (1998). Hem et al., 2003 L.E. Hem, L. de Chernatony and N.M. Iversen, Factors influencing successful brand extensions, J Mark Manag 19 (78) (2003), pp. 781806 (August). Full Text via CrossRef Herrnstein, 1997 R.J. Herrnstein, The matching law: papers in psychology and economics, Harvard University Press, Cambridge, MA (1997). Iglesias and Vzquez, 2001 V. Iglesias and R. Vzquez, The moderating effects of exclusive dealing agreements on distributor satisfaction, J Strat Mark 9 (3) (2001), pp. 119. John et al., 1998 D.R. John, B. Loken and C. Joiner, The negative impact of extensions: can flagship products be diluted, J Mark 62 (1998), pp. 1932 (January). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (79) Keller, 2003 K.L. Keller, Strategic brand management: building, measuring, and managing brand equity (2nd Ed.), Prentice Hall, New York (2003). Keller and Aaker, 1992 K.L. Keller and D.A. Aaker, The effects of sequential introduction of brand extensions, J Mark Res 29 (1992), pp. 3550 (February).Full Text via CrossRef Keller and Sood, 2003 K.L. Keller and S. Sood, Brand equity dilution, MIT Sloan Manag extensions, J Prod Brand Rev (2003), pp. 1215 Autumn. Kim, 2003 J.Y. Kim, Communication message strategies for brand Manag 12 (7) (2003), pp. 462476. Full Text via CrossRef extensions and core brands,

Kim et al., 2001 C.K. Kim, A.M. Lavack and M. Smith, Consumer evaluation of vertical brand J Bus Res 52 (2001), pp. 211222. Article | PDF (140 K) | View Record in Scopus | Cited By in Scopus (16) Kirmani et al., 1999 A. Kirmani, S. Sood and S. Bridges, The ownership effect in consumer responses to brand line stretches, J Mark 63 (1999), pp. 88101 (January). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (57) Kline, 2005 R.B. Kline, Principles and practice of structural equation modeling (2nd Ed.), The Guilford Press, New York (2005). Klink and Smith, 2001 R.R. Klink and D.C. Smith, Threats to the external validity of brand extension research, J Mark Res 38 (2001), pp. 326335 (August). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (89) Lane and Jacobson, 1997 V. Lane and R. Jacobson, The reciprocal impact of brand leveraging: feedback effects from brand extension evaluation to brand evaluation, Mark Lett 8 (3) (1997), pp. 261271. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (16) Lee and Ulgado, 1993 M. Lee and F.M. Ulgado, Service extension strategy: a viable basis for growth?, J Serv Mark 7 (2) (1993), pp. 2435. Full Text via CrossRef Leong et al., 1997 S.M. Leong, S.H. Ang and J. Liau, Dominance and dilution: the effects of extending master brands, J Consum Mark 14 (5) (1997), pp. 380389.

Loken and John, 1993 B. Loken and D.R. John, Diluting brand beliefs: when do brand extensions have a negative impact?, J Mark 57 (1993), pp. 7184 (July). Full Text via CrossRef Low and Lamb, 2000 G.S. Low and C.W. Lamb, The measurement and dimensionality of brand associations, J Prod Brand Manag 9 (6) (2000), pp. 350368. Martin and Brown, 1990 G.S. Martin and T.J. Brown, In search of brand equity: the conceptualization and measurement of the brand impression construct. In: M.L. Childers et al., Editors, Marketing theory and applications vol. 2, American Marketing Association, Chicago (1990), pp. 431438. Full Textvia CrossRef Martnez and de Chernatony, 2004 E. Martnez and L. de Chernatony, The effect of brand extension strategies upon brand image, J Consum Mark 21(1) (2004), pp. 3950. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (28) Martnez and Pina, 2003 E. Martnez and J.M. Pina, The negative impact of brand extension on parent brand image, J Prod Brand Manag 12 (7) (2003), pp. 432448. Full Text via CrossRef Milberg et al., 1997 S.J. Milberg, C.W. Park and M.S. McCarthy, Managing negative feedback effects associated with brand extensions: the impact of alternative branding strategies, J Consum Psychol 6 (2) (1997), pp. 119140. Abstract | Scopus | Cited By in Scopus (56) Milewicz and Herbig, 1994 J. Milewicz and P. Herbig, Evaluating the brand extension decision using a model of reputation building, J Prod Brand Manag 3 (1) (1994), pp. 3947. Full Text via CrossRef Mitchell and Edelman, 2003 Mitchell, V., Edelman, D.J. How to successfully extend your brand. Marketing Profs 2003.http://www.marketingprofs.com/3/edelman1.asp (consulted in August, 2003). Morrin, 1999 M. Morrin, The impact of brand extensions on parent brand memory structures and retrieval processes, J Mark Res 36 (1999), pp. 517525 (November). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (46) Mueller, 1996 R.O. Mueller, Basic principles of structural equation modeling. An introduction to LISREL and EQS, Springer Texts in Statistics, New York (1996). Park et al., 1986 C.W. Park, B.J. Jaworski and D.J. MacInnis, Strategic brand concept management, J Mark 50 (1986), pp. 135145 (October). Full Text via CrossRef Park et al., 1993 C.W. Park, M.S. McCarthy and S.J. Milberg, The effects of direct and associative brand extension strategies on consumer response to brand extensions, Adv Consum Res 20 (1993), pp. 2833. Park et al., 1991 W.C. Park, S. Milberg and R. Lawson, Evaluation of brand extensions: the role of product feature similarity and brand concept consistency, J Consum Res 18 (1991), pp. 185193 (September). Pryor and Brodie, 1998 K. Pryor and R.J. Brodie, How advertising slogans can prime evaluations of brand extensions: further empirical results, J Prod Brand Manag 7 (6) (1998), pp. 497 508. Full Text via CrossRef Ries and Trout, 1993 A. Ries and J. Trout, The 22 inmutable laws of marketing, McGraw-Hill, New York (1993). Roehrich, 1995 G. Roehrich, Innovativits hdoniste et sociale: proposition d'une chelle de mesure, Rech Appl Mark 9 (2) (1995), pp. 1941. Roehrich, 2004 G. Roehrich, Consumer innovativeness: concepts and measurements, J Bus Res 57 (6) (2004), pp. 671677. Article | PDF (156 K) |View Record in Scopus | Cited By in Scopus (47) Rogers, 1995 E. Rogers, Diffusion of innovations (4th Ed.), The Free Press, New York (1995). Romeo, 1991 J.B. Romeo, The effect of negative information on the evaluations of brand extensions and the family brand, Adv Consum Res 18(1991), pp. 399406. PDF (1372 K) | View Record in

Sheinin and Schmitt, 1994 D.A. Sheinin and B.H. Schmitt, Extending brands with new product concepts: the role of category attribute congruity, brand affect, and brand breadth, J Bus Res 31 (1) (1994), pp. 110. Abstract | PDF (1353 K) | View Record in Scopus | Cited By in extensions on parent PDF (126 K) | View Record in Scopus (27) Sheinin, 2000 D.A. Sheinin, The effects of experience with brand brand knowledge, J Bus Res 49 (2000), pp. 4755. Article | Scopus | Cited By in Scopus (19) Siu et al., 2004 N.Y.M. Siu, K.S. Woo and T.H. Cheung, Service quality expectation towards selfservice technology: the case of Internet banking, Proceedings of the 38th academy of marketing conference (2004) Cheltenham (U.K.). Smith and Andrews, 1995 D.C. Smith and J. Andrews, Rethinking the effect of perceived fit on costumers' evaluations of new products, J Acad Mark Sci 23(1995), pp. 115124 (Winter). Swaminathan et al., 2001 V. Swaminathan, R.J. Fox and S.K. Reddy, The impact of brand extension introduction on choice, J Mark 65 (2001), pp. 115 (October). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (51) Tauber, 1988 E.M. Tauber, Brand leverage: strategy for growth in a cost-control world, J Advert Res 28 (1988), pp. 2630 (August/September). Taylor and Bearden, 2002 V.A. Taylor and W.O. Bearden, The effects of price on brand extension evaluations: the moderating role of extension similarity, J Acad Mark Sci 30 (2) (2002), pp. 131140. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (22) van Riel and Ouwersloot, 2005 A. van Riel and H. Ouwersloot, Extending electronic portals with new services: exploring the usefulness of brand extension models, J Retail Consum Serv 12 (3) (2005), pp. 245254. Article | PDF (249 K) | View Record in Scopus | Cited By in Scopus (11) van Riel et al., 2001 A. van Riel, J. Lemmink and H. Ouwersloot, Consumer evaluations of service brand extensions, J Serv Res 3 (3) (2001), pp. 220231 (February). Full Text via CrossRef Villarejo, 2002 A. Villarejo, La Medicin del Valor de Marca en el mbito de la Gestin de Marketing, CEADE, Sevilla (2002). Viswanathan, 1997 Viswanathan, An anchoring and adjustment model of brand extension effects on the core brand consumers. Philosophical Degree, The Ohio State University (1997). Vlckner and Sattler, 2006 F. Vlckner and H. Sattler, Drivers of brand extension success, J Mark 70 (2) (2006), pp. 1834. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (58) Voss et al., 2003 K.E. Voss, E.R. Spangenberg and B. Grohmann, Measuring the hedonic and utilitarian dimensions of consumer attitude, J Mark Res 40(2003), pp. 310320 (August). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (97) Weber and Crocker, 1983 R. Weber and J. Crocker, Cognitive processes in the revision of stereotypic beliefs, J Pers Soc Psychol 45 (1983), pp. 961977 (November). Abstract | (1635 K) | Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (245) Weiss et al., 1999 A.M. Weiss, E. Anderson and D.J. MacInnis, Reputation management as a motivation for sales structure decisions, J Mark 63 (1999), pp. 7489 (October). Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (57) The authors thank the following sources for their financial help: CICYT (Ref: SEJ2005-02315) and Government of Aragon (GENERES, Ref.S-09; PM0262/2006). PDF