Functional fit and image fit in consumer evaluation of brand extension and the moderating role of involvement

(1)

Functional fit and image fit in consumer evaluation

of brand extension and the moderating role of

involvement

Meng Ran

Student number: 10604839

MSc Business Studies

(2)

1

Acknowledgement

First of all, I want to thank my supervisor Dr. Karin A. Venetis for her insightful advices and suggestions on my thesis, and for her immense knowledge, patient listening and encouragement not only during my thesis period but also throughout my two courses (Theories of Marketing and Marketing Strategy) at UvA.

Secondly, I would like to express my sincere gratitude to all my survey participants. Without their participation and precious comments, I will not be able to collect enough data and finish my thesis.

Finally, I would like to thank my parents, family members and friends, who have always been there for me, sharing my sorrow and happiness. I feel proud and lucky to have them in my life. I am on the way to better myself to love and protect all my angels.

With best regards, Meng Ran

(3)

2

Abstract

This study explores the bases on which consumers perceive the functional fit and image fit between a brand extension product and the original brand/product. It has been pointed out that brand/product associations are the foundation for functional and image fits’ perception. The author proposes that the difference between functional fit and image fit lies in the levels of abstraction of the associations that reflect the corresponding fit, where involvement has a role in altering the importance of different levels of associations. The existing research therefore aims to find out the role of involvement in the consumers’ perception of extension-origin functional and image fit. The results of surveys conducted in the pretest and final main study reveal that both functional fit and image fit have positive effect on evaluation of brand extension products. In addition, the empirically study supports that, as the level of involvement goes up, the effect of functional fit on evaluation of brand extension products decreases while the effect of image fit increases.

(4)

3

1. Introduction

1.1 Background

Brand extensions, known as the use of an established brand name for the introduction of new products or services (Aaker & Keller, 1990) have been kept attracting attention in both academic and practical field. Currently, brand extension already becomes an ideal option for many companies to introduce its new products or services across different industries (Aaker 1991; Farquhar, 1989; Leuthesser 1988; Tauber 1988). Reasons may be that, brand extensions could reduce introductory marketing expenses and leverage its associations of original brands. Although brand extensions are popular branding strategy, customers are the ones that pay the final products or services; therefore, they are the final determinants of the success of a brand extension.

From getting to know the extension products or services to final purchasing behavior, consumers would go through some processes, and among those processes, mental processes play a crucial role. Usually, the outcome of mental process is consumers’ evaluations such as attitudes, beliefs, and preference (Zajonc & Markus, 1982), which would influence final purchasing behavior, and in return, contributes to the success of brand extension.

Although brand extension strategy has gained great popularity among a number of famous brands, not all brand extensions received the expected favorable consumer evaluation (Keller, 2002; Martinez et al., 2003; Carrillat et al., 2013). This phenomenon drives scholars to study intensively on the factors that are critical for consumer evaluation of brand extension. Among the various studied factors, one weights an important position: fit between an extension product and its original

(7)

6 brand/product, which is also called brand origin-extension consistency, namely, the similarity between the two involved products/brands (Aaker & Keller, 1990).

Brand origin-extension fit is not one-dimensional, scholars base their study of brand origin-extension fit on a variety of different dimensions: quality aspect (Aaker & Keller, 1990; Keller & Aaker, 1992), feature/function/product and concept/image (Carter & Curry, 2013; Carrillat et al., 2010; Park et al., 1991), category including FMCG, durables, services, and retailers (Bhat & Reddy, 2001; Park et al., 1991), attribute and usage situation (Ahluwalia, 2008). Although there seems to be a number of aspects of fits, they can be generalized into two levels: product level fit and brand level fit (Bhat & Reddy, 2001). In the existing paper, the author focuses on functional fit and image fit, with the former one at product level and latter one at brand level, one is more tangible/concrete while the other more intangible/abstract. Assume that Heineken is introducing two extensions, an extension that shares an image fit could be movie disc because they are both for entertainment and relaxation, on the other hand, the one with a functional fit could be mineral water since both of the products are drinks. More detailed and systematic comparison of functional fit and image fit will be presented later in this paper.

Prior research on functional fit and image has studied the two fits’ roles such as being mediator or moderator in consumer evaluation of extension brand (Carter & Curry, 2011). These research provide several evidences why fit is so important to an extension, and the most convincing ones are: (1) there will be an enhanced affect or belief transfer towards the extension when the brand extension reflects high degree of fit with the established brand; and (2) a low degree of original-extension brand fit may

(8)

7 cause undesirable beliefs and associations. For instance, poor fit, according to Aaker and Keller (1990), may lead consumer to question a food firm’s ability to make bicycle. However, only a few scholars study the elements that would alter the effects of functional fit and image fit on consumer evaluation and among those researches, one element – consumer involvement was omitted.

Involvement has been frequently studied within consumer behavior and marketing area since its first use (Krugman, 1965). Involvement with an issue has been concluded as one important variable that influences the extent to which the information could be considered (Petty et al., 1983). The difference between low and high involvement lies in the motivation and ability. Consumer under high involvement has the motivation and ability to seek and process information while those under low involvement lack either motivation or ability to do the seeking and processing. Evidence concludes that brand attitudes might vary under different levels of consumer involvement (Mulvey & Olson, 1994; Keller, 1993; Keller, 2001; Petty & Cacioppo, 1983). Recent research shows that involvement is positively related to more favorable brand attitude by influencing consumers’ information searching, processing and attitude formation (VonRiesen & Herndon, 2011).

There exists a gap in the literature regarding the interaction between involvement and the two specific dimensions of fit: functional fit and image fit. Therefore, the existing paper attempts to relate involvement with consumers’ seeking and processing of brand origin-extension functional fit and image fit, to see whether the variable involvement alters the effect of functional fit and image fit on consumer evaluation of brand extension product. Furthermore, if involvement do alter the effect of functional

(9)

8 fit and image fit on brand extension evaluation, what is the specific role of involvement, will involvement positively or negatively influence the relationship between functional/image fit and consumer evaluation of brand extension product? In addition, which dimension of fit is more important for the evaluation of brand extension product under different degrees of involvement?

1.2 Research Question

Against above background, this paper aims to answer the following research question:

What is the effect of functional/image fit on consumer evaluation of brand extension product? Does the level of consumer involvement moderate the effects of functional fit and image fit on consumer evaluation of an extension? If there is a moderating effect, how does it work?

In order to answer the research question, several sub-questions are in need to be answered:

• What is brand extension?

• How does consumer form evaluation of brand extension product, and what is the role of functional/image fit in the evaluation of brand extension product? • What is the difference between functional fit and image fit in terms of being

inputs for consumer evaluation of brand extension products?

• How does involvement influence the consumers processing with functional fit and image fit?

(10)

9 1.3 Structure

In the second chapter, a literature review of the core concepts mentioned in the introduction and related theory is given. Afterwards, hypotheses will be built up. The third chapter will go into the methodology of this research, including pretest, initial design and follow-up experiment. Data analysis and results will be presented in the fourth chapter. Conclusions will be drawn in the last chapter, where the author will also discuss the implications, limitations and recommended avenues for future research.

2. Theoretical Framework

Consumer response reflects the state changes that a consumer experiences – either temporally or on a more permanent basis – as a result of exposure to a marketing communication.

Keller, 2011

There are two core elements in the quote from Keller (2001): (1) Consumer response processes, and (2) Consumer response outcomes. Consumer response processes could be peripheral/heuristic, or central/systematic. Response outcomes then include attitudes, preferences, behaviors etc.

In this chapter, the author will start with reviewing literature on consumers’ brand attitude formation theories, details on the evaluation basis will be provided. In the next, the concepts of functional fit and image fit will be elaborated, the nature and characteristics of functional and image fit is compared. Afterwards, mainly based on associative network theory, categorization process theory, the difference between consumers’ processing with functional fit and image fit will be presented.

(11)

10 Furthermore, the author will introduce the variable involvement, and apply consumer involvement theory to the functional/image fit’s role in consumer evaluation of brand extension. Finally, the hypotheses about under which level of consumer involvement is the image/functional fit be more important will be built up.

2.1 Brand extension

Brand extension, is an approach used by practitioners to introduce new products into either a new market segment or an entirely different product class (Aaker & Keller, 1990). Since 1970s, leveraging brand names to expand business has gained considerate attention in the business world (Aaker & Keller, 1990). There are several reasons that why brand extension is so appealing to companies. An established brand name enjoys a certain degree of consumer familiarity, knowledge and positive image. On the one hand, the advantages of the established brand could reduce the risk of introducing a product no matter into a new market segment or to a different product category. On the other hand, the aforementioned advantages could reduce the promotional expenses of introducing new product and increase the distribution efficiency (Schuiling & Kapferer, 2004).

2.2 Brand associations and brand origin-extension fit

According to associative network theory, an associative network in one individual’s mind is a cognitive model that is made up of interconnected informational nodes (Anderson & Bower, 1973; Keller, 1993). For a familiar established brand, consumers usually have such a well-recognized structure (Anderson, 2013), brand or product associations serve as the nodes and are frequently used as the basis to evaluate the brand’s new extension when they are activated by stimuli such as advertisement.

(12)

11 Brand name then could serve as a cue, which could activate brand images that have been built up in consumers’ mind due to previous usage or experience (Geuens & Pecheux, 2006).

Brand associations therefore are the bases on which consumers evaluate a brand or product. Consistent with Means-End Chain and Laddering Theories, which have been frequently used by scholars (Graeff, 1997; Gutman, 1982; Pieters et al., 1995) to analyze consumer attitude formation process, attributes, benefits, and brand attitudes are three different levels of brand associations that vary in the level of abstraction (Walker & Olson 1991).

Attributes include product related attributes and non-product related ones, which also equal to concrete and abstract attributes (Keller, 1993; Reynolds & Gutman, 1984; Walker & Olson 1991). Usually product’s physical characteristics refer to the former type of attributes, and scholars use features and product related attributes interchangeably. For latter type of attributes, they are external to a product’s consumption. Price, packaging, usage context and identity of the consumer are those making up non-product related attributes. Benefits, which are also called as consequences (Keller, 1993; Reynolds & Gutman, 1984; Walker & Olson 1991), include functional benefits and psychological ones. Functional benefits are the consequences of product related/concrete attributes. Psychological benefits then are higher order consequences that lead to how consumers feel to use the product (Keller, 1993; Reynolds & Gutman, 1984; Walker & Olson 1991). Consumer’s brand attitude is an overall evaluation of brand attributes and benefits associations (Keller, 1993; Reynolds & Gutman, 1984), and therefore is the highest order of brand association.

(13)

12 Instrumental attitudes pertain to concrete product attributes and benefit, while terminal attitudes accrue to abstract characteristics and benefits (Keller, 1993; Reynolds & Gutman, 1984; Walker & Olson 1991). Below is a brief framework of brand associations:

Figure 2.1: Brand Associations

In terms of evaluation of brand extension product, fit has been identified as the most important factor (Volckner & Sattler, 2006). Fit in the existing paper refers to the shared associations between the extension product and original product/brand (Carter & Curry, 2013). As mentioned above, associative network of brand or product associations is the basis for consumers to make evaluation of brands and products. A broadly accepted paradigm is that, the fit between an extension product and its original brand will increase as the similarity between two brands’ associations increases.

The reason why brand origin-extension fit is crucial is that, regardless of specific dimensions of fit, a brand origin-extension fit makes it possible for consumers’ previous brand evaluation to transfer to the extension. The higher the fit, the more

Brand attitudes Terminal Instrumental Benefits Psychological Functional Attributes Abstract Concrete Abstract Concrete

(14)

13 enhanced the transfer will be (Aaker & Keller, 1990); in contrast, low fit would inhibit such a transfer of either affect or cognition from an established brand to an extension. As Boush and Loken (1991) proposed and proved, that a high fit positively influences brand extension evaluation while a poor fit negatively influences brand extension evaluation.

2.3 Brand origin-extension functional and image fit

Functional fit accrues to the product associations that represent the product function while image fit pertains to the non-product associations that make up a brand image. Due to the difference in the nature of functional fit and image fit, the evaluation transfer processes will differ.

2.3.1 The concept functional fit

Functional fit, according to Keller (1990), refers to the substitute or complement nature of the extension relative to the established brand. To be more specific, functional fit is defined as the degree to which an extension and the original brand have functional overlap, and therefore satisfy the same consumer needs (Dacin & Smith 1994; Park et al. 1991). In this paper, the degree to which product associations of the extension are similar to those of the original product at the concrete level determines the degree of functional fit. For instance, if an established brand sells computer, and it is introducing an extension brand that sells tablet, since computer and tablet share similar functions such as multimedia and office software, then tablet could be seen as a substitute for computer. This means that the benefits or consequences of using computer and tablet are similar. In this case, there is a high functional fit between the extension brand and the original brand.

(15)

14 Regarding the dimensions of functional fit, the author adopts the theory put forward by Keller (1990) - two are demand-side and one is supply-side. From the perspective of demand-side, there are two dimensions of functional fit: COMPLEMENT and SUBSTITUTE. COMPLEMENT refers to the degree to which people see two products could jointly satisfy some certain needs. For example, toothpaste and toothbrush could be regards as complement to each other since they are for the same need – tooth cleaning. SUBSTITUTE refers to the degree to which two products could substitute each other to satisfy the similar needs. One instance could be that, detergent and wash powder could replace each other in terms of clothes washing. The third dimension of functional fit TRANSFER, from the perspective of supply-side, refers to the degree to which that the people, skills, and facilities a company uses to produce original product could transfer to the production of extension product. However, it is important to point out that the scope of brand origin-extension functional fit in this paper is slightly narrower than that of category fit or similarity. Functional fit in the existing paper refers to the functional overlap between an original brand and its extension (Cater & Curry, 2013), only when an extension and original brand satisfy the similar consumer needs could there be a brand origin-extension functional fit. However, category fit or similarity between an original brand and the extension has its emphasize on the term “category”, which is broader. For example, razor and hair drier are both within the household electrical appliance category, we could say there is a high category fit between the razor and hair drier, however, the functional fit between the two products is less significant because they are for different consumer needs.

(16)

15

2.3.2 The concept image fit

In addition to the function dimension, an extension could also fit with the original brand in the abstract meanings and benefits, which represent a brand image. Being introduced about half a century ago, brand image has been a common topic in the research area of business since then. Based on prior scholarly papers, Dawn and George (1990) identify five categories within which researchers define brand image:

(1) Broad definitions (Herzog, 1973; Runyon & Stewart, 1987), some definitions are so broad that they just give a general sense of brand image; the rest four have emphases on (2) symbolism (Frazer, 1983; Levy, 1958); (3) meanings and messages (Durgee & Stuart, 1987; Friedmann & Lessig, 1987; Swartz, 1983) (4) personification (Kassarjian, 1977); (5) cognitive or psychological elements (Reynolds & Gutman, 1984).

In this paper, the author adopts a mixture of those categories to represent the meaning of brand image. More specifically, brand image is a concept interpreted or perceived and held by consumer (Keller, 1993) and reflected by the firm selected value associations contained in the brand. Accordingly, brand origin-extension image fit refers to the degree to which an extension brand shares concept feelings and associations with the established brand (Salinas & Perez, 2009). Thus, image fit leads consumers to believe that they are able to enjoy similar benefits from both the parent brand and its extension products (Park et al., 1986). Although scholars sometimes use different terms such as concept consistency (Park et al., 1991; Lanseng & Olsen, 2012) and prototype fit (Mao & Chrishnan, 2006), they together with image fit, refer the similar meaning. The existing paper studies brand image that pertains to brand

(17)

non-16 product associations, the image fit information therefore is at an abstract or conceptual level.

Although brand image is an abstract concept, there are specific brand image and more general brand image as well. For example, some researchers specifically term brand image as traditional, European, expensive, cheap and so on (Keller, 1990); while other scholars generalize brand image as functional and expressive (Riezebos, 2003), regarding the expressive brand image, scholars further categorize it as symbolic and experiential (Park et al., 1986), subjective and hedonic (Mittal et al., 1990). In the existing paper, the author adopts the former perspective of image fit categorization. The reason is that, functional and expressive image is so broad that any products could be paired together as having high image fit as long as they are both functional and expressive oriented. Due to the fact that brand image perception is a more subjective process, perceptions of one certain brand image then vary across different individuals. For example, some people may see the brand Red Bull as unhealthy while others may see the same brand as energetic. Therefore, in order to avoid individual differences when people perceiving the image of a brand, the existing study only adopts the most representative images of a brand when designing and manipulating the image fit between an original brand and its extension product in the following pretests and experiment.

Furthermore, it has been pointed out that conceptual cues or triggers could make products accessible to consumers (Anderson 1973; Berger & Heath 2005; Berger & Fitzsimons 2008), which makes it possible for consumers to assign one product to

(18)

17 some certain category within which consumers enjoy the similar benefits (Krishnan & Smith, 1999).

Against above elaboration on the concept of brand associations, functional fit and image fit, brand origin-extension fit could be generalized into two levels: product fit and brand fit, which accrue to product attributes fit and non-product attributes fit respectively, as functional fit belongs to the product fit and image fit is a member of brand fit . Due to the different abstraction level of brand associations, the scope of image fit is broader than that of functional fit (Bhat & Reddy, 2001; Carter & Curry, 2013; Loken et al, 2008).

2.3.3 Functional fit and image fit information processing

Categorization process theory has been used in some researches (Carter & Curry, 2013; Mao & Krishnan, 2006) whose topics are related to brand extension or co-branding. Here the term “categorization” includes but is not limited to the product level, it can also refers to brand level such as concept category. The author adopts the definition of category which is made by Loken et al. (2008), that category is one set of products/brands/services/marketing entities/events/states that appear to be related one another to consumers in some way. The theory (Loken et al., 2008) points out that consumers use categorical representations stored in their cognitive system to assign one new product to a consumer category, these representations actually are equal to the brand associations in this paper. Thus, consumers are likely to regard an original brand and its extension as one category when there is a fit between the two products.

(19)

18 However, there is a difference between functional fit and image fit information processing. Within categorization theory, scholars view similar products from two perspectives: the prototype view and exemplar view, prototype view posits that category is reflected by abstract and imaginary (Mao & Krishnan, 2006) instances while exemplar view assumes that category is reflected by specific and concrete composites (Medin & Schaffer, 1978; Rosch & Mervis, 1975, Loken et al, 2008). These two views are consistent with image fit and functional fit information processing.

Based on empirical evidence that judging one object with a group of objects at an abstract level is more difficult than judging the same object with only one object at a specific level (Malt et al, 1989), Mao and Krishnan (2006) proposed and proved that, cognitive resources or involvement could positively moderate the effect of prototype fit on brand extension evaluation. In addition, they do not find the moderating role of cognitive resource in the effect of exemplar fit on brand extension evaluation. In the experiment, Mao and Krishnan (2006) chose Johnson & Johnson as the original brand, they designed extensions with different dimensions of fits. Although different names of fit are used in Mao’s experiment, the general meanings of prototype fit and exemplar fit are consistent with those of image fit and functional fit. It has to be pointed out that, the scope of functional fit in this paper is narrower than that of exemplar fit, since Mao adopted hairdryer (shampoo as original product), sponge (bath wash as original product), sunscreen oil (oil as original product) and wound cleanser (adhesive bandage as original product) as extensions that reflect high exemplar fit. The finding of Mao and Chrishnan (2006) drives the author to think whether the level of involvement also moderate the effect of functional fit and image

(20)

19 fit on consumer evaluation of brand extension. Earlier to Mao and Chrishnan (2006), Meyvis and Janiszewski (2004) demonstrate that both category fit and image fit affect consumer evaluation of brand extension, however, the role of image fit depends on the accessibility and diagnosticity of benefit associations, which reflect the level of brand origin-extension image fit.

The concept accessibility, which belongs to the retrieval theory, has been studied when scholars analyzing consumer responses to brand extensions (Kokkinaki & Lunt, 1999; Lynch et. al., 1988; Meyvis & Janiszewski, 2004). Accessibility refers to the degree to which a piece of information can be retrieved from memory (Feldman & Lynch, 1988). An important conclusion of the retrieval theory is that consumers will only retrieve the most accessible information to make a judgment instead of retrieving all the information, the retrieved information could be either benefit associations or prior evaluation (Lynch et al, 1988). It has been concluded that when making subsequent decision, the accessibility of an input in the memory and alternative inputs determine the likelihood that the prior evaluation will be used as an input (Lynch et. al., 1988). Alba and Hutchinson (1987) propose that consumers usually initially assign a new product to one category at the basic level, those with more brand and product knowledge are more likely to have stronger ability to make categorization of a new product at an abstract level. They also assume that similarity is a function of all accessible attribute information, low/high involvement inhibits/promotes consumers to use more accessible information (Alba & Hutchinson, 1987). While according to prior research, involvement has been concluded as a critical variable influencing consumers’ processing with cognitive resources (Petty & Cacioppo, 1983; Zhang & Zinkhan, 2006).

(21)

20 2.4 The role of involvement

Since its introduction by Krugman (1965), involvement has been attracting scholars’ attention for around half century. Consumer involvement has been regarded as one of the most important variables that influence consumer responses to a brand or product (Asadollahi, 2011). Regardless of differences on definition of involvement across different academic areas, it is broadly acknowledged that involvement is related to personal needs/values/beliefs/feelings at an individual level under one specific situation. Therefore, the level of consumer involvement might vary by the situation and change over time. In previous articles, researchers also use personal importance and personal importance interchangeably with the concept involvement (Antil, 1984; Petty & Cacioppo, 1983; Zaichkowsky, 1985).

Involvement could have its influence in different phases, for instance, consumers may act differently in terms of mental routes when searching or processing information under different level of involvement (Zhang & Zinkhan, 2006). More specifically, under high involvement, consumer is motivated and equipped with the ability to process stimulus information, thus takes the central route to process information and form evaluation (Keller, 1993, Keller, 2001; Olson & Jacoby, 1974). In response to a brand advertisement, consumer is more focused on the information content, argument quality under high involvement (Zhang & Zinkhan, 2006). However, under low involvement, consumer lacks either motivation or ability to process stimulus information, and therefore takes peripheral route to evaluate a product or brand based on accessible signals or peripheral cues at the time (Keller, 1993). For example,

(22)

21 Kaulio (1998) once pointed out that low involvement choices are believed to involve only few and rather concrete attributes.

As mentioned in earlier paragraph, the difference between the nature of functional fit and image fit makes them differ in terms of accessibility and diagnosticity. Since functional fit pertains to the similarity between the concrete associations of an extension and its original product and image fit pertains to the more abstract associations of an extension and its established brand. Means End Chain model has been suggested as tool for consumers to make evaluations based on brand associations, however, researchers contend that under low involvement, consumers lack the motivation and ability to process abstract associations (Gutman, 1982). In other words, the more important a product is for consumer, the more involved the consumer will be, and the abstract brand associations (Mulvey & Olson, 1994) that reflect brand origin-extension image fit are more likely to be used as evaluation inputs. Thus the author proposes that:

(1) under low level of involvement, functional fit is more important for consumer evaluation of a brand extension product, because the concrete associations that reflect functional fit are more accessible to consumers compared to the abstract associations that reflect image fit, this leads to a stronger prior attitude transfer via functional fit;

(2) under high level of involvement, consumers are more motivated and knowledge equipped to seek and think about the non-product associations, which means that the possibility that consumers process abstract brand

(23)

22 associations increases, as long as the abstract associations are presented, the likelihood that they are used as input for evaluation increases and therefore the importance of image fit rises. Meanwhile, compared to the increasingly accessible associations that reflect image fit, the accessibility of concrete associations remains the same, which makes functional fit relatively less important.

2.5 Conceptual model and hypotheses

As can be seen in Figure 2.2, the conceptual model of this paper incorporates two independent variables – functional fit and image fit, one moderator - involvement, and one dependent variable – evaluation of brand extension product.

Figure 2.2: Conceptual model

Regarding previous discussion in this chapter, the existing paper proposes the following hypotheses: H2 H3 H4 H1 Functional fit Image fit Involvement Evaluation of brand extension product

(24)

23 Hypothesis 1: Consumer evaluation of brand extension product is positively related with the level of brand origin-extension functional fit.

Hypothesis 2: Consumer evaluation of brand extension product is positively related with the level of brand origin-extension image fit.

Hypothesis 3: The effect of functional fit on consumer evaluation is negatively moderated by the level of consumer involvement.

Hypothesis 4: The effect of image fit on consumer evaluation is positively moderated by the level of consumer involvement.

3. Methodology

In this chapter, necessary pretest and main study design are discussed.

Because of the nature of the research question of this paper, which focuses on people’s perception and evaluation of brand/product, survey was selected as the appropriate data collection method. In the survey, real brands were used so that the brand associations consumers already have with the brands could be modeled.

3.1 Brand selection

When choosing the proper brands, there are several aspects that should be considered. The first aspect to be considered is the breadth of the brand. Brand breadth, refers to the variation among the current products of a brand, is concluded to have effect on brand typicality of the extension, which in return affects the brand extension evaluation (Boush & Loken, 1991). Broad brands are those who have extended to

(25)

24 provide very different products from the original products and narrow brands always offer similar products. There are two opinions among different scholars. For broad brand, some researchers posit that the brand breath would dilute the perceived fit between an extension and original brand even if there is a high fit between, while brand breadth has little effect on the extremely discrepant extensions (Boush & Loken, 1991). However, there are also scholars pointing out that broad brands provide multiple references that serve as evaluation basis and thus make fit perceptions and attitude transfer processes more complicated (Mao & Chrishnan, 2006). To further study the complexity of the broad brand extension evaluation, Dacin and Smith (1994) suggest that consumers may evaluate an extension based on many products of a multiple brands. An extreme phenomenon is that, a brand has been continuously extended into different categories, consumers may perceive this broad brand being able to produce everything, which means that, if a brand is too broad, consumers do not evaluate its extensions based on product fit information any more. Meanwhile, as the broadness of an original brand decreases, brand/product associations become less influential and accessible (Meyvis & Janiszewski, 2004), which would affect consumer perception of functional fit and image fit. Therefore, in the existing experiment, when choosing brands to conduct experiment, the author excluded those brands are either too narrow or too broad.

The second aspect is the level of consumer involvement. Scholars (Laurent and Kapferer, 1985) use the antecedents of involvement as operational indicators. According to previous research, there are several antecedents of involvement, and they can be divided into: the characteristics of consumer, the characteristics of the stimulus and the characteristics of the situation (Zaichkowsky, 1994). Therefore,

(26)

25 involvement varies across different individuals, stimulus and situations (Warrington & Shim, 2000). Based on prior research, Laurent and Kapferer (1985) developed Consumer Involvement Profile (CIP) to measure consumer involvement on four dimensions: perceived sign/symbolic value, perceived hedonic/pleasure value, perceived risk, and perceived importance. Almost at the same time, Zaichkowsky (1985) developed a Personal Involvement Inventory (PII) consisting 20 items, which could be applied to measure consumer involvement with product and purchase. Later on, Zaichkowsky (1994) reduced the number of the items from 20 to 10, dividing them into two subgroups – affective and cognitive in nature, the former group is represented by interesting, appealing, fascinating, exciting and involving (Nelson et al., 1985), the latter group is composited by important, relevant, valuable, means a lot to me and needed. The reduced inventory has also extended its application area into advertisement, which makes it a context-free tool. Through empirical study, the PII has been proven as a useful tool for both manipulating and measuring involvement. This study will firstly manipulate people’s involvement with some certain product categories and then measure the involvement to select proper ones.

Involvement manipulation is a core part in the experiments conducted in this study. As aforementioned, antecedents of consumer involvement reflect consumers’ involvement with certain product or advertisement, which means theoretically, there are several ways to manipulate consumer involvement based on the characteristics of individual, product and situation. The author chose certain product categories to manipulate consumer involvement because there tends to be less variation among consumers’ perceptions of certain product characteristics (Traylor, 1981), which means that consumer involvement with some certain product categories are stable. In

(27)

26 terms of PII and CIP, both tools have been proven valid and useful, all items and constructs incorporated in CIP are also included in PII (Mittal, 1990; Mittal, 1995; Rodgers & Schneider, 1993). Therefore in the existing study, the author adopted PII (Zaichkowsky, 1994) as the tool to manipulate and check respondents’ involvement with product/brand. To be more specific, the existing study used 6 items for involvement manipulation and measurement: important, mundane, worthless, involving, appealing and means a lot to me.

Because this paper studies the moderating role of involvement in people’s evaluation of brand origin-extension functional/image fit, therefore, two independent variables are functional fit and image fit, one dependent variable is consumer evaluation of brand extension product, while variables that could affect the result should be controlled.

In addition, there are two variables that will be measured in pretest and be controlled in the main study. In the main study, different original brands should enjoy similar familiarity and similar degree of favorable attitude among all the respondents.

3.2 Pretest

3.2.1 Pretest 1

The purpose of this stage is to (1) obtain the level of respondents’ involvement with some certain product categories; (2) get respondents’ initial ratings of corresponding brands in terms of brand familiarity and attitude.

(28)

27 When using PII (Zaichkowsky, 1994) to check respondents’ involvement with certain products, items “exciting, boring and needed” were dropped. According to prior research, an object does not need to be “exciting” or “interesting”, or “needed” to be involving, in other words, the item “exciting”, “interesting”, and “needed” are not reflective of personal involvement (Mittal, 1995). This is because people have different motivations for the purchases of some certain products. For example, when buying a mobile phone, some pay more attention to the function to the product while some value the product appearance design, thus even if a mobile phone is not exciting or interesting, people still could get high involved with it due to its excellent function.

Nine product categories were chosen and thirteen corresponding brands were included in the survey.

Table 3.1: Initial product categories and corresponding brands Product category Corresponding brands

Toothpaste Colgate

Chocolate Ferrero

Soft drink Coca-Cola

Coffee Nestle

Detergent Ariel

Soap Nivea, Dove

Mobile phone Nokia, Apple

Personal computer HP, Apple

(29)

28 According to previous studies, toothpaste, chocolate, soft drink, coffee, detergent and soap are the products that have low consumer involvement; in contrast, mobile phone, personal computer and car are products that have high consumer involvement (Laurent & Kapferer, 1985; Zaichkowsky, 1994).

Due to the fact that brand familiarity various across different countries, and the pretest will be conducted among respondents from different countries, which makes it difficult to choose the proper brands that enjoy the same degree of consumer familiarity. Before choosing the brands for both low involvement and high involvement products, the author made an initial selection on the brands based on informal communication with people of different nationalities, which helps to prescreen the brands that enjoy similar familiarity across different nations. As shown in the table, low involvement products are usually fast consuming goods, people may not perceive significant difference regarding the positioning of different brands. As shown in the table, low involvement products are usually fast consuming goods, people may not perceive significant difference regarding the positioning of different brands. Thus the author chose one representative brand for each low involvement product category. On the contrary, moderate or high involvement products tend to have more divergent images among different people and positioning strategies (Czellar, 2002), accordingly, the author selected two brands with different images/positioning within each product category to select the brands that enjoy the ideal level of familiarity and attitude favorability. After the initial selection, thirteen brands that enjoyed global familiarity were selected (Table 3.1).

(30)

29 Regarding the fact that low score of brand attitude would bias the attitude towards brand extensions. While this research only studies the positive attitude transfer via image fit and functional fit, so the brands used in this pretest and the following tests should enjoy positive and similar attitude among different respondents. Below is a brief summary of the measurements and related academic sources:

Table 3.2: Measurement items and corresponding scales

Measurement items/scales Sources

Involvement Laurent and Kapferer (1985)

Zaichkowsky (1994) Seven point bipolar items with anchors:

1. Unimportant/important 2.Mundane/fascinating 3.Worthless/valuable 4.Uninvolving/involving 5.Unappealing/appealing

6.Means nothing/means a lot to me

Familiarity Simonin et al.(1998)

Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): I am familiar with the brand

Initial attitude D’Astous et., al (2007)

Hem & Iversen (2003) Park et al. (1991) Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)):

(31)

30 1. My attitude is positive

2. I like it

3.I think it is of good quality

4. I would buy it if I was looking for a product in this category

Result of pretest 1

The pretest was conducted in the form of online survey. A total number of 85 respondents participated and 61 individuals who were all between 16 and 51 years old (mean = 27.9, SD = 8.67) completed the survey.

A reliability analysis is performed in SPSS to test the measurement items of involvement with the selected product categories and attitude towards some corresponding brands. For the scale items that measure either involvement with product category or attitude towards some specific brands, except for soap, all the Cronbach Alphas were above 0.90. Table 3.3 and 3.4 summarize the product category involvement, brand familiarity and attitude of aforementioned product categories/brands.

Table 3.3: Product category involvement a

Product Involvement Toothpaste 2.45 (0.535) Chocolate 3.00 (1.376) Soft drink 2.92 (1.439) Coffee 3.30 (1.678) Detergent 3.88 (1.548)

(32)

31 Soap 2.87 (1.076) Mobile phone 5.66 (1.066) Personal computer 5.82 (0.928) Car 5.27 (1.633) a

Numbers in the parentheses are the SD

Table 3.4: Brand familiarity and attitude a

Brands Familiarity Attitude

Colgate (toothpaste) 5.47 (1.099) 5.06 (1.198)

Ferrero (chocolate) 5.76 (1.770) 5.11 (1.542)

Coca-cola (soft drink) 5.71 (10.38) 4.41 (1.650)

Nestle (Coffee) 5.37 (1.386) 4.87 (1.544)

Ariel (detergent) 5.07 (1.821) 4.84 (1.344)

Nivea (soap) 4.30 (1.889) 4.23 (1.202)

Dove (soap) 5.05 (1.669) 4.93 (1.336)

Nokia (mobile phone) 5.43 (0.918) 4.96 (1.680)

iPhone (mobile phone) 5.84 (1.245) 5.39 (1.382)

HP (personal computer) 5.45 (1.594) 4.54 (1.374)

Apple (personal computer) 5.52 (1.255) 5.23 (1.373)

Toyota (car) 5.14 (1.673) 4.50 (1.560)

BMW (car) 5.61 (1.810) 5.91 (1.657)

a

(33)

32 Among the nine product categories as shown in the tables, toothpaste, soap and soft drink enjoy relatively low involvement; in contrast, car, mobile phone, personal computer are relatively high involvement product categories.

For brand familiarity, except for Nivea, all brands enjoy relatively high level of awareness among respondents. Among the three low involvement product categories, Colgate, Dove, and Coca-Cola received ratings (mean value) of 5.06, 4.93, and 4.41 respectively. While within the three high involvement product categories, BMW (5.81) got higher ratings than Toyota (4.50) did, iPhone enjoyed more positive attitude than Nokia did, and respondents’ attitudes toward Apple (5.23) were more favorable than towards HP (4.54).

Since this research only studies positive attitude transfer through brand/product associations, according to above results, the low involvement product category - soft drink, was dropped due to the relatively neutral rating that the corresponding brand Coca-Cola received. Taking brand familiarity and attitude into consideration, the brands of low and high involvement product categories should enjoy similar level of familiarity and attitude favorability, thus two brands – Colgate and Apple were selected in advance to be adopted in the following tests.

3.2.2 Pretest 2

The purpose of this stage is to: (1) double-check respondents’ involvement with the brands Colgate and Apple; (2) have respondents to generate product function, brand image and other brand associations, which will be applied in the design of brand extension products.

(34)

33 This open-ended pretest was conducted in the form of face-to-face survey. 30 respondents participated in the survey, all aged from 18 to 55. During the survey, there was short discussion between the researcher and participant, aiming to get better understanding of the product/brand involvement. After that, participants were asked to generate up function, image and any other associations the two brands have in their mind. Finally, each respondent was asked to come up with possible brand extension products based on either original product function or original brand image, or both.

Result of pretest 2

Table 3.5 summarizes the brand/product associations collected from respondents, only those had been mentioned frequently are listed.

Table 3.5: Summary of brand/product associations

Colgate Apple

Function Tooth cleaning, tooth protection Communication, office/study tool, entertainment

Image Clean, fresh, reliable User friendly, hi-tech, innovative,

intelligent/smart, fashion design Possible extension products with

functional fit

Electronic toothbrush, floss, mouthwash

Mobile phone, computer, professional audio devices Possible extension products with

image fit

Make-up remover, detergent, tissue

Household electric appliance, smart car, music instruments Possible extension products with

both functional and image fit

Dental care center, chew gum Intelligent office tools, online shopping site, game console

(35)

34 As shown in the table presented above, one finding is noteworthy. That the number of potential extension products generated for the brand Apple was higher than that for the brand Colgate. One possible explanation behind the finding is that, compared to Colgate, Apple has extended more broadly, which enables respondents to have more bases to generate extension product ideas for Apple than for Colgate. In other words, Apple has more accessible product/brand associations among respondents than Colgate does.

There is also one finding that was not reflected in the table. Before associations generation stage, respondents were asked the overall level of involvement they have with the two the two brands, all participants were reported to have low involvement with Colgate toothpaste, while 29 out of 30 said they were highly involved with Apple computer leaving the rest one to have medium level of involvement. This in advance supports the result of previous pretest, that toothpaste and personal computer are low involvement and high involvement product categories respectively.

3.2.3 Pretest 3

The purpose of this stage is to test whether respondents could perceive fit between the potential extension products and original brands under different conditions.

The pretest was conducted in the form of face-to-face survey. 31 respondents participated in the survey, all aged between 21 and 48. Based on the results of pretest 2, the researcher generated a list of products that could serve as potential extensions for Colgate and Apple respectively. Respondents were asked to rate each of the

(36)

35 extension product based on items (Table 3.6) that measure the functional fit and image fit between two products (Keller, 1990; Park et. al., 1991).

Table 3.6: Measurement items and corresponding scales Measurement items/scales Sources

Functional fit Keller (1990)

Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): 1. Complement

I think the extension product and original product could be jointly used to satisfy the same need 2. Substitute

I see the extension product as a substitute for the original brand product 3. Transfer

I think the manufacturer of the original product is able to make the extension product

Image fit Park et al. (1991)

Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): I think the extension product and the original brand shares consistent concept

The score for functional fit is the average of the three items mentioned above. Based the extension products generated by respondents, the researcher selected products that fit with original brand product on either only functional or image dimension, or both dimensions. In order to exclude the influence of favorability of the extension product itself, relatively neutral extension products were chosen.

(37)

36 Below is a table summarizes the respondents perceptions of the degree to which the extension products fit with the original products/brands.

a

Numbers in the parentheses are the SD

As shown in the table, according to respondents’ perceptions, for extension products of the brand Colgate, both floss and chew gum were perceived to possess some degree of functional fit and image fit, however, the fits were rated neither high nor low; detergent was rated as highly fit with the original brand on image dimension while received more divergent rating (with SD of 1.24) on the functional fie dimension. In addition, Colgate Electronic toothbrush was perceived to possess high functional fit and relatively low image fit, Colgate tissue received low score on functional fit and

Table 3.7: Perceptions of extensions’ functional fit and image fit a

Brands Extension products Ratings on functional fit Ratings on image fit

Colgate Floss 4.81 (0.54) 3.33 (1.12)

Electronic toothbrush 5.26 (0.94) 2.17 (0.38)

Tissue 1.29 (0.81) 5.13 (0.88)

Detergent 3.21 (1.24) 5.52 (0.51)

Chew gum 2.23 (1.32) 4.22 (0.78)

Dental care center 5.31 (0.66) 5.28 (0.49)

Apple Cheap and simplified mobile phone 5.34 (0.48) 1.31 (0.62)

Intelligent fan 1.42 (0.98) 5.28 (1.07)

Smart car 1.35 (1.33) 4.98 (1.16)

Online shopping site 2.25 (1.42) 4.32 (1.53)

(38)

37 high score on image fit, Colgate Dental Care center received high ratings on both functional fit and image fit. Regarding extension products for the brand Apple, respondents perceived the cheap and simplified mobile phone to be highly fit with the original product on functional dimension while lowly fit with the original brand on image dimension. On the contrary, intelligent fan got low score on origin-extension functional fit while high score on origin-extension image fit. Apple online shopping site was thought to have low functional fit with the original Apple product and moderate image fit with the brand Apple. Apple smart car and Apple smart printer both were regarded highly fit with the original Apple product/brand on function and image dimension, but Apple smart car was dropped as the respondents tended to have more diverse ratings on the smart car’s functional fit and image fit, compared to Apple smart printer and Colgate dental care center.

Above comparison considered, six extension products of Colgate and Apple were adopted in the following study, as shown in table 3.8.

3.3 Experiment design

The study is a 2 (involvement: low/high) x 2 (functional fit: low/high) x 2 (image fit: low/high) between subject design. According to involvement manipulation and check

Table 3.8: Selected extension products

Colgate Apple

Low functional fit & low image fit Printer paper Printer paper

High functional fit & low image fit Electronic toothbrush Cheap and simplified mobile phone Low functional fit & high image fit Tissue Intelligent fan

(39)

38 in the previous pretests, Colgate and Apple enjoy low and high involvement respectively. Based on the results of pretest 3, six extension products with different combinations of fit(s) were adopted in the main study.

3.3.1 The brand extension products group scenarios:

The experiment was conducted in the form of online survey. In the introduction part, participants were informed with strict anonymity, and were encouraged to answer each question truthfully. The scenarios of 8 experiment groups are described as below:

Scenario 1: Extension product with low functional fit and low image fit under low involvement

Colgate is selling its new extension product - Colgate printer paper.

Scenario 2: Extension product with high functional fit and low image fit under low involvement

Colgate is introducing its new extension product – Colgate electronic toothbrush. The durable electronic toothbrush offers you oral care such as teeth cleaning and protection.

Scenario 3: Extension product with low functional fit and high image fit under low involvement

Colgate is introducing its new extension product - Colgate tissue, which will make your life more clean and fresh.

(40)

39 Scenario 4: Extension product with high functional fit and high image fit under low involvement

Colgate is opening Colgate Dental Care Center. The center offers you reliable and professional oral care, which helps you to build your healthy and clean teeth.

Scenario 5: Extension product with low functional fit and low image fit under high involvement

Apple is selling its new extension product - Apple printer paper.

Scenario 6: Extension product with high functional fit and low image fit under high involvement

Apple is introducing its new extension product - iSimple. iSimple is a simplified mobile phone with functions of only phone calls and message sold at low price.

Scenario 7: Extension product with low functional fit and high image fit under high involvement

Apple is selling its new extension product - iFan. The intelligent electric fan could adjust its wind power according to the environment temperature.

Scenario 8: Extension product with high functional fit and high image fit under high involvement

Apple is introducing its new extension product - iPrinter. iPrinter's friendly and innovative design enables you to print, copy, and scan document in a convenient way, which would make your study and work more easy.

(41)

40

3.3.2 Measurement

Respondents first answered questions regarding demographic characteristics including age, gender, education and nationality. In the subsequent page, subjects were asked to answer questions in the following order: (1) familiarity with the original brand, (2) involvement with the original brand, (3) their perceptions of functional fit and image fit that the extension products have with original product/brands and (4) attitude towards the extension products. Ratings were all measured by Seven Likert items, with 1 represents the lowest score and 7 represents the highest. Below is a brief summary for the measurement items of extension product regarding attitude and functional/image fit:

Table 3.9: Measuring Items

Familiarity Simonin et al.(1998)

Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): I am familiar with the brand

Involvement Laurent and Kapferer (1985)

Zaichkowsky (1994) Seven point bipolar items with anchors:

1. Unimportant/important 2.Mundane/fascinating 3.Worthless/valuable 4.Uninvolving/involving 5.Unappealing/appealing

6.Means nothing/means a lot to me

(42)

41 Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): 1. Complement

I think the extension product and original product could be jointly used to satisfy the same need 2. Substitute

I see the extension product as a substitute for the original brand product 3. Transfer

I think the manufacturer of the original product is able to make the extension product

Image fit (only for experiment groups) Park et al. (1991)

Seven Likert items in response to statement (Strongly disagree (1) to Strongly agree (7)): I think the extension product and the original brand shares consistent concept

Attitude towards original/extension product D’Astous et., al (2007)

Hem & Iversen (2003) Park et al. (1991) Seven Likert items in response to original/extension product (Strongly disagree (1) to Strongly agree (7)):

1. My attitude is positive 2. I like it

3.I think it is of good quality

4. I would buy it if I was looking for a product in this category

(43)

42

4. Results

4.1 Sample

4.1.1 Response analysis

The experiment is conducted in the form of online survey that was created on the platform Qualtrics. Potential respondents were invited by email, social media and offline activity carried out in the city and school library. A total number of 371 individuals participated in the survey, respondents who click on the web link to the survey were randomly assigned to one of the eight scenarios. Among all the surveys started by respondents, 254 participants completed the survey, others either dropped out in the middle of the survey or submitted incomplete answers.

4.1.2 Descriptives of the sample

Age and gender

The final data set consists of 254 valid surveys, all respondents are between 16 and 50 with an average of 25.25 years old. Gender distribution condition among 8 scenarios was listed as following:

Table 4.1: Distribution of respondents

Scenarios Female male total

1. Low involvement, low functional fit, low image fit _{18 (60.00%)} _{12 (40.00%)} ₃₀ 2. Low involvement, high functional fit, high image fit _{16 (57.14%)} ₁₂_(42.86%) ₂₈ 3. Low involvement, low functional fit, high image fit 20 (60.61%) 13 (39.39%) 33 4. Low involvement, high functional fit, low image fit 15 (44.12%) 19 (55.88%) 34 5. High involvement, low functional fit, low image fit 18 (62.07%) 11 (37.93%) 29 6. High involvement, high functional fit, low image fit _{15 (45.45%)} _{19 (55.88%)} ₃₄

(44)

43 7. High involvement, low functional fit, high image fit 17 (54.84%) 14 (45.16%) 31 8. High involvement, high functional fit, high image fit _{22 (62.86%)} _{13 (37.14%)} ₃₅

As shown in the table, although the percentage of female respondents is slightly higher than that of male respondents, respondents were almost equally distributed into 6 scenarios during the survey.

In addition, pairwise Chi-Square Tests were performed among the 8 groups, all p > 0.10, so there was not gender difference between the eight groups at 0.1 level of significance.

Education

Only 12% out of 254 respondents received high school or less than high school education. Over half have received college degree of education, and the rest 34% have received higher degree of education than college. The pie chart below shows the distribution of education situation of the respondents. Also pairwise Chi-Square Tests were performed, all p > .0.1, which means that there was not education difference between the 8 groups at 0.1 level of significance.

(45)

44 Figure 4.1: Education situation

Nationality

Among the 254 respondents who completed the survey, Chinese together with Dutch represent over half of the total number, followed by Australian, American and German. Below is a pie chart illustrates the nationality distribution.

Figure 4.2: Nationality distribution

1% 11% 7% 47% 32% 1% 1%

Less than High School High School / GED 2-year College Degree 4-year College Degree Masters Degree Doctoral Degree Professional Degree (JD, MD) China, 46% 28% 8% 6% 3% 9% China Netherlands Australia America Germany Others

(46)

45 As shown in the pie graph, respondents come from Asia, Europe and North America, covered three continents and two economies (developing vs. developed), which was regarded as representative.

4.2 Pre-analysis result

4.2.1 Scale reliability

In order to test the internal consistency of each scale, reliability analyses were conducted for the items measuring respondents’ attitudes toward extension product under 8 different conditions. According to Table 4.2, the items measuring attitudes have high internal consistency.

Table 4.2: Scale Reliability (Cronbach’s Alpha Coefficients)

Attitude towards product

Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6 Scenario 7 Scenario 8 .924 .926 .940 .933 .935 .964 .940 .945

The five items that measure respondents’ attitudes toward different extension products were highly correlated, the author adopted an average rating of the five items to represent the dependent measure.

(47)

46

4.2.2 Manipulation checks

As mentioned in the pretest, brand familiarity and brand attitude are the two control variables. In the final survey, respondents were asked to rate on the familiarity and involvement with the original brand, the results are consistent with that of pretest. For familiarity, there was no significant difference across four different groups per brand Overall, Colgate and Apple got ratings of 5.49 and 5.53 respectively. Regarding the involvement with the brands, overall Colgate and Apple received mean ratings of 2.45 and 5.74 respectively.

4.3 Analysis results

Table 4.3 below briefly reports the results of the analysis.

Table 4.3: Treatment Means and Standard Deviations for Evaluation a

Scenarios Ratings

1. Low involvement, low functional fit, low image fit _{3.01 (0.33)} 2. Low involvement, high functional fit, low image fit 4.51 (0.18) 3. Low involvement, low functional fit, high image fit _{3.85 (0.16)} 4. Low involvement, high functional fit, high image fit _{5.30 (0.16)} 5. High involvement, low functional fit, low image fit 3.06 (0.20) 6. High involvement, high functional fit, low image fit 4.25 (0.21) 7. High involvement, low functional fit, high image fit 4.56 (0.19) 8. High involvement, high functional fit, high image fit _{5.36 (0.15)}

Note: a Numbers in the parentheses are the SD. All items were measured on a seven – point scale.

(48)

47 A preliminary comparison shows that functional fit and image fit have positive effect on respondents’ ratings on brand extension products respectively. For example, regarding extension products of both brands Colgate and Apple, the one that reflected either high functional fit or high image fit or both received higher ratings than the one that possesses low functional fit and low image fit.

In order to formally test the hypotheses, a correlation analysis was firstly performed, all the ratings used in the analysis have been standardized beforehand. The correlation matrix below is based on the 254 responses for the extension products.

Table 4.4 Correlation analysis results

Variables 1 2 3 4 5 6

1. Extension attitude 1

2. Initial brand familiarity .122 1

3. Initial brand attitude .137 .308* 1

4. Functional fit .546** .049 .046 1

5. Image fit .348** .078 .076 -.127 1

6. Involvement .000 .000 .000 .000 .000 1

*

. Correlation is significant at the 0.05 level (2-tailed). **

. Correlation is significant at the 0.01 level (2-tailed).

According to the table, extension attitude was correlated with functional fit and image fit at 0.01 level of significance, which made functional fit and image fit two influential factors for extension attitude. In addition, brand familiarity was correlated with initial brand attitude at 0.05 level of significance.

(49)

48 To better detect and understand multicollinearity between the independent variables, the author conducted stepwise regression. In each regression, one original independent variable was chosen and regarded as a new dependent variable that explained by the rest independent variables. Two indicators were used to check the multicollinearity. Firstly, all the VIFs (variance inflation factors) of the all independent variables in every regression were below 2, which is much lower than the cut-off recommended by Hair et al. (2006). Secondly, eigenvalues of all the independent variables in every regression were higher than 0. Taken both VIFs and eigenvalues into account, one conclusion to be drawn is that, multicollinearity would not threaten the results of the existing study.

In addition, to check whether the selection of two brands (Colgate vs. Apple) and seven extension product categories (Table 3.8) shifts the result of above regression, the author conducted a regression, in which dummy variables were adopted, the model was formulated as following:

Y = β0 + β1D1iD2j + β2D2j + β3 D1iD3j+ β4 D3j +β5 D1iD4j+ β6 D4j where the dummy variables

D1i = 1 if the brand i is Colgate

0 if the brand i is Apple

D2j = 1 if the product category j is printer paper

0 if the product category j is others

D3j = 1 if the product category j is electronic toothbrush or mobile phone

(50)

49 D4j = 1 if the product category j is tissue or fan

0 if the product category j is others

The regression results show that all the coefficients were not significant (all p > 0.1, R2= .187, Adjusted R2= .168). Thus, it can be concluded that, the selection of two

brands and eight extension product categories does not bias respondents’ attitudes toward brand extension product.

After the above analyses, the author estimated a regression model. In the model, the attitude towards brand extension product, the independent variables follow the four hypotheses and were listed in table 4.5. Initial brand attitude and initial brand familiarity are two control variables. Next are two fit variables from H1 and H2, functional fit and image fit. The following variable is the moderator involvement. The final two items reflect the interactions of two fit variables with the involvement, from H3 and H4. The author ran regressions separately, to include more variable subsequently.

Table 4.5 Regression model of attitudes toward brand extension products

Independent Variables 1 2 3 4

Initial Brand familiarity .085

** (1.079) .039* (.663) .039* (.663) .047* (.815) Initial brand attitude .109

** (1.377) .066** (1.130) .066** (1.130) .063* (1.096) Functional fit .594 *** (10.770) .594*** (10.770) .598*** (11.025) Image fit .415 *** (7.505) .415*** (7.505) .412*** (7.564)

Functional fit and image fit in consumer evaluation of brand extension and the moderating role of involvement