Is between brand- brand fit enough in co-branding? : assessing an interaction between the fit among brands and between the co-brand

(1)

Is between brand- brand fit enough in co-branding?

Assessing an interaction between the fit among brands and between the co-brand.

Master thesis

MSc in Business Administration- Marketing

Moska Lodin-10594175 University of Amsterdam

Master thesis

MSc in Business Administration- Marketing track Supervisor: Karin Venetis

(2)

2

Statement of Originality

This document is written by Moska Lodin who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Acknowledgements

Writing this thesis was one of the most challenging experiences I have faced academically. Finishing this project has taught me to be critical but mostly to keep confidence in my capabilities. Without the support of my family, friends, partner, fellow students this experience would have been more a strain. In special, I want to thank my supervisor, Karin Venetis, you are truly an exceptional mentor. The valuable feedback and critics has not only pushed be to be better, but also supported me tremendously during this process. I could not ask for a better guidance.

(4)

4

Abstract

Previous research has found significant cases that identifies the importance of brand fit between partners in co-brand evaluation. Also, the similarity between the parent brands product category with the co-brand has found tremendous support, not only in co-branding, but especially in brand extension literature. However, up till now no research has measured both dimensions of fit in one study. The aim of this study was to identify how these two dimensions of fit influence each other. It was expected that the brand fit between partner brands weakens the positive relationship of the product fit (between the base brand and the co-brand) and the co-brand evaluation. The accessibility-diagnosticity framework suggest that brand fit is more abstract and therefore more accessible. This study suggests, that when two brands collaborating a high brand fit, the product fit between the base brand and the newly launched co-brand will be less important and thereby weakened. This was examined by a 2x2 experimental between-subject design among 187 respondents. Like previous research in co-branding, positive relationships for the attitude towards the parent brand, product fit between the base brand and co-brand, brand fit between partners and the co-brand evaluation was found. The results showed an almost significant negative moderation effect. In contrast of what was expected at a high brand fit between partners, the effect of product fit between the base brand and the co-brand becomes stronger. Thus, not only is it important to find a partner that fits your brand concept, the fit between product categories stays significantly important.

(9)

9

1. Introduction

1.1.Preface

In 2011 Microsoft emerged into the mobile market in collaboration with Nokia, who has been present in the mobile industry for decades. The collaboration enabled Microsoft to enter a highly competitive market and provided an opportunity to leverage Microsofts brand equity in a new market. Dynamic markets, intense competition and higher costs to enter new markets force companies to adapt new branding strategies. To enable companies to enter new markets and exploit their brand equity, co-branding has emerged as a popular strategy. Co-branding is a strategy of two or more brands that are paired to form a separate and unique product (Park et al., 1996). Two brands collaborating together is not a new phenomenon, but has received increased interest in recent years in branding literature.

The strategy has also reached increased interest from companies, because it enables gaining more marketplace exposure by entering new markets through alliances. Moreover, co-branding enables brands to leverage on the complementary features of the partnering brands and transfer positive associations from one brand to another (Washuburn, Till & Priluck, 2000; Simonin & Ruth, 1998).

Recent marketplace examples include: Stratos developed by GoPro and Red Bull, and the Art of Travel Luggage line by Louis Vuitton and BMW. These are examples of co-branded products that have been a success. Their success can be explained by how these brands signal similar brand concepts. Red Bull and GoPro are lifestyle brands that both pursue extreme, fearless and an action driven lifestyle. Both BMW and Louis Vuitton value luxury and are well- known, traditional brands that are known for high-quality craftsmanship.

Imagine a luggage line in collaboration between Louis Vuitton and Toyota. Would this have been as successful as the collaboration with BMW? The associations of both brands and their match, play a significant role on how consumers evaluate co-brands.

(10)

10 1.2.Co-branding and fit

Research on co-branding supports the importance of fit (Simonin & Ruth, 1998; Lafferty, Goldsmith & Hult, 2004; Walchli, 2007). Findings suggest that for co-brands to be successful, the fit between the partners is crucial. When confronted to a new co-brand consumer are not only faced with the fit between the two partners, they are also cued with the fit between each brand and the new co-brand. Fit in a co-branding context mostly refers to customers’

perceptions of the compatibility or similarity of the two product categories of the partner brands and their brand concepts (Simonin & Ruth, 1998). Researchers in co-branding

literature find the degree of fit between two brands to be positively related to the evaluation of the co-brand (Simonin & Ruth, 1998; Hadjicharalambous, 2001; Lafferty et al., 2004). Hereby the focus lies only on how the partner match with each other. Explanation for this positive effect lies the categorization theory. A newly launched product is evaluated by the

information the consumer knows of the parent brand (Boush & Loken, 1991). When they signal the same category, the positive associations towards the parent brand can transfer to the new co-brand.

In co-branding consumers are confronted with the fit between the two brands that are collaborating and the fit between each brand and the new product. Also fit between each brand and the co-brand should have some influence. Significant amount of research has found that when introducing a new product there should be similarity with the new product category and the original product category (Aaker & Keller, 1990; Völckner & Sattler, 2004; Park et al., 1991). Thompson and Strutton (2012) did look at the fit between each partner and the new brand. But they disregard the possible effect of fit between the partners on how the co-brand is evaluated. The two dimensions of fit is not investigated in one study.

BMW and Louis Vuitton are two brands that are active in different product categories, but can be perceived as fitted. These two brands can be evaluated to be fit on brand level as

(11)

11 they are both premium and luxurious brands. They fit on brand level. Brand fit refers to the degree to which consumers are comfortable to the pairing of the two brands (Simonin & Ruth, 1998).

For BMW, launching a luggage line is a whole different product category then its’ original products. Product fit, refers to the extent to which consumers perceive the product categories in which the brands are active in, related to each other (Simonin & Ruth, 1998; Aaker & Keller, 1990). Research on product fit suggested, that launching a new product will be evaluated more positively when there is similarity in the product categories (Boush & Loken, 1991; Park, Jun & Shocker, 1996). For consumers to evaluate the new product they rely on the attitudes of the parent brand. Based on the categorization theory, the transfer of these positive associations is enables by fit between the two product categories. BMW launching a new product, such as luggage, indicates a low product fit. But is the effect of low product fit overcome by its’ collaboration with Louis Vuitton?

1.3. Problem statement & Research gap

Research on co-branding has extensive evidence for the importance of fit between partnering brands (Simonin & Ruth, 1998; Park et al., 1996). In co-branding, but more so in brand extension literature, the fit between each brand and a newly launched product is highlighted. Brand extension literature implies that an extension in a different product category, that is not perceived to be compatible to the parent brand, will be evaluated negatively (Spiggle, Nguyen & Caravella, 2012; Völckner & Sattler, 2006; Thompson & Strutton, 2012).

Does this indicate that a co-brand that is in a different product category and thereby is perceived to be low on product fit, a bad idea? Or can a brand still make a big jump by

collaborating with a different brand? What for role has the brand fit between partners now, on the co-brand evaluation? No research until now has examined the fit between a brand and the co-brand, and the fit between the partnering brands together.

(12)

12 The purpose of this study is to examine whether a brand can introduce a new product in a far product category, indicating a low product fit, and still have a positive co-brand evaluation, due the engagement with a favorable brand alliance that fits its brand concept. Which type of fit is more important? The product fit that arises from brand extension literature that has found tremendous support. Or is it for co-branding more important to engage with a brand that fits your brand concept? Finally, how do these two types of fit interact with each other? Can the level of brand fit between partners, weaken the positive relationship product fit between the brand and the co-brand have on co-brand evaluation? 1.4. Research question and sub-questions

This study will attempt to examine the following research questions:

‘’ How does the degree of brand fit between the partners influence the relationship of product

fit between the base brand and the co-brand and the co-brand evaluation? Which type of fit has a stronger effect on co-brand evaluation?

In order to answer the research questions, the literature review will answer the following sub-questions:

- What is co-branding?

- What are the drivers for co-brand success?

- What is the effect of product fit between the base brand and the brand on the co-brand evaluation?

- What is the effect of brand fit between partners on the co-brand evaluation? - Which of the two fits has a stronger effect on co-brand evaluation?

- How does the degree of brand fit between the partnering brands influence the relationship between the degree of product fit, between the base brand and the co-brand?

(13)

13 1.5. Contributions

By answering these questions this study makes primarily contributions to co-branding research: First, it examines what type of fit is more important in co-branding, the product fit between the brand and the co-brand or, the brand fit between the partners? Secondly, it introduces a three-way interaction model that examines, whether the brand fit between partners can weaken the relationship of product fit between the brand and the co-brand, and the co-brand evaluation.

These contributions are especially relevant for marketing practice. The findings will help managers in making decisions concerning co-branding. The findings will suggest in which circumstances partnering the right way will matter, when introducing a co-brand. The results will give an indication to which extent, market exposure could extend through co-branding while being compromised by the type of brand alliance.

1.6. Outline

The remainder of this study is as followed: The study will start with an overview of existing literature on co-branding, and how consumers evaluate co-brands. This chapter will be closed with the conceptual framework and hypotheses, based on the literature review. This is

followed by chapter two, which discusses the methodology and research design. Results and the testing of the hypotheses is discussed in the fourth chapter. These results are further elaborated in chapter six. This chapter is closed by the implications, contribution and limitations of this study. Finally, a summary of the study is given in the conclusion.

(14)

14

2. Literature review

2.1. Introduction to co-branding

Since the current century (21th), researchers started to have a growing interest in co-branding as a strategy to leverage brand equity (Simonin & Ruth, 1998). The philosophy behind co-branding is that it is a strategy that enables a bigger market share, leads to more profitability due to the new product introduction, and leads to a competitive advantage through the higher customer awareness (Chang, 2009). The aim is that a product has a greater value when introduced by two brands, then it would when introduced solely. It is a strategy that considers

how two brands can align their value chain to deliver a greater value and jointly capitalize on

the rewards (Blackett & Road, 1999).

Co-branding refers to a form of co-operation between two or more brands that are paired to form a separate unique product (Park et al., 1996). Co-branding differs from promotional alliances, because co-branding concerns a new product, which signals to consumers that the partners are committed to a long-term relationship (Leathesser, Kohli & Suri, 2003).

Co-branding has emerged as a new way of leveraging a brands’ equity. Each brand has an own and specific equity, that is defined as the given value that is endowed on a product by the brand (Herr, Farquhar, & Fazio, 1996). From a consumer perspective brand equity is referred as the differential effect of brand knowledge on consumer response to the marketing of the brand (Keller, 1993). In a co-branding context, this means that a firm uses the associations of their brand and those of the collaborating brand to leverage on. For example, take Philadelphia a cream cheese brand that decides to collaborate with Milka, a chocolate brand, to introduce a new chocolate spread together. Hereby the good chocolate flavour of Milka, and the spread expertise of Philadelphia are leveraged. Another popular way to stretch a brands equity is by introducing brand extensions. Hereby the brands equity is borrowed and extended

(15)

15 to another product. An established brand name is then used to introduce a new product (Völckner & Sattler, 2006).

2.2. Types of co-branding

In the current literature regarding branding, the terminology concerning the types of co-brands varies greatly. This study, only focus on co-brand types were both co-brands contribute to launch the co-brand. Hereby, the potential association transfer of both brands to the co-brand are included. Therefore, this section describes ingredient branding and complementary co-branding.

In ingredient co-branding, there is an asymmetrical combination of two established brands. One brand is the dominant brand, where the other brand provides an important component too (Uggla, 2004). A key attribute of a brand is integrated into a new product hosted by the constituted brand. A widely known example is the use of Intel processors into Dell computers. Hereby, Dell is the host brand, while Intel provides an important attribute. Intel is a supplier of processors for Dell computers. An important attribute of ingredient co-branding, is that is the ingredient brand does not introduce a new product, but adds a brand attribute as an ingredient into an existing product. The co-brand is a new product introduced by the host brand (Vaidyanathan & Aggerwal, 2000).

Compared to ingredient co-branding, complementary co-branding, does not infer an asymmetrical relationship. For a complementary co-brand, the skills and competences of two brands are combined to introduce a new product, which is new to both brands. Hereby there is a more symmetrical commitment (Uggla, 2004). For example, the Nike + is a newly

introduced co-brand that enables to track your running statistics through special shoes on an iPod. Nike introduced a Nike + shoe line and Apple developed a tracking chip that fitted the Nike’s running shoes. This enabled consumers to get the information of the chip into their iPod. Hereby two target groups, people who love running and people who love listening to

(16)

16 music, are combined. Complementary co-branding is considered an appropriate strategy were the competences and reputation are used in a new product (Uggla, 2004). This means, that the collaboration will likely create synergies between each brands’ unique associations and attributes (Leuthesser et al., 2003)

2.3. Advantages and disadvantages of co-branding

In co-branding, two brand names are involved. Hereby the potential exists to achieve synergies between the associations of both brands. There is an opportunity capitalize on the unique brand characteristics of each brand (Rao & Ruekert, 1994). They suggest when two brand names of high value are combined, these values are added up, signalling a potential higher benefit. Chang (2009) lists potential benefits that occur through co-branding. First, co-branding can lead to a bigger customer base by combining the customers of each brand. It also enables to meet potential needs customers have that aren’t met yet, through an extension of the product line. Furthermore, by combining two brands their market share grows which strengthen their competitive position. Finally, it yields operational benefits, by the possibility to share costs. According to Blackett and Road (1999) co-branding is also considered as risk reducing, it leads to greater consumer reassurance having to brands working on one product (Blackett & Road, 1999)

However, it can not be assumed that co-branding is all about high rewards, and low risks and costs. There are considerable risks involved towards the brands in co-branding. One of the risks is the loss of some controle because two parties are involved. As a brand you can not control incidents happening to the partnering brand. Chosing the wrong partner can be detremental. Simonin and Ruth (1998) explained that in co-branding new evaluations and associations of both brands influence how a co-brand is evaluated. When a partner brand faces bad publicity these negative associations can transfer to the collaborating brand (Boad, 1999). Geylani, Inman & Hofstede (2008) suggests that negative spillover effects occur to the brand

(17)

17 by a partnering brand. When Intel was experiencing quality problems with the Pentium microprocessors, Dell as a partner was concered that these negative quality asssociations would spillover to their brand.

2.4. Co-branding and brand extensions

This section compares two leveraging strategies: co-branding and brand extensions. The main objective of co-branding is to launch a new product; therefore, it is also referred to as a special type of brand extension (Park et al., 1996). Brand extension is a relatively old strategy compared to co-branding in branding literature. Therefor a lot of mechanism in co-branding is based on theories that have emerged from brand extension literature. Because the two have similarities on how they work and are effected they are discussed in this section. This ensures why co-branding can be explained by brand extension literature.

Co-branding and brand extensions are both leveraging strategies where, at different ways the firms brand equity is leveraged. Where for co-branding the brand equity of two brands are leveraged. Using the associations of both brands to transfer to a new product (Rao & Ruekert, 1994). For brand extension, only the equity of one brand is leveraged. Hereby, an established brand name is used, to introduce a new product in the same category, referred to as line extension, or in a different category, referred to as brand extension (Völckner & Sattler, 2006). Similar to co-branding, brand extensions take the advantage of the brand recognition and brand image of the parent brand in the introduction of the new product (Aaker & Keller, 1990). How the quality associations of the parent brands are transferred to the new product, are similar to both strategies. The role of pre-existing attitude and fit plays a

significant role. Only for co-branding the attitude of two brands need to be taken into account. Also, the fit between the two brand emerges as a significant factor.

(18)

18 2.5. Brands as associative networks

How a co-brand is evaluated finds its’ foundation in brand associations.Therefore this section describes brands as associative networks. Kotler (1991, pp.244) defined brand as ‘‘… a name, term, sign, symbol, design or combination of them which is intended to identify the goods and services of one seller or group of seller to differentiate them from those of competitors’’. The equity of a brand is the outcome resulted from the marketing of a product or service marketing, due to mentioning the brand name (Keller, 1993). But, Keller (1993) argues that the actual value of a brand resided from the way customers view brands. He introduces the customer based brand equity model (CBBE), which refers to the knowledge consumers have of a brand that has a differential effect on how consumers respond to the marketing of the brand. The CBBE model suggest that the power of a brand lies in what consumers have learnt, felt, seen and heard of the brand over time.

How the knowledge of a brand is conceptualized in the memory of consumers is represented in the associative network model (Keller, 1993). The associative network model views how a memory consists of nodes, which are stored parts of information which can vary in strength, that are linked to each other. Links between nodes in the memory of the consumer are named brand associations (Keller, 1993). For example, a node can represent a brand (BMW), a product (cars), or an attribute (speed). A link between any of two nodes suggests an association in consumers’ mind (Krishnan, 1996). According to Aaker (1991), a brand has an underlying value which consists of a set associations. An association is the meaning of the brand firmly established in consumers mind (Aaker, 2001). For example, Apple is associated with user friendliness, Chanel with luxury and Timberland with durability.

Associations can come in any form and can suggest characteristics directly related to the brand or product itself or indirectly related (Cheng-Hsui Chen, 2001). These can differ from specific attributes but, they can also be abstract such as type of users, usage situation and the

(19)

19 history and heritage of the brand (Aaker & Jacobson, 2001). The evaluation of co-brands is influenced by perceptions of the initial brands (Vöckler & Satller, 2006; Simonin & Ruth, 1998; Aaker & Keller, 1990). So, the meaning consumers derive from brands, through brand associations, is a core component on how co-brands are evaluated.

2.6.Success determinants

In branding literature two factors are considered to be the most important drivers of co-brand success (Ruth & Simonin, 1998; Thompson & Strutton, 2012). The pre-existing attitude toward each parent brand and fit, are considered valuable in co-branding. Fit in co-branding can be the fit between each parent brand and the co-brand and the fit between the partnering brands. These factors are discussed in the following sections.

2.6.1. Brand attitude

In both co-branding and brand extension literature one of the determinants of product success is the attitude consumers have of the initial brand(s) (Simonin & Ruth, 1998; Völckner & Sattler, 2004; Thompson & Strutton, 2012). Extensions and co-brands appear to be more favourable when strong positive beliefs and attitudes exists towards the parent brand (Simonin & Ruth, 1998; Bath & Reddy, 2002; Vöckler & Sattler, 2006). The research of Simonin and Ruth (1998) measured the effect of the attitude towards both parent brands on the co-brand evaluation. Both attitudes had a direct positive relationship with how the newly launched product is evaluated. When there is a positive attitude towards the parent brands, this attitude is transferred to the co-brand. The same effect counts for a negative attitude. This effect was confirmed a decade later by Thompson and Strutton (2012).

Different from brand extension, is that the attitude of two brands are considered to have an effect on the co-brands’ evaluation. According to the cognitive consistency, theory consumers will try to maintain consistency internally among their attitudes. Thus, when two

(20)

20 brands that are possibly conflicting are introducing a co-brand, consumers will try to adjust and average out their attitude towards the parent brands to shape their attitude (Pimentel & Reynolds, 2004). Based on the information integration theory consumers combine their beliefs or attitudes of the brands, as people receive, interpret, evaluate, and then integrate brand information with existing beliefs or attitudes (Simonin & Ruth, 1998). This theory suggests that the attitude in a co-branding context is modified by the information about the alliances that is added.

The effect of the attitude towards each parent brand on co-brand evaluation finds most support in the categorization theory which is widely used to explain the transfer of attitude in a brand extension context. The categorization theory proposes that a parent brand constitutes a category, where the co-brand due to the names inevitably belongs to (Cohen & Basu, 1987, Boush & Loken, 1991). When consumers are exposed to the co-brand, the category of the parent brands are activated. After activation, the category provides cognitions and affective judgements that are already associated with the parent brands. These existing judgements for the category form a baseline for the co-brand (Cohen & Basu, 1987, Thompson & Strutton, 2012). This enables the transfer of attitudes and formations of judgements to the co-brand (Thompson & Stutton, 2012). When evaluating a co-brand just after launch, consumers have little knowledge of the extension. Consumers retrieve judgements and beliefs of the co-brand by categorizing the brand to help them understand.

Keller (1990) states that the categorization suggests two ways in how consumers evaluate a brand. First a brand can be evaluated by piecemeal processing, hereby the co-brand is evaluated by the inferred co-brand attribute beliefs and the evaluative importance. Secondly, a co-brand is evaluated by the overall attitude towards the original brand. This is called category-based processing. Because a co-brand in unfamiliar to consumers, they are likely to base their evaluation of the knowledge they have of the original brand. The

(21)

21 categorization theory confirms how associations of the parent brands are transferred to the extension evaluation.

The pre-existing attitude towards the parent brand has an influence on how a co-brand is evaluated. In the context of co-branding, not only has the attitude of the base brand an influence on how the co-brand is evaluated, but also the attitude of the partner brand has influence on the co-brand evaluation. This mechanism is explained by the categorization theory. Therefore, the following hypothesis is developed:

H1) Prior attitude towards the base brand and partner brand is positively related to the evaluation of the co-brand.

2.6.2. Perceived fit

In brand extension fit, `` the perceived similarity ‘' between the parent brand and extension is perceived one of the success drivers of extension success (Aaker & Keller, 1990; Park et al., 1991; Broniarczyk & Alba, 1994; Völckner & Sattler, 2004). In a co-branding context, the role of fit is more complex. In co-branding consumers are faced evaluating the fit between the parents and the fit between each parent brand and the co-brand (Simonin & Ruth, 1998; Park et al., 1996).

The general definition of fit is the degree of similarity between an extension (or co-branded) product and the parent brands’ existing product category (Aaker & Keller, 1990). Similarity between the product categories can be the degree in which they satisfy the same needs, the situation in which they are used, and the skills needed to manufacture them (Aaker & Keller, 1990; Tauber, 1998; Park et al., 1991). Regardless of how fit is conceptualized, research suggest, that when fit increases, consumers are more able to transfer the positive associations of the parent brands to the co-brand (Aaker & Keller, 1990). Subsequent research conceptualized this broad concept of fit more detailed, proposing two main types of fit. These are discussed is the following sections.

(22)

22 2.6.2.1.Product fit

The first type of fit is referred to product fit. Product fit refers to similarity of product categories between each parent brand and the co-brand (Simonin & Ruth, 1998). This type of fit is extensively discussed in the brand extension literature. Therefore, it will form the base for the application in a co-branding context. Similar in co-branding, each brand is faced introducing a product in a product category which is similar or dissimilar to their original product category. In brand extension literature, the ability of established brand to introduce successful extensions depends on how consumers can generalize the brand beliefs to a new product category (Park et al., 1996; Aaker & Keller, 1990; Boush & Loken, 1991). The ability for consumers to transfer brand beliefs and attitudes, is affected by the degree to which the extension (or co-brand) is perceived to be fitted to the brands’ existing product category (Aaker & Keller, 1990; Park, Milberg & Lawson, 1991). When there is fit between the brand and the category the parent brand is able to reduce uncertain associations of the new product category (Smith & Andrews, 1995). Launching products to a far category involves losing brand differentiation and credibility (Aaker, 2004). Also, a far extension can harm the original brand. This is called feedback effect, which refers to the potential brand dilution that arises from an unfavourable evaluation of brand extension due to low fit (Milberg et al., 1997).

Fit is described as the similarity of two product classes, but disregards the dimension that fit can have. In the research of Aaker and Keller (1992) a firm producing potato chips extending their product category into cheese crackers is considered to have a high product fit. Because, similar to potato chips, cheese crackers have associations such as salty, dry and is a snack. As a low product fit, they introduced an ice-cream, which can be considered as far due to the sweet and creamy associations it has. A similar example is given in co-branding. Thompson and Strutton (2012) introduced a co-brand, a sports drink, by different parent brands. In this case each parent brand was established in one of three product categories. Nike, which

(23)

23 is a sport brand, is assumed to have a high product fit with the sport drink category, due to the similar sport associations. A moderate fit between a sport drink was the water brand Evian. Evian can have similar associations with sport drinks, because it can be consumed in the same usage situation, namely during exercising. Finally, a mouth water brand, Listerine, had a low product fit. Mouth water has no comparable sport association, it can only be perceived similar, because they both are liquid.

Early extension research has supported the importance of fit to extension success. Perceived fit is referred as the ability of consumers to perceive the new item to be consistent with the parent brand (Tauber, 1988). Other researchers have examined how the relatedness of two product categories mediates the brand extension evaluation and purchase intentions (Chakravarti, Maclnnis & Nakamoto, 1990; Farquhar, Herr & Fazio, 1989). Overall, extension research has found a positive relationship between product category similarity and brand extension evaluations and purchase intentions.

A reason for the success is that, fit enables the transfer of perceived quality of the parent brand to the extension. This can also be explained by the categorization theory (Cohen & Basu, 1987; Fiske, 1982). When the consumer perceives a fit between the original product class and the extension product class, consumers will activate category based processing. This enables consumers to transfer quality perceptions to the new brand extension (Aaker & Keller, 1990). An extension product category similar to the original product categories enables the activation of the category. Hereby associations of the original category are more likely to be transferred to the extension. Also, when extending to a close product category, consumers perceive the brand to be more capable to produce the new product. Researchers confirm the positive relationship between product similarity and brand extension evaluation (Keller, 1993; Farquhar et al., 1990).

(24)

24 the co-brand extension. They found there to be a positive relationship, also in a co-branding context. Thus, research has confirmed that the effect of product fit in brand extension also can be applied in co-branding. Therefore, the following hypothesis is developed:

H2) The perception of product fit between the base brand and the co-brand is positively related to the co-brand evaluation.

2.6.2.2.Brand fit

In a co-branding condition, consumers are confronted with three possible product categories and their level of fit. First a fit between each brand and the co-brand arises, and the fit between the two brands is cued. Therefore, in a co-branding context, not only fit between each partner brand and the co-brand should be taken into account, but also the fit between the brands that are collaborating to launch the co-brand. Simonin and Ruth (1998) describe two dimensions of fit between two brands whom are collaborating. First, they describe fit as the ‘’extent to which consumers perceive the two product categories to be compatible’’ as product fit (p.33). In a co-branding context, product fit refers to the relatedness of the product categories implied by collaborating brands (Simonin & Ruth, 1998). Besides brand extension research confirming the significant role of product fit, also brand alliance research confirms, the importance of similar product categories between collaborating brands (Harlam et al., 1995).

However, the similarity among product categories denies a brand specific fit. A partnership does not only trigger product association, but also more abstract association such as luxury. Simonin & Ruth (1998) establish a higher level of fit between partners, namely the fit between brand images or brand concepts. The second type of fit is referred to as brand fit. It refers to the similarity between the brand concepts of the partnering brand (Simonin & Ruth, 1998). This type of fit is applied in brand extension as in co-branding research. Various definitions are given, such as brand concept consistency, but in this study, it will be referred to as brand fit and solely be the fit between the brand images of the partnering brands in a

(25)

co-25 branding context.

Brand concept is the position that products have in the mind of consumers and differentiate products from several brands in the same product category (Park, Jaworski, and Maclnnis 1986). Products can be differentiated from other brands in the same product category on concrete and abstract level. Park et al., (1991) compared Rolex and Timex, which are both watch brands. The two brands, are similar in product category, both differ in terms of brand concept. Rolex is more likely to be associated with concepts of luxury and high status. Whereas the Timex brand is more likely to be associated with functional attributes, such as, durability and reliability. Their research supported the importance of brand fit. Indicating that an extension that is fitted to the brand concept will be evaluated better compared to a lower fit in brand concept. Fit in a brand alliance context, underpins the logic between the partnership (Pruppers, Dawar & Oversloot, 2006).

As in a brand extension research, fit can facilitate the transfer of associations between the partnering brands, which also enables the justification of the collaboration. Partner brands collaborating together to launch a co-brand should be consistent in their brand concepts, to be able to justify the collaboration. Simonin and Ruth (1998) emphasize the importance of fit on a brand image level, beyond just product fit in co-branding. They measure the fit between each brand on a brand level and product level. Their results confirmed the importance of brand concepts being cohesive with each other when introducing a co-brand. Based on this and previous research confirming the importance of brand fit in a brand alliance context, the following hypothesis states:

H3) The perception of brand fit between the partnering brands is positively related to the co-brand evaluation.

(26)

26 2.7. Interaction between product and brand fit

Both extension and co-branding literature confirmed the importance of brand fit and product fit. However, no published research measured the fit between the co-brand and each brand with the fit between the partnering brands. Therefore, there is no knowledge on how these two variables interact with each other. Can for example, partnering with a brand that fits your brand on brand level, overcome any negative effects of product fit between your brand and the co-brand? This section describes how the access of the types of fit in the memory of the consumer may affect consumers’ evaluation of a co-brand.

Cognitive economy suggests that consumers exert as little cognitive effort to make judgements of the cues they receive (Wyer & Srull, 1986). When introducing a co-brand several cues are directed to the consumers, among which, their feelings of each brand, the level of fit between each parent brand and the co-brand and how the images of the two parent brands fit. To make a judgment according to the accessibility-diagnosticity framework, consumers will select cues that are easily retrieved from their memory. Pruppers et al., (2006), applies the accessibility-diagnosticity framework to co-branding. They suggest that there are two types of fit between the co-brand and between partners. Namely, product fit, which refers to the associations between the current product categories and the new product category (Aaker & Keller, 1990). This can be between the original product category of the parent brand and the co-brand. Brand fit, relates to the similarity between the brand concepts of a parent brand and the co-brand and between parent brands concepts. They suggest that the fit that is at brand concept level, is more abstract than fit at a product level.

Based on the accessibility-diagnosticity framework, the more abstract the more accessible and diagnostic, brand fit is perceived. Brand concepts, such as luxury, are more general cues and give a more richer information to make a judgement. In this study, the accessibility-diagnosticity framework can be applied, explaining the moderating role of parent

(27)

27 brand fit. This study suggests, that because the many cues consumers face when making a judgement of a co-brand, the more abstract cues are important. This study expects that the brand fit is so important, that the effect of product fit between the base brand and the co-brand on the co-brand evaluation, becomes weaker. Hereby, the existing literature regarding the effect of product fit on a (cobranded) extension gets refuted. Signalling that brands can make a far jump to a new product category if they partner with a brand that fits the base brands concept. Based on the accessibility-diagnosticity framework, this study suggests the following hypothesis. H4) The perception of brand fit between the partnering brands will negatively moderate the relationship between the perception of product fit between the base brand and the co-brand, and the co-brands’ evaluation.

2.8. Conceptual model

The first step measures the effect of the attitude towards the parent brands on the co-brands’ evaluation. Afterwards, the effect of product fit between the base brand and co-brand, on the co-brand evaluation is assessed. Also, the effect of brand fit between partners on the co-brand evaluation is tested. Finally, an interaction effect is set up, between the two fits (figure 2.1).

(28)

28 Figure 2.1:

H1) Prior attitude towards the base brand and partner brand is positively related to the evaluation of the co-brand.

H2) The perception of product fit between the base brand and the co-brand is positively related to the co-brand evaluation.

H3) The perception of brand fit between the partnering brands is positively related to the co-brand evaluation.

H4) The perception of brand fit between the partnering brands will negatively moderate the relationship between the perception of product fit between the base brand and the co-brand, and the co-brands’ evaluation.

(29)

29

3. Methodology

This chapter describes how the experiment is conducted, which measurements are used for all variables and it discusses the pre-test that is conducted. Finally, it describes how the data is obtained and how the data will be analysed.

3.1. Research design

The objective of this paper is to investigate which type of fit is more important in a co-branding context, the fit between the brand and the extended product or the fit between the partner on an image based level? Another objective is, how does the fit between the partnering brands influence the relationship of the product fit between the base brand and brand and the co-brand evaluation? Is it possible to introduce a co-co-brand that is far from your initial product category, but is still evaluated positively by partnering the right way? In order to assess the role of partner brand fit on product fit, and to compare its level and impact on the established relationship for product fit (between the base brand and the co-brand) and the co-brand evaluation, an experimental design is set up.

This method is chosen, because it allows researchers to show stimuli, which enables the investigation of changes in the variables of interest (Blumberg, Cooper & Schindler, 2008). The independent variables are manipulated and then observed whether the dependent variable in the study is affected by the changes in the independent variable. For this study, an online-experiment is used. This makes it possible to reach a high number of participants in a limited amount of time.

Different scenarios were developed where a fixed base brand (Rolex), forms a co-branding alliance with four other brands (brand A, brand B, brand C and brand D). These combinations represented different types and levels of fit in co-branding. The manipulations were based on the level of product fit between the base brand and the co-brand and the level of

(30)

30 parent brand fit. The experiment will employ a 2 x 2 (Level of product fit between the base brand and the co-brand [ low product fit, high product fit] x Level of Partner brand fit [ low partner brand fit, high partner brand fit]) full factorial, between subject design, where the level of product fit between the base brand and co-brand, and the level of partner brand fit are the independent variables. The combination of the level of product fit between the base brand and co-brand (PF-BB-CB) and the level of partner brand fit (PBF) accounted for the 2 x 2 between subject design, is described in table 3.1 below. This experiment tested the effects of the independent variables on the dependent variable, consumer evaluation of the co-brand. Variables that were expected to have an effect on these relationships were taken into account and were assessed in the pre-test: brand familiarity (partner brands). The design stimuli are developed with the help of a pre-test, which is further discussed in this chapter.

Table 3.1: Levels of the independent variables that display the four experimental conditions Experimental scenarios

The level of product fit between base brand and co-brand

The level of brand fit between partners

High PBF Low PBF

High PF-BB-CB Condition 1 (45) Condition 2 (41)

Low PF-BB-CB Condition 3 (44) Condition 4 (57)

3.2. Design stimuli development

The design stimuli for the main study were developed by a pre-test. The set-up and the results are described in this section. The respondents selected for the pre-test, were recruited from a close social network and were excluded from the main study.

(31)

31 3.2.1 Pre-test

The pre-test was conducted to select the brands for the main study. This study aims to find evidence for how the brand fit between partners can influence the relationship of the product fit between the base brand and the co-brand. This pre-test was setup to help the author decide which brands in a co-branding context represented high and low brand fit. Due to time limitations, the product fit between one parent brand and the co-brand is measured. The product fit between the other partner brand and the co-brand, was constant under all conditions, with a high level of product fit.

The pre-test also acted as a tool, to select brands that are similar in their degree in which consumers viewed them as familiar and positive. Previous research in co-branding and brand extension has found a positive effect between the degree of familiarity and attitude on product evaluation (Simonin & Ruth, 1998; Thompson & Strutton, 2012, Völckner, & Sattler, 2006).

Based on previous research on brand extensions, the author generated a list of potential brand extensions which were proven to be considered high or low in their level of product fit (Park et al., 1991).

A base brand is chosen to have 10 potential collaborations, which represented the 4 conditions. The base brand should have strong unique associations to make sure that the brand can be easily differentiated from competitors. Strong unique associations are important to manipulate fit on a brand image level. Because luxury is a specific brand association that is easily observed with little knowledge, a luxurious base brand is chosen. This luxury brand should be able to extend its’ products in a category that implies a low product fit with the co-brand and a category that implies a high product fit. The co-brand that represents all of this is found in the previous research of Park et al., (1991).

(32)

32 to have a high product fit and perfumes are considered to have a low product fit. In this study Rolex acts as the base brand. Similar to their study, Rolex considers to launch two products: a bracelet and a perfume. However, in this study these co-brands are launched in collaboration with other brands that are already operating in these product categories. Hereby, to the author’s knowledge, the fit between a base brand and co-brand, and between partners, are for the first time examined together.

Overall, a list of 12 potential collaborations are presented, wherein 6 collaborations consider to launch a bracelet in collaboration and 6 collaborations consider to launch a perfume. The brands that were selected varied in their premium status within their product category. This enabled to find collaborations to be fit or misfit on a brand image level.

3.2.2. Pre-test sample and measures

A questionnaire was developed in which 15 respondents (Male=7, Female=8, Xage =23) were confronted with 10 potential collaborations with the Rolex brand (attachment 1). First a short description of the Rolex brand was given, that highlighted the premium position of Rolex in the watch industry. Followed by five items that assessed the perceived extension fit (product and brand fit) for Rolex launching a bracelet and a perfume (Aaker & Keller, 1990; Pryor & Brodie, 1998). After the extension fit questions the potential collaborations were mentioned. For each partner brand, a short description was given and the degree of familiarity and the attitude were tested by three items, which were assessed by a seven point Likert-scale developed by Simonin and Ruth (1998). For each collaboration, the brand fit was tested by three items which were assessed from a seven point Likert-scale developed by Aaker and Keller (1990).

(33)

33 3.2.3. Pre-test data analysis

Table 3.2 below provides an overview of the pre-test results. The brand extension product fit results confirmed that bracelets are considered to be high on product similarity (M=5.864) and a perfume is considered to be low on product similarity (M= 2.636).

Table 3.2: Outcome of Pre-test: Mean scores and standard deviations in parameters (N=15)

Pandora Cartier Tiffany Lucardi Guess Blvgari

Attitude 5.33 (1.23) 5.18 (1.15) 5.64 (1.10) 3.88 (1.46) 4.67 (1.91) 5.58 (1.15) Familiarity 6.09 (1.14) 5.45 (1.37) 5.36 (1.57) 6.56 (0.52) 5.73 (1.79) 4.91 (1.97) Brand image fit 3.24 (1.42) 5.36 (1.55) 5.18 (0.78) 1.61 (0.79) 2.82 (1.35) 5.63 (0.86)

Chanel Ritual Dior Bruno

Banani Victor & Rolf Etos Attitude 6.58 (0.62) 6.52 (0.58) 6.27 (0.73) 3.76 (1.51) 6.06 (0.61) 5.21 (1.38) Familiarity 6.45 (0.69) 6.64 (0.51) 6.55 (0.52) 5.27 (1.79) 6.27 (0.79) 6.09 (1.51) Brand image fit 4.45 (1.63) 1.54 (0.65) 5.00 (1.51) 2.24 (1.58) 3.91 (1.98) 1.12 (0.31)

Out of all the potential brand alliances that were presented to the respondents, Cartier and Blvgari were considered the most fitted on brand image when there is a collaboration in a high product extension situation. Eventually Cartier is chosen to represent condition 1: high PF-BB-CB and high PBF, because the respondents were more familiar with Cartier compared to Blvgari. For condition 2, which entails a high PF-BB-CB and a low PBF, Guess and Lucardi scored the lowest on brand fit. Based on the more positive attitude associated with Guess, Guess was chosen to represent the second condition. Dior was chosen to represent condition 3. For the brand alliances that represent a low PF-BB-CB, Dior scored the highest on a brand fit and scored high on familiarity and attitude. Rituals and Etos scored the lowest on brand fit, but Rituals was considered more positive. Therefore, Rituals represented the final condition, which

(34)

34 represent a low PF-BB-CB and a low PBF.

Based on these results, there is controlled for brand attitude and brand familiarity. Brands are chosen that are perceived to be positive and familiar.

3.2.4. Pre-test reliability

All data was collected through Qualtrics, analysed and coded in SPSS 24. To check the validity of the measures a reliability test is conducted. The reliability test, tested the following continuous measures: brand attitude, PF-BB-CB and PBF. The results validated the internal consistency of the items used to measure the variables, Cronbach’s Alpha > 0,700. Brand attitude was measured by a set of three items (positive/negative, favourable/unfavourable, good/bad) on a 7-point Likert scale, with a Cronbach’s Alpha= 0,986. PBF was measured by a set of three items (fit together/does not fit together, makes sense/does not make sense, consistent/inconsistent) on a 7-point Likert scale, with a Cronbach’s Alpha= 0,986. PF-BB-CB was measured by a set of two items (similar/not similar, resources are helpful/resources are not helpful) on a 7-point Likert scale, with a Cronbach’s Alpha =0,848.

3.3. Sample

To be able to draw conclusion a sample size of at least 160 participants were needed. This research was conducted among 230 participants, of which 187 were completely filled out. The data of these 187 respondents is used (XAge= 25,76; 32% Male). They conducted the online-experiment, which was hosted by Qualtrics. They were approached through multiple social media channels and asked to fill in the questionnaire, this method is referred as the snowball method (Biernacki & Waldorf, 1981). The participants were also asked to share the survey link, to expand the reach of the type of participant beyond the existing social network. All observations were valid and are used in the final analysis.

(35)

35 People whom participated in the pre-test could get aware of the research goal. This could threaten the internal validity of this research (Schreuder & Peter, 2011).

3.4. Procedure

The participants, participated by using an online link hosted by Qualtrics. First, they were exposed to an introduction, which shortly described the research goal. The participants were then asked to read the questions and possible answers carefully, and to choose the answer that is closest to their opinion. To test the four conditions, four questionnaires were developed. After the introduction, the participants were randomly assigned to one of the four questionnaires. Four different co-brands were created and operated as stimuli. The participants were exposed to either a high PF-BB-CB & high PBF, high PF-BB-CB & low PBF, low PF-BB-CB & high PBF, or low PF-BB-CB & low PBF condition.

All conditions started with a short description of the base brand (attachment 2), followed by questions measuring familiarity and attitude. Subsequently, the potential launch of a brand extension was described. This was either a bracelet, a high PF-BB-CB, or a perfume, which represented a low PF-BB-CB. Participants were asked how the launch of the brand extension was appropriate/similar to Rolex. Afterwards, a potential collaboration was shown with either a brand with comparable brand associations as Rolex, or brand that did not trigger these luxurious associations. Finally, images are shown that could indicate how the future co-brand should look like. For the low PF-BB-CB condition, three pictures instead of two are shown, because there was a feeling that the picture of a luxurious bracelet did not trigger enough Rolex associations like it did for the perfume bottle. A luxury bracelet, a bracelet of the partnering brand and the band of the Rolex watch is shared to the associations of Rolex in the stimuli. After being exposed to this, participants were asked how they would evaluate the co-brand and how fit the parent brands are on brand level. Figure 1 gives an overview of the four manipulated co-brands.

(36)

36 Figure 3.1: Final design of stimuli

3.5. Measurements

3.5.1. Independent variables

Each questionnaire describes one out of the four possible co-brands, which differed in their level of PF-BB-CB and PBF. Thus, each questionnaire manipulated the levels of the two independent variables, PF-BB-CB (high/low) and PBF (high/low).

Product fit between base brand and co-brand is assessed by two items measured on a 7-point Likert scale developed by Aaker & Keller (1990) assessed in Salinas and Perez (2009). Participants were asked in which extent they perceived the extension (a bracelet of perfume) to

High PBF Low PBF L ow PF -BB -CB Hig h P F -BB -CB

(37)

37 be similar to other Rolex products and whether it is perceived that Rolex has the resources that are helpful for the launch of the extension.

Partner brand fit was measured by a set of three items. Participants were asked in which extend they agreed with the following statements: ‘’Rolex and brand X brand image fit well together’’, ‘’a brand alliance between Rolex and brand X makes sense’’, and ‘’cooperation between Rolex and brand X is consistent with both brand images’’, on a 7-point Likert scale developed by Aaker and Keller (1990).

To confirm findings of previous research also the attitude of each brand is assessed. Brand attitude was measured by a set of three items (positive/negative, favourable/unfavourable, good/bad) on a 7-point Likert scale, developed by Simonin and Ruth (1998).

3.5.2. Dependent variable

To test the main objective of this study, the variable, co-brand evaluation is measured by a set of four items developed by Park et al., (1996). Participants were asked to indicate the extend they agreed with the statements: ‘’the quality is high of the …’’, ‘’I feel favourable about the…’’,’’ I like the idea of the ...’’, and ‘’I would buy the… if needed one’’. The answers were assessed by a 7-point Likert scale. The distribution of co-brand evaluation was normally distributed having a skewness -1.0 and +1.0 and kurtosis between - .5 and + .5.

3.5.3. Control variables

Prior research regarding brand extension and co-branding, confirm the positive effect of parent attitude on the evaluation of a (co-branded) extension (Simonin & Ruth, 1998; Aaker & Keller, 1990; Vöcklner & Sattler, 2006; Thompson & Strutton, 2012). The study of Simonin and Ruth (1998) found that brand familiarity moderated the effect of brand attitude on the evaluation of the co-branded extension. To test the main objective of this study, to solely asses the effects of

(38)

38 fits on the co-brand evaluation, the familiarity of the brands is controlled for in the pre-test. Based on the results of the pre-test all brands with high familiarity were chosen

Brand familiarity was measured by a set of three items (recognize/do not recognize, heard from/did not hear from, familiar/unfamiliar) on a 7-point Likert scale, developed by Simonin & Ruth (1998).

3.6. Data analysis

As initial step, the raw data is exported to SPSS version 24. The data set is cleaned up, all the useless information, such as IP-address were deleted. The questionnaires that were not fully filled in were deleted and all variables are checked for missing data or outliers. Due to the force to response option, which was active for all question in Qualtrics, no missing data was detected. The measures: PF-BB-CB, PBF, co-brand evaluation, brand attitude and brand familiarity consisted out of multiple indicators. Before computing an average for each measure, for all items a reliability analysis was run indicating all measures being internally consistent (Cronbach’s Alpha >0.700, see Table 4.5). To use the data for the analysis, an average of each variables is computed to form a new variable. To be able to compare the means of the different conditions, each condition was coded with 1, 2, 3, or 4.

The data is analysed by computing descriptives, such as means, a correlation matrix, to test for multicollinearity and to find relationships among variables. The different conditions are checked by a One-way ANOVA. Finally, the hypotheses are tested using a hierarchical regression analysis and a PROCESS analysis.

(39)

39

4. Results

In this section, a detailed description of the results is provided. The data is analysed, using the IBM SPSS Statistics 24 software.

4.1.Sample

The questionnaire was distributed through social media networks. In total, there were 230 responses, however only 187 questionnaires were fully filled in. These 187 responses left are used for the analysis. Since there are four different conditions, four questionnaires are distributed. The participants that clicked on the link were randomly assigned to one of the four questionnaires (conditions). The distribution of the respondents over the four conditions is as following: condition 1 (High PF-BB-CB & High PBF) N=45, condition 2 (High PF-BB-CB & Low PBF) N=41, condition 3 (Low BB-CB & High PBF) N= 44, and condition 4 (Low PF-BB-CB & Low PBF) N=57.

4.1.1. Descriptive of sample & Group difference

The sample of 187 respondents, consisted out of 126 females (67%) and 61 males (33%), with an average age of 25.76 years old. The base brand used in this study can be regarded familiar and moderately positive (M=6.07; SD=0.84; M=4.78; SD=1.27).

It is important to make sure that the respondents in the four conditions are homogeneous in their knowledge and perception of the core brand Rolex, in their age and gender. Inconsistent profiles would have precluded the comparability of the results across the different conditions. Looking at table 4.1, it already can be deducted that these four factors are quite similar for all 4 conditions. A One-way ANOVA is conducted to statistical confirm that these groups are homogenous among all conditions. The analysis confirmed that the four factors did not differ significantly from each other in the different conditions. The familiarity and attitude towards the Rolex brand did not differ among groups (p=0.605; p=0.121). Also, all groups were similar in their distribution of age (p=0.320) and gender (p=0.537).

(40)

40 Table 4.1: Group difference

Age Gender (female) Rolex familiarity Rolex attitude Condition 1 26.578 62.2% 6.000 5.000 Condition 2 24.854 73.2% 6.089 4.512 Condition 3 26.955 72.7% 6.212 5.038 Condition 4 24.842 63.2% 6.012 4.626 4.2.Pre-analysis results 4.2.1. Scale reliability

Before testing the hypothesises, the validation of scales is assessed. All scales were considered reliable, all Cronbach’s Alpha’s were above the 0.700 directive (Keller & Aaker, 1992). An overview of the Cronbach’s Alpha values is displayed in Table 4.5. The familiarity and attitude of parent brands are assessed using the same scale as for the base brand familiarity and attitude. The correlations between the items were all above .30, this indicate that all items are highly correlated with the total score of the scale.

4.2.2. Manipulation check

This section aims to verify whether the four conditions are manipulated correctly. To do so, the mean of PF-BB-CB and PBF should be different among the four conditions. Conditions 1 & 2 are manipulated to have a high PF-BB-CB and should be significantly different from the two other conditions. For the level of PBF, conditions 2 & 4 are manipulated to be low. Thus, it is expected to be significantly different from conditions 1 & 3 in terms of PBF. First, an overview of the means for PF-BB-CB are given in Table 4.2.

(41)

41 Table 4.2: Perceived PF-BB-CB M SD Condition 1: High PF-BB-CB & High PBF (45) 4.790 1,338 Condition 2: High PF-BB-CB & Low PBF (41) 4.878 1,413 Condition 3: Low PF-BB-CB & High PBF (44) 3.553 1,510 Condition 4: Low PF-BB-CB & Low PBF (57) 3.553 1,375

A more in-depth analysis is conducted to confirm the manipulations are significantly different in this study. The analysis is conducted using a One-way ANOVA. Levene’s test confirmed the homogeneity of variances assumed (p=0.787). The One-way ANOVA confirmed that there are significantly differences among the four conditions (F (30.293) = 15.289, p<0.000). To get a more detailed overview of differences between the groups a Tukey’s post hoc is conducted. The results confirm the manipulation of the conditions on PF-BB-CB. Conditions 1 and 2, which were manipulated to be high on PF-BB-CB, are significantly different from the conditions 3 and 4, which were manipulated to be low on PF-BB-CB (cond.1 and cond. 3, p<0.000; cond.1 and cond. 4, p<0.000; cond. 2 and cond. 3, p<0.000; cond. 3 and cond. 4, p<0.000). The analysis does indicate differences between the groups, however the objective of this paper is to measure the effect of low and high conditions of PBF and PF-BB-CB. Based on the means described in table 4.2 the ‘’low’’ conditions with a value of approximal 3.5 (out of 7), indicate moderate low-neutral levels. Therefore, in this study it is not considered to be a low PF-BB-CB. The same counts for the ‘’high’’ condition with a value of approximal 4.8 (out of 7), it indicates a moderately high fit. In this study, this cannot be accounted for a high PF-BB-CB.

To check whether the levels of PBF are manipulated correctly first an overview of the means of PBF for the four conditions is presented in table 4.3.

(42)

42 Table 4.3: Perceived PBF M SD Condition 1: High PF-BB-CB & High PBF (45) 4.815 1,377 Condition 2: High PF-BB-CB & Low PBF (41) 3.724 1,730 Condition 3: Low PF-BB-CB & High PBF (44) 4.720 1,429 Condition 4: Low PF-BB-CB & Low PBF (57) 3.076 1,480

The same analysis is conducted to check for the manipulation of PBF. Levene’s test confirmed the homogeneity of variances assumed (p= 0.136). The One-way ANOVA confirmed PBF is significantly different among the four conditions (F (34.787)=15.390, p<0.000). To get a more detailed overview of differences between the groups a Tukey’s post hoc is conducted. The results confirm that the conditions based on PBF is manipulated correctly. The conditions 1 and 3, which were manipulated to represent a high PBF, were significantly different from the conditions 2 and 4, which were manipulated to represent a low PBF (cond.1 and cond. 2, p<0.01; cond.1 and cond. 4, p<0.000; cond. 3 and cond. 2, p<0.05; cond. 3 and cond. 4, p<0.000). The analysis does indicate differences between the groups, however the objective of this paper is to measure the effect of low and high conditions of PBF and PF-BB-CB. Based on the means described in table 4.3 the ‘’low’’ conditions with a value of approximal 3.5 (out of 7), indicate a moderate low-neutral level. Therefore, in this study it is not considered to represent a low PBF. The same counts for the ‘’high’’ condition with a value of approximal 4.7 (out of 7), it indicates a moderately high fit. In this study, this cannot be accounted for a high PBF.

Based on this analysis it can be assumed that the conditions are significantly different from each other, but do not represent the high and low levels of PF-BB-CB and PBF. Therefor

(43)

43 the scale that measures the degree of perceived brand and product fit is used to test the hypotheses, rather than the manipulated stimuli.

4.2.3. Assumptions regression analysis

To test the model, by a hierarchical regression analysis, a few assumptions should be met. First, to test the overall model and the individual predictors, the sample size should at least be 122 (Green, 1991). With a sample size of 187 respondents, this criterion is met. Non-collinearity is tested by detecting tolerance and VIF among the independent variables (table 4.4.). These results indicate that there is no multicollinearity, because no values lower than 0.20 are found for tolerance and no values above 5 are found for VIF.

Table 4.4: Multicollinearity analysis.

Collinearity statistics

Tolerance VIF

Familiarity base brand 0.759 1.318

Attitude base brand 0.742 1.347

Familiarity parent brand 0.566 1.768

Attitude parent brand 0.542 1.846

Brand extension product fit 0.898 1.113

Partner brand fit 0.872 1.146

4.2.4. Correlations

Before the hypotheses are tested a correlation analysis among all variable is executed. The correlations are presented in the correlation matrix below (table 4.5). The matrix shows, the correlations among the variables. As predicted, the attitude towards the parent brands have a significant positive correlation with the co-brand evaluation (r=0.242, p<0.005; r=0.365, p<0.005). Indicating that there is a positive relationship between the attitudes towards the brand and how consumers evaluate a co-brand. The results show that both type of fits have a positive significant correlation with the co-brand evaluation. Interesting is that PBF has a stronger correlation with the co-brand evaluation (r=0.631, p<0.05) PF-BB-CB (r=0.271, p<0.005). Suggesting parent brand fit being more important for co-brand evaluation than the product fit between the base brand and the co-brand.