Can new brands start from scratch? : a study on association transfer from categories to new brands

(1)

Can new brands start from scratch?

A study on association transfer from categories to new brands.

Master’s Thesis - final

Name Luuk Ballhaus

Student number 10273530

Supervisor Jorge Labadie

(2)

Statement of Originality

This document is written by Luuk Ballhaus, 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)

Preface

Amsterdam, 24/06/2016

This thesis finalizes my MSc Business Administration at the University of Amsterdam. The topic is related to the track I chose: Marketing. During my Master, I found the courses on consumer behavior and branding the most interesting, which is why I chose to write my thesis on this topic.

I am very pleased with the guidance and supervision I received from Jorge Labadie. He put in a large amount of effort to help me finish this thesis and with his enthusiasm he motivated me during some setbacks in the research process. I would like to thank Jorge for his supervision during the thesis process. I also want to thank Roger Pruppers for providing valuable feedback on the design and analytical sections of my study.

Additionally, I would like to thank my close friends and family who helped keeping me motivated while writing my thesis and helped me distribute my surveys. A special thanks goes out to my girlfriend Kim, who provided substantial support to finishing my thesis by motivating me. Now that I am finalizing my Master, I am proud and satisfied with my performance throughout the year. I am looking forward to the opportunities that are ahead in the future.

Kind Regards,

(4)

Abstract

This study is concerned with the relation between new brands and the categories they enter. The goal is to look closer into the process of association transfer from categories to new brands. This is done by analyzing the impact of association strength and favorability on the number and type of associations that transfer. In order to analyze the transferred category-associations effectively, a fictional brand name is used and associations were collected using free association methods. The results brought to light that stronger associations are more likely to transfer to new brands. Another interesting fining is that categories with either high or low association favorability transfer more associations compared to categories with mid favorable associations. Additionally, the results show that some types of associations are more likely to transfer than others. Our results provide new opportunities for further research and delivers managers of both start-ups and new brands with valuable insights concerning their branding strategies. Since associations transfer in all conditions, managers should understand that entering a category is a strategic decision which has an effect on how a brand is positioned in the mind of consumer.

(5)

-Appendix C

- 89 -

Appendix D

- 96 -

Appendix E

- 104 -

Appendix F

- 108 -

Appendix G

- 113 -

Appendix H

- 114 -

(7)

(8)

1. Introduction

1.1 New brand introductions 1.1.1 Startup difficulty

Eight out of ten startups fail within 18 months (Wagner, 2013). Many drivers have been investigated to identify why startups tend to fail, but few of these studies focus on the associations and attitudes consumers have with new brands. Brand associations form the basis for brand attitudes, which in turn are a predictor for consumer behavior towards the brand (i.e. purchase intention). So managing the associations, or image, of a new brand is essential. However, the image of a new brand is for a significant degree influenced by external factors.

In a broader sense, when people analyze unknown things (like brands) or situations, they try to associate it with something familiar. This enables them to form judgments and feelings based on prior knowledge and experiences. However, placed in the context of encountering new brands, this does imply that brands do not really have the opportunity to make a first impression from a clean starting point, since they are subject to a transfer of associations, called image transfer (Loken, Barsalou & Joiner, 2008) or association transfer. These associations that are transferred to the brand for free are called ‘autonomous associations’. Image transfer has multiple definitions, but for this study we specifically define image transfer as the transfer of associations from one object to another, made by an individual (McCracken, 1989; Gwinner, 1997; Loken et al., 2008). Image transfer mostly happens unconsciously and in many different situations. Marketers learned about this phenomenon and actively use it to build brand image effectively, in celebrity endorsements and event sponsorships for example.

1.1.2 Classification of unknown bands

Prior research shows that associations can transfer from the product category to new brands in that category. Evidence suggests that category associations are more likely to extend to the new

(9)

brand when there is a high degree of perceived similarity between the category and the new category member (Loken et al., 2008). Subway is a Fastfood Companies company which instantly got associated with being unhealthy, simply by entering the Fastfood Companies category. Another example is one of the fastest growing tech-startups of this moment, DoubleDutch, which makes revolutionary event-apps for business conferences, by introducing new functionalities as real-time lead and performance management. The problem they encountered however, is that their clients associate them with regular event-app builders. Therefore, they teamed up with a company that positions itself as a “category creation agency” and together they launched a new category, in which they now are the only player. This had a significant effect on their business performance (Doubledutch, 2016). However, not every company has the resources to follow this path, but it highlights the relevance and importance of this subject. Entering a category is a strategic decision. By entering a category, brands choose their competition, which they have to differentiate themselves from. However, no studies focused on determinants of the process of image transfer differs between categories and new brands, and whether consumers in a category are affected in the same way. Fortunately, image transfer has been extensively studied in other contexts. Think of top-sportsmen endorsing brands in commercials, brands leveraging their country of origin and brands sponsoring the world cup soccer or the super bowl for example.

The major difference, however, is that image transfer normally is the goal of the process and thus something positive. When looking at categorization theory, new brands can either benefit or suffer from category associations. The associations that transfer can be strong or weak, they can be favorable of unfavorable, but they can never be unique since they are category associations (Keller, 1993). This implies that new brands can never gain a competitive advantage based on the transfer of category associations. Managers are in need of better understanding of image transfer in categories, in order to manage it to their advantage.

(10)

1.2 Image transfer in product categories

1.2.1 The influence of category type on association transfer

This study firstly aims to find whether the degree of image transfer differs among different

category types. Therefore, the question this thesis aims to answer is:

What is the impact of association strength and favorability on the number and type of associations that transfer from a category to a new brand in that category?

The context in which this is investigated will be by using a new brand, since new brands are expected to evoke less associations the first time consumers are exposed to them than existing brands. Existing brands already have an associative network of their own, which make feedback effects possible. Therefore, it is assumably better to use new brands to measure the effects on transferred category associations. Additionally, since it is expected that certain personal characteristics related to the category might influence the results, some additional variables are assessed as well. These variables are explained more elaborately below.

1.2.2 Determinants of category type and personal category characteristics

First, the concept ‘category type’ needs to be elaborated. Since the aim of this study is to find differences in the transfer of associations, the type of category is defined by the strength (weak/strong) and the favorability (low/mid/high) of associations concerned with the category. Combining these two characteristics leads to six distinct category types. A category can be classified accordingly in one of the six conditions. One could argue that categories can also be divided based on type of association, for example on symbolic vs functional associations. However, this study aims to stay close to Keller’s (1993) framework and the characteristics of

(11)

associations, and therefore focuses on association strength and favorability. The associations that are analyzed can be functional or symbolic, but are not analyzed separately on these criteria. Subsequently, association formation and transfer is also influenced by other factors, which should be accounted for in this study. After reviewing relevant literature, three personal category characteristics are defined that are expected to make an impact, namely: Category Expertise, Category Involvement and consumers’ Attitude towards a category (Sujan & Deklava, 1987; Celsi & Olson, 1988; Gwinner, 1997). These characteristics and their expected impact are discussed in larger detail in the next chapter.

1.2.3 Delimitations

Since this thesis aims to investigate whether and how brand image is affected by the product category, this study focuses on image transfer from the category on a newly introduced brand. Existing brands that extend to a different product category are not incorporated in this study, since these brands already have associations based on their original products or services. Furthermore, this study focuses on a set of three personal category characteristics that are expected to moderate the image transfer from a category to a brand. It is likely that there are more factors influencing this effect, although these three have been identified by reviewing the relevant literature. Furthermore, as stated earlier, this study does not analyze the differences between functional and symbolic associations and their likelihood of transfer.

1.3 Contributions

1.3.1 Theoretical contributions

Although prior research suggests that associations can transfer from category to new brands in that category, the factors influencing this transfer remain untouched. This thesis builds on existing categorization theory, by borrowing concepts of image transfer from prior research on

(12)

leveraging sources of secondary associations. Researchers can use this study as a foundation to analyze how image transfer from a category to a brand is influenced by the category type and personal characteristics. A next step can be to check whether, for example, positive associations can be transferred while rejecting negative associations. However, before this can be studied it is necessary to first understand which associations transfer more easily and whether there are other variables that have an effect on the process. Additionally, this thesis also adds on the literature about leveraging sources of secondary associations by adding new insights on using the product category as a source of secondary knowledge.

1.3.2 Managerial contributions

For managers, knowing which kind of categories are influenced most by image transfer is essential for crafting brand image strategies for newly introduced brands. For example, when a manager considers to enter a certain category, he or she has to understand what potential opportunities and threats are accompanied with such a move. In order to manage the transfer of category associations to the brand, one must first understand the full process and the factors that have an influence. Knowing which kind of associations are most likely to transfer, and what influence category association strength and favorability have on the number of associations that transfer is of utter importance.

1.4 Outline

This study is structured in the following chapters. The next two chapters of this thesis focus on reviewing the relevant literature on the key concepts about brand image and category image, image transfer and categorization theory. Furthermore, relevant literature about personal category characteristics and their role on the process is discussed. The fourth chapter is dedicated to the conceptual framework and hypothesis are formulated and presented.

(13)

Subsequently, the fifth chapter clarifies the chosen methodology and explains the research design in detail. The results of this study are presented in chapter six. Finally, the seventh chapter discusses the empirical findings in the discussion chapter, followed by a section about implications for managers and for researchers, limitations of the study and suggestions for future research. This thesis ends with a conclusion.

(14)

2. Image and association transfer

The following chapter focuses on reviewing the theoretical background on the individual concepts that are used in this study. First, a foundation is build by reviewing the core literature on brand image and associative networks. Then, the concept ‘association transfer’ is discussed in different contexts, in order to understand how associations normally transfer, and whether it can be assumed that this works for product categories in the same way. Finally, an overview of the current knowledge on categorization theory is provided, which is used as a starting point for this study.

2.1 Image and associative networks

Brand image has been studied extensively and one of the main scholars on this particular topic is Keller. According to Keller (1993), the image of a brand consists of an interlinked network of associations that consumers hold in memory (Till et al., 2011). Brand associations come in different forms: attributes, benefits and attitudes. Brands can distinguish themselves from competition based on the uniqueness, strength and favorability of these associations (Keller, 1993). All three features need to be satisfied in order to build CBBE (Keller, 2001). The favorability element hints that associations can be positive as well as negative of nature. These associations form the basis for attitude formation, which in turn is one of the main predictors for consumer behavior (i.e. consideration or purchase intention) (Keller, 1993; James, 2005). Till et al. (2011) extend this view by adding the relevance and number of associations as important characteristics of brand associations.

Strength of association represents the intensity of the relationship between the

association and the brand node. Stronger brand associations come to mind faster and might lead the consumer to other (closely related) associations as well. Stronger associations are more easily accessible via spreading activation (Keller, 1993; Till et al., 2011). Favorability is concerned with the valence of associations. Some associations are more positive or negative

(15)

than others, which reflects to the brand’s overall image (Till et al., 2011). Uniqueness is defined as the degree to which an association is distinct from associations that other brands within the same product category hold (Keller, 1993; Till et al., 2011). According to Keller (1993), in order to differentiate from competition, associations also have to be unique. Although many scholars found results supporting the importance of brand differentiation, Romaniuk, Sharp & Ehrenberg (2007) show in their study that perceived differentiation is a questionable driver of buyer behavior. They conclude that it is better to focus on being different and unique. The importance of association relevance is introduced by Till et al. (2011), who state that some associations might show a strong link to a certain brand but fail to add any value due to irrelevancy for consumers. Finally, the number of associations are also included as a feature of associative networks by Till et al. (2011), which is simply the amount of associations in a consumers’ associative network.

Apart from Keller, multiple other scholars have been studying brand equity and brand image. Cian (2011) reviews multiple techniques to measure brand image, both qualitatively and quantitatively. The variety in available methods shows that brand image can be hard to conceptualize for consumers and that different methods might be needed to bring all associations to light. These associations then need to be structured in order to analyze them (Keller, 2013). When assessing the structure of brand associations, abstraction and complexity must be taken into account (Hsieh, 2002). The role of abstraction level in classifying associations is reflected in for example means-end chain theory (Gutman, 1982), since multiple associations may be linked through different levels of abstraction. Associations in higher levels of abstraction clarify what consumers perceive the brand to do for him/her, which makes these associations to have a more immediate impact on purchasing behavior than lower-level associations (Hsieh, 2002). Subsequently, complexity refers to the number of dimensions needed to describe the beliefs a person holds about a brand (Hsieh, 2002).

(16)

2.2 Managing brand image

Associations form the basis of brand image, and brands can build and adjust their associative networks in multiple ways. Keller (2001) introduced the customer-based brand equity (CBBE) pyramid, which is used by managers as a tool to make the brand resonate in the mind of consumers. The pyramid consists of four distinct stages a brand needs to go through. The first step represents the brand identity or salience of the brand. This consists of two dimensions, namely (1) depth: how easily the brand is recalled, and (2) breadth: whether consumers think of the brand in certain situations (Keller, 2001). The second step in the brand pyramid is about brand meaning, which is related to sources of the brand’s associations. Keller (2001) divides the second step in an emotional and rational part, namely imagery and performance respectively. For this study, deeper understanding on which sources can be used to build associative networks is important. The third step takes the consumer one level further and focuses on responses and attitudes, which are also used by consumers to form associations. Here, Keller makes the distinction between an emotional and rational part again, feelings and judgments respectively in this case. The final and ultimate step in the pyramid is brand resonance and is about loyalty and brand communities (Keller, 2001).

As stated above, the second and third step in the pyramid are about attitudes and associations. In his 1993 article, Keller provides multiple sources of brand associations, which can be product related or non-product related. The latter focuses on price, packaging, user imagery and usage imagery. In his 2001 article, Keller adds on this view by making a distinction between an emotional and rational side. However, in 2005 Keller writes about branding shortcuts. These tools help the marketer build brand identity, even prior to consumption. Keller (2005) talks about choosing primary brand elements and leveraging secondary brand associations, in order to build brand image in an efficient way. Primary brand elements are communication devices the brand can choose. Web URL’s, logos and symbols are primary

(17)

brand elements for example (Keller, 2005). There are some criteria for choosing brand elements effectively, which focuses on both how to build the brand equity and on how to defend it once it has been established.

One of the most impactful primary brand elements is the brand name itself (Keller, 2005). Choosing a memorable, meaningful or likable brand name is an easy way to help consumers to identify what the brand is about. Subsequently, once the consumer has become familiar with the brand, elements of transferability, adaptability and protection of the brand name can help to leverage and preserve brand equity from the brand name (Keller, 2005). Meaningful brand names like ‘Flu-Relief’ or ‘Clean-All’ are suggestive of the category and help the consumer identify the brand. Research also shows that meaningful brand names are usually rated higher on perceived quality and perceived as better suitable than similar brand that have a more abstract name (Kohli, Harich & Leuthesser, 2005). This makes meaningful brand names an efficient tool to immediately generate favorable associations, while non-meaningful brand names require more investment of time and money to match that. The downside with meaningful brand names is that they can not be extended to new categories and are less suitable for name-changes than non-meaningful brand names (Kohli et al., 2005).

Additionally, brand managers can manage the associations of a brand by linking the brand to external entities with their own associations, which is known as leveraging sources of

secondary brand associations. The goal is that associations transfer from the external entity

to the brand (Keller, 2005). Many sources of secondary brand knowledge exist, but they can be grouped in four distinct sources, namely: people, places, things and other brands (Keller, 2005). This brand building strategy can be a fast and lucrative manner, although the lack of control over these external entities also make it a potentially dangerous tool to use. Once a clear connection has been established between the brand and a source of secondary brand associations in the mind of the consumers, any negative attention for the external entity can

(18)

also be reflected in the consumers’ evaluation of the brand (Keller, 2005). Similar to choosing a brand’s primary brand elements, marketers can choose which sources of secondary associations they want to emphasize, in order to shape the brand’s image to their desire. The next chapter about image transfer elaborates more on how leveraging secondary associations works.

2.3 Association transfer

In order to understand how category-associations might transfer to new brands, existing theory on association transfer and leveraging secondary brand associations is reviewed. Association transfer is usually the goal when leveraging secondary associations (Keller, 2005), and has been studied extensively in that context. However, association transfer in categories is still hardly researched, and the major difference is that association transfer in this context is a phenomenon that happens regardless of intentions. As stated before, the sources of secondary association can be grouped in four distinct sources: people, places, things and other brands. The following section focuses on the process of image transfer in each of these different contexts, in order to find differences and similarities. For each of the four groups of sources, one source identified by Keller (2005) is shortly reviewed, namely: celebrity endorsement, sponsorship, brand alliances and country of origin.

2.3.1 Celebrity endorsement and image transfer

Brands use celebrity endorsement to transfer image or meaning from the celebrity to the brand, which is based on the concept of classical conditioning (McCracken, 1989). One can think of professional soccer players using a brand in commercials for example. The goal of this process is to transfer certain characteristics from the celebrity, like attractiveness, likeability or

(19)

trustworthiness, to the brand (McCracken, 1989; Erdogan, 1995). McCracken (1989) conceptualized the process of association transfer in the following way:

This model implies that associations are first transferred to a certain product, after which consumers can access these positive associations by using the product.

Erdogan (1999) presents a literature review on the effectiveness of many celebrity endorsement studies and lists a series of moderators, of which the most impactful ones are product-celebrity fit, message and product type, level of involvement, number of endorsements by the celebrity, characteristics of consumer and the overall meaning attached to the celebrity (personality and values). These moderators might also prove to have an effect in category association transfer.

2.3.2 Image transfer in sponsorship

The concept of association transfer is also extensively covered in sponsorship literature, and is developed using the celebrity endorsement literature as a starting point. The goal of sponsorship is usually to increase (brand) awareness or to change the brand image by transferring positive association from the sponsee, the person or entity that is being sponsored, to the sponsor (Cornwell et al., 2005). Gwinner (1997) developed a conceptual model for image transfer from events to the brand via sponsorships. Firstly, the determinants of event image are proposed,

(20)

which are: Event type, event characteristics and individual factors. Proposed by Gwinner (1997), the image transfer process is further moderated by: Degree of similarity, level of sponsorship, event frequency and product involvement. These moderators show similarities with the moderators in celebrity endorsement theory.

Extensive research has been done to find the underlying cause of sponsorship success (Cornwell et al., 2005). The most frequently studied factor influencing sponsorship success is congruence or fit. The proposed theory is that an evident match between sponsor and sponsee is more easily remembered. On the one hand, researchers argue that congruency is more easily remembered due to resemblance with expectations (Cornwell et al., 2005). On the other hand, however, another stream of researches state that incongruence requires more elaborate processing, and would require consumers to follow the central processing route (Petty, Cacioppo & Schumann, 1983; Cornwell et al., 2005). This implies that a low level of fit is more easily remembered.

2.3.3 Brand extensions and image transfer

Another area where secondary brand associations are leveraged is with brand extensions. Similar to endorsement and sponsorship, the goal of a brand extension is to transfer the positive associations from the established brand to the new brand (Park, Milberg & Lawson, 1991; Meyvis & Janiszewski, 2004). Brand extensions have been studied extensively, since it can be both beneficial as dangerous to use, for both the established and the new brand. When a brand extension fails, both the extension and the established brand can suffer from this (Aaker, 1990; Park et al., 1991; Meyvis & Janiszewski, 2004). Although leveraging the power of the established brand’s name is tempting, the success depends on whether consumers think that the extension is credible. This can be achieved in two ways, either by extending the brand to a category that has similar associations with the established brand, or by extending the brand to

(21)

a category that is in line with the concept of the established brand (Park et al., 1991). Both of these ways underline the role of fit, or at least perceived fit, especially for considering concept consistency.

2.3.4 Country of origin and image transfer

The first three sources of secondary brand associations that are discussed all showed great importance for the degree of fit between the external entity and the new brand. This is in line with the ‘match-up hypothesis’, which has been studied extensively as well (Kamins 1990; Gwinner, 1997; Till & Busler, 2000). For the country of origin as a source of secondary associations, the role of fit can be seen as that the brand name has to match the country of origin, or at least match the perceived country of origin. An interesting study shows that the role of perception is crucial in leveraging the country of origin as a secondary association (Magnusson, Westjohn & Zdravkovic, 2011). The results showed that the brand’s image in the mind of the consumer is affected by the perceived country of origin, regardless of accuracy. Additionally, the image is updated, positively or negatively, when the consumer learns the actual country of origin of the brand (Magnusson et al., 2011).

Perception clearly plays a role here, which was also a big part of the image transfer process in brand extensions (Meyvis & Janizewski, 2004). Marketers know that perception is something that can be influenced via marketing tactics or articulation (Cornwell et al., 2005), but this has both upsides as well as downsides. On the one hand, the article by Magnusson et al. (2011) shows that consumers update their image of a brand after their perception changes. This means that brand image is not persistent in the mind of consumers, and that brands can alter their image when it turns out that consumers think negatively of them. On the other hand, the attraction of making the brand look nicer than it factually is can backfire once consumers learn the truth, for example by choosing a brand name that suggests a certain country of origin

(22)

while in fact that is not true (Darke & Ritchie, 2007; Magnusson et al., 2011). In the long run, this might cause consumers to process future brand-related information with suspicion.

2.3.5 Concluding remarks

Taken together, it can be concluded that association transfer has been studied extensively in multiple contexts. Linking a brand to an external entity causes associations to transfer. Although it can be used to build brand associations quickly, secondary associations need to be managed properly since a brand has no control over the external entity that it is being linked to. Fit seems to play a large role, as well as perception. Repetition is important for using secondary associations effectively, as respondents form stronger connections between the entity and the brand when they are exposed to that link more often. Although not discussed in Keller’s article on leveraging sources of secondary associations (2005), the product category a brand belongs to can be seen as a source of associations that affects new brand that enter in that category.

(23)

3. Categorization and image transfer in categories 3.1 Categorization Theory

3.1.1 Categorization

In order to understand how category associations are transferred to brands, it is essential to know how categories are represented in the mind of consumers. Categorization is the process of identifying and comparing new things or ideas (Cohen & Lefebvre, 2005). The brain uses prior information about a certain category in order to identify and make judgments about new category members, which are called category inferences (Loken et al., 2008). This process helps customers understand new brands and products by placing it in a relevant frame of reference (Bloch, 1995). Loken et al. (2008) and provided an overview of four different theories on how categories are represented in the mind of consumers. First, the classical approach, which assumes all categories are perfectly clear, and that categorization can take place when there is a clear view on the subject that needs to be categorized. Second, the prototype view is based on category prototypicality: meaning that the features of the new product or brand overlap with the category features (Rosch & Mervis, 1975). Third, the exemplar view assumes consumers compare a new stimulus to products or brands from prior experience (Medin & Schaffer, 1978). And fourth and final, the connectionist or ‘knowledge’ approach, which states that consumers connect a new product or brand to a category based on prior knowledge (McLelland & Rumelhart, 1985).

Once a product or brand is categorized in the mind of the consumer, information and knowledge is transferred from the category to the entity (Waldmann, Holyoak & Fratianne, 1995). It can be proposed that the category acts like a source of secondary associations. Using the category image can be an efficient way to build or alter brand image, although it can also harm the brand fairly easy.

(24)

3.1.2 Leveraging category associations

In order to use the category associations to an advantage, establishing the frame of reference is the first step according to Keller, Sternthal and Tybout (2002). To be able to beat competition, that competition first needs to be defined. Subsequently, the second step is to leverage the ‘points of parity’ (PoP’s) that are accompanied in the category (Keller et al., 2002). There are two kinds of PoP’s (Keller, 1998): category PoP’s and competitive PoP’s. The category PoP’s define which category the brand is in, and which competitors it competes with. In order to pop up in the mind of consumers in usage or purchase situations, the category PoP’s are needed in order to be considered as a viable option (Keller et al, 2002). Secondly, the competitive PoP’s are used to challenge the PoD’s of competitors. This updates the PoP’s of the category, meaning that brands constantly have to find new ways to add value, in order to stay ahead of competition (Keller et al, 2002). The third and final step is to find compelling points of difference, even if they seem contradictory (Keller et al., 2002). Additionally, Meyvis & Janiszewski (2004) state that these PoP’s and PoD’s result in associations about product- or service-related features and usage situations.

The notion that a brand’s image consists of an interlinked network of associations (Keller, 1993; Till et al., 2011) can also be extended towards the entire product category. Alternatively, this ‘category image’ is a collection of more general associations consumers have with all products or services that belong to the specific category (Loken et al., 2008). Therefore, associations of a category can vary in strength and favorability, but associations of the category can never be attributed to one single brand only in the category. Prior studies showed that associations in product categories transfer between multiple entities. Brand image can be affected by the product category, the category image can be affected by brands that enter the category (Loken et al., 2008) and associations transfer between brands within a category as well (Lei, Dawar & Lemmink, 2008). This association ‘spillover’ is often not symmetric, which

(25)

makes it hard to predict to what extent associations will transfer. Moreau, Markman & Lehmann (2001) state that adding a category label when introducing a new product or brand enhances consumers’ cognitive processing. This can result in consumers basing their associations on the stated category, when product information is missing (Gelman & Markman, 1986), which implies association transfer from the category to the brand (Gregan-Paxton & Moreau, 2003). These associations can be positive as well as negative, which makes it essential to understand what factors influence this process, in order to manage it to an advantage.

3.2 The role of category type

There are multiple ways to make distinctions in product categories. For example, categorization can take place on the product level, brand level, or based on usage situations. As stated above, brand image is affected by the category image. For this study, one of the goals is to investigate the influence of category type on this process. Therefore, a distinction in categories is made on strength (high/low) and favorability (high/mid/low) of category associations, which describes the overall image of the category (Loken et al., 2008). For example, a category can display strong negative associations, like Fast-food.

It has been proven that negative information and associations are more persistent than positive ones in the mind of consumers (Mittal, Ross & Baldasare, 1998; Colgate & Danaher, 2000; East, Hammond & Lomax, 2008; Arbore & Busaca, 2009). According to these findings, it is therefore likely that negative associations are more easily transferred from the category to the brand entering the category than positive associations However, other studies indicate that positive information is stored and retrieved easier. The TraceLink theory depicts that information that is stored under high arousal is better for long-term retention of the information, making it more persistent and easier accessible in the mind of consumers (Meeter & Murre, 2005). High arousal emotions are either very positive (excitement and awe) or very negative

(26)

(anxiety and anger). Similarly, strong associations are formed under more elaborate cognitive processing (Pitta & Katsanis, 1995). For example, brand associations can become stronger when they are presented consistently over time (Keller, 1993). Therefore, it is likely that strong associations are more persistent in the mind of the consumer, resulting in easier access to these associations than weak ones.

To measure this, a selection method is needed that identifies product categories that have typical associations. In other words, a typical combination of strength (high/low) and favorability (high/mid/low) of the associations accompanied with the category. For example, think of the fast-food category that has strong and negative associations. Although such a selection method is not readily available, there is a study that might prove to fit the occasion particularly well. Percy & Rossiter (1992) identified four distinct advertising strategies for brand attitude based upon type of motivation and type of decision, and they also identified some product categories as examples for each of the four conditions. Although their study focuses on advertising strategies, the distinction made in their experiment can also be used for this particular study if it turns out that these categories are also divided based on associative networks.

Percy & Rossiter (1992) describe the type of motivation as either to enhance something positive (i.e. seeking extra enjoyment/stimulation), or to reduce something negative (i.e. seeking solutions/alternatives). This distinction can be extended to the respective associations, meaning that positive motivation is accompanied by positive associations, and vice versa for negative motivation. Additionally, the type of decision can be seen as a high-involvement or a low-involvement decision (Percy & Rossiter, 1992). Pitta & Katsanis (1995) highlight the notion that high-involvement situations require more elaborate processing, resulting in that the brand association strength correlates with the quality and quantity of cognitive processing a consumer devotes to it (Petty et al., 1983). However, a brand that consumers buy in

(27)

high-involvement situations is not necessarily accompanied with strong associations. Therefore, dividing categories based on purchase motivation and type of decision can possibly resemble a division based on strength (weak/strong) and favorability (low/high) of associations concerned with the category, but needs further testing first.

3.3 The role of personal category characteristics

Another goal of this study is to test whether consumers are affected differently by the image transfer effect, based on personal characteristics with the category. The proposed factors that have a moderating influence on association transfer are Category involvement, Category knowledge (expertise) and Attitude towards the category. The following three subparagraphs elaborate on their respective influences. These factors are likely to differ between consumers in similar situations, which is why it is relevant to incorporate them in this study.

3.3.1 The role of involvement

The term ‘involvement’ has been defined as: “an individual, internal state of arousal with intensity, direction, and persistence properties” (Andrews, Durvasula & Akhter, 1990). Celsi & Olson (1988) describe this state as ‘felt involvement’, and discriminate between two types of involvement: personal relevance (intrinsic involvement) or environmental cues (situational involvement). The results of their study indicate on a strong influence of felt involvement on consumers’ attention and comprehension processes, and the extent of cognitive elaboration during comprehension (Celsi & Olson, 1988). The role of involvement has been studied extensively in advertising literature, where it clearly shows to have a strong influence (Petty et al., 1983; Muehling, Laczniak & Andrews, 1993; Zhang & Zinkhan, 2006).

Petty et al. (1983) show that in low involvement situations, consumers are influenced significantly stronger by peripheral cues, which provide consumers a ‘reason to believe’ when

(28)

they lack the ability or motivation to process an argument. Alternatively, when consumers have both the motivation and ability to cognitively process an argument, they question the reason to believe the peripheral cue, and rationally review an argument based on factual information. This process is described as following the central route, which is based on cognitive elaboration (Petty et al., 1983). In advertising literature, central and peripheral cues are studied to find whether and when (i.e. in which situations) they have a persuasive effect on consumers. However, whether consumers process a message via the peripheral or central route also has an effect on attitude and association formation.

Using the evidence shown by Celsi & Olson (1988) on the influence of felt involvement, it can be stated that the degree of involvement differs between consumers in similar situations. Additionally, by adding the Elaboration Likelihood Model (ELM) introduced by Petty et al. (1983), it can be stated that consumers form their attitudes and associations differently, because of the individual differences in involvement (Celsi & Olson, 1988). The ELM also depicts that arguments processed via the central route are more persistent and stronger represented in the mind of consumers, since they required more elaborate processing and updating of cognition. Since these results indicate that consumers form associations differently based on involvement, it is expected that consumers transfer different associations as well.

3.3.2 The role of category expertise

Another personal factor that is likely to have an effect on whether associations transfer from a category to a brand is the degree of category expertise. Sujan & Dekleva (1987) found evidence for a moderating effect of category expertise on the number and quality of inferences that consumers can make. In another study, evidence was found that consumers with high category expertise make consistent and more stable product evaluations of concepts with minor innovations (Schoormans, Ortt and de Bont, 1995). The results of these studies imply that

(29)

consumers make evaluations based on their prior knowledge. Additionally, Shane (2000) and Rao & Monroe (1988) showed that consumers generally use their prior knowledge to evaluate new things or opportunities. The results from the studies described above indicate a pattern, where expertise (prior knowledge) is used as a basis for new evaluations.

3.3.3 The role of attitude towards the category

Associations are often formed under influence of attitudes towards the object (i.e. a brand, product or a category). Keller (1993) indicated that brand associations are based on attitudes, benefits and attributes. Measuring attitude towards the category in this is study is mainly incorporated as a control variable. For example, people with a positive attitude towards a company are expected to have more positive associations with the category. So when they form associations for a brand in that category based on their attitude with the category, it is likely that they recall the positive associations they have with the category since that is the most easily accessible in their associative network (Till et al., 2011). The same holds for consumers with a negative attitude towards the category, which would imply that these consumers transfer significantly more negative associations. Therefore, attitude towards the category is expected to have an effect on associations transfer from a category to a brand.

3.3.4 The role of fit

When looking at theory on leveraging sources of secondary associations, the role of fit is repeatedly indicated as the main influencer of image transfer (Aaker, 1990; Gwinner, 1997; Erdogan, 1999; Keller, 2005; Magnusson et al., 2011). In categorization theory, Loken et al. (2008) also state that congruence is the most important factor influencing whether associations are transferred from the category to a brand or vice versa. Similar to how a brand name can fit the country of origin, a brand name might also differ in the degree of fit with a category, which

(30)

might have an effect on the results of this study Therefore, in order to measure the effects of category type on the number and type of associations that transfer, the degree of fit has to be assessed as well, since we are only interested in the category associations that transfer.

(31)

4. Proposition, hypotheses and conceptual model

The following chapter describes the conceptual model of how association transfer from categories to brands is proposed and hypothesized. The chapter ends with a visual representation of the conceptual model.

4.1 Category image transfer

Prior research provided substantial evidence for the notion of image transfer in product categories. Consumers categorize brands in order to compare them with brands they already know (Loken et al., 2008). Prior research suggests that associations transfer when brands are strongly linked in the mind of consumers (Keller & Aaker, 1992). This notion also finds support from the TraceLink theory, which states that related information is stored in a consolidated way, following the connectionist approach of categorization (McLelland & Rumelhart, 1985; Meeter & Murre, 2005). This does imply, however, that consumers need to be able link a brand to a category first in order to transfer associations, with the category serving as a reference point. Then, once a product or brand is categorized in the mind of the consumer, information and knowledge is transferred from the category to the entity (Waldmann, Holyoak & Fratianne, 1995). One could argue that this transfer of knowledge works in a way similar to leveraging sources of secondary associations, which was introduced by Keller (2002). Furthermore, consumers base their associations on the stated category, when product information is missing (Gelman & Markman, 1986), which again implies that associations transfer from the category to the brand (Gregan-Paxton & Moreau, 2003). Therefore, the following expectation is made which serves as a foundation for this study:

Similar to association transfer from sources of secondary knowledge, category associations will transfer to a brand when this brand is linked to that particular category.

(32)

4.2 Number of associations that transfer

However, not all associations are expected to be transferred to the same extent. Substantial evidence has been found that strong associations are more persistent in the mind of consumers than weak associations, since they required more elaborate processing to form in the first place (Petty et al, 1983; Pitta & Katsanis, 1995). Besides that, strong associations are also more easily accessible in a respondent’s associative network (Keller, 1993; Till et al., 2011). Conceptually, stronger associations are more likely to be recalled via spreading activation, which is the underlying mechanism in mental maps (Anderson, 1983) Therefore, the following hypothesis is formed:

H1: When category association strength is high, the number of associations that transfer is also high, compared to when category association strength is low

Similarly, multiple studies show that negative information and attributions weigh stronger in the mind of the consumer than positive information or attributions (Mittal, Ross & Baldasare, 1998; Pullig, Netemeyer & Biswas, 2006). However, other studies indicate that positive information is stored and retrieved easier. The TraceLink theory depicts that information that is stored under high arousal is better for long-term retention of the information, making it more persistent and easier accessible in the mind of consumers (Meeter & Murre, 2005). High arousal emotions are either very positive (excitement and awe) or very negative (anxiety and anger). When we review this information in the context of category associations, it is assumed that extreme favorable/unfavorable associations are more easily accessible in the associative network of consumers than moderately favorable associations, similarly to how strong associations are more accessible than weak associations (Till et al., 2011). Therefore, the second hypothesis is formulated as:

(33)

H2: When category association favorability is either high or low, the number of associations that transfer from a category tot the brand is also high, compared to when category association favorability is medium.

Additionally, one could argue that association strength plays a more dominant role in the transfer of associations than favorability. For example, when you think of Fastfood Companies companies or Racing cars, the strong associations are the ones that are the most easily accessible, where in the case of Fastfood Companies it probably is an unfavorable association and in the case of Racing cars a favorable one. This is in line with Keller’s (1993) line of reasoning, stating that the importance of association strength, favorability and uniqueness work in this particular order. If Keller’s theory is true, meaning that favorability only becomes relevant when associations are strong, then this indicates a moderating effect. Strength defines the conditions for when favorability becomes important. Therefore, the third hypotheses is formulated as follows:

H3: The relation between association favorability and the number of associations that transfer is moderated by association strength, so that the effect of association favorability is stronger when association strength is high, compared to when association strength is low.

4.3 The types of associations that transfer

Another goal of this study is to analyze the types of associations that transfer. The three hypotheses formed focus on the overall strength and favorability of a category. However, it is also interesting to look at what these associations are based on. No relevant literature could be found on what types of category associations are most likely to transfer. Therefore, we decided to qualitatively look into the types associations that transfer and assess whether differences can be found between the conditions. Due to the exploratory nature of this part of the study, no real

(34)

expectations can be formed. Therefore, a proposition rather than a hypothesis is formulated. The proposition is formulated as follows:

P1: Are some association types more likely to transfer than others? And does this differ among the different conditions?

4.4 Conceptual model Strength of category associations Number of category associations that transfer to brand H1 + Favorability of category associations H2 + / - / + Strength of category

associations Type of category

associations that transfer Favorability of category associations Strength of association Favorability of association Likelihood of transfer from category to brand H3 + P1 P1

(35)

5. Methodology

This section focuses on justifying the research methods used. The chapter starts with the initial design choice and indicates why some adjustments were made to that. The adjusted design along with the procedure is discussed in the second section. Subsequently, both the pre-tests are explained, including their goal, procedure, sample and results. After this, the measures used are operationalized

5.1 Experimental design

5.1.1 Initial research design and problems

The main goal of this thesis is to shed light on the transfer of associations from product categories to a new brand in that category. Since categories have differences in the strength and favorability of their associative networks, multiple categories should be tested ideally in order to find whether these differences have an effect on the transfer of associations. Therefore, an experimental design was chosen since it allows to manipulate a stimulus (i.e. the category), after which the distinct groups can be compared.

As stated in the Hypotheses section, it is expected that category association strength is positively related to the number of associations that transfer on to a new brand in that category. A similar effect is expected for category association favorability, although this variable is hypothesized to show a larger effect for more extreme values, so for either significantly positive or negative values. This results in six testable conditions (2 Strength x 3 Favorability). A pre-test was conducted to analyze the associative networks of 12 product categories, of which the procedure and results are discussed further on in this section. We expected to have tested two product categories that match the characteristics of each respective condition, after which the most suitable category per condition could be selected. The hypotheses could then be analyzed quantitatively.

(36)

However, after analyzing the results of the first pre-test, not all six conditions were found. However, we did manage to find categories with three distinct levels of favorability, and categories with two distinct levels of strength. We therefore decided to test the effects of strength and favorability of the category associations separately. However, a large disadvantage of this method is that we can no longer test for an interaction effect of Strength and Favorability, and thus prohibits us from testing our third hypothesis:

H3: The relation between association favorability and the number of associations that transfer is moderated by association strength, so that the effect of association favorability is stronger when association strength is high

5.1.2 Design and procedure

The previous section provided the rationale for choosing an experimental design, and how we plan on analyzing the hypotheses. This section focuses on the procedure and methods needed for the main study. The data is collected by using an online survey, in which six different categories are used that match the corresponding conditions.

Respondents are first asked for their associations with one of the six categories, and are also asked to rate each association they provide on a seven-point scale on strength and favorability. Which category they are exposed to is randomized. This first step is needed to analyze the associative networks of the categories, which we use to make predictions about the associative network of a new brand in that category. Secondly, respondents are asked to fill out their associations with the brand name ‘Wepo’. This fictional brand name was tested in the second pre-test and yielded very few associations from respondents. This ‘blank brand’ is needed so that we can state with certainty that the associations respondents provide in the next question originate from the category and not the brand name. Fictional brand names are frequently used in brand association studies to heighten experimental control (Boush and

(37)

Loken, 1991; Keller & Aaker, 1992). Subsequently, respondents are then asked for their associations with the new brand ‘Wepo’ as a member of one of the five other categories, and also asked to rate each association on a seven-point scale on strength and favorability again. Which category the respondent is exposed to now is randomized, although the same category as in the first question should not be used here since respondents might be ‘primed’. This might result in respondents not providing associations for a second time, so that they only state new associations here which is a bias for the results. This means that there is a between-subjects design used here.

The first part of the questionnaire focused on free association, in order to conceptualize the associative networks of (1) the categories, (2) the blank brand and (3) the brand as new member of a category. Analysis of the associative networks of the categories (first question) can also be used as a manipulation check, since we can analyze how respondents scored the category’s associations on strength and favorability, and whether that matches the experimental condition. The same holds for the blank brand (question 2), which is supposed to evoke few associations. The second part of the questionnaire focuses on personal characteristics, which are expected to have an effect on the the number and variety of associations that transfer. Respondents are asked to fill out scales measuring the degree of category involvement and the degree of category expertise. Additionally, they are also asked to state their attitude towards the category and their perceived fit of the fictional brand name with the category, which are included as control variables. The survey ended with a few questions regarding demographic variables, in order to check whether the six conditions were similar on these terms, which is needed for a between-subjects design.

(38)

5.2 Pre-tests

Before an online experimental study can be conducted to test the hypotheses made in the previous chapter, a series of two pre-tests are necessary to identify usable stimuli. The primary goal of the first pre-test is to identify product-categories with significant differences in their associative networks on association strength and favorability. The main goal of the second pre-test is to find a fictional brand name that can be used in the experiment as a new brand in a certain category. An elaborate description of the two pre-tests is given in the following paragraphs.

5.2.1 Pre-test 1

The first pre-test consisted of an online questionnaire amongst 90 Dutch participants, of whom 85 completed the survey. Respondents were gathered via the convenience sampling method, mainly via Facebook. The goal of the first pre-test is to conceptualize the associative networks of a set of 12 product categories, in order to find differences in the characteristics of their associative networks. The analyzed characteristics are number, strength and favorability of the associations with the category. Additionally, the associations are also categorized in 16 types of associations using the ProBAR technique (Labadie, 2016), which allows to investigate where these associations are based upon. An overview of the categorization criteria is presented in appendix A. Categorization of the associations was performed by two independent people unrelated to the study, who received instructions from the author and discussed inconsistencies in categorization they had. Notably, these association-categories provide the opportunity to relate the associations to the CBBE-model by Keller (2001), as well as to other sources of associations, like primary brand elements or secondary associations. The used sampling method is a suitable method for this pre-test, since the goal is to find associative networks for

(39)

product-categories, which is unlikely to differ significantly between the conditions, as long as the sample composition is similar.

For the main experiment, six experimental groups are needed, for which six appropriate product categories need to be identified by this pre-test, following a 2x3 factorial design (Low/High Strength vs. Low/Medium/High Favorability). A total of 12 product categories are analyzed in the first pre-test, in order to find six suitable product categories for the experiment. An overview is provided of the analyzed product categories in Table 1 and their expected outcomes.

Table. 1 Expected outcome of pre-test 1

Favorable Mid-Favorable Unfavorable

Strong Music Venues _{Electrical Cars} Housing Agents _{On-demand Video} Fastfood Companies _{Oil Companies}

Weak Detergents Museums Insurance Companies

Mineral Water Pain Relievers Notary Offices

Respondent’s associations with the product category were asked via a free association method. An advantage of free association methods, as opposed to closed-statement questioning, is that free association does not force associations, and therefore identifies more salient associations. Additionally, it also finds weak spots and negative associations and provides the opportunity to uncover blind spots (Timmermans, 2001). A disadvantage of free association is the time-consuming analysis needed and difficulty to interpret the results. Each respondent is exposed to three of the twelve categories, which were presented in randomized orders. This procedure is chosen to prevent biased results caused by respondent depletion, and resulted in a mean completion time of 5 minutes. Respondents did not fill in less associations in the last category they were exposed to compared to the two previous ones. Subsequently, respondents are asked to score each association they provide on strength and favorability on a seven-point

(40)

scale. The aim is to identify six appropriate product-categories that can be used for the experiment. An overview of the results is presented in table 2.

Table. 2 Favorability scores of testes categories N MEAN SD Insurance Companies 21 2.88 0.95 Oil Companies 21 3.33 0.84 Fastfood Companies 21 3.64 1.31 Pain Relievers 20 3.82 0.66 Notary Offices 20 4.37 0.77 Housing Agents 21 4.44 0.91 Detergents 21 4.84 0.88 Mineral Water 22 5.14 0.89 Electrical Cars 20 5.28 1.08 Museums 21 5.35 1.03 On-demand Video 20 5.41 0.87 Music Venues 21 5.51 0.58

Both the pre-tests and the first part of the main experiment is based on a free association method. For the first pre-test and the first and third question in the main experiment, respondents are asked to provide all their thoughts, when thinking about category X. In the question it is stressed that no ‘wrong’ answers can be given here, and respondents are encouraged to provide as many thoughts that come up. A maximum of 10 answer fields is available. Subsequently, respondents are asked to score each association on strength and on favorability. After these three questions, a comprehensive image of respondent’s associative networks regarding the category can be made.

The results are analyzed by computing the mean strength and mean favorability per respondent. The category-strength and category- favorability were then computed by taking the mean of all respondent’s scores. However, after analyzing the results of the first pre-test, not all six conditions were found. The mean strength of the category associations showed a positivity bias, resulting in scores ranging from 4.8 to 5.6 on a seven-point scale. The first noticeable finding implies that respondents only state strong associations, as the mean strength

(41)

of all product categories is significantly higher than the midpoint of the scale. A possible explanation is that respondents find it difficult to interpret the question regarding how strong their association is. Another explanation might be that, since respondents take 1:30 minutes to complete one category, that all associations that come to mind are indeed strongly present. Nevertheless, the pre-test yielded some variety in category association strength, which makes it possible to analyze what the effect is of an increase in strength, rather than testing the differences between high and low conditions.

As described in the experimental design, the hypotheses are tested by analyzing the effects of strength and favorability separately. Six product categories could be identified that match the conditions in the following way: to match the three levels of favorability, two categories are identified for each of the three levels, in order to minimize the chance of finding category-specific results and to make the findings more robust. We decided not to use the mid point of the scale to identify the medium-favorable condition, since their is a slight positivity bias (M = 4.50). Therefore, a one-sample T-test is performed to test which categories are not significantly different from the test value 4,5. This is the case for ‘Notary Offices’ (M = 4.37, p = .477) and ‘Housing Agents’ (M = 4.44, p = 0.769). These categories are therefore combined as the mid favorable condition.

For the other two conditions, the high favorability condition is formed by combining Museums and Electrical Cars and the low condition by combining Insurance Companies and Oil Companies. Subsequently, a One-way ANOVA is performed on the three conditions F (11, 237) = 20.066, p < .001. Planned contrasts revealed that the low condition is significantly lower than the mid condition (p < .001) and the high condition (p < .001). The mean favorability was also lower in the mid condition as compared to the high condition (p < .001).

Can new brands start from scratch? : a study on association transfer from categories to new brands