The impact of brand love on brand extension evaluations and the roles of variety seeking tendency and extension product involvement: do they differ between extension types?

The impact of brand love on brand extension

evaluations and the roles of variety seeking tendency

and extension product involvement: do they differ

between extension types?

By

Edgar van Dam

University of Groningen

Faculty of Economics and Business

Msc. Business Administration, Spec. Marketing Management

Boterdiep 117-9

9712LM Groningen

0628042289

e.l.van.dam@student.rug.nl

Student number: 1766163

Supervisor: Dr. J.C. Hoekstra

2

Supervisor: D.A. Naydenova MSc

2

Abstract

In this study, the impact of the relatively new construct of brand love on the evaluations of brand extensions is investigated. Also, is looked if this impact differs among different levels of variety seeking tendency and extension product involvement. This impact is researched for both line extensions and category extensions, to see if the effects differ from one another. The results of the study showed that no significant effect was found of brand love on the evaluation of line extensions. For category extensions however, the results showed that brand love positively influences the evaluation by consumers. From these results, implications were found that brand love diminishes the impact of similarity between the brand extension and the core brand on evaluations. Also, variety seeking tendency was found to enhance the effect of brand love with category extension evaluations. Another finding was that Instead of the hypothesized positive moderating effect, extension product involvement was found to have a positive direct effect on the evaluation of category extensions.

Keywords: brand love, brand extension evaluation, variety seeking tendency, extension product

3

CONTENT

Abstract ...2

1. INTRODUCTION ...5

1.1. Introduction and problem specification ...5

1.2. Academic and managerial relevance...7

1.3. Structure of thesis ...8

2. THEORY ...9

2.1. Brand love and brand extensions ...9

3. CONCEPTUAL MODEL AND HYPOTHESES ... 10

3.1. Conceptual model ... 10

3.2. Hypotheses ... 11

3.2.1. Brand love ... 11

3.2.2. Variety seeking tendency ... 12

3.2.3. Extension product involvement ... 14

4. METHODOLOGY ... 17

4.1. Design ... 17

4.2. Data collection ... 19

4.3. Measurement of constructs ... 19

4.4. Method of analysis ... 20

4.4.1. Normality, linearity and correlation ... 20

4.4.2. Regression analysis ... 21

4.4.3. T-test ... 22

5. RESULTS ... 23

5.1. Data collection descriptive ... 23

5.2. Correlation analysis ... 25

5.3. Regression... 25

5.3.1. Line extension evaluation – Premium light beer ... 25

5.3.2. Category extension – Perfume ... 29

5.3.3. Testing of the hypotheses ... 32

5.4. T-test ... 33

6. DISCUSSION ... 35

4

6.2. Variety seeking tendency ... 36

6.3. Extension product involvement ... 36

7. MANAGERIAL IMPLICATIONS ... 37

8. LIMITATIONS AND DIRECTIONS FOR FURTHER RESEARCH ... 38

9. CONCLUSION ... 39

REFERENCES ... 41

Appendix A: Linearity and normality check ... 46

A.1. Linearity check ... 46

A.2. Normality check ... 47

5

1. INTRODUCTION

1.1. Introduction and problem specification

In order to increase their brand equity, firms may enter new markets with their established brand name. This practice is defined by the term ‘brand extension’ (Keller, 2007). Brand extensions can be divided into two types, namely line extensions and category extensions. Line extensions are brand extensions with which the company uses a current brand name to enter a new market segment within their existing product class. Category extensions on the other hand are extensions where a company uses a current brand name in order to enter a completely different product class with a new product (Keller, 2007). The term “brand extension” was originally used only for describing category extensions (Aaker & Keller, 1990), however in later literature (e.g. John, Loken & Joiner, 1998; Suphellen, Eismann & Heim, 2004), the term has been used for both category extensions and line extensions. The success of these brand extensions is largely determined by the evaluations of consumers. In literature, consumers’ evaluations of a brand extension is deemed a joint function of how much the brand is liked in its original category and the similarity between original and extension product categories (Czellar, 2003; Grime, Diamantopoulos, & Smith, 2002).

Studies concerning the success of brand extensions have shown that introducing new products under established brand names have a positive effect on the ease of adoption of the new product by consumers when compared to new products with new brand names (Aaker & Keller, 1990; Keller, 2007; Smith & Park, 1992). Furthermore, several studies found that a positive fit, in terms of product category similarity and a positive parent brand effect following from a positive attitude of consumers towards the brand, can provide a great opportunity for brand extensions to increase brand equity and revitalize flagship products (Hem & Iversen, 2003; Klink & Smith, 2001; Supphellen, Eismann, & Hem, 2004; Taylor & Bearden, 2002; Zhang & Sood, 2002). A flagship product is defined by Suphellen, Eismann and Hem (2004) as being the product that is seen as most closely related to the brand name by the consumers. When the interest in this product or the overall brand is declining, successful brand extensions thus have the potential to enhance and bringing new life to the status of the brand and the flagship product. Keller (2007) also concludes such potential positive effects of brand extensions, by stating that a brand extension is an effective way for firms to leverage their brand equity thereby increasing the overall value of their brand.

6 brand dilution, which entails the negative affect or negative associations that could be projected back to the parent brand (Loken & John, 1993; Milberg, Whan Park, & McCarthy, 1997). Therefore brand extensions need to be considered carefully (Ng, 2010).

Previous studies on brand extensions and their evaluations show when and in what way different positive and negative effects of brand extensions are likely to occur. In these studies, the focus lies mainly on the effect of similarity of the product category and the attitude towards the original brand on brand extension evaluations. However, the attributes of consumers may also be a key influencer of the evaluation of brand extensions (Sood & Keller, 2012). Sood and Keller (2012) indicate that research on individual differences that may affect the interpretation of experiences when evaluating brand extensions, thereby affecting brand dilution, would be worthwhile. These individual differences are of importance because consumers base evaluations of a brand extension on their subjective affective reaction to the core brand without considering any specific features that the extension might have (Yeung & Wyer, 2005). The importance of individual consumer differences is also implicated by the difference that was found in age, which influenced the way certain attributes of a brand are prioritized and processed, thereby determining the evaluation of consumers (Zhang & Sood, 2002).

7 This research aims to gather empirical data to gain and expand insights on the nature of the relationship between brand love and the evaluation of brand extensions by consumers. In researching this relationship, the two different types of brand extensions (line extension and category extension) are also considered to see if there is a difference in the effect of brand love between the two. In order to look at the influence of individual differences of consumers, moderating roles of variety seeking tendency and extension product involvement are also studied. Within these moderating roles of variety seeking tendency and extension product involvement is also looked at the difference between the two brand extension types. The research combines the relatively new marketing concept of brand love with brand extension evaluations and two individual consumer aspects as moderators.

The research question that is central in this study is:

“What is the effect of brand love on brand extension evaluations, is this effect influenced by the variety seeking tendency of consumers and their extension product involvement and do these effects differ between line and category extensions?”

This research question can be divided in the following four sub-questions: - “What is the effect of brand love on brand extension evaluations?”

- “Is the effect of brand love on brand extension evaluations influenced by the type of

extension?”

- “Is the effect of brand love on brand extension evaluations influenced by the variety seeking

tendency of the consumers?”

- “Is the effect of brand love on brand extension evaluations influenced by extension product

involvement?”

1.2. Academic and managerial relevance

8 contribute to the overall academic knowledge on the subject and thus proves the academic relevance of this research.

Besides the academic relevance of the research, there is also relevance from the managerial perspective. Bearing in mind the potential impact that brand extensions could have on maintaining or increasing the brand equity of a firm, decisions on brand extensions are important. Understanding how consumers evaluate the extensions and why is vital. Research aimed at clarifying this understanding thus is of great importance for managers. Better understanding the reactions of consumer who love the original brand may lead to better retaining these consumers and identifying dangerous practices that could lead to dissatisfaction with these consumers. Brand extensions that are negatively evaluated could lead to brand dilution, which could especially have an impact on these brand loving consumers. This as brand love consumers generally lead to positive WOM, higher brand loyalty and a willingness to pay a price premium (Bergkvist & Bech-Larsen, 2010; Carrol & Ahuvia, 2006). So lessons in order to retain these important consumers are fundamental. This research will potentially lead to insight in circumstances where these consumers are more likely to positively or negatively evaluate brand extensions. Using the knowledge on these circumstances, managers can make well informed decisions for introducing brand extensions.

Also, in the understanding of the relationship between brand extensions and the evaluations of the consumers, the relationship between individual differences of the consumers and brand extension evaluations are still somewhat unclear. Identifying differences in reactions of individual consumers on brand extensions could potentially lead to important insights, based on which more accurate targeting decisions of brand extensions could be made.

1.3. Structure of thesis

9

2. THEORY

2.1. Brand love and brand extensions

Brand love is defined as the degree of passionate emotional attachment a satisfied consumer has for a particular trade name, which includes passion for the brand, attachment to the brand, positive evaluation of the brand, positive emotions in response to the brand and declarations of love for the brand (Carroll & Ahuvia, 2006). The concept is related to the concepts of brand affect and brand liking that are most commonly used in literature, but differs from those concepts. Brand Liking and Brand affect are constructs to describe a favorable attitude towards the parent brand (Czellar, 2003; Bhat & Reddy, 2001). The concept of brand love is not just a stronger form of interpersonal liking but also involves integration of the brand into the sense of identity of the consumer. Also, it is conceptualized as a mode of satisfaction of which the lower bound is defined as being the absence of satisfaction as an emotional response. This conceptualization thus rules out negative feelings like brand disliking or hate (Carroll & Ahuvia, 2006). Brand love is different from regular satisfaction and the related constructs of brand liking and brand affect due to its stronger affectionate focus, no requirement of expectancy or disconfirmation, its willingness to declare love and the integration of the brand into the consumers’ identity. Another factor in which it is different than other constructs is that it results from a long-term relationship between the consumers and the brand (Carroll & Ahuvia, 2006).

10

3. CONCEPTUAL MODEL AND HYPOTHESES

3.1. Conceptual model

In this section, the conceptual model of this study is explained and is presented in figure 3.1. After this, the variables that are central in the model and the reasoning behind the relationships in the model including the hypotheses are explained.

FIGURE 3.1: CONCEPTUAL MODEL OF BRAND LOVE AND BRAND EXTENSIONS

In figure 3.1., the conceptual model of this study is presented. Brand love (independent variable) is presented as positively influencing the brand extension evaluations of consumers (dependent variable) which is hypothesis H1a. Reasoning behind this is that consumers who love the brand are thought to be likely to transfer their love for the original brand to a more positive evaluation of the extension product based purely on their love for the original brand. For example, when a consumer does not like the actual extension product or its attributes, he may positively evaluate the extension product due to his love for the original brand.

11 The second moderator, extension product involvement is the extent to which the consumer is involved in the extension product. As can be seen in the model, this moderator is hypothesized to have a positive impact on the relationship between brand love and brand extension evaluations (H3a). Reasoning behind this is because high extension product involvement leads to a smaller need for complementarities between the attributes of the original brand and the extension product and a more affect based evaluation of the extension.

A third moderator included in the model is the type of brand extension. This moderator manipulates the extent of fit of the brand extension with the original brand. It is assumed that this last moderator not only influences the relationship between brand love and brand extension evaluations (H1b), but also moderates the moderating effects of variety seeking tendency (H2b) and extension product involvement (H3b). The effect of the moderators is proposed to be stronger with either line extensions or category extensions, due to the difference in similarity towards the original product. Lastly, perceived quality of the original brand is used as a control variable in this model, as the perceived quality of the original brand has shown to have a positive impact on the evaluation of extensions (Zeithhalm, 1988; Aaker & Keller, 1990; Boush & Loken, 1991; Chowdury, 2007). Zeithalm (1988) concludes that when the original brand is perceived as having a high quality, the extension should benefit from this. Accordingly, Chowdury (2007) concludes, based on empirical research that perceived quality directly effects the evaluation of consumers on brand extensions in a positive way.

3.2. Hypotheses

12

H1a: Brand love positively influences brand extension evaluation.

Considering the type of extension as moderator on the effect of brand love on brand extension evaluations the following can be found. Firstly, literature concludes that the success of a brand extension is not only due to the previously mentioned factor of how much the brand is liked in its original category, but also the similarity between original and extension categories (Czellar, 2003; Grime, Diamantopoulos, & Smith, 2002). When forming their evaluations of a brand extension, consumers are less likely to incorporate core brand evaluations with extensions that are dissimilar to the core brand (Barone, 2005).This stresses the importance of the similarity of the extension to the core brand to transfer the positive or negative feelings of the core brand to the extension. This is in congruence with the observation of Boush, Shipp, Loken & Gencturk et al. (1987) who describe that a higher similarity between the core brand and the brand extension product leads to a more positive or negative effect to the evaluation of the new product. Their research also suggests that when a brand has a reputation for excellence in a specific product area it might have a negative impact on the evaluations of the consumers of their new product in a dissimilar category. Additionally, it has been found that when a parent brand is extended into a similar category, the consumers are more likely to perceive the extension more favorably than when the extension falls into a more dissimilar category (Aaker & keller, 1990; Bhat & Reddy, 2001; Meyvis & Janiszewski, 2004). Despite that within both line extensions and category extensions similarity and dissimilarity can occur to the core brand, it is more likely to perceive a dissimilar extension in category extensions than in line extensions as line extensions are extensions within the current product class and the category extension occurs in a completely new product class. When applying the insights of the findings of previous research it can be assumed that line extensions will be more positively evaluated than category extensions due to their higher similarity to the core brand. According to these findings the following hypothesis is formulated:

H1b: The positive effect of brand love on brand extension evaluation is stronger for line extensions than for category extensions.

3.2.2. Variety seeking tendency

13 change in order to resolve boredom associated with a brand. This view is supported by research of Inman (2001) who suggest that the variety seeking tendency appears among consumers who have become satiated with the specific characteristics of a product.

Other research however, suggests that consumers who have a variety seeking tendency derive the pleasure purely from the change itself when switching between brands or products, so without taking in account or resulting from the brand they are switching from (Givon, 1984; McAlister, 1982). The variety seeking tendency is also regarded by literature as the consumers’ search for the optimal level of stimulation (Ratner, 2006). This tendency of behavior leads to consumers being less likely to choose something in a specific situation when they have just made a similar choice in an earlier situation, even when the consistent alternative is clearly better than other alternatives (McAlister, 1982; Ratner, Kahn & Kahneman, 1999; Read & Loewenstein, 1995). Especially in public situations, consumers tend to seek more variety in order to show that they are unique, open-minded and interesting (Ratner & Kahn, 2002; Kim & Drolet, 2003). From these studies, it can be concluded that there are two main views to the variety seeking tendency. One in where the consumers are driven to seek variety as a result from the original brand and one in which this tendency is based in the personality and the behavior of the consumer purely seeking variety for the pleasure of change. In this research, the focus lies more on the second view on variety seeking as being an individual personality trait of consumers that results in a certain tendency to seek variety.

Compared to non-variety seeking consumers, enthusiast consumers who have a high variety seeking tendency tend to use significantly more information sources when evaluating products (Dodd, Pinkleton & Gustafson, 1996). By using more information sources, these consumers are less likely to be affected by affective or emotional factors that might impact their evaluation of a brand extension. Therefore it is assumed that consumers that have a high variety seeking tendency will evaluate brand extension evaluations less favorably than they would when loving the brand and not having the tendency to seek variety. This results in the following hypothesis:

H2a: Variety seeking tendency mitigates the effect of brand love on brand extension evaluation.

14 product class which occurs in line extensions, it is assumed that consumers with a variety seeking tendency will react more favorable to category extensions and more negative to line extensions. Based on this implication the following hypothesis is formulated:

H2b: The negative moderating effect of variety seeking tendency on the relationship between brand love and brand extension evaluation is stronger for line extensions than for category extensions.

3.2.3. Extension product involvement

In literature, researchers differ on the definition of the involvement concept. On the one hand researchers find that involvement differs among consumers for a specific product class (Houston & Rothschild, 1978; Lastovicka & Gardner, 1979) and on the other hand researchers who argue that it is not the specific product that is involving but the personal meaning towards the features of that product that results in involvement. Zaichkowksy (1994) states that these views are equally right because involvement results from the interaction between individuals and products. Hence, both products and individuals are important aspects of involvement. This shows that involvement can be conceptualized in many different ways. In research on consumer behavior, involvement generally refers to the perceived relevance of the object or in this case product on the needs, values and interests of the consumers (Bloch & Richins, 1983; Mitra, 1995; Zaichkowsky, 1985; 1994). In this research the focus lies on the involvement of the consumer with the extension product, which is why in this study involvement is defined as the perceived relevance of the extension product to the consumers’ needs and values.

The effect of involvement and its importance for the evaluation of brand extensions has been researched by several studies (e.g. Hansen & hem, 2004; Barone, 2005; Maoz & Tybout, 2002; Nkwocha, Bao, Johnson & Brotspies, 2005). Research done by Barone (2005) studied whether the influence of mood on extension evaluations depends on the level of involvement. Results of this research showed a significant direct effect of mood on brand extension evaluations for high and low levels of involvement. However, the processes responsible for the effect were found to be different. With high involvement, the relation between the core brand and the extension was considered more, while mood had a more direct impact in low involvement levels. From this research can be seen that different levels of involvement influences the way the consumer looks at the extension and evaluates it. This shows the theoretical evidence to base the link between involvement as a moderator between brand love and brand extension evaluations.

15 Based on this finding the authors state that it is important for marketers to consider the involvement level of consumers in the target segment for a brand extension. Insights from this study reveal that potential consumers who are involved in the actual product and have a perceived need for it are more likely to buy the extended brand. Involvement in the extension category was found to affect the intention to buy a brand extension in a positive way. As it can be concluded that marketing activities would be more effective in providing success when it is targeting high involvement consumers, the importance of involvement for brand extensions is shown. Because of the relatedness between purchase intention and brand extension evaluation it is assumed that involvement can also be an important influencer for brand extension evaluations.

Nkwocha, Bao, Johnson and Brotspies (2005) conducted an experimental study containing multiple brand extensions. In their research they propose product involvement as a situational motivation for consumers to discount the importance of product fit when they evaluate brand extensions. They find that perceived complementarities between the parent brand and extension product is significantly positively influenced by involvement. Specifically in the case of low involvement, complementarity is found to be an important factor for brand extension evaluations. High involvement consumers were found to tend to discount the factor of complementarity between the core brand and the extension product. The evidence from these studies show that involvement could act as a moderator on the relation between brand love and brand extension evaluations (Barone, 2005; Hem & Iversen, 2004; Nkwocha, Bao, Johnson & Brotspies, 2005). In line with these conclusions, research suggests that successful extensions of strong brands can be found in product categories that are seemingly dissimilar to each other. This under the condition that the consumers are motivated and are able to identify meaningful similarities between the core brand product and the extension product (Maoz & Tybout, 2002). These kinds of similarities can be spontaneously identified by consumers that have high brand knowledge and are involved with the product category (Muthukrishnan & Weitz, 1991).The nature of this effect would be positive, following the assumption of relatedness of purchase intention with brand extension evaluation and the creation of spontaneous similarities due to high involvement.

16 extension, rather than a superficial analysis of readily available, emotional and salient stimuli which is the case with the “peripheral route” during low involvement.

Because of these two different views on the effect of consumer involvement with the brand extension products, this hypothesis is more exploratory of nature. However, the view for a positive effect of involvement in the extension product on the relationship between brand love and brand evaluations is taken as starting point for this hypothesis. Also it is expected that when consumers love the brand they are more likely to evaluate the brand extension even more positively when that specific product is relevant to their interest, needs and values. Based upon this, the following hypothesis is formulated:

H3a: Extension product involvement enhances the effect of brand love on brand extension evaluation.

Evidence for the difference in the effect of extension product involvement between the two extension types in the form of a line extension or a category extension can be found in literature. In their study, Nkwocha, Bao, Johnson and Brotspies (2005) found that when product involvement of consumers is low, complementarity is especially important for brand extension evaluations, as product involvement acts as a situational motivation to discount the importance of product fit. This means that when consumers are highly involved they are more likely to discount the importance of product fit and evaluate the brand extension based more on their relevance and need for the product than the similarity between the extension category and the core brand. It can be concluded that for consumers whose involvement in the product is high, there should be no difference between either line extensions or category extensions. However, when consumers are low involved with a product, similarity is important which means that line extensions would be more favorably evaluated than category extensions.

These findings are in line with research that has been conducted by Maoz and Tybout (2002), who suggest that a more similar brand extension is more favorably evaluated by consumers than a dissimilar extension when involvement for the consumer is low. Contrasting to the study of Nkwocha, Bao, Johnson and Brotspies (2005) they also conclude in their research that when consumers are highly involved, similar extensions are in general more favorably evaluated than dissimilar extensions.

17 findings can be concluded that when the consumer is high involved with the product, the evaluations would be more favorably towards line extensions than category extensions.

Implications of research thus range from no differences at all between the two extension types with high involvement, only a difference between the two extension types for consumers who are low involved and a more favorable positive effect of highly involved consumers in the case of line extensions. Because of this variety in implications of previous research, this is a more exploratory hypothesis. Following the findings from Barone (2005) the assumption is made that when consumers are highly involved with an extension that has a high similarity to the core brand, this will be more favorably evaluated than a more dissimilar extension. In this case this means that the assumption is made that high involvement consumers will have more favorable evaluations of brand extensions when they evaluate line extensions than when they evaluate category extensions. The following hypothesis then is formulated:

H3b: The positive moderating effect of extension product involvement on the relationship between brand love and brand extension evaluation is stronger for line extensions than for category extensions.

4. METHODOLOGY

This section elaborates on the design of the study, the way the data has been collected, the constructs that were measured in the study and the method of analyzing the collected data.

4.1. Design

The aim of this study is to answer the main question: “What is the effect of brand love on brand

extension evaluations, is this effect influenced by the extension product involvement and variety seeking tendency of consumers and do these effects differ between line and category extensions?” In

18 The type of extension was the variable that was manipulated in this study. This was done by letting the respondents evaluate two types of extensions, both a line extension and a category extension. Heineken was chosen as the brand of which examples of extensions were shown to the respondents, as Heineken is a global brand which is widely known by consumers. It was expected that using this brand would yield a high probability of providing respondents that love the brand in order to be able to get significant results on this variable. For the line extension example that was presented in the survey is hypothesized that Heineken will introduce a light version of its beer (Heineken Premium Light Beer). As a line extension, this extension was designed as being very close to the original brand as this is a small variation in the beer product category. The other example, that of a category extension, was hypothesized that Heineken is introducing a perfume range (Heineken the Fragrance). This extension lies outside the current product class and was expected to have less perceived fit. Despite the less expected fit, the product of perfume was chosen due to the fact that it would be an extension that would not be totally implausible. As Heineken already has several lifestyle products like clothing and bags, a perfume product might also be seen as an appropriate extension of the Heineken brand. An overview of these examples of the type of extension that were used in this study can be found in table 4.1.

TABLE 4.1.: EXAMPLES OF THE EXTENSION TYPES

Extension examples used in the survey:

Line extension: Heineken Premium Light

Category extension: Heineken Perfume (Heineken The Fragrance)

19 higher fit. As the objective of the study was to manipulate and create the different extensions in the form of a line extension and a category extension.

4.2. Data collection

This research was conducted using a survey in the form of an online questionnaire which was made available via thesistools.com. This survey can be found in appendix B. The respondents were found using convenience sampling, which means that the questionnaire was sent to as many potential respondents as possible. This was done by distributing the link to the online questionnaire via e-mail and Facebook to fellow students, friends and family. Using this data collection method, 117 respondents fully completed the survey.

4.3. Measurement of constructs

For the measurement of the variables in this study, existing scales were used. These measurement scales can be found in table 4.2. All of the variables were measured using either a 7-point Likert-type scale or a 7 point semantic differential scale. Results of the analysis of reliability and consistency of the scales can also be found in table 4.2. These results show that all scales have a Cronbach alpha higher than .7, which is the critical value indicated by Nunnally and Bernstein (1991) for a good internal consistency reliability. From this can be concluded that all measurement scales have a high reliability in measuring the variables used in the model of this study.

TABLE 4.2.: MEASUREMENT SCALES OF VARIABLES

Variable Items Scale Cronbach alpha (α) Brand Love

(Carrol & Ahuvia, 2006)

1. This is a wonderful brand. 2. This brand makes me feel good. 3. This brand is totally awesome.

4. I have neutral feelings about this brand. (–) 5. This brand makes me very happy.

6. I love this brand!

7. I have no particular feelings about this brand. (–) 8. This brand is a pure delight.

9. I am passionate about this brand. 10. I’m very attached to this brand.

20 Variety seeking tendency (Donthu & Gilliland, 1996)

1. I generally like to try out something different. 2. I like to try different things.

3. I like a great deal of variety. 4. I like new and different styles.

All 7-point Likert-scale (1 = Definitely disagree, 7 = Definitely agree) .893 Extension product involvement (Beatty & Talpade, 1994)

1. In general I have strong interest in this product category.

2. This product category is very important to me. 3. The product category matters a lot to me. 4. I get bored when other people talk to me about this product category. (–)

All 7-point Likert-scale (1 = Definitely disagree, 7 = Definitely agree) Line: .814 Category: .922 Manipulation: Perceived fit (Dens, de Pelsmacker, 2009)

1. Good fit/Bad fit 2. Logical/Illogical 3. Appropriate/Inappropriate All 7-point semantic differential scale Line: .890 Category: .902 Control: Perceived quality of original brand (Chowdury, 2007)

1. Very low/Very high 7-point

semantic differential scale

4.4. Method of analysis

In this section, the statistical methods that are used to be able to accept or reject the hypothesis of the model and eventually answering the main question of the study are explained. A correlation analysis, regression analysis, analysis of the moderators and finally a t-test were performed. An explanation of these methods and the reason why was chosen for the specific method is described in the sections below.

4.4.1. Normality, linearity and correlation

In order to be able to analyze whether there is a relationship between the dependent variable and the independent variable, a correlation analysis is performed. The correlation analysis shows if the two specific variables that are used in the analysis are related, how strong they are related and if this relationship has a positive or negative tendency.

To be able to choose which correlation test (Pearson or Spearman) is appropriate for the analysis, the data needs to be checked for normality and linearity.

21 of the brand love on the x-axis and the specific independent variable on the y-axis. Also, scatterplots of the predicted values versus the observed values of both extensions are shown. The results of these scatterplots can be seen in Appendix A.1. Based on the shape of these scatterplots it can be concluded that linearity can be assumed.

Checking for normality of the of the error terms of the dependent and the independent variables is done by checking normal probability histograms and looking at the skewness and kurtosis of the variable distribution. This is done by calculating the ratio of the skewness and kurtosis statistic to their respective standard error values, which must be smaller than 2 or larger than -2. When calculating these ratios for Brand love (skewness ratio = -.26 and kurtosis ratio = -1.51), line extension evaluation (skewness ratio = -.33 and kurtosis ratio = -.56) and category extension evaluation (skewness ratio = 1.10 and kurtosis ratio = -.25) was found that none of these exceeded the ratio range of 2 and -2. From this finding it can be concluded that all the variables are approximately normal distributed. Also, to control whether the normal probability assumption was met, a histogram and a p-p plot of the regression standardized residuals following from the regression model were made. This was done for both the line extension model as for the category extension model. These figures can be seen in Appendix A.2. The figures show that the normal probability assumption was met for both the line extension regression model and the category extension regression model. Since now is established that the variables that are used in the correlation analysis are both linear and approximately normal distributed, a Pearson test for correlation is appropriate to be performed. 4.4.2. Regression analysis

Next, in order to test the hypotheses a regression analysis is performed.

A regression analysis is performed for the two separate models, analyzing the relation between brand love and line extension evaluations (premium light beer) and the relation between brand love and category extension evaluations (perfume).

22 of the moderator variables variety seeking tendency and extension product involvement are researched.

In order to measure whether the proposed moderators of variety seeking tendency and extension product involvement with the extension product have an impact on the relationship between brand love and brand extension evaluations, a moderator analysis is performed. This is done using the approach described by De Vries & Huisman (2008). In this approach, the interaction term is computed by multiplying the centralized value of the moderator variable with the centralized independent variable value. By doing this an interaction term variable is created, which can be added to the regression analysis. An analysis of this regression then shows whether or not there is a moderator effect. The nature of this effect then is further analyzed by categorizing the moderator variable in a state of low, mid and high scores. This categorization is performed in order to see how the regression coefficient of brand love changes for different levels of variety seeking tendency and extension product involvement. These categorized scores are then computed as new variables and put into a regression model in order to calculate the regression coefficients for those variables at a 95% confidence interval. Finally, to be able to present a clear view of the moderating effect, this change in regression coefficients is depicted in a graph.

The total regression equation then becomes:

Y

= α + β

BL + β

POQ + β

VST

+ β

I

+ β

BL

VST

+ β

BL

I + ε

In which:

Y= line extension evaluation (premium light beer) or category extension evaluation (perfume). BL = Brand love of the consumer.

VST = Variety seeking tendency of the consumer. I = Extension product involvement.

POQ = Perceived overall quality.

BL_VST = Interaction term brand love and variety seeking tendency. BL_I = Interaction term brand love and extension product involvement. ε = Error term.

4.4.3. T-test

23 the values of these two models in the following formula: (β1 - β2) / √ (var1 + var2). The result of this formula is a t-test value which, when it is higher than 1.96 (for p < 0.05), is significantly different.

5. RESULTS

In this results section, the results of the correlation analysis, regression analysis and the performed t-test are discussed. First however, an overview of the descriptives of the data collection is given. The results of the testing of hypotheses H1a, H2a and H3a are discussed after presenting the regression results of both the line extension model and the category extension model. This is done because the hypotheses are dependent of the results of both of these models. After this, the results of the t-test are presented and the results of testing hypotheses H1b, H2b and H3b are discussed.

5.1. Data collection descriptive

The data for this research was provided by a survey with a sample of 117 respondents (n = 117). Descriptive data on the respondents is presented in tables 5.1. – 5.4. From table 5.1. can be seen that the sample consisted of slightly more males (66) than females (51). Table 5.2. shows that most of them are between 20 and 30 years old (63.2%) followed by the 50-60 year old group (18.8%). For the largest part these are people with a low income which is <15.000 euro’s a year (54.7%), followed by the 50.000-100.000 group (12.8%), shown in table 5.3. Lastly, the distribution data in table 5.4. shows that education level of the respondents is best represented in the HBO/WO – Bachelor group (44.4%) but is closely followed by the WO – Master group (39.3%). Considering these distributions it can be concluded that the respondents are mainly young, highly educated and have a low income. This was to be expected as the convenience sample survey was primarily communicated among undergraduate students.

TABLE 5.1.: GENDER DISTRIBUTION Gender

Male Female

Frequency (%) 66 (56.4) 51 (43.6) TABLE 5.2.: AGE DISTRIBUTION

24 TABLE 5.3.: INCOME LEVEL DISTRIBUTION

Income <15.000 15.001-30.000 30.001-50.000 50.001-100.000 >100.000 Don't want to answer Frequency (%) 64 (54.7) 8 (6.8) 10 (8.5) 15 (12.8) 8 (6.8) 12 (10.3)

TABLE 5.4.: EDUCATION LEVEL DISTRIBUTION

Education

MBO HAVO/VWO Bovenbouw/ WO propedeuse

HBO/WO – Bachelor WO - Master

Frequency (%) 11 (9.4) 8 (6.8) 52 (44.4) 46 (39.3)

After looking at the descriptives of the respondents that are in the sample, the results from the completed surveys of these respondents can be seen in table 5.5. In this table the mean and the standard deviation of the variables that were included in the survey are shown.

TABLE 5.5.: VARIABLE DESCRIPTIVES

The results in this table show that the line extension evaluation (mean = 4.39) scores slightly better than the category extension evaluation (mean = 3.13). Also, the mean of the variety seeking tendency variable was quite high (mean = 5.16). On average, the respondents were more involved with the category extension product (mean = 3.75) than with the line extension product (mean = 2.90). However, this probably results from a select group of respondents that were highly involved with the category extension which can be seen from the higher standard deviation with the category extension (1.70) compared to that of the line extension (1.25). Finally, a quite large difference can be seen between the mean score of the perceived fit with the line extension (mean = 4.85) and the category extension (mean = 2.27).

Variable Mean Standard Deviation

Brand Love 3.68 1.19

Perceived overall quality 4.97 1.16

Line extension evaluation 4.39 1.20

Category extension evaluation 3.13 1.16

Variety seeking tendency 5.16 1.16

Line extension product involvement 2.90 1.25

Category extension product involvement 3.75 1.70

Perceived fit line extension 4.85 1.38

25

5.2. Correlation analysis

As previously stated, a Pearson correlation measure was used as the variables are both normally distributed and linear. This analysis showed that brand love correlated insignificantly with the line extension evaluation – premium light beer (p = .174), as this value is higher than p = .05. With the category extension evaluation – perfume however, brand love was found to be significantly correlated (p = .004). Since the correlation between brand love and evaluation for perfume has been found to be significant, can be concluded that there is a relationship between the dependent variables and the independent variable. The correlation value for the evaluation of the category extension – perfume is fairly low at .266, which indicates that the relationship is relatively weak as it is closer to 0 than to 1. No significant relationship is found between brand love (IV) and line extension evaluation – premium light beer (DV).

5.3. Regression

Next, a regression analysis was performed. Firstly, the control variable of perceived overall quality was analyzed in order to check if the variable indeed has a significant relationship with the dependent variable. Perceived overall quality was found to be significantly related to the evaluation of the line extension (B = .286 and p = .003) and not significant with the evaluation of the category extension (B = .170 and p = .066). Despite the fact that overall perceived quality was found non-significant for the evaluation perfume, the control variable of perceived quality was added to both the regression model for line extensions and category extensions.

In the following sections the results of the performed regressions are discussed, starting with the model for the evaluation of line extensions and then the model for the evaluation of category extensions.

5.3.1. Line extension evaluation – Premium light beer

In this section, the results from the regression on the dependent variable of the evaluation of the line extension (premium light beer) are discussed. The regression analysis was performed building up to the full regression model in 5 steps. A summary of the results of these five steps are presented in table 5.6.

26 TABLE 5.6.: REGRESSION RESULTS LINE EXTENSION – PREMIUM LIGHT BEER

Line extension Hypothesis effect Model 1 Model 2 Model 3a Model 3b Model 4 Main variable Brand Love 1a (+) .128 -.076 -.074 -.087 -.089 Control variable Perceived overall quality .335a .361a .252b .274a Moderators Variety seeking tendency .180b .184b Extension product involvement .444a .447a Interaction effects Variety seeking tendency * Brand Love 2a (-) .148b _.135b Extension product involvement * Brand Love 3a (+) .157b .145b R2 .016 .080 .158 .352 .424 Adjusted R2 .007 .064 .127 .329 .393 R2 change 0,016 .064a .078a .272a .344a F-value 1.867 4.959a 5.236a 15.238a 13.510a Note: a p-value <.01; b p-value <.05; c p-value <.10

For the next step, the control variable was added to the model (model 2). Perceived overall quality shows to have a significant positive effect on the evaluation of the line extension (B = .335 and p = .006). Also, this addition leads to a significant change of the explanatory value of the model (R2 change = .064) and makes the model significant (F-value = 4.959 and p = .009).

In the third and fourth step the moderator variables of variety seeking tendency and extension product involvement are added separately. This is done in order to be able to see the pure effect of the specific moderator, without letting them interfere with each other.

27 value of the regression coefficient of brand love on the evaluations of line extensions is shown for different levels of variety seeking tendency. This value of the regression coefficient is presented in the form of a range of values based on a confidence interval of 95%. The figure shows that the range of regression coefficients is higher for respondents with a high level of variety seeking tendency, thus showing a positive moderation effect. The addition of these variables contributed to a significant increase in the explanatory value of the model and now explains 15.8% (R2 = .158 and R2 change = .078).

FIGURE 5.1.: GRAPHICAL REPRESENTATION OF THE MODERATING EFFECT OF VARIETY SEEKING TENDENCY

28 FIGURE 5.2.: GRAPHICAL REPRESENTATION OF THE MODERATING EFFECT OF EXTENSION PRODUCT INVOLVEMENT

The fifth and final step was to include all of the variables in the regression model (model 4, table 5.6.). Brand love is still insignificant (B = -.089 and p = .360) and the perceived overall quality remains positive and significant (B = .274 and p = .006). Variety seeking tendency (B = .184 and p = .017) and extension product involvement (B = .447 and p = .000) both show to have significant positive direct effects on the evaluation of the line extension. The interaction effect of variety seeking tendency is found to be positive and significant (B = .135 and p = .015). The interaction effect of extension product involvement also shows to be significantly positive (B = .145 and p = .020). The results of this final regression model show that the overall model is significant (F-value = 13.51 and p = .000) and has a quite good explanatory value (R2 = .424). Important to acknowledge here is that while significant interaction effects were found, the main effect was found to be insignificant. While implications of the effects are given, this makes that the found interaction effects are not reliable to make hard conclusions on the nature of this interaction effect. This was confirmed by also estimating this same regression model using the log of the variables, which resulted in the finding of significant negative interaction effects of variety seeking tendency (B = -.015) and extension product involvement (B = -.025).

29 means that multicollinearity is significantly present (Cohen et al. 2003). The VIF values for the regression model are presented in table 5.7.

TABLE 5.7.: LINE EXTENSION – PREMIUM LIGHT BEER REGRESSION MODEL VIF SCORES

Line extension - Premium light beer VIF

Brand Love 1.753

Perceived overall quality 1.024

Variety seeking 1.071

Variety seeking tendency *Brand love 1.185

Extension product involvement 1.210

Extension product involvement *BrandLove 1.712

As can be seen from the VIF values in this table, none of these scores even come close to exceeding the critical VIF value of 10. From this then is concluded that no significant multicollinearity influences the regression model.

5.3.2. Category extension – Perfume

The same five steps that were performed for the line extension are now used for the category extension. Results of these steps are shown in table 5.8.

For the first step, the results are shown in table 5.8. under model 1. These results show that brand love has a significant positive effect (B = .260 and p = .004) on the evaluation of the category extension (perfume). However, this model only explains 7.1% (R-square = .071) of the proportion of the total variation in evaluations for the variation in brand love, which is relatively low. This relatively low value of the R-square implies that there are many other variables that influence the evaluation of the category extension.

30 TABLE 5.8.: REGRESSION RESULTS CATEGORY EXTENSION - PERFUME

Category extension Hypothesis effect Model 1 Model 2 Model 3a Model 3b Model 4 Main variable Brand Love 1a (+) .260a .255b .258b .255b .261b Control variable Perceived overall quality .007 .032 -.032 -.012 Moderators Variety seeking tendency .168c .118 Extension product involvement .174a .139b Interaction effects Variety seeking tendency * Brand Love 2a (-) .138b _.111c Extension product involvement * Brand Love 3a (+) .058 .024 R2 .071 .071 .144 .146 .182 Adjusted R2 .063 .055 .113 .115 .137 R2 change .071a .000 .073b .075a .111a F-value 8.785a 4.356b 4.705a 4.782a 4.065a Note: a p-value <.01; b p-value <.05; c p-value <.10

31 FIGURE 5.3.: GRAPHICAL REPRESENTATION OF THE MODERATOR EFFECT OF VARIETY SEEKING TENDENCY

Analyzing the results that are presented in figure 5.3., shows that a higher variety seeking tendency corresponds to a higher value of the regression coefficient for brand love. From this can be concluded that the higher the variety seeking tendency, the higher the positive effect on the regression coefficient for brand love will be. Thus, there is a positive moderating effect of variety seeking tendency on the relationship between brand love and the evaluation of the category extension. The explaining quality of the model has increased significantly, the model now has an explanatory value of 14.4% (R2 = .144 and R2 change = .073).

Next, the fourth step is to include the moderator variable and interaction term of extension product involvement rather than the variety seeking tendency variable and its interaction. The results of this regression can be found in table 5.8. (model 3b). Extension product involvement is found to have a significant positive direct effect on the evaluation of the category extension (B = .174 and p = .005). The interaction term of extension product involvement shows an insignificant effect (B = .058 and p = .136) When looking at the explanatory value of this model, can be seen that this step increases the explanatory value of the model significantly (R2 = .146 and R2 change = .075).

32 confidence level (p = .096). Now, the interaction term is only significant for a p-value that is smaller than .1 (p = .096). This means that, considering a 90% confidence level, variety seeking has a significant positive interaction effect on the evaluation of the category extension. This change in significance means that the moderator effect of variety seeking tendency is influenced by the other moderator variable of extension product involvement. So, variety seeking tendency is found to positively moderate the effect of brand love on the evaluation of category extensions, but in the overall model this effect has been influenced by the other moderator variable of extension product involvement. In turn, extension product involvement is found to have a significant positive direct effect on the evaluation of the perfume extension (B = .139 and p = .029). Looking at the interaction effect of extension product involvement, can be seen that this effect is not significant (B = .024 and p = .607). The total explanatory value of the regression model is 18.2% (R2 = .182). Also the overall regression model is found to be significant (F-value = 4.065 and p = .001).

Just as with the regression model for line extensions, this regression model on the evaluation of category extension needs to be checked for significant influences of multicollinearity. The VIF scores can be found in table 5.9.

TABLE 5.9.: CATEGORY EXTENSION - PERFUME REGRESSION MODEL VIF SCORES

Line extension – Perfume VIF

Brand Love 1.646

Perceived overall quality 1.080

Variety seeking 1.194

Variety seeking tendency *Brand love 1.132

Extension product involvement 1.168

Extension product involvement *BrandLove 1.741

Looking at the VIF scores of the regression model in table 5.9., it can be seen that none of these values exceed the critical value of 10, or come even close to the value of 5.9. From this can be concluded that the regression model is not significantly influenced by multicollinearity.

5.3.3. Testing of the hypotheses

The first hypothesis (H1a) of the brand love variable was: “Brand love positively influences brand

extension evaluation.” According to the found results, brand love was found to have an insignificant

33 the line extension, hypothesis 1a is partially supported since a significant positive effect has been found of brand love on the evaluation of category extensions.

Hypothesis (H2a) of variety seeking tendency was described as follows:”Variety seeking tendency

mitigates the effect of brand love on brand extension evaluation”. For category extensions, the

interaction effect of variety seeking tendency was found to enhance the effect of brand love on the evaluation, however this effect was found to be only significant at a confidence interval of 90% (B = .111 and p = .096). When measured separately from the moderator of extension product involvement, variety seeking tendency showed a significant positive effect on the effect of brand love on evaluation on a 95% confidence interval (B = .138 and p = .030). This means that the moderator of extension product involvement slightly influences the effect of the variety seeking tendency moderator. With line extensions, results implicated that variety seeking tendency significantly enhances the effect of brand love on the evaluation (B = .135 and p = .015). However, this finding cannot be build upon as the main effect was found to be insignificant. Because there were no significant negative effects found, hypothesis 2a cannot be supported. Although insignificant, the results implicate a positive effect rather than a negative effect of variety seeking tendency.

For extension product involvement the main hypothesis (H3a) was: “Extension product involvement

enhances the effect of brand love on brand extension evaluation”. With line extensions, the results

showed that although brand love had an insignificant effect on line extension evaluations, implications were found that extension product involvement had a significant positive moderating effect (B = .145 and p = .020). The results for the category extension however, showed an insignificant moderation effect of extension product involvement (B = .024 and p = .607). Because of the latter insignificance and the insignificance of the main effect of brand love with the line extensions, it cannot be accepted that extension product involvement enhances the effect of brand love on brand extension evaluation and therefore hypothesis H3a is not supported.

5.4. T-test

34 TABLE 5.10.: REGRESSIONCOEFICCIENTS AND RESULTS T-TEST

T-test _b1 Premium light beer b2 Perfume var1 Premium light beer var2 Perfume T-test value Brand Love _{-.089 .261 .009 .012 -2.41}

Variety seeking tendency _{.135 .111 .003 .004 .279}

Extension product

involvement .145 .024 .004 .002 1.56

In this table, the regression coefficients of the variables are presented for both the line extension and the category extension (b1 and b2). In the next column the variances of these regression coefficients are presented (var1 and var2). These numbers are then used in the formula in order to compute the t-test value. When the t-test value of the t-test is higher than 1.96 or lower than -1.96, there is a significant difference between the effects with the two different extension types. When looking at the results in table 5.10., it can be seen that the t-test value of brand love is lower than -1.96 as this value is -2.41. Thus, the effect of brand love for line extensions and category extensions differ significantly from each other. When the effects are compared, the effect of brand love on line extension evaluation was found to be insignificant and has a regression coefficient of -.089, which thus is significantly lower than the found significant positive regression coefficient (.261) of brand love on category extension evaluations. From this can be concluded then that brand love has a significant more positive effect on category extension evaluations than on line extension evaluations. Based on this result from the t-test, conclusions can be made on hypothesis H1b: “The positive effect

of brand love on brand extension evaluation is stronger for line extensions than for category extensions.” The t-test shows that the difference between the effect of brand love on the line

extension and the category extension differs significantly. However, contrasting to what was hypothesized, brand love was found to have a more positive impact on the evaluation of category extensions than for line extension evaluations. From this can be concluded that hypothesis 1b is not supported.

The t-test value of the variety seeking tendency does not exceed the value of 1.96 or -1.96 from which can be concluded that there is no significant difference in the moderator effect of variety seeking tendency between the line extension and the category extension. Based on this result, some conclusions can be made for the secondary hypothesis H2b: “The negative effect of variety seeking

tendency on the relationship between brand love and brand extension evaluation is stronger for line extensions than for category extensions.” Based on the finding in the regression analysis that no

35 Lastly, when looking at the t-test value for the extension product involvement moderator, can be seen that this value (1.56) comes close to being significant, however as it still does not exceed the critical values this difference is not significant. So no significant different extension product involvement moderator effect exists between the line extension and the category extension. This result then is used for testing hypothesis (H3b): “The positive moderating effect of extension product

involvement on the relationship between brand love and brand extension evaluation is stronger for line extensions than for category extensions”. Based on the finding that there is no significant

difference between the moderator effect of extension product involvement between line extensions and category extensions this hypothesis cannot be supported.

6. DISCUSSION

In this discussion section, the findings in the results section are discussed. This begins with the findings on brand love, after which the findings of variety seeking tendency and extension product involvement are discussed.

6.1. Brand love

36 consumers who do not love the brand. This reasoning of brand love, that especially positively influences evaluation of dissimilar extensions, implicates that brand love diminishes the impact of similarity between the brand extension and the core brand on evaluations, found by Aaker and keller (1990), Bhat and Reddy (2001), Boush, Shipp, Loken and Gencturk et al. (1987) and Meyvis and Janiszewski (2004). These authors found that a higher similarity between the extension and the core brand leads to more favorable evaluation of the extension. While this is found to be true for ‘normal’ consumers, consumers who experience brand love show to be less influenced by the similarity as they are found to significantly evaluate the dissimilar extension more favorably than consumers who do not experience brand love.

6.2. Variety seeking tendency

The findings implicate that variety seeking tendency has a significant positive moderating effect in both evaluations of line extensions and category extensions. This instead of the proposed negative effect of the variety seeking tendency on the effect of brand love on extension evaluations. This proposed negative effect was based primarily on the finding of Dodd, Pinkleton and Gustafson (1996) who found that compared to non-variety seeking consumers, high variety seeking consumers tend to use significantly more information sources for the evaluation of products. By using more information sources, these consumers were found less likely to be affected by affective or emotional factors that impact their brand extension evaluation. However, results in this study contradict this statement as positive moderating effects of variety seeking were found. An explanation for this could be that the extensions, even the similar line extensions, are already seen as a form of extra variety by the consumers. Thus, consumers who love the brand but are seeking variety find fulfillment of this variety seeking in the extension of their beloved brand. Because of this fulfillment, the effect of variety seeking tendency strengthens the brand love of the consumers with the brand even more and has a positive impact on the evaluation of the brand extension. This reasoning is in line with the study of Helmig, Huber and Leeflang (2007), who found that consumers are more likely to seek variety within a familiar brand than seeking variety within different brands. Also, the positive effect of variety seeking tendency could be resulting from a more affective and emotional reasoning of consumers to find more variety, rather than making a rational decision based on more sources of information.

6.3. Extension product involvement

37 complementarity between the core brand and the extension product. Consumers with high involvement thus are believed to overlook differences between the core brand and the extension product and thus evaluate the extension more favorably. With line extensions, the results implicated that although the positive effect of brand love on line extension evaluations was found insignificant, extension product involvement has a significant positive moderating effect. The results for the category extension however, showed an insignificant positive moderation effect. Furthermore, it has been found that extension product involvement has a significant positive direct effect on the evaluations of both line extensions and category extensions. This direct positive effect of extension product involvement on the evaluation of brand extensions may be the reason that no significant moderating effects were found for category extensions. The direct effect explains the difference in evaluation of the extension more than the moderation effect of extension product involvement, which results in an insignificant effect of this moderator for category extensions. This direct positive effect is quite logical because, when consumers perceive the extension product as relevant to their needs, values and interests, they are more likely to evaluate the extension more favorably.

As both effects of extension product involvement were found to be positive, this implicates that involvement would indeed have a positive effect rather than a negative effect on the evaluation of brand extensions. This implicates that the view of a the Elaboration Likelihood model, according to which extension product involvement would have a negative moderating effect on the effect of brand love on evaluations of brand extensions, does not apply to the extension product involvement for evaluation of brand extensions. This model stated that when consumers are highly involved they process information via the ‘central route’ and focus more on attribute and performance information rather than emotional stimuli (e.g. Petty and Cacioppo 1984).

7. MANAGERIAL IMPLICATIONS