Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations

JBR-08057; No of Pages 9 Journal of Business Research xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Business Research

Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations Géraldine Michel a,⁎, Naveen Donthu b,⁎⁎ a b

Sorbonne Graduate Business School, France Georgia State University, USA

a r t i c l e

i n f o

Article history: Received 21 June 2012 Received in revised form 18 March 2014 Accepted 21 March 2014 Available online xxxx Keywords: Brand extension Brand image Central brand associations Peripheral brand associations

a b s t r a c t This research introduces the concept of brand association centrality and distinguishes central brand associations (CBAs) and peripheral brand associations (PBAs). Study 1 shows that the consistency with CBAs and PBAs influences perceived brand extension fit. However, the brand extension consistency with CBAs positively affects consumer–brand extension evaluations, while the brand extension consistency with PBAs does not have a significant effect. Study 2 demonstrates the mediating role of perceived fit between brand extension consistency with CBAs and brand extension evaluations. The study shows that the more inconsistent the product extension is with CBAs, the more the brand's image is altered and the faster brand evaluations deteriorate. In contrast, the brand extension inconsistency with PBAs does not affect brand evaluations. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Brand extension failure does not always have a negative impact on the parent brand; for example, the Bic perfume did not negatively affect attitudes toward Bic. Conversely, brand extension success does not systematically cause positive feedback on the parent brand; the success of Calvin Klein brand extensions (underwear, perfume) did not positively affect the Calvin Klein brand image. How can this paradox be explained? Are the causes of brand extension success similar to those of a positive brand extension impact? Many studies examine the process of brand extension evaluation (Bhat & Reddy, 2001; Dens & De Pelsmacker, 2010; Kapoor & Heslop, 2009), identifying brand-related variables such as brand concept (Monga & John, 2010), quality level (Völkner & Sattler, 2007) or consumer–brand relationship quality (Kim, Park, & Kim, 2014). Other studies investigate the brand extension impact on brand attitudes (Loken & John, 1993; Martinez, Montaner, & Pina, 2009), identifying consumer related variables such as styles of thinking (Monga & John, 2010). Finally, a central variable explaining brand extension evaluations and brand extension reciprocal effects is the relationship between parent brands and brand extensions. In general, this

⁎ Correspondence to: G. Michel, Sorbonne Graduate Business School, 21 rue Broca, 75240 Paris cedex 05, France Tel.: +33 1 53 55 28 00 ⁎⁎ Correspondence to: N. Donthu, Department of Marketing, Georgia State University, Atlanta, GA 30303, USA. Tel.: +1 404 413 7662. E-mail addresses: [email protected] (G. Michel), [email protected] (N. Donthu).

relationship derives from brand extension consistency, commonly referred to as the “fit” between the extension and the parent brand, and reflects consumer perceptions of the logical connection between the product and the brand (Tauber, 1988). Research usually measures fit using direct and holistic methods (Ahluwalia & Gürhan-Canli, 2000; Desai & Keller, 2002; Keller & Aaker, 1992; Park, Milberg, & Lawson, 1991). As Broniarczyk and Alba (1994) note, associations are a key factor in assessing the consistency between brands and brand extensions. However, no study demonstrates the influence of brand associations on perceived brand extension fit or examines brand extension evaluations and their impact on parent brand attitudes. What is the role of brand associations on consumer perceptions of a new product's consistency with the brand? What are the roles of brand associations on consumer evaluations of an extension? How are brand associations altered? This study examines the specific role of brand associations on both perceived brand extension fit and brand extension evaluation and also investigates the brand extension impact on the parent brand. This study applies social representation theory (Abric, 1994; Moscovici, 1988) to introduce the concepts of central brand associations (CBAs) and peripheral brand associations (PBAs). “Central” and “peripheral” are the terms originally used in the social representation theory (Abric, 1994; Moscovici, 1988), they are no link with the dual processing models (Petty & Cacioppo, 1986). Brand associations are central if the majority of consumers perceive them as indispensable to express the brand meaning (Abric, 1994). Conversely, peripheral associations are strong but not entirely inseparable from the core brand. This framework distinguishes the brand extension consistency with CBAs and PBAs.

http://dx.doi.org/10.1016/j.jbusres.2014.03.020 0148-2963/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

2

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

Study 1 analyzes the link between the brand extension consistency with CBAs and PBAs and perceived brand extension fit. The findings reveal that the brand extension inconsistency with both CBAs and PBAs negatively influences fit. In addition, the holistic fit measure does not capture the brand association centrality in the brand meaning. Study 1 also shows that the consistency with CBAs but not with PBAs influences brand extension evaluations. Study 2 analyzes brand extension evaluations and their impact on four major food industry brands. The study reveals that though the brand extension consistency with CBAs influences brand extension evaluations, the fit explains these evaluations better. Brand extension inconsistencies with CBAs and PBAs also explain brand attitude deterioration and brand image dilution. Finally, this research shows that fit is a mediating variable between brand extension consistency with CBAs and PBAs and brand extension evaluations, but perceived fit does not have a mediating effect on the brand extension impact. This study offers an explanation of the contradictions in the results of previous research on the brand extension fit impact (Jap, 1993; Milberg, Park, & McCarthy, 1997). That is, consumer perceptions of a weak brand extension fit might be due to the inconsistency between the brand extension and PBAs. By definition, PBAs are not connected with brand meaning, and therefore this low fit does not disturb the brand meaning. In contrast, the low fit due to the inconsistency between the brand extension and CBAs disturbs the brand meaning and provokes a negative brand impact. 2. Conceptual background and hypotheses 2.1. Brand image and brand associations In general, research uses the theory of semantic networks to understand how consumers perceive brands (Keller, 1993; Rosch, 1975). According to this theory, brand knowledge is a set of associations organized around the brand in the consumer's memory (Loken, Barsalou, & Joiner, 2007). The consumer's mental representation of the brand's image then determines the strength, valence, uniqueness, and cohesion of associations attributed to the brand (Keller, 1993). The primary interest of this theoretical framework helps identify the power of brand equity, which is represented by brands with strong, positive, unique, and coherent associations connected with the product (Yoo, Donthu, & Lee, 2000). However, no studies account for the role of associations in the change of brand image. What is the role of different types of associations on perceived inconsistency between the brand extension and the brand image? Are some associations more stable than others? This study applies a social representation approach to address these questions. 2.2. Brand category: a concept structured around CBAs and PBAs Perceived category depends on goals that are salient at a particular time or in a particular context (Barsalou, 1982). As an extension of previous studies demonstrating the stability and flexibility of consumer category representations, this study proposes a new framework for examining brand categories and associations—that is, social representations. According to Moscovici (1988), social representations are formed and shared socially because they are based on experiences and shaped by information, knowledge, and thought processes received and transmitted through tradition, education, and social communication. Social representation aims to establish a common vision of reality within a social ensemble (group, class, or cultural). Thus, this approach considers the brand category organized around a core and peripheral system (Abric, 1994). Using this approach, this study distinguishes brand associations according to their degree of centrality (CBAs and PBAs). CBAs are indispensable for expressing the brand meaning. They differ from strong associations, which are linked to the brand in memory (Keller, 1993), in that people can have two strong associations, but only one is central. For example, in the case of Malibu brand, although respondents considered both associations as

strong, “exotic” was a central element and “sweet” was peripheral (Ambroise & Valette-Florence, 2005). Conversely, the peripheral system integrates associations that are strong but not central (Abric, 1994). For example, “Clothing” association is strongly linked to the respondents' Zara image, but is not central. Even if the product is not clothing, it can still be considered a Zara product as long as it is consistent with the central brand associations “feminine” and “fashionable” (Ambroise & Valette-Florence, 2005). With this social representation approach, we distinguish between CBAs and PBAs (Michel, 1999).

2.3. The link between brand extension fit and brand extension consistency with CBAs and PBAs Mandler (1982) defines the concept of congruence as a structural correspondence between two entities. Thus, perceived consistency between a brand and an extension means that the consumer perceives the product as “logical with the image of the brand” (Tauber, 1988, p. 28). To understand the basis of perceived fit, previous research suggests taking into account the brand and brand extension image (Murphy & Medin, 1985; Park et al., 1991). In line with social representation theory (Abric, 1994), this study differentiates the consequences of brand extension consistency with CBAs from brand extension consistency with PBAs on perceived fit. Because CBAs are central to the brand meaning, when brand extension consistency with CBAs is weak, consumers are more likely to perceive a weak brand extension fit. In contrast, because PBAs are contextual, inconsistency with these associations does not influence perceived brand extension fit. Thus: H1a. Brand extension fit is higher when the brand extension is consistent rather than inconsistent with CBAs. H1b. Brand extension fit is not significantly different when the brand extension is consistent rather than inconsistent with PBAs.

2.4. Brand extension evaluation A high degree of fit is associated with more positive brand extension evaluations (Aaker & Keller, 1990; Boush & Loken, 1991). Again building on social representation theory (Abric, 1994), this study differentiates the influence of brand extension consistency with CBAs from brand extension consistency with PBAs on brand extension evaluations. When central consistency is weak, consumers are more likely to evaluate the extension negatively. In contrast, because PBAs are contextual, inconsistency with these associations does not affect brand extension evaluations. Thus: H2a. Brand extension evaluation is greater when the brand extension is consistent rather than inconsistent with CBAs. H2b. Brand extension evaluation is not significantly different when the brand extension is consistent rather than inconsistent with PBAs.

2.5. Mediating role of brand extension fit Brand extension consistency with CBAs and PBAs influences perceived brand extension fit. The brand extension consistency with CBAs influences brand extension evaluations. In contrast, the effect of PBAs on brand extension evaluations is not significant. This study proposes that brand extension fit mediates the relationship between brand extension consistency with CBAs and brand extension evaluations.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

3

H3. Perceived fit has a mediating effect on the relationship between the brand extension inconsistency with CBAs and brand extension evaluations.

each association with reference to their brand image. They formulated their answers using a seven-point strength scale (1 = weakly connected, 7 = strongly connected). Strong brand associations received an average of 5 or more on the scale.

2.6. The impact of brand extension on brand attitudes

3.2. Identification of CBAs and PBAs

To clarify the brand extension influence on the parent brand, this study defines the brand extension inconsistency with the parent brand according to two dimensions: the inconsistency between the brand extension and (1) CBAs and (2) PBAs of the brand. When new information is inconsistent with important elements of the person's selfconcept, his or her attitudes will change (Moliner, Joule, & Flament, 1995). Furthermore, the attitude change depends on the importance of cognitions that cause the cognitive dissonance (Festinger, 1957), such that the dissonance due to the inconsistency with CBAs will be greater than that due to the inconsistency with PBAs. In parallel, perceived brand extension fit does not account for the importance of brand associations that are inconsistent with the brand extension. Thus:

The identification of CBAs is based on the principle of refutation developed in central core theory (Moliner, 1995, 1996). If questioning one association leads a majority of consumers to reject the brand, this suggests that the association is central to the brand's image. For example, do consumers view the word “X” associated with the [brand name] as central? Respondents replied on a seven-point scale (1 = hardly probable, 7 = very probable) to the question: “If the product is not X, can it be a [brand name] product?” If a significant number of respondents reply between 1 and 3 (inference χ2: significant difference at 1% with a theoretical distribution), this means that “X” is a central association. Otherwise, the association is peripheral. 3.3. Measuring brand extension inconsistency with CBAs and PBAs

H4. The brand extension inconsistency with CBAs influences the brand evaluation change more negatively than (a) the brand extension inconsistency with PBAs or (b) perceived brand extension fit.

Two dimensions capture the brand extension consistency with the brand: brand extension consistency with (1) CBAs and (2) PBAs. The Euclidean distance between the associations' strength attributed to the brand and that attributed to the extension is as follows:

H5. The brand extension inconsistency with CBAs influences the change in CBA strength more positively than (a) the brand extension inconsistency with PBAs or (b) perceived brand extension fit.

(1) InconsistencyBE/CBA = √Σ(Sci/brand − Si/BE)2, where Sci/brand = strength of CBA i and Si/BE = strength of brand extension association i; (2) Inconsistency BE/PBA = √Σ(Spi/brand − Si/BE)2, where Spi/brand = strength of PBA i and Si/BE = strength of brand extension association i.

H6. The brand extension inconsistency with CBAs influences the change in PBA strength more positively than (a) the brand extension inconsistency with PBAs or (b) perceived brand extension fit. The hypotheses related to the impact of brand extension on brand attitude are presented in Fig. 1.

3.4. Measuring perceived brand extension fit 3. Method and measures Studies 1 and 2 test the hypotheses. Study 1 is an experiment with a real brand and false brand extensions to test H1a, H1b, H2a and H2b. Study 2 uses real brand extensions to test H3–H6. Both studies use the same measures.

In line with Monga and John (2010), the respondents report their perceptions of brand extension fit on seven-point scales (e.g., doesn't fit with the brand/fits with the brand, inconsistent with the brand/ consistent with the brand). The scale is reliable (Study 1 α = .91; Study 2 α = .82). Given the high internal consistency of the scale, the averaged perceptions form a mean score for each respondent.

3.1. Identification of brand associations and strength

3.5. Measuring brand extension image and evaluation

In order to generate brand associations and detect the changes of brand associations, we use the free association method (Broniarczyk & Alba, 1994) and in-depth interviews. All interviews were taperecorded and transcribed before the thematic content analysis. An expert panel (Three assistant professors in the marketing and linguistic department assisted in this process) reviewed a list of brand associations to eliminate any redundant associations, to prevent inflating the importance of certain associations to the detriment of others. Using brand associations lists, the respondents evaluated the strength of

Respondents first express their thoughts about the brand extension, indicating the strength of brand associations with the product: “Even though you have never tried this product, what went through your mind when you saw this product; that is, what is the association's strength connected with this product?” (1 = weakly connected, 7 = strongly connected). In line with prior research (Bhat & Reddy, 2001; Klink & Smith, 2001), they indicate their brand extension evaluations on seven-point scales with two questions: “I like this [extension product]” and “I have a good opinion of this [extension product].” The

BE inconsistency with PBAs

H1

H3

Perceived fit

H3

BE evaluation

H2

BE inconsistency with CBAs

Brand attitude change H4 H5

CBAs strength change PBAs strength change

H6 Fig. 1. Global model of the research.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

4

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

scale is reliable (α = .80). The averaged evaluations form a mean score for each respondent.

4.1. Procedure An experiment with a before-and-after design was conducted with 101 students (female 52%, age average: 27.5). The questionnaires were administered to several university classes from an executive program before the beginning of the course. The respondents took part in the main experiment using a 2 (brand extension consistency vs. brand extension inconsistency with CBAs) × 2 (brand extension consistency vs. brand extension inconsistency with PBAs) between-subjects design. The students completed a questionnaire before and after exposure to the brand extension. The stimulus materials describe the characteristics of a fictitious brand extension. To manipulate the inconsistency with CBAs, the camera is described as non-disposable (vs. disposable), inexpensive, and practical; the dishes are presented as expensive (vs. inexpensive), disposable, and practical. To manipulate the inconsistency with PBAs, the camera is described as metallic (vs. plastic), functional, and simple; the dishes are presented as made of cardboard (vs. plastic), functional, and simple (see Table 2). Each respondent evaluated two brand extensions (one camera and one dish), for a total of 202 (101 × 2) responses.

3.6. Measuring the brand evaluation and its changes In line with similar research (Kapoor & Heslop, 2009; Völkner & Sattler, 2007), the respondents report their brand evaluations on seven-point scales with two items: “Generally, I think that brand X is a very good make” (1 = totally disagree, 7 = totally agree) and “Circle the score that best corresponds to the opinion you have of the brand” (1 = negative, 7 = positive). The scale is reliable (α = .83). The measure of brand evaluation change is calculated by the difference between the evaluation before and after brand extension presentation (T1 and T2). 3.7. Measuring brand association changes Consumers assess the brand associations' strength twice in relation to their idea of the brand on a seven-point strength scale (1 = weakly connected, 7 = strongly connected). The brand associations' change is measured by calculating the difference between the brand associations' strength before and after exposure to the brand extension (Martinez & Pina, 2009). The global change of the brand's core represents the sum of differences in strength for all CBAs. The global change of the brand's peripheral system represents the sum of differences in strength for all PBAs.

4.2. Results The manipulation check results (ANOVA) indicate that the disposable camera and the inexpensive dishes have a higher consistency with the CBA association than the non-disposable camera and the expensive dishes (inconsistency with CBAIC/CP M = 3.19 vs. inconsistency with CBACC/CP M = 2.23; F(1, 96) = 9.50, p b .001; inconsistency with CBAIC/IP M = 3.50 vs. inconsistency with CBACC/IP M = 2.33; F(1, 96) = 12.67, p b .01). Regarding brand extension consistency with PBAs, the manipulation check results (ANOVA) indicate that the plastic camera and the plastic dishes have a higher consistency with the PBA association than the metallic camera and the cardboard dishes, respectively (inconsistency with PBACC/CP M = 6.25 vs. inconsistency with PBACC/IP M = 7.43; F(1, 99) = 5.08, p b .05; inconsistency with PBAIC/CP M = 6.53 vs. inconsistency with PBAIC/IP M = 8.13; F(1, 83) = 7.21, p b .01). A 2 (brand extension consistency with CBAs) × 2 (brand extension consistency with PBAs) between-subjects ANOVA analyzes fit. As predicted, brand extension fit is higher when the brand extension is consistent rather than inconsistent with CBAs (M = 4.18 vs. 3.70; F(1, 202) = 5.07, p b .05), in support of H1a. Unexpectedly, brand extension fit is higher when the brand extension is consistent rather than inconsistent with PBAs (M = 4.20 vs. 3.67; F(1, 202) = 6.27, p b .05); thus, H1b is not supported. Fig. 2 depicts the link between fit and brand extension consistency with CBAs and PBAs.

4. Study 1: Brand extension evaluation, fit vs. consistency with CBAs and PBAs Study 1 is a study about the Bic brand. In general, people are familiar with the Bic brand in different markets (e.g., pen, razors, lighters), which helps them imagine fictive new products of Bic in other markets. The investigation begins by conducting a qualitative study to identify the CBAs and PBAs of Bic brand. 75 students, taking graduate management classes at a French University, were interviewed (female 60%, age average = 29). Ten associations have a score equal to or higher than 5 (see Table 1) and thus are strong associations. Table 1 illustrates the Bic image by distinguishing CBAs (“functional”, “inexpensive”, “disposable”) and PBAs (“practical”, “simple”, “plastic”, “democratization”, etc.). Next, the brand extension consistency with CBAs and PBAs is manipulated by selecting fictitious brand extensions. We selected camera being a credible as a disposable/non-disposable or plastic/metallic product, and we selected dishes being credible as a expensive/inexpensive or cardboard/plastic product. Table 1 CBAs and PBAs of Bic Brand (Study 1). Association strength

Centrality Total sample

N = 101

N = 101

1 to 7 scale

Centrality Sample divided by 2 N = 50

N = 51

1 to 3

X2

1 to 3

X2

1 to 3

X2

CBAs Functional Inexpensive Disposable

5.55 5.93 5.50

88** 84** 79**

55.69** 46.24** 32.16**

36 35 35

9.68** 8.00** 8.00**

39 49 34

15.68**

PBAs Practical Simple Plastic Democratization Lighters Pen Razor

5.66 5.74 6.05 5.04 5.83 6.45 5.71

63 58 58 55 46 40 28

4.36* 2.22 2.22 .80 .80 4.36* 20.00**

32 30 27 29 23 20 15

3.92 1.58 .32 1.28 .32 2.00 8.00**

31 28 31 26 23 20 13

2.88 .72 2.88 .80 .32 2.00 11.52**

** significant at .01 level, * significant at .05 level.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

between brand extension evaluation and brand extension consistency with CBAs and PBAs.

Table 2 Descriptive variables of the experiment (Study 1). Brand extension

Consistent with central associations

Inconsistent with central associations

Consistent with peripheral associations

CC/CP Disposable & plastic camera

IC/CP Non-disposable & plastic camera Fit = 3.87 Inconsistent CBA = 3.09 Inconsistent PBA = 6.55 BE evaluation = 3.25 Expensive & plastic dishes Fit = 4.15 Inconsistent CBA = 3.34 Inconsistent PBA = 6.51 BE evaluation = 3.85 Total N = 44 Fit = 4.05 σ = 1.53 Inconsistent CBA = 3.19 σ = 1.82 Inconsistent PBA = 6.53 σ = 2.96 BE evaluation = 3.48; σ = 1.32 IC/IP Non-disposable & metallic camera Fit = 2.98 Inconsistent CBA = 4.55 Inconsistent PBA = 8.61 BE evaluation = 3.85 Expensive & cardboard dishes Fit = 3.38 Inconsistent CBA = 3.03 Inconsistent PBA = 7.92 BE evaluation = 3.20 Total N = 44 Fit = 3.35 σ = 1.40 Inconsistent CBA = 3.50 σ = 1.90 Inconsistent PBA = 8.13 σ = 2.44 BE evaluation = 3.16, σ = 1.04

Inconsistent with peripheral associations

5

Fit = 4.64 Inconsistent CBA = 2.06 Inconsistent PBA = 6.44 BE evaluation = 4.32 Inexpensive & plastic dishes Fit = 3.94 Inconsistent CBA = 2.38 Inconsistent PBA = 6.06 BE evaluation = 4.20 Total N = 55 Fit = 4.36 σ = 1.54 Inconsistent CBA = 2.23 σ = 1.17 Inconsistent PBA = 6.25 σ = 2.39 BE evaluation = 4.26; σ = 1.33 CC/IP Disposable & metallic camera Fit = 3.86 Inconsistent CBA = 2.40 Inconsistent PBA = 6.94 BE evaluation = 3.33 Inexpensive & cardboard dishes Fit = 3.79 Inconsistent CBA = 2.26 Inconsistent PBA = 7.90 BE evaluation = 4.48 Total N = 59 Fit = 4.00 σ = 1.47 Inconsistent CBA = 2.33 σ = 1.32 Inconsistent PBA = 7.43 σ = 2.79 BE evaluation = 3.89; σ = 1.45

A 2 (brand extension consistency with CBAs) × 2 (brand extension consistency with PBAs) between-subjects ANOVA analyzes brand extension evaluation. As predicted, brand extension consistency with CBAs is significant (F(1, 202) = 16.27, p b .001). Brand extension evaluation is greater when the brand extension is consistent rather than inconsistent with CBAs (M = 4.08 vs. 3.32), in support of H2a. Brand extension evaluations does not differ significantly when the brand extension is consistent or inconsistent with PBAs (M = 3.87 vs. 3.53; F(1, 202) = 3.40, p = .07), in support of H2b. Fig. 3 depicts the link

4.3. Discussion The findings show that the higher the inconsistency between the brand extension and PBAs, the lower is the fit, regardless of the consistency level between brand extension and CBAs. Similarly, the higher the inconsistency between the brand extension and CBAs, the lower is the fit, regardless of the consistency level between brand extension and PBAs. These findings indicate that perceived fit can be weak even when the brand extension consistency with CBAs is high; conversely, perceived fit can be high even when the brand extension consistency with CBAs is weak. In parallel, the brand extension consistency with CBAs explains brand extension evaluation better than the brand extension consistency with PBAs. The multiple regression shows a positive, significant impact of perceived brand extension fit and brand extension consistency with CBAs on brand extension evaluations and a nonsignificant influence of brand extension consistency with PBAs (βfit = .507, p b .001; βconsistency CBAs = − .409, p b .001; βconsistency PBAs = −.209, p = .834; R2 = .42). To further explore this issue, Study 2 analyzes the mediating role of perceived brand extension fit and the influence of the brand extension consistency on brand attitude changes. 5. Study 2: Brand extension evaluation and its impact on the parent brand 5.1. Study 2a: Identification of brand associations 5.1.1. Procedure Study 2a aims to identify the associations with the brand's current image, the brand extension image, and the potential changes in brand image after exposure to the brand extension. The study included four well-known brands (Andros, McCain Foods, Weight Watchers, and Yoplait) and their extensions (Andros fruit juices, McCain iced teas, Weight Watchers yogurts, and Yoplait meal substitutes). The four brands have the same product portfolio size (a maximum of four product ranges) to avoid brand dilution effects (slightly diluted vs. highly diluted), which could influence the brand extension impact (Kardes & Allen, 1991). Identifying the impact of a brand extension assumes that consumers do not know the brand extension product but know the brand extension category. Consumers were selected according to their answers to three filter questions: (1) “Are you familiar with brand X?” (2) “Here is a list of products. Which ones are commercialized by brand X?” and (3) “Do you buy the [product of extension]?” According to the saturation principle, eight people were interviewed for each brand (32 interviews in all; 84% female and age average = 35).

Fig. 2. Link between fit and brand extension consistency with CBAs and PBAs.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

6

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

Fig. 3. Brand extension evaluation.

5.1.2. Results The entire process generated a list of approximately 30 associations for each brand. The same expert panel as previous helped eliminate any redundant associations; the process generated four lists from 20 to 23 associations for each brand. These lists contain associations connected with the brand's image, the characteristics of its existing brand extensions, and new potential brand associations. These lists are part of the quantitative phase. 5.2. Study 2b: Brand extension evaluation and its impact on the brand 5.2.1. Procedure A before-and-after survey was conducted, including the use of a control group (Jap, 1993; Keller & Aaker, 1992; Park et al., 1991). This method offers the advantage of analyzing intra-individual changes in attitude after presenting the brand extensions. Data collection occurred, at the workplace, through two self-administered questionnaires over a 10day interval, to limit history effects (Zikmund & Babin, 2007). The respondents read the text describing the characteristics of the product and saw the photo showing the real product with its packaging and physical details. A control group evaluated the brand on two occasions, 10 days apart, without being exposed to the brand extension. Collected data came from a sample of 520 people, 120 of whom made up the control group. The respondents were familiar with brand X and did not cite the brand extension in consideration (filter question). The respondents are mostly women (84%; age average = 33.6). According to chi-square tests, the experimental and control groups are no different in terms of household size, gender, and socio-professional categories. However,

more people aged 25 to 34 were in the control group (59.2% compared with 39.1%) than those aged 15–24 and 35–50. 5.2.2. Results The associations are deemed part of the brand image if they receive an average of 5 or more on a seven-point scale. The results show 10 to 12 strong associations for each brand, and each brand has four CBAs (see Table 3). The identification of CBAs and PBAs is systematic for the two sub-samples randomly split. A t-test compares the change of brand image and brand evaluation between the experimental group and the control group. Use of a control group ensures that changes in brand evaluation are due to the brand extension presentation. Changes in brand evaluations are significant for the four brands (t-values = 2.20–3.67, all significant). Hierarchical regressions are used with brand evaluation as dependent variable, and with 3 groups of independent variables (inconsistency brand extension/CBAs, inconsistency brand extension/PBAs, and fit; see Table 4). The findings reveal that the brand extension inconsistency with CBAs more negatively influences brand evaluations than either the brand extension inconsistency with PBAs or the perceived fit for the total sample (βCBA = − .301, βfit = .057, βPBA = .030, R2 = .098) and for the four brands (Averages: βCBA = − .244 [− .191, − .301]; βPBA = .054 [.022, .085]; βfit = .077 [.041, .081]; R2 = .07). These results confirm H4a and H4b. Changes in brand image (CBA, PBA) are significant for the four brands (t-values = 3.88–to 5.15; all significant). Hierarchical regressions are completed with CBA change as the dependent variable. As expected, the brand extension inconsistency with CBAs influences the CBAs change more than either the brand extension inconsistency with

Table 3 Centrality of brand associations (Study 2b). Andros

McCain

1 to 7 scale

1 to 3

4 to 7

CBAs Fruits Pleasure Natural Flavor

96a 73a 71a 64a

3 26 28 35

53 49 48 48 47 47 45

46 50 51 51 52 52 54

PBAs Jam Vitamins Practical Good humor Family Child

a

Weight W 1 to 3

4 to 7

American Modern Abundance Practical

92a 80a 74a 71a

6 18 24 27

Pleasure Youthful Chips Dynamic Original Pizzas Convivial Family

62 62 61 61 60 59 58 55

36 36 37 37 38 39 40 43

Yoplait 1 to 3

4 to 7

Slimming Fat-free Balanced Women

97a 95a 78a 83a

2 4 21 16

Healthy American Modern Variety Guilt-free Expensive Top range Frozen ml.

61 60 51 55 52 38 40 29

38 39 48 44 47 61 59 70

1 to 3

4 to 7

Milk Freshness Health Pleasure

91a 87a 82a 67a

9 13 18 33

Agreeable Family Natural Dessert Child Indulgence Yogurt

58 57 56 55 52 42 38

42 43 44 45 48 58 62

X2 significant at .01 level.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

7

Table 4 Hierarchical regressions (Study 2b). β standardized regression coef.

Hypothesis BE evaluation BE inconsistency with CBA BE inconsistency with PBA Fit between BE and brand Brand evaluation change H4 BE inconsistency with CBA H4a BE inconsistency with PBA H4b fit between brand and BE CBA strength change H5 BE inconsistency with CBA H5a BE inconsistency with PBA H5b fit between brand and BE PBA strength change H6 BE inconsistency with CBA H6a BE inconsistency with PBA H6b fit between brand and BE

ANOVA

R² and ΔR² test

F(3,393) = 19.11, p b .001

R2 = .128

F(3,393) = 14.21, p b .001

ΔR² ΔR²** R2 = .098

F(3,393) = 41.86, p b .001

ΔR² ΔR² R2 = .243

F(3,393) = 14.11, p b .001

ΔR² ΔR² R2 = .097

−.164** .037 .281** −.301** .030 .057 .516** −.001 .097* .020 .303** .004

ΔR²** ΔR²

Note. BE = brand extension. **p b .001, *p b .005.

PBAs or perceived fit for the total sample (βCBA = .516, βfit = .097, βPBA = .001, R2 = .243) and for the four brands (Averages: βCBA = .460 [.385, .513]; βPBA = .050 [.038, .063]; βfit = .066 [.001, .139]; R2 = .218). These results confirm H5a and H5b. Finally, hierarchical regressions are conducted with PBA change as the dependent variable. Unexpectedly, the influence of brand extension inconsistency with CBAs does not explain the PBA changes better than the brand extension inconsistency with PBAs for the total sample (βCBA = .020, βfit =− .004, βPBA = .303, R2 = .097) and for the four brands (Averages: βCBA = .104 [.020, .119]; βPBA = .250 [.218, .336]; βfit = .054 [.052, .153]; R2 = .097). These results do not support H6a and H6b. H3 predicts a mediating effect of brand extension fit between the brand extension consistency with CBAs and, respectively, brand extension evaluation and brand attitude change. To test this mediation, a full analysis of the covariance structural model (Amos) is conducted using the Bootstrap test of mediation (Zhao, Lynch, & Chen, 2010). The fit indices show that the comparative fit index is .89, which exceeds suggested limits (Bentler, 1990). Furthermore, the root mean square error of approximation (.11) is above recommended levels (Hair, Black, Babin, Anderson, & Tatham, 2006). The chi-square value for the model is 315.44, with 49 degrees of freedom (p b .01). To test the mediating ability of perceived fit, the indirect effect of brand extension consistency with CBAs on (1) brand extension evaluation, (2) brand evaluation change, (3) CBAs strength change, and (4) PBAs strength change through perceived fit is examined. As Table 5 shows, the indirect effects through perceived fit are significant and help explain brand extension evaluation (inconsistency with CBAs → brand extension evaluation through fit: β = −.087, p b .05). Thus, the higher the brand extension consistency with CBAs, the higher is the perceived fit and the more positive is the brand extension evaluation, in support of H3. In addition, fit does not mediate the impact of brand extension inconsistency with CBAs on brand attitudes.

5.2.3. Discussion Study 2 shows that the brand extension inconsistency with CBAs more negatively influences brand extension evaluations and better modifies the brand image than either the brand extension inconsistency with PBAs or fit. These findings confirm H4 and H5 (see Table 6). However, the results do not confirm H6; the brand extension inconsistency with CBAs does not influence PBA change more than the brand extension inconsistency with PBAs. Finally, the study confirms H3, showing the mediating role of brand extension fit between the brand extension inconsistency with CBAs and brand extension evaluation. 6. General discussion 6.1. Theoretical implications This research introduces the notion of centrality of brand associations to enhance research that distinguishes brand associations according to their strength, valence, uniqueness, and cohesion (Keller, 1993). As a complement, this research analyzes brand associations according to their role in the internal organization of brands. CBAs are both strong and inseparable (Abric, 1994) from the brand's image, distinct from strong associations, which are connected with the brand through intense links (Keller, 1993). The conceptual framework distinguishes two dimensions of brand extension consistency—that with CBAs and PBAs—to identify the brand associations that play a specific role in brand extension evaluations and the process of brand image change. The findings show that consumers can positively evaluate a brand extension even if that extension is inconsistent with PBAs. In contrast, inconsistency with CBAs negatively affects brand extension evaluation, because CBAs are inseparable from the brand meaning. These results are in line with the work of Mandler (1982), which introduces the notion of moderate congruence

Table 5 Summary of standardized indirect, direct and total effects (study 2b, test of H3). Sample n = 396

Std effects

BE inconsistency CBAs → BE evaluation through fit Indirect effects −.087* Direct effects −.145* Total effects −.231**

Confidence interval

Sample n = 396

[−.131, −047] [−.240, −.036] [−327, −.124]

BE inconsistency CBAs → PBAs strength change through fit Indirect effects .005 [−.023, .033] Direct effects .158* [.038, .276] Total effects .163** [.046, .273]

BE inconsistency CBAs → brand evaluation change through fit Indirect effects −.014 [−.047, .012] Direct effects −.288** [−.385, −.195] Total effects −.302** [−.385, −.126] 2 CFI = .89; RMSEA = .11; X = 315.44 df = 49 (p b .01)

Std effects

BE inconsistency CBAs → CBA strength change through fit Indirect effects −.027 Direct effects .527** Total effects .501**

Confidence interval

[−.057, .001] [.422, .625] [.403, .590]

*p b .05, **p b .01. Significance levels are based on BC bootstrap confidence intervals.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

Not supported

Supported

Experiment with one real brand and 4 false brand extensions, manipulating the consistency between brand extension with CBAs and PBAs.

H1a. The brand extension fit is higher when the brand extension is consistent with CBAs than when it is inconsistent with CBAs H1b. The brand extension fit is not significantly different when the brand extension is consistent with PBAs than when it is inconsistent with PBAs H2a. Brand extension evaluation is greater when the brand extension is consistent with CBAs than when it is inconsistent with CBAs. H2b. Brand extension evaluation is not significantly different when the brand extension is consistent with PBAs than when it is inconsistent with PBAs. H3. Perceived fit has a mediating effect on the relationship between the brand extension inconsistency with CBAs and brand extension evaluations H4. The brand extension inconsistency with CBAs influences the brand evaluation change more negatively than (a) the brand extension inconsistency with PBAs or (b) the perceived brand extension fit H5. The brand extension inconsistency with CBAs influences the change in CBA strength more positively than (a) the brand extension inconsistency with PBAs or (b) the perceived brand extension fit. H6. The brand extension inconsistency with CBAs influences the change in PBA strength more positively than (a) the brand extension inconsistency with PBAs or (b) the perceived brand extension fit

Survey with 4 real brand extensions in the food industry.

Methods Hypotheses

Table 6 Summary of hypotheses and findings.

Supported Not supported Supported Supported Supported Supported

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

Results

8

and assumes a non-linear link between the congruence and object evaluation. Thus, the inconsistency between the brand extension and PBAs provokes weak comprehensible incongruence, which does not negatively affect brand extension evaluations. This research also reveals that consumers do not use the same thinking styles for brand extension evaluations (more holistic with perceived fit) as for the process of brand attitude changes (more analytic fit distinguishing the consistency with CBAs and PBAs). Regarding the reciprocal effects of brand extension on brand attitude, the findings show that brand extension inconsistency with PBAs explains the changes in PBAs but does not influence changes in brand evaluation. In parallel, brand extension inconsistency with CBAs negatively influences brand extension evaluation and positively influences the changes of CBAs. The greater power of brand extension consistency with CBAs in explaining brand attitude changes confirms prior research on attitude change (Mandler, 1982; Petty & Cacioppo, 1986), which affirms that the importance of a questioning element is central to understanding attitude change. Finally, the results for the significant reciprocity effects of brand extension confirm the work of Kardes and Allen (1991) and Lane and Jacobson (1997). However, they contradict the work of Jap (1993), which asserts that brand extensions do not influence brand evaluations. The divergence of these results might be due to the choice of brands and the methodology used. On the one hand, the brands examined herein have a narrow range of products. Therefore, the addition of a new product represents a major event that entails a change in brand evaluation. On the other hand, Jap uses a method based on comparing evaluations of the brand in a control group and an experimental group. This technique assesses interindividual changes in brand evaluation, while the present study is based on an intra-individual analysis. Finally, in comparing the findings of brand extension evaluations and the brand extension impact on the parent brand, this research reveals that positive brand extension evaluations (caused by perceived fit) do not necessarily lead to a positive impact on the parent brand (caused by the brand extension consistency with CBAs), and vice versa. 6.2. Methodological implications From a methodological standpoint, first this research introduces a measure to identify the central brand association questioning the added value of association in the brand meaning. This measure allows identifying the inseparable brand associations that are essential in the brand essence. Second we introduced the indirect measure to identify the brand extension consistency with CBAs and PBAs. This indirect measure has the advantage of going beyond the rational discourse of the respondent and relying on an analytical comparison between the strength of associations with the brand and that with the extension. 6.3. Managerial implications This study provides an approach that enables managers to identify the image of their brands in the context of an important product extension. The gradual disappearance of single-product brands and the increased complexity of product ranges mean that a brand's image evolves when offering new products. Today, brands are compared with different competitors according to the product category under consideration. In this context, treating a brand as representation that integrates not only associations specifically supported by its products (PBAs) but also associations independent of its products but shared by all (CBAs) can shed light on how brand image is defined in different product categories. Managers can use this new approach to understand the internal organization of the brand's image around a core and peripheral system. For managers, the central core ensures the brand's consistency. Furthermore, this study shows that an extension can modify the CBAs, PBAs and brand evaluations. However, brand extensions do not

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020

G. Michel, N. Donthu / Journal of Business Research xxx (2014) xxx–xxx

have a short-term effect on the core of each brand studied, while they are responsible for the disappearance of PBAs and the appearance of new ones. To the extent that any brand transformation must begin with a change in the brand's peripheral system, an analysis of this system is vital in helping managers predict brand transformations. The peripheral system represents an indicator of future changes in the brand and an unequivocal symptom of its evolution. These different conclusions bring new impetus to the field of research concerned with studying evolutions in brand image—notably the impact of brand extensions on the parent brand. 6.4. Limitations and further research Through an experimental approach, this research demonstrates the organization of brands around a central core. However, this demonstration is only the first step toward defining the brand as a social representation. A next step would be demonstrating that representations of brands vary between groups of individuals. In this light, further research could examine whether groups of individuals, defined according to their lifestyles, culture, or social network, have different images of brands. Regarding internal validity, Study 1 and Study 2b introduce a text and photo to present the brand extension. These conditions encouraged evaluation of the extension through categorization (holistic or analytical) rather than a multi-attribute process, and thus respondents might have focused on some characteristics more than they would in real conditions. Further research could use a different protocol to allow respondents to discover the brand extension in more realistic conditions. Study 2's assessment of changes in attitudes toward the brand is based on immediate reactions that could evolve over time. To reinforce the internal validity of these results, research should measure attitudes directly after presentation of the extension and again at a later date. Doing so would offer insight into whether changes in attitudes toward the brand last. Regarding external validity, the population studied was heterogeneous in education and standards of living compared with previous studies based exclusively on student samples. However, insofar as most of the brands studied target women, the sample of the quantitative study consisted mostly of women (82%). Furthermore, the study was conducted in France. These characteristics limit the possibility of generalizing the findings. Further research could use a more balanced sample in terms of gender and populations in other countries. References Aaker, D., & Keller, K. L. (1990). Consumer evaluations of brand extensions. Journal of Marketing, 54(1), 27–41. Abric, J. C. (1994). Les représentations sociales: Aspects théoriques. In J. C. Abric (Ed.), Pratiques sociales représentations (pp. 11–35). Paris: Presses Universitaires de France. Ahluwalia, R., & Gürhan-Canli, Z. (2000). The effects of extensions on the family brand name: An accessibility diagnosticity Perspective. Journal of Consumer Research, 27(3), 371–381. Ambroise, L. Michel G., & Valette-Florence, P. (2005). Measurement of brand central core and brand personality: Comparison of first results on predictive validity towards brand engagement and buying intention. Cons Persona and Res Conference, Dubrovnick, Croatia, September 2005. Barsalou, L. (1982). Context-independent and context-dependent information in concepts. Memory & Cognitions, 10(1), 82–93. Bentler, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238–246. Bhat, S. K., & Reddy, S. K. (2001). The impact of parent brand attribute associations and affect on brand extension evaluation. Journal of Business Research, 53(3), 111–122. Boush, D., & Loken, B. (1991). A process-tracing study of brand extension evaluation. Journal of Marketing Research, 28, 16–28. Broniarczyk, S. M., & Alba, J. W. (1994). The importance of the brand in brand extension. Journal of Marketing Research, 31, 214–228.

9

Dens, N., & De Pelsmacker, P. (2010). Attitudes toward the extension and parent brand in response to extension advertising. Journal of Business Research, 63(11), 1237–1244. Desai, K. K., & Keller, K. L. (2002). The effects of ingredient branding strategies on host brand extendibility. Journal of Marketing, 66(1), 73–93. Festinger, L. (1957). A theory of cognitive dissonance. Evanston, IL: Row, Peterson. Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006). Multivariate data analysis (6th ed.). Englewood Cliffs, NJ: Prentice Hall. Jap, S. D. (1993). An examination of the effects of multiple brand extensions on the brand concept. In L. McAlister, & M. L. Rothschild (Eds.), Advances in consumer research., Vol. 2. (pp. 607–611). Provo, UT: Association for Consumer Research. Kapoor, H., & Heslop, H. A. (2009). Brand positivity and competitive effects on the evaluation of brand extensions. International Journal of Research in Marketing, 26, 228–237. Kardes, F. R., & Allen, C. T. (1991). Perceived variability and inferences about brand extensions. In R. H. Holman, & M. G. Soloman (Eds.), Advances in consumer research., Vol. 18. (pp. 392–398). Association for Consumer ResearchProvo, UT. Keller, K. L. (1993). Conceptualizing, measuring, and managing customer-based brand equity. Journal of Marketing, 57(1), 1–22. Keller, K. L., & Aaker, D. A. (1992). The effects of sequential introduction of brand extensions. Journal of Marketing Research, 29, 35–50. Kim, K., Park, J., & Kim, J. (2014). Consumer–brand relationship quality: When and how it helps brand extensions. Journal of Business Research, 67(4), 591–597. Klink, R. C., & Smith, D. C. (2001). Threats to the external validity of brand extension research. Journal of Marketing Research, 38, 326–335. Lane, V., & Jacobson, R. (1997). The reciprocal impact of brand leveraging: Feedback effects from brand extension evaluation to brand evaluation. Marketing Letters, 8(3), 261–271. Loken, B., Barsalou, L. W., & Joiner, C. (2007). Categorization theory and research in consumer psychology. In C. P. Haugtvedt, P. Herr, & F. Kardes (Eds.), Handbook of consumer psychology (pp. 133–163). Mahwah, NJ: Erlbaum. Loken, B., & John, D. R. (1993). Diluting brand beliefs: When do brand extensions have a negative impact? Journal of Marketing, 57(3), 71–84. Mandler, G. (1982). The structure of value: Accounting for taste. In M. C. Clark, & S. T. Fiske (Eds.), 17th annual Carnegie symposium on cognition (pp. 3–36). Hillsdale, NJ: Erlbaum. Martinez, E., Montaner, T., & Pina, J. M. (2009). The brand extension feedback: The role of advertising. Journal of Business Research, 62(3), 305–313. Martinez, E., & Pina, J. M. (2009). Modeling of brand extensions' influence on brand image. Journal of Business Research, 62(1), 50–60. Michel, G. (1999). L'évolution des marques: l'approche par la théorie du noyau central. Recherche et Applications en Marketing, 14(4), 33–51. Milberg, S. J., Park, C. W., & McCarthy, M. S. (1997). Managing negative feedback effects associated with brand extensions: The impact of alternative branding strategies. Journal of Consumer Psychology, 6(2), 119–140. Moliner, P. (1995). A two dimensional model of social representation. European Journal of Social Psychology, 25, 27–40. Moliner, P. (1996). Images et représentations sociales. Grenoble: Presse Universitaire de Grenoble. Moliner, P., Joule, R. V., & Flament, C. (1995). Essai contre-attitudinal et structure des représentations sociales. Les Cahiers Internationaux de Psychologie Sociale, 27, 44–55. Monga, A.B., & John, D. R. (2010). What makes brands elastic? The influence of brand concept and styles of thinking on brand extension evaluation. Journal of Marketing, 74(3), 80–92. Moscovici, S. (1988). Notes towards a description of social representations. European Journal of Social Psychology, 18(3), 211–250. Murphy, G. L., & Medin, D. L. (1985). The role of theories in conceptual coherence. Psychological Review, 92, 289–316. Park, C. W., Milberg, S. J., & Lawson, R. (1991). Evaluation of brand extensions: The role of product feature similarity and brand concept consistency. Journal of Consumer Research, 18(September), 185–193. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology. , Vol. 19. (pp. 124–206). Orlando: Academic Press. Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology: General, 104(3), 192–233. Tauber, E. M. (1988). Brand leverage: Strategy for growth in a cost control world. Journal of Advertising Research, 28(4), 26–30. Völkner, F., & Sattler, H. (2007). Empirical generalizability of consumer evaluations of brand extensions. International Journal of Research in Marketing, 24, 149–162. Yoo, B., Donthu, N., & Lee, S. (2000). An examination of selected marketing mix elements and brand equity. Journal of the Academy of Marketing Science, 28(2), 195–211. Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of Consumer Research, 37(August), 197–206. Zikmund, W. G., & Babin, B. J. (2007). Exploring marketing research (9th ed.). Mason, OH: Thomson/South-Western.

Please cite this article as: Michel, G., & Donthu, N., Why negative brand extension evaluations do not always negatively affect the brand: The role of central and peripheral brand associations, Journal of Business Research (2014), http://dx.doi.org/10.1016/j.jbusres.2014.03.020