Multiple celebrity endorsement : an empirical study looking at the effectiveness of multiple celebrities endorsing a single brand, while taking the role of repetition and type of fit into account

Multiple Celebrity Endorsement

Master Thesis of Elise Tijssen

Student number: 5947057

Date: 24-06-2016

MSc Business Administration – Marketing

Faculty of Economics and Business

STATEMENT OF ORIGINALITY

This document is written by student Elise Hyacintha Sylvia Tijssen 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.

ABSTRACT

The use of celebrity endorsement is a popular advertising strategy, nowadays. Especially, since the digital era made the celebrity lifestyle a lot more transparent. Most companies use multiple celebrities to endorse their brand. However, little research has been done to this phenomenon. Based on the associative learning theory, this research also investigates the role of repetition in a celebrity

endorsement context. According to this theory, repeated pairings between two stimuli are needed to establish a strong link between the two. The third variable that is investigated in this research is type of fit. Derived from the brand extension theory, two types of fit are distinguished: substitute fit and complementary fit. A substitute fit between endorsers represents two endorsers that express the same association of the brand. A complementary fit represents two endorsers that both fit the brand, but express other associations of the brand and have no overlapping associations with each other. An experiment was conducted to test the hypotheses. To test model generalizability, the study was conducted with two different product categories: clothes and watches. The results show that repeated exposures of one endorser work, as well as different endorsers with a substitute fit. Therefore, this research demonstrated the added value of repetition. Finally, the complimentary fit between the endorsers appeared to be perceived by consumers as a misfit. The concept of complementarity of brand associations on an abstract level appeared to be more complex than initially was thought. The main findings were found in both product categories. That implies that the findings are not specific to a product category, but apply to different categories.

TABLE OF CONTENT

1. Introduction 1

1.1. Celebrity endorsement 1

1.2. Multiple celebrity endorsement 2

1.3. Problem definition 3

1.4. Delimitations of the study 4

1.5. Theoretical contribution 5

1.6. Managerial contribution 5

1.7. Research outline 6

2. Brands as associative networks 7

2.1. Brand associations 7

2.2. Customer-Based Brand Equity 9

2.3. Connecting associative networks 10

3. Leveraging secondary brand knowledge 11

3.1. Sources of secondary brand knowledge and fit 11

3.2. Celebrity endorsement as source of secondary brand knowledge 14

3.3. Endorser-brand fit 15

4. Celebrity endorsement 17

4.1. Advantages and risks of celebrity endorsement 17

4.2. Single celebrity endorsing multiple brands 17

4.3. Multiple celebrities endorsing a single brand 18

4.4. Repetition in a celebrity endorsement context 20

4.5. Endorser brand fit and multiple celebrity endorsement 21

4.5.1. Type of fit 22

5. Hypotheses & Conceptual framework 23

5.1. Hypotheses 23 5.2. Conceptual framework 27 6. Method 28 6.1. Research design 28 6.2. Qualitative pre-test 30 6.3. Quantitative pre-test 34 6.4. Main study 43 7. Results 48 7.1. Data preparation 48 7.2. Sample profile 48 7.3. Reliability analyses 50

7.4. Control variable checks 51

7.5. Manipulation checks 54

7.6. Hypotheses testing 69

7.7. Additional analyses 75

8. Discussion 78

8.2. Implications 84

8.2.1. Theoretical implications 84

8.2.2. Managerial implications 86

9. Conclusion 89

9.1. Summary and answer to the research question 89

9.2. Limitations 90

9.3. Recommendations for future research 91

Literature 93

Appendix 97

1. Example pre-test questionnaire 97

2. Pre-test results 105

3. Example main study questionnaire 106

4. Output manipulation checks 119

5. Output hypotheses testing 130

1

1.1. Celebrity endorsement

The use of celebrities to endorse products, services, or brands is a way of marketing that has been done for decades (Erdogan, 1999). McCracken (1989) defined a celebrity endorser as “any individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement” (p.310). In 1983, Michael Jackson changed the game in the advertising world by signing a five-million-dollar contract with Pepsi to be the endorser of the successful ‘New Generation Campaign’. This endorsement deal meant that the pop star was tied to brand for ten years. Regarding money, this deal shattered the record of any other endorsement deal that was ever made (Herrera, 2009). Today, the money Pepsi and other companies spend on celebrity endorsement is even much higher. In 2012, Beyoncé signed an endorsement deal with Pepsi for 50 million dollars (Makarechi, 2012).

Even though celebrity endorsement is an incredibly expensive way to create awareness and equity for a brand, it appears a profitable advertising strategy. Just the announcement of an endorsement deal of a company with a celebrity already has a positive effect on stock returns of the endorsed brand (Agrawal & Kamakura, 1995). Six months after the endorsement deal announcement, stock returns of brands that pair up with athletes increase with an average of four percent (Elberse & Verleun, 2012). Today, celebrity endorsement is viewed as a worthwhile investment in advertising (Misra & Beatty, 1990; Agrawal & Kamakura, 1995; Erdogan, 1999; Elberse & Verleun, 2012).

Celebrity endorsement has been investigated by many. Most of this research builds upon the assumption that brands and celebrities can be seen as associative networks. When a celebrity is used to endorse a brand, those associative networks become linked to each other. Associations attributed to celebrity endorsers transfer from the celebrity endorser to the endorsed brand and vice versa, when the endorser and the product are combined in one advertisement (Gwinner, 1997). This way, secondary

2

brand knowledge can be leveraged. For example, if a consumer sees Lionel Messi in an advertisement of an Adidas shoe, the associations that the consumer has of Messi are transferred to the Adidas shoe. For this reason, it is important that companies do not randomly pick a celebrity to endorse their brand. A company should pick an endorser that has the associations they want to associate with the brand. In other words, there must be a fit between the endorser and the brand. This is in line with the match-up hypothesis (Kamins, 1990), which suggests that celebrity endorsement is only effective if the endorser has a congruent fit with the endorsed product or brand. This fit is a key factor in the success of celebrity endorsement in advertising (Hsu and McDonald, 2002).

1.2. Multiple celebrity endorsement

Many companies elect to have more than one celebrity endorsing their brand. Rolex is a good example of a single brand using multiple endorsers. At one point in time, they used 42 athletes to endorse their watches (Rice, Kelting & Lutz, 2012). With the high amounts of money spent on endorsement deals, it is important for companies to know if using multiple celebrity endorsers is an effective way of advertising.

Surprisingly, very little empirical research has been done to the effectiveness of using multiple celebrities to endorse a single brand, compared to using only one endorser (Rice, Kelting & Lutz, 2012; Hsu & McDonald, 2002). However, there are two exceptions: Hsu & McDonald (2002) and Rice et al. (2012) investigated ‘multiple celebrity endorsement’. Still, many questions are unanswered, as the results of both of these researches cannot be generalized to all advertising that uses multiple celebrity endorsers.

The research of Hsu and McDonald (2002) examined the multiple endorser strategy with a case study about the ‘Milk Moustache Campaign’. This campaign aimed to “change consumer attitudes that caused decline in per capita consumption of milk and increase public awareness of nutritional facts” (Hsu & McDonald, 2002, p. 21). The campaign was used to create awareness for a certain category of

3

consumption, but not to advertise a brand or product. Furthermore, the research of Rice et al. (2012) showed results that suggested that multiple celebrity endorsement is effective. However, their research could not exclude that the effect they found is a result of repetition. The present study investigates the importance of repetition in the context of multiple celebrity endorsement. In particular, the study will answer the question whether the usage of multiple celebrity endorsers across different ads is favorable for a brand while the number of exposures remains the same.

Rice et al. (2012) used multiple endorsers that expressed the same association in their research. However, one may question the benefit of multiple endorsers if every celebrity endorser communicates the same story. Multiple endorsers across different ads could prevent audience boredom. It prevents consumers getting tired of seeing the same endorser repeatedly. Despite that, could it not be more valuable for a brand to use multiple celebrity endorsers communicating different elements of the brand? That way, more aspects of the brand can be covered. Multiple brands use this strategy. Take Puma for example: Usain Bolt and Rihanna are both celebrity endorsers for the Puma brand. Usain Bolt probably adds the association of expertise to the brand, which may make Puma credible as a sports brand. Rihanna adds more of a symbolic meaning to the brand with her style, as she is not a professional athlete. Both celebrity endorsers add different associations to the brand. In other words, the celebrity endorsers communicate different elements of the brand. This study further investigates the role of the type of fit between endorsers in the context of multiple celebrity endorsement.

1.3. Problem definition

The aim of the present study is to investigate the effectiveness of using multiple celebrity endorsers to endorse a single brand, taking the role of repetition and their fit with each other into account.

4

What is the effect of number of celebrity endorsers on brand attitude and how is this influenced by number of exposures and type of fit?

In order to resolve the problem definition stated above, the following sub-questions are answered first: • What are associative networks?

• What does it mean when brands are seen as associative networks? • How can associative networks be connected?

• What is secondary brand knowledge and how do you leverage it?

• What is known about celebrity endorsement? What are its advantages and disadvantages? • What is known about single vs. multiple celebrity endorsers used for a single brand? • What is the role of repetition in a multiple celebrity endorsement context?

• What is the role of type of fit between multiple celebrity endorsers in a multiple celebrity endorsement context?

1.4. Delimitations of the study

First, secondary brand knowledge can be leveraged through different sources: other brands, places, things and people (Keller, 2003). This study focuses on celebrity endorsement as a source of leveraging secondary brand knowledge. Other sources are only briefly described in this study. Second, this study examines the effectiveness of using multiple celebrity endorsers across a series of advertisements whereby only one endorser is shown per ad. The use of multiple celebrity endorsers together in one ad will not be researched in this study. Finally, this study focuses on the effectiveness of using multiple celebrities in endorsing a single brand on consumers’ brand attitude. The effectiveness of specific product attitudes could be examined in further research.

5

Despite the extensive research that has been done in the field of celebrity endorsement, many questions remain unanswered when it comes to multiple celebrity endorsement. Most of the existing literature focused on the transfer of images of a single endorser to a single brand (Kahle & Homer, 1985; McCracken, 1989; Ergodan, 1999; Till & Busler, 2000; Miller & Laczniak, 2011). The effectiveness of using multiple celebrities endorsing a single brand has only been investigated by a few. The study of Rice et al. (2012) addressed this topic. However, their study design did not take number of exposures into account. This may have influenced their results. Therefore, the present study builds upon the results of Rice et al. (2012) by replicating their research while taking the number of exposures into account. Furthermore, this research builds upon the research of Rice et al. (2012) by looking into the type of fit between multiple celebrity endorsers. The type of fit between celebrity endorsers has not been investigated before. The research of Rice et al. used identical celebrity endorsers that all communicate the same message. However, the effect on the brand attitude might be very different with multiple celebrity endorsers that complement each other by communicating different messages. On the other hand, the use of multiple endorsers communicating different characteristics of the brand could be

counterproductive with too different endorsers. This experimental study will provide answers to these questions.

1.6. Managerial contribution

This study is also relevant from a managerial perspective, as it will contribute to the current managerial knowledge about celebrity endorsement as a form of advertising. A better understanding of the

usefulness of multiple celebrity endorsers is of value for brand managers. First, this research can provide companies guidance in decision-making about whether to use multiple celebrity endorsers or just a single celebrity endorser. Furthermore, this research gives insight into the role of the type of fit between

6

multiple celebrity endorsers. It is noteworthy that the type of fit between endorsers has never been investigated before, while differences in type of fit between endorsers do really exist. It may also be helpful in taking decisions about adding more endorsers to an existing advertising campaign. Is it better to add an endorser telling the same story as the original endorser, like for example Nathalie Portman and Charlize Theron for Dior fragrances? Alternatively, would a new endorser associated with another aspect of the brand be more effective, like in the example of Puma, Usain Bolt and Rihanna as

mentioned above? The answers to these questions provide brand managers with more understanding of this topic. As a result, brand managers will be better equipped when taking decisions about their

celebrity endorsement strategy.

1.7. Research outline

This thesis consists out of nine chapters. In this chapter, celebrity endorsement was introduced and the research question was formulated. In order to provide a better understanding of this topic, a review of the literature is provided in chapters two to four. In the second chapter, the phenomenon of brands as associative networks is described and customer-based brand equity is explained. Chapter three provides an overview of how secondary knowledge can help building a brand and it is leveraged. Chapter four zooms in on celebrity endorsement, repetition and the type of fit between endorsers. The conceptual framework and the hypotheses based on this framework are formulated in chapter five. An overview of the methodology that is used for this experiment is provided in chapter six. The analysis and results are described in chapter seven. Finally, discussion and conclusions are formulated in chapter 8 and 9, which also sketches the limitations of the study and the suggestions for further research.

7

Insight into a brand as an associative network is pivotal in understanding the process of endorsement. Therefore, an explanation of the concept of brand knowledge, brands as associative networks and how associative networks are connected is provided in this chapter.

2.1. Brand associations

It is commonly known that a brand is more than just a name with a logo. The underlying value of a brand name is the set of associations linked to it (Aaker, 1991). Keller (1993) created a model of the

dimensions of brand knowledge based upon the perspective that the memory is structured as an associative network, that consists out of a set of nodes and links. Each node represents a piece of information. “Nodes are stored information connected by links that vary in strength” (Keller 1993, p.2). An overview of the concept of brand knowledge is provided in Figure 1.

8

Understanding the concept of brand knowledge is important, because it influences what comes to mind when a consumer sees or thinks about a certain brand. However, for this research the concept of brands as associative networks is specifically relevant. This can be explained by the dimensions of brand image. Therefore, the explanation of Figure 1 emphasizes on the four dimensions of brand image.

Keller (1993) defined brand image as “perceptions about a brand as reflected by the brand associations held in the memory of the consumer" (p.3). Brand associations reflect the consumers’ perceived image of the brand; the associations a consumer has while thinking about a brand. From a network memory perspective, nodes can be brands, products or attributes. The term ‘association’ means a link between any two nodes (Krishnan, 1996). Therefore, brand associations are informational nodes linked to the brand in the memory of the consumer. In response to the brand name, the

association set is activated in the memory of the consumer (Krishnan, 1996). Practically speaking, consumer brand associations can be perceptions, preferences, and choices in the memory linked to a brand (Aaker, 1991). An example of an associative network is provided in Figure 2.

9

Brand associations take different forms and can therefore be divided into three categories: attributes, benefits, and attitudes. First, attributes are the consumers’ thoughts on what the product or service is. As seen in Figure 1, the attributes can be product-related or non-product related. The latter can be further divided into price, packaging, user imagery and usage imagery. Secondly, benefits refer to what the product or service can do for the consumer. These can be functional benefits (what it functionally does for you using the product or service), experiential benefits (how you feel while using the product or service) or symbolic benefits (social benefits). Thirdly, brand attitude refers to the overall evaluation of the brand by the consumer (Keller, 1993).

2.2. Customer-based brand equity

Brand associations are important for a brand, as they can provide value. Among the ways brand associations can create value are: “help to process/retrieve information, differentiating the brand, generate a reason to buy, creating positive attitude/feelings and providing a basis for extension” (Chen, 2001, p.440). The added value of a brand can be explained by the concept of brand equity. Brand equity consists out of the marketing effects uniquely attributable to the brand (Keller, 2009). From a consumer-based perspective, brand equity can be described as “the differential effect of brand knowledge on consumer response to the marketing of the brand” (Keller, 1993, p.1). Brand associations are a core asset for building strong customer-based brand equity. A full understanding of brand associations is necessary in order to build strong brands (Chen, 2001). The strength, favorability and the uniqueness of the brand associations play an important role in the differentiation of a brand and its’ competitors. In other words, strong customer-based brand equity is created when the consumer is familiar with the brand and holds favorable, strong and unique brand associations (Keller, 1993).

10

Associative networks can be linked to each other. Congruence is an important factor in the context of connecting associative networks. In this context, congruence can be defined as “the extent to which a brand association shares content and meaning with another brand association” (Keller, 1993, p.7). The congruence between two brand associations has influence on two things. First, on how easily an existing brand association can be recalled and secondly, how easily other associations can become linked to the brand node in the memory (Keller, 1993). Information that is consistent in meaning with the existing brand association will become easier connected to the brand compared to inconsistent information (Keller, 1993). In other words, the more similar two associations are, the easier the associations will become integrated within an associative network (Till & Busler, 2000).

So far, we know the meaning of brands as associative networks and the fact that brand

associations can create value. Now, it is important to understand how this value can be created. This is explained in the next chapter.

11

It is important to understand how customer-based brand equity can be build and which tools are

needed to connect associative networks to each other. There are three ways to build brand equity. First, through the initial choice of brand elements, like the logo, brand name, packaging and slogan. Second, through marketing activities and developed marketing programs. Through product, pricing, distribution and communication strategies, these marketing programs will help to remind the consumer what the brand has to offer. Third, customer-based brand equity can be built through the leverage of secondary associations that link the brand to other entities (Keller, 2003, p.638). This chapter zooms in on the latter, as celebrity endorsement is a tool that can be used to leverage secondary associations to a brand.

3.1. Sources of leveraging secondary brand knowledge and fit

In order to leverage secondary brand knowledge, companies often try to link their brand to other liabilities and assets. This way, customer-based brand equity can be build. Keller (2003) defined four categories of sources that could leverage secondary brand knowledge: other brands, places, things and people. An overview of the secondary sources of brand knowledge is provided in Figure 3 (Keller, 2003). To show the universality of the phenomenon, illustrations from a few different sources are provided.

12

A source of secondary brand knowledge that falls under the category ‘places’ is country of origin. In this case, the country of origin of the brand is linked to the brand in the mind of the consumer and can create secondary associations to the brand (Keller, 2003). For example, the brand Bertolli has

associations linked to it that represent the Italian cuisine or an Italian lifestyle, based on its country of origin: Italy. Besides the country of origin, other geographic locations could also generate brand associations, such as states or cities. To establish an association based on country of origin or

geographies, the location is either named in the brand name (Bailey’s Irish Cream), in the language of that certain location (La Vache qui rit) or strongly communicated in the advertising of the brand. The latter is the case with Bertolli; in all of their advertising, they try to communicate the Italian lifestyle.

One of the secondary sources of brand knowledge that falls under the category ‘things’ is events. The reason for companies to engage in event sponsorship is to build the brand through

leveraging of secondary associations that are linked to that specific event. When a brand is linked to an event, some of the associations with the event (enjoyable, fun, youthful) become linked to the brand in

13

the memory of a consumer (Keller 2003). This suggestion is in line with the research of Gwinner (1997), who investigated the effect of image transfer in the field of event sponsorship and found that the image of the sponsored event is an important factor as it may transfer to the image of the brand. Event sponsorship can help building brand equity by increasing brand awareness, improving the strength, uniqueness and favorability of the brand associations or adding completely new associations to the brand (Keller, 2003). Fit between the sponsored event and the brand is a crucial factor for the success of the image transfer from the event to the brand (Gwinner & Eaton, 1999).

A second source of secondary brand knowledge that falls into the category of ‘things’, is causes. In this case, a non-profit organization is linked to a commercial brand, which is also called a cause-brand alliance. When a brand starts an alliance with a cause, the associations that are linked to the cause can become linked to the brand. An example of a cause-brand alliance is UNICEF that is linked to Pampers (Procter & Gamble). Every year, for a period for three months Pampers has a special activation for UNICEF: with every purchase of a pack of diapers Pampers donates one vaccine against tetanus to UNICEF (http://www.unicef.nl/wat-kun-jij-doen/betrokken-ondernemen/unicef-partnership/pg-en-unicef/). An ad with an implemented cause-related marketing message results in more favorable attitudes towards the company compared to an ad without (Nan & Heo, 2007). However, research shows that also with cause-brand alliances, the level of fit between the brand and the cause plays a crucial role; for consumers that are conscious about brands, attitudes towards the brand are high with a cause-brand alliance with a high cause-brand fit compared to a low cause-brand fit (Nan & Heo, 2007).

An example of a source of secondary brand knowledge that falls under the category ‘other brands’ is extensions. Apple is a good example of a brand that launched multiple successful brand extensions, with among others the iPhone, iPad and iTunes. Just as the other examples described above, the associations linked to, in this case the extension, can transfer to the brand and vice versa (Aaker & Keller, 1990). However, also in the case of extensions, in order to connect the associative networks of

14

the extension and the brand, the level of fit between them is crucial. Furthermore, a poor fit may not only detract from the transfer of positive association, but may actually cause an increase in the transfer of negative associations (Aaker & Keller, 1990). For example, when a computer company like Dell starts producing juices, a consumer could become confused about the expertise of the company and could start doubting the quality of the original products the company produced.

3.2. Celebrity endorsers as sources of leveraging secondary brand knowledge

The fourth category of sources of secondary brand knowledge is ‘people’, which is further divided into employees and endorsers (Keller, 2003). The topic of this research is celebrity endorsement and therefore this source of secondary brand knowledge will be explained more extensively.

Celebrity endorsers can be used to create new associations or strengthen current associations of a brand. McCracken (1989) defined a celebrity endorser as “any individual who uses his or her public recognition on behalf of a consumer good by appearing with it in an advertisement” (p.310). Celebrity endorsers can be entertainers – actors, singers, models – athletes, politicians and business people (Callcott & Lee, 1995; Hsu & McDonald, 2002). Within the context of associative networks, both brands and celebrities can be seen as nodes (Till, 1998). The endorsement process builds an associative link between the endorser and the brand (Keller, 1993). This way, the associative networks of the endorser and the brand can be connected to each other. When consumers are familiar with the celebrity endorser, they already have knowledge and therefore associations of this endorser in their memory. Because the brand becomes identified with the endorser, consumers will ‘assume’ that the brand shares associations with the celebrity endorser. In this case, the associations of the endorser are transferred from the endorser to the brand. They become a part of the association set of the brand. This happens also the other way around, whereby the associations of the brand transfer to the endorser and become a part of the endorsers’ association set. An example is Nespresso, endorsed by George Clooney.

15

Nespresso uses George Clooney consistently in their advertisement. When consumers think about Nespresso, they think about George Clooney. However, this also works the other way around. When consumers think about George Clooney, they think about Nespresso. They have become part of each other’s association set. Consistently paring the brand and the endorser helps building a stronger link between the two (Till, 1998). Ideally, the associations that become related to the brand through the endorser are positive. A celebrity endorser could for example build credibility for the brand, because of its’ trustworthiness and expertise (Keller, 1993).

3.3. Endorser-brand fit

Much research into the fit between a celebrity endorser and the brand or product has been done. In the same way as the fit is important with linking other sources of secondary brand knowledge to the brand, these studies concluded that a fit between the endorser and the brand is crucial for positive evaluations of the endorsed brand (Kahle & Homer, 1985; Kamins, 1990; Till, 1998; Gwinner & Eaton, 1999).

The fit between celebrity endorsers and the brand can be based on different characteristics. Kahle and Homer (1985) conducted research to the endorser-brand fit based on attractiveness. According to their research, the effectiveness of the advertisement increases when the image of the celebrity endorser is in line with the image of the endorsed product. However, the research failed to demonstrate that the attractiveness of the celebrity endorser would not have an effect on the advertisement when it would endorse an attractiveness-unrelated product. The research of Kamins (1990) showed that the attractiveness of the celebrity endorser did not have an effect when the endorser was linked to an attractiveness-unrelated product.

Till and Busler (2000) conducted research to the fit effect based on expertise. The results of their research imply that effective celebrity endorsers should be picked based on their expertise of the product they endorse. According to the research of Hsu and McDonald (2002), the fit between the

16

endorser and the brand should be based on attractiveness, credibility or other highly relevant characteristics.

Till (1998) used the associative network theory to explain the role of the endorser-brand fit in the image transfer process. He stated, “The greater the perceived fit between the brand and the celebrity endorser, the more quickly an associative link between the two can be expected to develop” (Till, 1998, p.403). According to Till (1998), the endorser should possess the associations the brand already has or possibly could have. If the endorser has the same associations as the brand currently has, the celebrity endorser strengthens the existing associations. When the celebrity endorser has

associations the brand desires to have, the celebrity endorser can develop these associations for the brand.

Empirical studies have shown that a good fit between an endorser and a brand positively affects consumers’ perception of the credibility of the endorser, attitude towards the brand, brand recall, attitude towards the advertisement, expertise of the endorser, recognition, purchase intentions and willingness to pay (Kahle & Homer, 1985; Gwinner & Eaton, 1999; Till & Busler, 1998; Hsu & McDonald, 2002). Therefore, it is truly important for a company to pick a celebrity that has a fit with their brand.

17

4.1. Advantages and risks of celebrity endorsement

The use of celebrity endorsement has some great advantages. Firstly, when a celebrity endorses a brand, attention automatically increases because of the endorser being famous. Secondly, a celebrity endorser can polish the image of the brand with his or her reputation. Thirdly, when a new brand is introduced, a celebrity endorser can give the brand a flying start. When the brand wants to reposition itself, a celebrity endorser can help to communicate this to the consumer. Finally, celebrity endorsers who are known all over the world can help a brand grow internationally with their international reputations (Erdogan, 1999).

However, there are also risks that come along with linking a celebrity to your brand. A celebrity endorser can overshadow a brand when the attention of consumers is focused on the celebrity endorser and the brand is moved to the background. Secondly, public controversy could be harmful to the brand when for example the celebrity endorser gets involved in a scandal. A third disadvantage is the loss of public recognition of the celebrity endorser, for instance when a tennis player loses its’ top ranking position. It is likely that one would prefer to dissociate the brand from a fallen celebrity endorser. Fourthly, celebrity endorsement is an expensive marketing tool (Ergodan, 1999; Roberts, 2014). Finally, image change and overexposure happens when a single celebrity endorses too many different brands.

4.2. Single celebrity endorsing multiple brands

When a celebrity endorses too many brands, blocking can occur. Blocking can be explained based on the association network theory that was explained in the previous two chapters (Till, 1998). The more products or brands the celebrity endorses, the bigger the association set of the celebrity becomes. Therefore, the association with a certain brand can become displaced by the other associations of the endorser with the other brands. The consumer has too many associations linked to the celebrity

18

endorser and therefore the link between the celebrity endorser and the brand weakens. When the set of associations of both the celebrity endorser and the brand is small, there is a bigger chance that the consumer develops a positive attitude towards the brand. The link between the celebrity endorser and the brand is better activated, when compared with a celebrity endorser and brand with both large sets of associations (Till, 1998). For example, soccer player Lionel Messi has multiple endorsement deals. A consumer interested in soccer probably links Adidas and Messi. However, Messi also endorses Pepsi. Because of the many endorsement deals, the link between Pepsi and Messi is not easily made. However, if Messi would only endorse Pepsi, without endorsing other brands, the link with Pepsi would probably be easier established. Blocking can have some negative consequences for both the brand and the endorser. Consumers perceive the brand as less favorable and are less interested in buying the product when they know celebrity endorsed multiple brands (Mowen and Brown, 1981, cited in Tripp, Jenson & Carlson, 1994). Furthermore, consumers’ perceptions of celebrity credibility, celebrity likability and attitude towards the advertisement become less favorable when the number of brands endorsed by one celebrity increases (Tripp, Jenson & Carlson, 1994). Based on these findings, one may conclude that marketers should select celebrity endorsers that do not have too many strong associations with other brands, products, or services.

4.3. Multiple celebrities endorsing a single brand

Like mentioned in the introduction, companies often choose using multiple celebrity endorsers for their brands. However, very little empirical research has been done to the phenomenon of multiple celebrity endorsement.

Saleem (2007) conducted research into multiple celebrity endorsers that appear in the same ad, under both low and high involvement circumstances. The results of this research show that this strategy has a positive effect on attitude towards the ad, attitude towards the brand and purchase intentions in

19

low involvement ads. This could be explained based on the Elaboration Likelihood Model (ELM) of persuasion developed by Petty and Cacioppo (1984). This model claims that a receiver of a persuasive message processes stimuli via one of two routes: the central or the peripheral route. The central route represents the situation when the elaboration likelihood is high and the peripheral route when it is low (Petty & Cacioppo, 1984). “Under the central route, attitude changes result from a person's careful attempt to evaluate the true merits of the advocated position. Under the peripheral route, however, attitude changes occur because the person associates the attitude issue or object with positive or negative cues or makes a simple inference about the merits of the advocated position based on various simple cues in the persuasion context” (Petty & Cacioppo, 1984, p.668). The results of the research of Saleem (2007) showed that when receivers are low in their involvement and the ad is processed via the peripheral route, having multiple celebrities in one ad has a positive influence on attitude towards the ad, brand attitude and purchase intentions. However, when receivers are high in involvement and the ad is processed via the central route, the number of endorsers does not have an influence, as the argument of the ad is more important than the number of endorsers.

The research of Saleem focused on multiple celebrities in one ad. However, the present research is focused on using multiple celebrity endorsers across different ads. The research of Hsu & McDonald (2002) used multiple celebrities across different ads, whereby a single endorser per ad was shown. In the research of Hsu and McDonald (2002), three reasons were found for a brand to use multiple endorsers to endorse their brand. First, it reduces the risk of having a negative image transfer of a certain endorser to the brand. If the negative image is transferred to the brand, still other endorsers may transfer positive images to the brand. Second, using multiple celebrity endorsers avoids audience boredom. When people always see the same celebrity in combination with the brand, they might get bored. Having multiple celebrity endorsers linked to a brand could solve this problem. Third, with multiple celebrity endorsers, multiple audiences can be addressed. Different audiences may like

20

different celebrities. In general, marketers assume that the celebrity endorsers should relate to the target audience, for example in age or gender, to have the target audience relate to the advertisement (Wells et al., 1989, cited in Hsu & McDonald, 2002). When a brand has a wide target audience, multiple celebrity endorsers are needed to cover the whole target audience. L’Oreal is an example of a brand that tries to reach multiple segments with different endorsers. L’Oreal has a wide range of endorsers, ranging from young, like Blake Lively (1987), to older celebrity endorsers, like Diana Keaton (1947), to address both a young and older audiences (http://www.celebrityendorsementads.com/celebrity-endorsements/product-index/#loreal). Like mentioned in the introduction, the research of Hsu & McDonald (2002) was a case study about the Milk Moustache Campaign’. This campaign was developed to “change consumer attitudes that caused decline in per capita consumption of milk and increase public awareness of nutritional facts” (Hsu & McDonald, 2002, p.21). The present study will focus on a multiple celebrity endorsement strategy used for a brand that produces consumer goods.

4.4. Repetition in a celebrity endorsement context

Establishing a link between a brand and a celebrity can be explained by the theory of associative learning. A well-known example of associative learning is classical conditioning. Repeatedly pairing two stimuli together is an essential step in the process of building an associative link between two stimuli. In the context of celebrity endorsement, a company tries to establish an associative link between the celebrity and the brand by pairing the two together in communications towards the consumer. This way, companies aim that the associations a consumer has of the celebrity will be transferred to the brand. Repetition is a determent factor in the process of establishing and strengthening the link between the celebrity and the brand (Till, 1998). “Repeated pairings of the endorser with the brand increases consumers’ recognition that the brand is a good predictor of the presence of the celebrity, strengthening the link between the brand and the celebrity” (Till, 1998, p.402).

21

The study of Rice et al. (2002) researched the effect of multiple celebrity endorsement on brand attitude. The results of this research showed that using multiple celebrity endorsers has a positive influence on brand attitude. However, it cannot be excluded that the results they found were caused by a difference in repetition. In their research, the participants were divided into a single endorser

condition and a multiple celebrity condition. The participants in the latter condition were shown three ads, with three different endorsers, while the participants in the ‘single celebrity condition’ were only shown one ad with one endorser. Therefore, one cannot distinguish between effects driven by the difference in number of endorsers and those driven the difference in number of exposures (repetition). Therefore, it is interesting to look into the phenomenon of multiple celebrity endorsement, while taking the role of repetition into account.

4.5. Endorser brand fit and multiple celebrity endorsement

Even though multiple celebrity endorsers might be beneficial for a brand, selecting multiple celebrity endorsers should be managed carefully. Using multiple celebrity endorsers often means that consumers also see ads with celebrity endorsers that are meant for different audiences. When these celebrity endorsers do not have anything in common, consumers might get confused. In the example of L’Oreal, the target audience of Blake Lively consists of girls and young women between 15 and 30 years old. This target audience might also see ads of L’Oreal with Diana Keaton, which are not necessarily meant for them. However, both Blake Lively and Diana Keaton are beautiful women that always look flawless and neat and for that reason, both have a fit with L’Oreal. In this case, the target audience of Blake Lively for L’Oreal will probably not get very confused when they see an ad with Diana Keaton, due to their

overlapping associations. However, if they would see an ad with an endorser that has nothing in common with Blake Lively and does not have a fit with L’Oreal, they could get confused. Therefore, it is

22

important that every single celebrity endorser has a fit with the brand to prevent damage to the brand image.

Rice et al. (2012) also researched the importance of fit in the field of multiple celebrity endorsement. They showed that under low involvement conditions, brand attitude becomes more favorable with multiple endorsers. Under low involvement conditions, the participants process the message via the peripheral route. Multiple celebrities create more attention and more associations linked to the brand. Under high involvement conditions, this positive effect was only found when there was a fit between the endorsers and the brand. As was shown in the previous paragraph, the

participants that were high in involvement processed the message via the central route. If a participant is very involved and sees ads with multiple endorsers, whereby an endorser does not fit the brand, the participant will notice this, which causes the fact the brand attitude will not become more favorable with multiple celebrity endorsers.

4.5.1. Type of fit

The research of Rice et al. (2012) used three different endorsers in the ‘multiple celebrity endorser category’: Pierce Brosnan, Sean Connery and Anthony Hopkins. They chose these three celebrities because they all express ‘sophistication’. However, as mentioned in the introduction, if every celebrity endorser communicates the same story, what is then the added value of using multiple celebrity endorsers? It could be a solution for audience boredom. However, would it not be more valuable for a brand to use multiple celebrity endorsers to communicate different elements of the brand? In other words, would it not be more valuable to use endorsers that complement each other instead of

endorsers that express exactly the same association of the brand? The present study looks into the type of fit in the context of multiple celebrity endorsement. The following chapter explains the hypothesis of this research.

23

5.1. Hypotheses

Only a few studies have researched the field of multiple celebrity endorsement. Tripp, Jensen and Carlson (1997) researched multiple exposures to message sources. The results of their research showed a negative relationship between number of exposures to a single celebrity and ad attitude, which is explained by consumers getting tedious when constantly being exposed to the same endorser. If the attitude towards the ad is negative, it is likely the effect will translate into a negative brand attitude (Shimp & Yokum, 1982). By using multiple endorsers across different ads, tediousness could be

postponed. Furthermore, Hsu & McDonald found three reasons for companies to use multiple celebrity endorsers instead of only a single endorser. Rice et al. (2012) stated that under certain conditions, brand attitude is more favorable when using multiple endorsers versus a single endorser. In line with these findings, the following hypothesis is formulated:

H1: Brand attitude is more favorable when a consumer is exposed to two endorsers versus one endorser.

The research of Rice et al. (2012) showed that when there is a high endorser-brand fit, using multiple celebrity endorsers instead of a single celebrity endorser results in an increase of the brand attitude. However, their research shows something interesting in their study design. The study of Rice et al. (2012) used two conditions: in the first condition, the participants were shown one ad with one endorser, in the second condition participants were shown three ads with each a different endorser. This means, there was a difference in exposure between the two conditions. This could have been a possible confounder in their manipulation. Therefore, it is arguable that the effect that was found relates to the difference in the number of endorsers. The research of Rice et al. (2012) cannot rule out

24

the fact that the effect that was found could also be caused by the difference in the number of exposures.

When examining the theoretical base of multiple celebrity endorsement, one must carefully study previous research that could help explain why companies would pick multiple celebrities to endorse their brands. One phenomenon that is important in this context is advertising repetition, which can cause ‘wear in’ and ‘wear out’. ‘An ad is said to have worn in at a particular level of repetition if, when consumers are exposed to it, it has a significant positive effect on them’ (Pechmann & Stewart, 1988; p.286). If an ad ‘wears in’, it happens immediately after a single exposure. Furthermore, after every exposure, favorable thoughts and brand attitudes increase and peak at three exposures (Cacioppo & Petty, 1979; Pechmann & Stewart, 1988). However, when the number of exposures keeps increasing, ’wear out’ sets in because of tedium, which results in a decrease in favorable thoughts and brand attitudes (Pechmann & Stewart, 1988). In other words, they state that there is an inverted

U-relationship between number of exposures and brand attitude and favorable thoughts. Using a single endorser for a series of ads will initially enhance brand attitudes, but after three exposure decrease brand attitudes because of tedium. An illustration from the article of Pechmann & Stewart (1988) is provided in Figure 4 to get a better understanding of the phenomena ‘wear in’ and ‘wear out’.

25

In the study of Rice et al. (2012), brand attitude was measured after a single exposure to an ad with one endorser (condition A) versus three exposures to three ads with each a different endorser (condition B). As shown in Figure 4, positive thoughts outweigh the negative thoughts after three exposures, while negative thoughts outweigh the positive thoughts after one exposure. Based on the phenomenon ‘wear in’, this study suggests that a part from the difference in number of endorsers, brand attitudes would possibly have been higher in condition B anyway. Therefore, the following hypothesis is formulated:

H2: Brand attitude is more favorable when a consumer is exposed to an endorser twice versus when exposed to an endorser only once.

The importance of fit is a widely researched topic within the celebrity endorsement context.

Furthermore, literature from brand extensions explains that a poor fit between a parent brand and an extension may not only result in less positive associations of the brand, but also actually stimulate undesirable associations about the brand. If there is a misfit, consumers may see the extension as humorous or ridiculous (Aaker & Keller, 1990). For example, when a clothing brand starts building mobile devices, a consumer may question their expertise in making quality clothing. This study suggests that the brand extension theory is applicable to multiple celebrity endorsement as well. This is in line with the research of Hsu & McDonald that claim that every celebrity endorser should have a fit with the endorsed brand. Therefore, the following hypothesis is formulated:

H3: When consumers are exposed to two endorsers, brand attitude is higher when the two celebrity endorsers have a fit with the brand versus a misfit with the brand.

26

This research zooms in on the type of fit endorsers have with the brand. In their multiple endorsement condition, Rice et al. (2012) used three endorsers that all added the same association to the brand. The participants of the multiple celebrity condition saw Pierce Brosnan, Anthony Hopkins and Sean Connery in three separate ads for a watch brand, whereby all three endorsers added the association of

‘sophistication’ to the brand. An interesting question then is: ‘What is the benefit of using multiple endorsers instead of a single endorser if they all add the same association to the brand? Would it not be even more effective if the multiple endorsers are used to add more than one positive association to the brand?’ This study will try to answer that question.

Literature about brand extensions can help explaining the different dimensions of fit between celebrity endorsers. Aaker and Keller (1990) divided fit between a parent brand and the extension into three measures: substitute, complement and transfer. The first two measures take a demand-side perspective, while transfer takes a supply side perspective. The two dimensions substitute and complement can be relevant in explaining the difference in fit between endorsers. Substitute fit in the brand extension context can be defined as ‘the extent to which consumers see two product classes as substitutes’. The second dimension, complement fit, can be defined in the brand extension context as ‘the extent to which consumers see two product classes as complements’ (Aaker & Keller, 1990). In the celebrity endorsement context, an additional celebrity endorser can be a substitute of the first endorser, when they add the same association to the brand as the first endorser. Because the three endorsers that are used in the research of Rice et al. (2012) all add the same association to the brand, the endorsers can be seen as substitutes. However, a second endorser can also be complementing the first endorser, when it adds a different, positive association to the brand. Take Puma for example, they have Usain Bolt and Rihanna as endorsers for their brand. Usain Bolt probably adds the association of expertise to the brand, which may make Puma credible as a sports brand. Rihanna adds more of a symbolic meaning to the brand with her style, as she is not a professional athlete. Therefore, Rihanna and Usain Bolt

27

complement each other as celebrity endorsers for the brand Puma. This study suggests that by adding more positive associations to the brand by complementary celebrity endorsers, may result in a bigger associative network build upon more positive associations. This is in line with the brand extension literature, where it has been proven that complementarity plays a more important role in effectiveness of brand extension evaluations compared to substitutability (Aaker & Keller, 1990; van Riel & Oversloot, 2005). Therefore, the following hypothesis is formulated:

H4: When consumers are exposed to two endorsers, brand attitude is more favorable when the celebrity endorsers possess complementary associations versus overlapping associations.

5.2. Proposed conceptual framework

Based upon the literature review and the hypotheses that are formulated, the following two conceptual models are proposed.

Model 1

Model 2

Within a multiple celebrity endorser context:

Number of

endorsers

Brand attitude

Number of

exposures

Brand attitude

Type of fit

H1

H2

H3 + H4

28

Now that the literature is thoroughly analyzed, the hypotheses are formulated, and the conceptual framework is presented, the methodology that is used for this research is explained in this chapter. First, the research design is described. Second, the qualitative pre-test is conducted and explained in order to develop the stimuli that are needed for this research. Third, the quantitative pre-test of the study is conducted to test the developed stimuli. Fourth, the main study is described.

6.1. Research design

The aim of this study is to investigate the effectiveness of multiple celebrity endorsement on brand attitude and the role of repetition and type of fit in this relationship. The literature covers a great amount of research into brand attitudes and celebrity endorsement in general. Therefore, this study uses a deductive approach.

6.1.1. Experimental design

An experimental research design was used to test the proposed hypotheses. The following three independent variables were manipulated: ‘number of endorsers’, ‘number of exposures’ and ‘type of fit’. These variables were manipulated to investigate the effect they have on the dependent variable: ‘brand attitude’. By manipulating these three independent variables, one between-subjects factor was developed: ‘endorsement strategy’, with five levels. A between-subjects factor was chosen to limit the chance of hypotheses guessing.

The variables ‘number of endorsers’ and ‘number of exposures’ both had two levels (1 vs. 2). The variable ‘type of fit’ had three levels (substitute vs. complementary vs. misfit). Table 1 shows how the levels of the three variables were divided over the five levels of the between-subject factor ‘endorsement strategy’.

29

Table 1. An overview of the distribution of the levels of the between subjects factors

Next to the between-subjects factor, one within-subject factor with two levels was used in the

experiment: ‘context’ (brand 1 vs. brand 2). That means each participant answered two questionnaires, one of each brand. By using ‘context’ as a within-subjects factor, the study is actually repeated. The reason to repeat the study by using a second brand, it to test model generalizability . This enabled us to show that the effect we may find also applies for different categories. Each participant was randomly assigned to one of the five conditions. In Table 2 an overview of the whole experiment is provided.

Table 2. Overview of the main study

To help develop the stimuli that are needed for this research, two pre-tests were conducted. First, a qualitative pre-test, in the form of interviews, was conducted in order to develop two sets of stimuli. Second, a quantitative pre-test was conducted to test whether the two sets of stimuli,

Endorsement strategy Number of _endorsers Number of _exposures Type of fit

1 1 1 -

2 1 2 -

3 2 2 Substitute

4 2 2 Complementary

5 2 2 Misfit

Endorsement strategy Context

Brand 1 Brand 2

1. End1 – 1 exposure Endorser A 1x Endorser A 1x

2. End1 – 2 exposures Endorser A 2x Endorser A 2x

3. End1 + End2 (Sub) Endorser A + Endorser B Endorser A + Endorser B 4. End1 + End2 (Compl) Endorser A + Endorser C Endorser A + Endorser C 5. End1 + End2 (Misfit) Endorser A + Endorser D Endorser A + Endorser D

30

developed in the qualitative pre-test, were actually reliable enough for the main study. Finally, the main study was conducted in the form of an experiment. The proposed hypotheses from the latter chapter were tested in this experiment.

6.2. Qualitative pre-test

The first pre-test was a qualitative pre-test in the form of interviews. Four female and four male participants were interviewed, all within the age of 20 and 30. An informal discussion about various brands and endorsers was held between the participant and the researcher. The objective of this pre-test was to come up with two sets of stimuli, so that the research could be conducted twice with two different stimuli sets. Each set of stimuli consisted of one brand and four celebrity endorsers, all meeting the requirements that are provided in the following two sections. All participants were informed about the stimuli requirements.

6.2.1. Brand requirements

The brands that were used in the main study had to be real brands, because the study then could employ the associative networks the brands already had. To use those associative networks, the brands had to be known by the participants. If the participants would not have known the brand, the fit between the endorser and the brand could not have been judged properly. Furthermore, the brands had to appeal to male and female participants to the same degree (unisex). The brands had to have two clear associations (Association A and Association B) linked to the brand. For the sake of this research, those two associations were clearly distinguishable, as this research investigates if two endorsers can express two different associations that are linked to the brand. To make it a realistic situation, the brands had to not have an extreme connection to another celebrity endorser already. Furthermore, the brand had to have a positive image. If consumers would have a negative attitude towards the brand, it

31

would not have been possible to research if the fit were the determinant of the negative attitude or not. Finally, the reason to use two sets of stimuli was to show whether the results we may find also apply to different categories. Therefore, we picked the two brands in two different categories. In the following section, the requirements for the four endorsers are provided.

6.2.2. Endorser requirements

First, each set of stimuli consisted of four celebrity endorsers (Endorser A, Endorser B, Endorser C and Endorser D). For the same reason as for the brand, all four endorsers had to be known celebrities that in a broad sense had a positive image. Furthermore, the selected celebrities had to be not seen as

endorsers for the selected brands already. This could have influenced the research, if a connection between the associative networks of the endorser and the brand already had existed in the mind of the participants.

Three of the four endorsers (Endorser A, Endorser B and Endorser C) had to have a high fit with the brand. The fourth endorser (Endorser D) had to have a low fit with the brand. Furthermore, two of the endorsers (Endorser A and Endorser B) had to express one (Association A) of the two associations of the selected brand and not the other (Association B). Endorser A and endorser B had to be of the same sex. The third endorser (Endorser C) had to express Association B and not Association A. The fourth endorser (Endorser D) had to express neither of the two associations (A or B), but a third association (Association C), which was not expressed by the brand or the other three endorsers.

6.2.3. Outcomes qualitative pre-test

The goal of the qualitative pre-test was to come up with two sets of stimuli, including both one brand and four endorsers that met the requirements described above, that could be used for the final experiment. The stimuli that were created are shown in Table 3.

32

6.2.3.1. Chosen brands and associations

The two brands that were selected for the experiment were fashion brand ‘Ralph Lauren’ and watch brand ‘Omega’. Multiple brands from different categories were discussed during the pre-test, such as electronic brands, car brands, food brands and liqueur brands. It proved to be difficult for participants to distinguish two clear associations for brands in these categories. It appeared that many brands only had one clear association or the brand had two associations that overlapped each other. The two identified associations for ‘Ralph Lauren’ were Classiness (Association A) and Fashion (Association B). For Omega, the two associations were Luxury/Status (Association A) and Time (Association B). For both of the chosen brands, one association had a symbolic character (Classiness and Luxury/Status) and the other had a functional character (Fashion and Time). The main association of the endorsers that represented the misfit (Ed Sheeran and Jamie Oliver) were Singing (for Ed Sheeran) and Cooking (for Jamie Oliver). The qualitative pre-test showed that Singing was not associated with Ralph Lauren and that Cooking was not associated with Omega.

6.2.3.2. Chosen endorsers Ralph Lauren

For the first stimuli set the brand, ‘Ralph Lauren’ was chosen. The four endorsers of the first stimuli set that came out of the qualitative pre-test were Colin Firth (Endorser A), Jude Law (Endorser B), Kendall Jenner (Endorser C) and Ed Sheeran (Endorser 4). According to the qualitative pre-test, both Colin Firth and Jude Law expressed Classiness and did not express Fashion or Singing. Kendall Jenner expressed

Brands Endorser A Endorser B Endorser C Endorser D

1 Ralph Lauren Colin Firth Jude Law Kendall Jenner Ed Sheeran

33

Fashion and did not express Classiness or Singing. Finally, Ed Sheeran did not express Classiness or Fashion, but did express Singing. Table 4 provides an overview of the brand, the endorsers, and the associations they did and did not express.

Table 4. Overview Ralph Lauren and proposed endorsers and their associations

6.2.3.3. Chosen endorsers Omega

For the second stimuli set the brand, ‘Omega’ was chosen. The four endorsers of this stimuli set were: Julia Roberts (Endorser A), Emma Watson (Endorser B), Usain Bolt (Endorser C) and Jamie Oliver

(Endorser D). Both Julia Roberts and Emma Watson expressed the first association Luxury/Status and did not express the association Time or Cooking. Usain Bolt expressed the association Time and did not express Luxury/Status or Cooking. Finally, Jamie Oliver did not express Luxury/Status or Time, but did express Cooking. Table 5 provides an overview of the brand, the endorsers, and the associations they did and did not express.

Table 5. Overview Omega and proposed endorsers and their associations

Classiness Fashion Singing

Ralph Lauren + + -

Colin Firth + - -

Jude Law + - -

Kendall Jenner - + -

Ed Sheeran - - +

Luxury/status Time Cooking

Omega + + -

Emma Watson + - -

Julia Roberts + - -

Usain Bolt - + -

34

During the qualitative pre-test, it became clear that the symbolic association of the brand was stronger compared to the functional association. For Ralph Lauren, this was Classiness (versus Fashion); for Omega, this was Luxury/Status (versus Time). As shown in Table 2, the main study contained of different conditions where some participants saw one endorser and some participants two endorsers. However, per stimuli set, one endorser (Endorser A) was shown in every condition. That endorser expressed the strongest association: the symbolic association. For Ralph Lauren, we chose Colin Firth over Jude Law as he scored higher on Classiness during the qualitative pre-test. For Omega, we chose Julia Roberts, as she scored higher on ‘status/luxury’ compared to Emma Watson. That means that Colin Firth and Julia Roberts were shown in all conditions of the experiment. Based on the results of the qualitative pre-test, couples were made that represented the type of fit between endorsers. The proposed couples are provided in Table 6.

Table 6. Proposed endorser couples

6.3. Quantitative pre-test

A quantitative pre-test was conducted to test the stimuli that had been created based on the qualitative pre-test. More specifically, the quantitative pre-test tested the stimuli that are given in Table 4 and Table 5. Furthermore, the quantitative pre-test tested the following as well:

• Endorser A, B and C all have a fit with the brand and endorser D does not have a fit with the brand;

Type of fit Ralph Lauren Omega

Substitute fit Colin Firth & Jude Law Julia Roberts & Emma Watson Complementary fit Colin Firth & Kendall Jenner Julia Roberts & Usain Bolt

35

• The brands and endorsers used in the experiment are known;

• The brands and endorsers used in the experiment do not have a negative image in general.

6.3.1. Sample, data collection and data analysis

To quantify the points above, a questionnaire was sent out to 40 participants. The ‘context’ (Ralph Lauren vs. Omega) was the in between-subjects factor, whereas the ‘endorser’ was the within subjects factor (End A, B, C and D). The questionnaire was made using Qualtrics and sent out to participants through email. The data was collected and analyzed with the use of SPSS. An example of the pre-test questionnaire can be found in Appendix 1.

6.3.2. Measures quantitative pre-test 6.3.2.1. Manipulation checks

Whether the brands and the endorsers expressed the right associations was tested with the following measure: I associate [brand or endorser} with […}. Every association was tested with three items. An overview of the measures is provided on the next page in Table 7.

Table 7. Overview manipulations checks associations

6.3.2.3. Control variables Attitude towards the brand

A great amount of literature has been written about the construct brand attitude. In both the pre-test as

Association

Classiness Fashion Singing Status/Luxury Time Cooking

Classiness Clothing Music Class Time Cooking

Traditional Outfits Songs Prestige A clock Food

36

the main study, brand attitude was measured with the help of three items, on a 7-point Likert scale ranging from strongly disagree to strongly agree. The three items were: ‘I think Brand X is a good brand’, ‘I like Brand X’ and ‘I am positive towards Brand X’ of Spears and Singh (2004).

Attitude towards the endorser

The endorsers that had been chosen had to be generally liked by the participants. When participants would have had a bad attitude towards the endorsers, it could have influenced the attitude towards the brand. This could have biased bias the experiment. Attitude towards the endorser was measured with three items, on a 7-point Likert scale ranging from strongly disagree to strongly agree. The three items were based on the items that measured the brand attitude: ‘I think Endorser X is a good man/woman, ‘I like Endorser X’ and ‘I am positive towards Endorser X’.

Brand and endorser familiarity

Two variables that needed to be controlled for were ‘brand familiarity’ and ‘endorser familiarity’. The brand and the endorsers that were chosen had to be known by the participants, to make a valid judgment of the collaboration between the brand and the endorser(s). Both variables were measured with one item on a 7-point Likert scale ranging from strongly disagree to strongly agree.

6.3.3. Results Ralph Lauren

To test whether the stimuli met the requirements, for every brand and endorser a One Sample T-Test and a One-Way Repeated Measure ANOVA were performed. Because all of the variables were measured on 7-point Likert scales, the means of the scores were compared to the average, 4. All of the One Sample T-Tests tested whether the mean was significantly smaller or larger than 4. Table 8 shows the expectations, the means, the standard deviations and the significance of the results of the One Sample T-Test for the first stimuli set.