Perception becomes reality : the impact of a celebrity endorser scandal and the effect of a powerful response on consumer attitudes towards the brand

(1)

1

Faculty Economics and Business

“Perception becomes Reality”

The Impact of a Celebrity Endorser Scandal and the Effect of

a Powerful Response on Consumer Attitudes towards the Brand

Author: Isabelle van Doorne (10455639)

MSc in Business Administration-Marketing Track

Date: 17.12.2014

1st_{Supervisor: Drs. Jorge Labadie MBM} 2nd_{Supervisor: Drs. R.E.W Pruppers}

(2)

2

STATEMENT OF ORIGINALITY

This document is written by Isabelle van Doorne who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

ABSTRACT

The use of celebrity endorsement has become very popular in the last few decades. Unfortunately, more and more of these icons in our society have been become involved in some kind of scandal (Fong and Wyer, 2012). This could have a significant impact on consumer attitudes and therefore also on brand equity. Does the effect of this scandal depend on the type of scandal (cheating vs doping) and the type of brand (family vs sport)? And what can a brand do to minimize the potential negative impact of this scandal on consumer attitudes towards the brand? This study focuses on the impact of a scandal, and how a brand can best respond to this scandal in order to increase the effect on consumer attitudes. An experiment was conducted where the participants were exposed to a news feed about the scandal and later about the brand response to this scandal. The findings show that a scandal has a negative impact on consumer attitudes towards the brand, regardless of the type of scandal and type of brand. Furthermore the findings show that attitudes towards the brand can become more positive by responding effectively. When responding, the brand should always keep its core values in mind and when these are inflicted, it is better to discontinue the contract with the celebrity endorser. This study concludes by providing some interesting recommendations for future research.

(4)

4

1. INTRODUCTION

When reading or watching the news these days, it is hard to ignore the involvement of celebrity endorsers in scandals. Many icons in our society have become involved in scandals such as the use of doping, adultery and violence. This could be because they are extensively reported and discussed in popular media (Muda et al, 2013). The first celebrity magazine was launched in the US in 1916 and started to reveal the life of high society and A-listers (www.adage.come). Now, celebrity news is so popular that

“MailOnline has grown into one of the most visited newspaper websites in the U.S. and the world, through a mix of splashy stories about celebrities, scandals and breaking news” (www.adage.com). Celebrities are under the magnifying-glass of our society and are a constantly “hunted” like a rare species bij journalists world wide. On the other hand news travels very fast these days (social media networks). Celebrities are valuable instruments for corporations to enhance their image, but are also a risk concerned the above mentioned issues.

In general, companies invest millions of dollars in endorsement contracts, but is this worth the investment and money? Does it really bring what they expect? Reading about this more and more, my fascination for this topic started to grow. In addition, working for a few multi-national corporations (Unilever and L’Oréal) who invest heavily in the use of celebrities to be the face of their brand, made this thesis an interesting and challenging case for me. It made me realise there is more than just the scandal these people generate, it must have a higher impact around them. How is the brand they are representing dealing with it? Subsequently, what is the consumer thinking about the brand? How important is it to the consumer what the brand’s reaction will be to the scandal? Do they even bother? These questions remain relatively unanswered until now, but everyone has an opinion about it. In this thesis I will give support what, or not, the impact of a celebrity endorser scandal is and what the benefit is of the right response from the brand. It all has to do with managing perceptions, as William Shakespeare (1564-1616) already said a few centuries ago. Managing perceptions is an art that, when done right, may have an important impact on the brand’s image. That’s what this thesis is all about.

(7)

7 Celebrity endorsement

Throughout the history of advertising, many companies have started to use endorsers to be the face of their brand. Especially the use of celebrity endorsement has become very popular in advertising over the last few decades. Some examples of these brands are Nike (Tiger Woods), Pepsi (Beyoncé), H&M (David Beckham), Chanel (Kate Moss), L’Oréal (Eva Longoria) and Nespresso (George Clooney).

It is however not a recent phenomenon since celebrities have been endorsing brands since the late nineteenth century (Erdogan, 1999). One of the examples from these early days involves Queen Victoria in association with Cadbury’s Cocoa (Erdogan, 1999). This type of advertising started to increase rapidly and in 1979 one in every six

advertisements used a celebrity (Erdogan, 1999). Today, celebrity endorsement is booming. A Global Consumer Trends Survey from Euromonitor revealed that for 10% of the consumers, celebrities are their key role model in life (Euromonitor International, 2014). This could explain why a company like Nike has invested $909 million on endorsements contracts in 2012 (Euromonitor International, 2014).

Celebrities are used in advertising campaigns to stimulate the selling of a product or service (Kamins & Gupta, 1994). In today’s highly competitive environment, the use of celebrity endorsers can give a competitive advantage in differentiating the firm from competitors (Doss, 2011). “Celebrities are believed to enhance the subject’s

attentiveness to the ad, make the copy more memorable, credible, and desirable, and effectively glamorize the product” (Kamins & Gupta, 1994: 569). Another study by Atkin and Block (1983) shows that a celebrity in advertising leads to more positive product evaluations (Atkin & Block, 1983). It can also be said that (celebrity) endorsers have benefits such as increasing awareness of the brand, message recall and higher sales (Bailey, 2007).

Brands & Celebrity Endorser Scandals

Famous endorsers include Tiger Woods, Lance Armstrong, Kate Moss, Michael Phelps, and many more. They are, or have been the face of strong brands such as Nike, Omega, Burberry, Chanel, and Rolex. Despite this, however, these endorsers all have something in common. They recently have been involved in a public scandal. Kate Moss for example

(8)

8 was photographed using hard drugs, Lance Armstrong was accused of using

performance-enhancing drugs, and Tiger Woods was accused of adultery. The involvement of these endorsers in major scandals could be a huge drawback for the companies they represent as being an endorser. People’s image and confidence in the brand can be destroyed in a blink of an eye, and the companies can suffer major losses (Till & Shimp, 1998). But is this also the case for the impact on consumer attitudes towards the brand? After Tiger Woods’ scandal, Nike suffered major losses and these losses did not stop until the date that Tiger Woods announced his retirement in golf (Knittel & Stango, 2010). After having invested millions of dollars in the endorsement deal, the companies are faced with the important decision to either continue the sponsorship or to disassociate from the endorser.

Michael Phelps is the face of several brands such as Omega, Speedo, Visa and Kellogs. When photos of him smoking marijuana showed up in the press, they spread around the world quickly. Despite this scandal, Omega, Speedo and Visa decided to stick with

Michael Phelps, whereas Kellogs did not (Advertising Age, 2009). This was because, according to their website, Michael Phelps’ image is not consistent anymore with the image of Kellogs (www.kelloggs.com). When looking at Nike, they decided to stick with their endorser Tiger Woods, but they did not continue their endorsement deal with Lance Armstrong. Tiger Woods was also a brand endorser for Gilette at the time of his scandal, but they ended the contract after hearing about his engagement in extramarital affairs. So Gilette ended their contract with Tiger Woods, whereas Nike continued the sponsorship.

These examples of brand responses show that it is interesting to find out under which condition a brand should (dis) continue the endorsement deal to avoid negative

spillover effects. The fact that some companies have begun to use specialised insurance plans in their endorsement contracts could show that celebrity endorsement could be risky (Euromonitor International, 2014).

The aim of this thesis is to find out under which conditions celebrity endorsers damage a brand’s image the most and whether the brand should continue, or discontinue their endorsement contract. Little research has been done about the fit between a type of scandal and a type of brand, so it will therefore be interesting to link a specific type of

(9)

9 scandal to a specific type of brand, and see if consumers’ attitudes differ per link. This so-called brand-scandal fit can be of importance for companies when they make a continuation decision with regard to the endorser scandal. Ofcourse each company makes its own decision regarding the scandal. Michael Phelps for example lost Kellogs as a sponsor, but Visa, Speedo and Omega decided to keep him as the face of their brand. Why is that? What makes the one stop and the other continue the contract? Does the decision to stop sponsoring depend on the type of scandal? And how does the business management decision influence the attitude to the brand of the consumer? It is a difficult decision for companies to take and it may depend upon several factors. This thesis will research the relevance of fit between the brand and the scandal for the consumer’s attitude towards the brand. Furthermore, it will be researched under which conditions a brand’s (dis) continuation response, can limit a potential decrease in customer attitude towards the brand.

The following research question has been formed:

Can a brand minimize the negative impact on consumer attitudes when they respond effectively to a celebrity-endorser scandal?

A considerable amount of research has been conducted focusing on the positive sides of celebrity endorsement, such as increasing awareness of the brand, message recall and higher sales (Bailey, 2007). Exisiting research also shows the impact negative

information can have on a brand and evaluations of a brand (Till & Shimp, 1998). This thesis will contribute to existing knowledge about celebrity endorsement and its risks, by focusing on the impact of a scandal on consumer attitudes and how the brand can best deal with this scandal to limit a potential decrease in these attitudes.

Due to a limit of scope and time, this thesis is narrowed down to the following starting points:

 one celebrity endorser, endorsing two brands, involved in two types of scandals  change in consumer attitude towards the brand after the scandal

(10)

10 Not only theoretical contributions are made, but this thesis may also lead to managerial or strategical contributions. This thesis can serve as a supportfor companies who are considering the use of endorsers for their brand. These firms should well keep the historical facts of endorser scandals, in mind. Celebrity endorsement is proven helpful, but can also involve risks. It is important for managers to know how they externally communicate and should deal with endorsement scandals. Should they terminate the contract? Or is it better to continue the contract? Does the type of scandal influence this decision? How will the consumers react to their decision? The outcome of this research can help firms in making the right decisions regarding their endorser in order of

(11)

11

2. LITERATURE REVIEW

This study is focused on the impact of a celebrity endorser scandal on brand equity. The increasing use of celebrities as brand endorsers makes it an important research topic since many of these endorsers are often engaged in some sort of scandal. Especially the effect a scandal can have on brand equity is a highly interesting topic to investigate. Because potentially, brand equity could be reduced when the brand-endorser is involved in a scandal.

The literature review will first discuss prior academic work on the importance of brand equity, and how brand equity is used to create an associative network in the mind of the consumer. Then the concept of celebrity endorsement is reviewed as well as risks involved when using endorsers.

Brand Equity and Celebrity Endorsement

Customer-Based Brand Equity

Prior academic work has highlighted the importance of brand equity in creating a strong brand. Many academics and practioners have regarded brand equity as an important concept. Brand equity arises from the marketing that is done by a particular brand, and the different consumer responses to this marketing (Keller, 1993). Consumers respond differently to the marketing efforts of a brand because they all have their individual experiences and knowledge about the brand (Keller, 2012). The power of a brand

depends highly on what is going on in the mind of the consumer, it is how he perceives a brand and how that perception was built. In his book, Keller (2012) has introduced the customer-based brand equity (CBBE) model, which is a model that highlights the importance of consumers and their perspectives in building brand equity. How a consumer thinks about a brand is highly important and marketers have to ensure that consumers have the right experiences to form positive opinions about a brand, which then may result in buying the brand and being loyal to the brand. Keller (2012) defines CBBE as “the differential effect that brand knowledge has on consumer response to the marketing of that brand” (Keller, 2012: 54). When looking at this definition, it can be noticed that brand knowledge is an important influencer of responses to the marketing of a brand. It is the key to building a strong brand, as it exists in the consumers’ memory as a positive associative network. Every person has its own associative network in their

(12)

12 brain, consisting of stored information and linkages between this stored information. Keller (1993) identified three types of brand associations. These are 1)brand attributes 2)brand benefits and 3)brand attitudes. Brand attributes refer to what the product has to offer to the consumer. Brand benefits refer to what people think a product can do for them. And brand attitudes are the overall evaluations of the brand (Keller 1993). This research will focus on the third type of association, i.e. brand attitudes.

CBBE occurs when associations in the mind of the consumer are strong, favourable and unique (Keller, 2003). Brand knowledge, which is central in building CBBE, is of high relevance for building a strong brand because people rely on what they know about the brand and how this knowledge is stored in their memory.

Brand knowledge consists of two components, which are 1)brand awareness and

2)brand image (Keller, 2012). Brand awareness is high when someone is able to identify a brand in different conditions and situations. For example brand image is the people’s perceptions about a brand. These perceptions result in certain consumer behaviours and the image must be positive to build brand equity.

When awareness of a brand is high, and the brand image consists of strong, favourable and unique associations, positive CBBE occurs (Keller, 2012). When CBBE is positive, consumers will for example be less sensitive to an increase in price and will be more loyal to the brand. They react more favourable to a brand and to its marketing activities, which may ultimately result in true loyalty (Buil et al, 2013). Buil (2013) introduces the traditional hierarchy of effects model. According to the author brand equity is built as result from the consumers learning process. Consumers learn about a brand and this learning process starts by forming brand awareness. Consumers’ awareness leads to attitudes and these attitudes may or may not result in brand loyalty (Buil et al, 2013). Advertising plays an important role in the consumer learning process, and thus also in building brand equity. So marketers are faced with the challenge to develop strong and effective marketing campaigns in order to create a brand as strong as possible in a highly competitive market.

It can already be concluded that in order to build a strong brand, marketers must manage CBBE so that it creates strong, favourable and unique associations in the mind of the consumers. The main driver of CBBE is brand knowledge.

(13)

13 Building brand equity

There are several ways to build CBBE. Firstly, the marketer has to choose the right brand elements. Brand elements include brand name, logo, slogan, jingle or character. Secondly, marketing strategies should be created that are consistent with the brand elements and how they want to be perceived as a brand. The third way is the use of secondary brand associations (Keller, 2012). Secondary brand knowledge can be created by linking the brand to another entity, such as other geographical areas, events or to a celebrity endorser. Within this research the focus will be on celebrity

endorsement and how risks involved can affect brand knowledge. When linking the brand to another entity it can evoke new associations, judgments and feelings in the consumers mind. What people know about the entity can be leveraged to the brand and this is a highly cognitive process (Keller, 2012). However, several factors are important in order to ensure the leverage of associations between the brand and the entity. The consumer must have awareness and knowledge about the entity, the entity must be meaningful to the consumer, and this knowledge should be easy to transfer (Keller, 2012). If at least one of these factors is present, the consumer will start to transfer associations.

With this in mind, it could be explained why celebrity endorsement is so popular. A celebrity is well-known and people often identify to them easily and look up to them. This will increase the likelihood that the consumer will form associations and link these to the brand. Especially in low involvement situations, where the consumer is not very motivated or able to process an advertisement, celebrity endorsement makes it more easy and interesting to process the advertisement and create associations and feelings (Petty & Caciappo, 1979).

Secondary associations may be very useful to the brand since they create associations that may otherwise not be present in the mind of the consumer (Keller, 1993). Celebrity endorsement is one of the ways to do this and it is also a very popular one. From all the commercials, around 20% of them use celebrities to promote the brand (Bradley, 1996). Some examples are Beyonce for Pepsi, Brad Pitt for Heineken, Nicole Kidman for Chanel, Georges Clooney for Nespresso and Lance Armstrong for Nike. A famous person has the ability to draw attention and provoke associations, especially in low involvement

(14)

14 situations (Petty et al., 1983). As noticed earlier, an associative network structures the memory of the consumer, and brand and celebrities are nodes in this network

(Martindale, 1991). When a link between these nodes is activated the brand and the entity become part of each other’s association set and play a role in how the brand is perceived and how consumers think about the brand. Associations of the celebrity become relevant to the brand and vice versa (Keller, 1993). It would be ideal if these associations are strong, favourable and unique since this contributes to creating positive brand image. Because in the end, a positive brand image has an impact on the brand knowledge which is key to building brand equity.

Risks of using celebrity-endorsers

Much research has been done about the effectiveness of using a celebrity endorser. Friedman and Friedman (1979) found in their research that celebrity endorsement success varies per product. From this view, the importance of fit was captured and is now referred to as the “match-up hypothesis” (Kamins, 1990). The fit between an endorser and a product is highly important. If a consumer cannot relate the endorser with the product, they are unlikely to think about the combination and form associations or judgments or feelings, and it will not contribute to their brand knowledge. Positive effects will only be driven when it is easy for someone to integrate similar concepts in their associative network (Till & Busler, 2000). Kamin (1990) studied the match-up hypothesis and found that it is more effective to use an attractive celebrity to endorse a product that enhances once attractiveness (i.e. cosmetics). The attractive celebrity has status and will increase the product its credibility (Kamin, 1990). However, Till and Busler (2000) conducted a study and found that expertise is even more important than physical attractiveness. This explains that famous sports athletes are used to endorse brands such as Nike, Adidas or Aesics. Their status as a great athlete and their significant expertise helps to make the product more credible. The consumers’ cognitive process considers that if the endorser uses this brand, it must be a good brand (Friedman & Friedman, 1979).

There are however a lot of risks involved in using celebrities as brand endorsers. It could be for example that a celebrity is endorsing so many different products that consumers get confused because the celebrity lacks product meaning (Keller, 2012).

(15)

15 Secondly, a marketer never knows if consumers see a match between the celebrity and the product. If consumers fail to see the match, the endorsement deal will not create a positive brand image (Keller, 2012). The third risk, and a highly relevant one in this research, is the risk that celebrity endorsers get into trouble or get involved in a scandal which significantly decreases the marketing value to the brand (Keller, 2012). A

marketer has hardly any control about the life of the endorser, so the chances are high that something will happen that will affect brand image, either in a good or a bad way. This is also a risk of using secondary associations to build knowledge. There is a lack of control about what happens to the entity and how this in transferred back to the brand in the mind of the consumer. A marketer can only control how they want to be perceived as a brand, but one will never know if this is really also how people perceive the brand in their minds.

Scandal involvement and impact

Celebrity Endorser Scandals

As stated in the previous paragraph, Keller (2012) defined that one of the risks for companies, that use celebrity endorsers, is that these endorsers can get involved in a scandal. The word scandal is used “to describe a sequence of actions and events which involve certain kinds of transgressions and which, when they become known to others, are regarded as sufficiently serious to elicit a response of disapproval or condemnation” (Thompson, 2000:4).

This research will focus on celebrity scandals, which pertain to famous individuals and its misbehaviour (Thompson, 2001). According to Geyser (2009), spokesperson misbehaviour and controversy is one of the sources that could harm a brand. Because ultimately, negative brand attitudes can damage the brand equity (Dawar & Pillutla, 2000).

Highly visible examples of endorsers involved in a scandal are Lance Armstrong, Tiger Woods, Oscar Pistorius, Kate Moss, Justin Bieber and many many more. These scandals range in types from ‘sport-scandal’ (Lance Armstrong) to ‘sex-scandal’ (Tiger Woods) and ‘violence-scandal’ (Oscar Pistorius) (Um, 2013).The effect a scandal has on the endorsers is highly different from the effect the scandal has on the brand. Kate Moss for example was caught sniffing cocaine, but this did not stop brands like Christian Dior and Rimmel from using her as an endorser. Her manger even said it ‘boosted her career’,

(16)

16 because she was able to double her income over the next five years (DailyMail, 2010). Endorsement scandals and how to react to it as a brand, is not lineair science. Also a variety of interests play their part that makes how brands react, or not, to scandals. In the case of Lance Armstong it is known that he was already suspected of doping long before Nike ceased the contract with him. There was sufficient inside information available and leaked out to the press, for Nike to take a decision. In the case of Nike relatedness to a scandal was not necessarily the trigger to stop the contract, but also the repeatedness of the act, probably played a part in brand decicions to a scandal. Nike only announced the termination of their sponsorship agreement with Lance Armstrong “due to the insurmountable evidence he participated in doping and misled Nike for more than a decade” (DailyMail Online, 2012). Many consumers protested already some time before the decision of Nike to drop Lance Armstrong. This indicates consumer

involvement with brands. Timing and the way to respond, is highly important for a brand to manage consumer attitude in crisis like these.

According to Greyser (2009), the situation for a brand can be ‘life-threatening’ when a company’s marketplace position and its brand meaning are seriously challenged

(Greyser, 2009:592).If they do not take any effective action, consumers may loose trust in the brand and perceive it as being less credible. This may create negative meanings associated with the brand and has a negative impact on brand equity (Till, 1998). Again, this highlights the importance for managers to manage brand equity as an ongoing and dynamic process. Till and Shimp (1998) found in their research that when celebrities are evaluated lower, it can result in lower brand evaluations. Negative news about the celebrity can reduce the appeal of the brand that is endorsed (Till & Shimp, 1998). Additonally, as found in their studies by Campbell & Warren (2012) show that “brands are more likely to acquire the negative than the positive personality traits associated with a celebrity endorser and that negative associations transfer even under conditions that inhibit the transfer of positive associations” (Campbell & Warren, 2012: 1).

Negative information often has a greater weight and will thus have more influence on brand image (Campbell & Warren, 2012). Furthermore, the impression formation literature provides evidence that discrepant information is perceived to be more interesting as it is often unexpected (Sujan, 1985). This implies that it is also more easy to remember and will therefore have an influence on brand awareness and thus also on

(17)

17 brand knowledge. Discrepant information also gives advantages in recall (Srull, 1981). But the question is whether this advantage in recall is also an advantage for the overall brand image.

When looking at the previous research that has been done, it can be concluded that managing brand equity involves the management of the constellation of meanings associated with a brand (Aaker, 1991). In order to build a strong brand, marketers must create a high brand value as well. Celebrity endorsement is a popular marketing tool to build brand equity. This combining of two different entities, for example the brand and an endorser, should create optimal positioning in the mind of the consumer so that they form a positive attitude towards the brands. However, if one of these entities, for

example the endorser, is involved in a scandal, managers should try to minimise the impact on their brand equity. This research will try to prove what a brand reaction can do to minimise this impact, since little to no academic research has been done about this.

Negative celebrity information: image transfer and spillover

Previous literature has led to much investigation on the impact of negative reputations of celebrity endorsers on brands. La Ferle (2009) found for example that negative information influenced celebrities’ reputations and the brands they endorse. Negative information has the biggest counter productive effect on endorsement

effectiveness, compared to endorsement expertise and trustworthiness. (Amos, 2008). This can be explained through the principle of associative learning. Associative learning is a mechanism by which links and relationships between concepts are established (Till & Busler, 2000).

The building of an associative link between the brand and the celebrity brings the added value to celebrity endorsement (Till and Shimp, 1995). There is however a risk involved when establishing this link, since negative information regarding the celebrity endorser can damage the link between the brand and the celebrity (Till, 1998). The brand and the celebrity are nodes, and negative information about the celebrity will activate in the memory and provides a path for this negative celebrity information to transfer to the brand (Till, 1998). The celebrity node is activated, which then activates the brand node as well, causing for the transfer of the reduced evaluation of the celebrity to the brand (Till, 1998: 405).

(18)

18 As Keller already stated, building linkages between a brand and a celebrity has a risk, since the image of the celebrity is transferred to the brand.

Campbell and Warren (2012) also highlighted in their research the risk of meaning transfer, stating that “negative associations are more likely to transfer to a brand than positive associations” (Campbell and Warren, 2012: 172). Furthermore their studies show that negative characterics of the celebrity endorser are remembered more easily and therefore are more likely to be connected to the brand than the celebrity endorsers positive characterics (Campbell and Warren, 2012). So meaning transfer can occur with both positive and negative traits.

Research furthermore shows that negative information is weighed more heavily than positive information. The impact of negative information is determined by the level of blameworthiness (Louie & Obermiller, 2002) and scandal severity (Bednall & Colling, 2010).

Blame to the celebrities increases with the perception that the image and the career of the celebrities were ruined as a result of the scandal. A scandal has the highest impact when the celebrity is seen as blameworthy and responsible for the scandal. Also, the more severe the scandal the more it can damage consumer attitudes of both the celebrity and the brand.

Crisis Management

The previous paragraphs show that there are many risks involved when using celebrity endorsers.

In todays business environment, crises are happening more severe and more frequently (Gruber et al., 2014). One of the reasons for this according to Gruber et al. (2014) is social media and the significant acceleration in the speed at which information is shared these days. Furthermore the reach for messages is high and can be accessed by almost everyone. Gruber et al. (2014) showed with their case study that a local event can turn into a national or even global event by using social media. This shows the powerful effect social media has in spreading messages and could also be one of the reasons for crises to become more frequent and severe (Gruber et al. 2014).

A crisis is “a business or organizational problem that is exposed to public attention, and that threatens a company’s reputation and its ability to conduct business” (Petkeviciene

(19)

19 2014: 453). It is agreed, by other scholars, that it is a sudden and unexpected event that can cause a threat financially and to a company’s reputation. According to Weiner (2006) there are different forms of crises. These can include criminal acts,

environmental spills, product failure or recall, corporate or spokesperson misbehaviour, labour disruptions and many more (Weiner, 2006).

One of the characteristics of a crisis is that they are unpredictable and non-routine events (Janssen et al., 2014). When the celebrity endorser of a brand misbehaves and is involved in a scandal, it is in general also an unpredictable and non-routine event that was not foreseen by the company. When a crisis (or a scandal) hits, it is of high

importance that the company responds to this crisis and to manage public perceptions (Janssen et al. 2014).

Clays and Cauberghe (2011) state that the reputational damage depends on how

responsible the organization is for a crisis. The less responsible they are, the smaller the reputational damage will be and the more responsible, the more damage the crisis will cause. According to Coombs (2004) this damage can be controlled or minimized

following the correct public relations strategy. Saying the right words at the right time is one of the main recommendations in order for the strategy to be correct (Coombs, 2004). Coombs (2007) developed a Situational Crisis Communication Theory (SCCT).According to this theory, “when the reputational threat as the result of a crisis is mild, it is

sufficient that an organization offers objective information to minimize the reputational damage. If the reputational threat is severe, it is strongly advised that the organization integrates reputation restoring crisis response strategies into its communication strategy (Claeys & Caubertge 2011: 83).

Coombs (2007) states that the SCCT guidelines can benefit crisis managers and their organisation by using these in their crisis response strategies.

Above literature on crisis management has focused mainly on protecting organistations reputations. Griffith and Ryans (1997) highlight that global corporate communications are also of importance to protect the brand equity.

Adding to the above literature Amos et al. (2008) propose that firms should develop response plans so they are prepared in case negative information or events occur (Amos, 2008). One research shows for example that it is better to end the endorsement-contract

(20)

20 after very severe scandals, since endorsement effectiveness dropped to very low values. Furthermore, Louis & Obermiller (2002) show that companies should disassociate with endorsers who have high levels of blame for negative events they are involved in. This thesis will try to show whether it is better, or not, for a brand to continue the endorser contract and if this decision results in higher consumer attitudes towards the brand.

(21)

21

3. Hypotheses Development

The preceding chapters have discussed the risks of using celebrity endorsers when building brand equity (Keller, 2013), how negative celebrity information can easily transfer to the brand (Till, 1998) and what a brand can do to manage the involvement in a crisis (Greyser, 2009). The following chapter will discuss the main variables of this study, i.e. brand-scandal fit and brand response fit, resulting in the hypotheses. The research question, as previously stated in the introduction is: “Can a brand minimize the negative impact on consumer attitudes when they respond effectively to a celebrity-endorser scandal?”

This study will try to find out under which conditions consumer attitudes towards the brand decrease the most, and if the brand response can minimize this negative impact on consumer attitudes.

Figure 1 presents the conceptual framework of the hypotheses that will be proposed.

Figure 1- Conceptual framework

Scandal

(before vs

after)

Attitude

towards the

brand

Brand-Scandal

Fit

(High vs Low)

Brand-Response

Fit

(High vs Low)

H1 H4 H3 H2

(22)

22

Main effect of a scandal

Existing research from several scholars has proved that negative information is likely to have a negative impact on the brand. La Ferle (2009) found for example that negative information influenced celebrities’ reputations and the brands they endorse. Greyser (2009) also agrees, stating that spokesperson misbehaviour is one of the sources that could harm a brand. Furthermore Campbell and Warren (2012) highlight that negative information is more likely to transfer to the brand than positive information. This research focuses on consumer attitudes towards the brand, and it is therefore expected that the attitudes towards the brand before the scandal are higher than after the scandal. The first hypothesis can be proposed:

H1: Attitude towards the brand before the scandal will be higher than attitude towards the brand after the scandal

Figure 2: Hypothesis 1

Brand-Scandal Fit

Brand-Scandal fit is a factor for which a little to no previous scientific research has been done. It can be determined as the impact on brands that are publicly being connected to bad news in relation to the celebrity who endorses them and when the brand’s core values are at stake.

There is a high brand-scandal fit when the scandal inflicts damage to the core values of the brand. On the other hand there is a low fit when the core values are not affected by the scandal. This does not imply that a low brand-scandal fit has no impact on customer attitude towards the brand, because it could still be affected. But with a low brand-scandal fit the values of the brand are not directly inflicted.

In this thesis, brand-scandal fit is used to see if this potential fit between a brand and a scandal, has a certain influence on consumer attitude towards the brand. It tries to

Before Scandal Higher attitudes towards the brand

(23)

23 analyse if the consumer attitude differs depending on different types of scandals in combination with different types of brands. So brand-scandal fit measures the level of fit between a type of brand and a type of scandal, where the endorser of the brand is

involved in. As stated in the literature review, the type of scandal could be for example a sport-scandal or a violence-scandal (Um, 2013).

Amos et al. (2008) researched that when there is a fit between the product and the athlete as an endorser, the expertise of the athlete is the most important factor in the endorsement to be effective. On the other hand, when there is no fit between the athlete and the product, other factors such as trustworthiness and attractiveness determine the effectiveness (Amos et al. 2008).

Research from Bednall and Collings (2010) shows that if the expertise of the athlete made the endorsement effective, and his expertise is according to the customer not affected by the scandal, the athlete might be suitable to continue endorsing. In other words, there was not fit between the scandal and the expertise, so the expertise might be rated at the same level as before the scandal. Therefore, it is expected in this research that when the type of brand is different from the type of scandal, the consumer brand attitude will be less influenced.

Other previous research has shown that for athletes, expertise was affected by the sport-related scandal, where it was not affected by the non-sport sport-related scandal.

When looking at endorsement scandals, the most serious situations are those that affect the distinctive attributes/characteristics of the brand, which Greyser (2009) terms as the essence of the brand. When this essence is affected, the brand’s marketplace position and meaning is seriously challenged (Greyser, 2009).Previous literature shows that the effect on a brand will be the worst when a scandal ‘hits’ the essence of the brand’s desired identity, and the characteristics that are most closely associated with the brands meaning and success (Greyser, 2009).

Building on this previous literature from Greyser (2009), the following hypothesis can be presented:

(24)

24

H2: When brand-scandal fit is high the negative effect of the scandal will be greater than when brand-scandal fit is low

Figure 3- Hypothesis 2

Brand-Response Fit

Another factor that will be investigated in this research is the brand response to an endorser-scandal. The intention of brand-response could be to control the ‘damage’. In this thesis, it means to reduce damage of the decline in consumer attitude towards the brand as much as possible.

As stated in the literature review, Amos et al. (2008) highlighted the importance for brands to have a response plan ready since there is a chance that negative information arises about their celebrity endorser. More specifically, they have to make a decision how to react when their endorser is involved in a scandal. The response of Nike to the scandal of Lance Armstrong is a good example to show why it is so important for a company to react. In the first months of all the ‘fuss’ about Lance Armstrong, Nike did react but they stayed neutral. After a while, consumers started to protest and expressing their feelings that they thought Nike was involved in the whole doping scheme from Lance Armstrong (BBC, 2012). This shows that it is critical for a company to react and respond openly to the scandal their endorser is involved in. Because if not, the attitude towards the brand could be negatively influenced.

Tybout and Roehm (2009) have created a 4-steps guideline to let the response of a company fit the scandal. They state that “the most effective responses are carefully and systematically calibrated to the characteristics of the brand, the nature of the scandalous event, and the company’s degree of seeming culpability (Tybout and Roehm, 2012:84).

Brand-Scandal Fit

High

Low

Greater negative effect of the scandal

(25)

25 According to their guideline, four steps are proposed to let the response fit the scandal. These four steps include:

1: Assess the incident

2: Acknowledge the problem 3: Formulate a response 4: implement the response

In this thesis, it is expected that if the brand responses in line with their core values, the attitude towards the brand will be less affected. In the response it is important to

centralise the core values the brand stands for, when communicating this to the consumer. This way, it could be that the attitude towards the brand is less damaged because the consumer agrees with the decision of the brand to either continue or discontinue the endorsement.

Building on the previous literature on crisis management, it is expected that the brand response is generally important when their endorser is involved in a scandal. However, it could also depend on the type of scandal and the type of brand they stand for what the best reaction of a brand should be. Tybout and Roehm (2014) suggest that a response should fit the scandal. If the scandal inflicts the core values of the brand (so when brand-scandal fit is high), it could be better for the brand to discontinue the contract with their endorser. On the other hand, when the scandal has little to no impact on the brands core values (so when brand-scandal fit is low), it could be better for the brand to continue the endorser contract. It is therefore expected that when the response matches with the brand, consumer attitudes towards the brand will be rated more positively. The next hypothesis can be presented:

H3: When brand-response fit is low the negative effect of the scandal will be greater than when brand-response fit is high

Brand-Response Fit

High

Low

Less negative effect of the scandal

(26)

26

Interaction Brand Scandal Fit and Brand-Response fit

Building on the literature from Tybout and Roehm (2014) that a response should fit a scandal in order to minimize the damage done to a company’s corporate reputation, it would be interesting to find out under which brand-scandal fit condition the impact of a brand-response fit will be the greatest. This thesis focuses on how to minimize the damage done to consumer attitudes, and it is thus expected that when the core values of the brand are inflicted (high brand-scandal fit) a high brand-response fit can potentially reduce the negative impact of the scandal, whereas the low brand-response fit will not. On the other hand, when the core values are not inflicted (low brand-scandal fit) it is expected that the type of response (continue or discontinue contract) is less important, and any response can have a positive effect. So under the high brand-scandal fit

condition, a correct response is expected to be the most effective. Hypothesis 4 can be presented:

H4: The interaction between the scandal and brand-response fit is more pronounced when brand-scandal fit is high than when brand-scandal fit is low

High brand scandal fit More pronounced interaction between _{scandal and brand-response fit}

Low brand scandal fit Less pronounced interaction between

(27)

27

4. METHODOLOGY

Design

This explanatory study investigates the impact of an endorser scandal on consumer attitude towards the brand. An explanatory study is intended to discover causal

relationships between key variables (Saunders et al. 2012). The study consisted of 1) a pre-test, in which the manipulations were tested, and 2) the main study. For the main study, a 2x2 factorial experiment design was employed to test the hypotheses that were developed. The dependent variable is attitude towards brand. The independent variable is brand-scandal fit (high and low). The moderating variable is brand-response fit (high and low). In order to create these variables, three factors will be manipulated i.e. the brands, the scandals and the brand response (continue or discontinue). The brand-scandal manipulation creates a high or low brand-brand-scandal fit condition. This way it can be found out under which condition the attitude towards the brand decreases the most. By adding a brand response to the scandal manipulation, a high or low brand-response fit condition is created. By doing so it can be researched under which condition attitudes towards the brands can become more positive.

A vignette experiment was conducted to answer the research question. In this type of experiment a hypothetical situation is presented to which research participants respond. This way their perceptions are revealed as well values and impressions of events (Caro et al. 2012). “The premise is that the judgments made about vignette situations can provide insights about the basis on which people, who are similar to research

participants, will act when faced with similar conditions” (F. Caro et al, 2012: 7). Short descriptions of realistic situations are usually used to elicit respondents’ judgment about the scenario that is shown (Atzmüller et al, 2010).

In this research, subjects were exposed to fictitious news feeds about the endorser. Four different questionnaire versions were developed. All the versions were identical in terms of look and feel. They differed from each other in type of brand, type of scandal, and brand-response to the scandal.

(28)

28

Stimuli

To test the impact of the scandal, participants in the questionnaire were exposed to fictitious news feed. In order to create this newsfeed, stimuli needed to be developed. To develop the stimuli, a small focus group was conducted. One of the assignments was that participants in the focus group had to come up with:

1) A well known celebrity that can act as a brand endorser 2) Two well known consumer brands

3) Two celebrity-endorser scandals

First of all, the criteria for the endorser were that it had to be someone internationally well-known, famous and most importantly, not recently involved in a scandal.

Furthermore, the type of brands had to be brands that people are familiar with and represent a high brand recognition. Finally, the type of scandals had to be something that can happen humanly speaking any day, and they had to be clear, simple and easy to understand. The endorser-brand combination and the brand-scandal combination should reflect real life situations, however they had to be fictitious. The choices made for the investigation are of such intention to minimise prejudice with the respondents, and pollution of the data. This research focuses to determine the link between a type of brand and a type of scandal and whether consumer attitudes are impacted and/or differ per combination. Therefore the scandal combinations not only had to be fictitious, but also celebrity choice and the situations should avoid associations with earlier

experiences in the minds of the respondents.

The goal of the focus group was to develop the stimuli used for the experiment. Roger Federer was chosen as the brand endorser, since he still belongs to the vast diminishing group of celebrities that have not been involved in a scandal yet.

After choosing the celebrity that can act as an endorser, two types of brands needed to be chosen. The first type of brand that came to discussion was a sports brand, as Roger Federer is not only a famous and successful sports man, but also someone who

represents fair play values. It is likely that Federer will endorse a sports brand that acts in line with his standards. Puma came out of the discussions as a well-established sports brand, with clearly communicated values. Their core values are: performance, honesty in the game and being inspirational (Marketing Plan Puma 2014). According to the

(29)

29 participants of the focus group, these values not only reflect the desire to create a brand with a higher standard of consciousness, but also to make an ethical statement to the world.

The second type of brand the participants came up with, was a family brand. This type of brand should be known in a family broad perspective. Also an obvious choice because it had to relate to Roger Federer because he is known as a real family man and having a close family relationship, for his wife and kids support him almost every match. They look like the ideal couple. For the majority of the participants in the focus group Nutella was one of the first family brands that came to mind. Almost everyone associated this brand with having breakfast together with the family. Nutella’s core values are family, tradition and quality (Nutella 2014). Both Puma and Nutella are well known brands around the world and Roger Federer has not been sponsored yet by one of them. After the process of choosing the type of brands, a sport brand and a family brand, two scandals had to be defined that could potentially inflict the brand’s core values i.e. desired identity. One scandal should be a sports related scandal and the other one a family related scandal. Meaning to define how consumers will react on these type of scandals.

The sports-related scandal that was chosen is the involvement of Roger Federer using doping in his tennis career, where he simultaneously won seventeen grand slam

tournaments. This is likely to be in contrast of his fair play image and more importantly corporate values of the brand.

The family-related scandal is the adultery of Roger Federer, admitting he cheated on his wife with multiple women after a marriage of 15 years. In this case this is to be expected contrary to his image as perfect family man, thus damaging the brand’s core values.

The brand responses to the scandal that were chosen were either to continue the contract with the endorser, or to discontinue the contract with the brand-endorser. According to the focus group participants these are the most common responses that brands usually give when their endorser is involved in a scandal.

(30)

30

Design of the news feed

In order to create a realistic situation, four news feeds were developed as fictive newspaper articles. An image was created that looked like a real front page of a

newspaper. Other articles and the name of the newspaper were blurred, for a reason not to influence the opinion of the respondents.

The news feed differed on two points:

1) Brand - scandal combination (brand-scandal fit manipulation) 2) Brand Response to the scandal (brand-response fit manipulation)

Brand-Scandal Fit

The respondent was either exposed to a news feed with a high brand-scandal fit or a low scandal fit. As said before, it is defined in this research that there is a high brand-scandal fit when the brand-scandal inflicts damage to the core values of the brand. On the other hand there is a low fit when the core values are not affected by the scandal. So when the sports brand (Puma) is linked to the sports scandal (doping), we define a high brand-scandal fit and it is to be expected that the attitude towards the brand will be negatively influenced. The same accounts for when the family brand (Nutella) is linked to the family scandal (extra-marital affairs). For the other combinations (Puma-cheating & Nutella- doping) brand-scandal fit is expected to be low. The figure below shows an overview of the different Brand-Scandal fit manipulations. For all these combinations Roger Federer is the celebrity endorser.

Figure 6- Brand-Scandal fit manipulation

Doping

(sport scandal)

Extra-marital affairs (family scandal) Puma (sport brand) High brand-scandal fit Low brand-scandal fit

(31)

31

Brand Response Fit

After being exposed to a news feed about the scandal, the respondent will also see a short news feed about the brand-response to the scandal. It is expected that when brand-scandal fit is high, so when the core values of the brand are inflicted, it will be better for the brand to end the contract with their endorser. For Puma, it is expected that when the endorser is involved in doping, it will affect their core values. The correct brand-response therefore is to discontinue the contract. When the endorser is cheating, it is less likely to inflict the core values, thus better to continue the contract.

For Nutella, the incorrect brand responses are added. So when the endorser is involved in doping, they discontinue the contract, eventhough core values are not damaged.

Figure 7- High fit responses for Puma Figure 8-Low fit responses for Nutella

Pre-test

After the development of the stimuli, a pre-test was conducted to test the stimuli and to determine if the manipulations were perceived as intended.

 16 respondents (four in each scenario) cooperated.  52% of these were male respondents.

 The overall age was 26.

The participants were randomly assigned to one of the four scenarios. The pre-test consisted of three parts.

Puma Doping _contractEnd Nutella Doping _contractEnd

Puma Cheating Continue

contract Nutella Cheating

Continue contract

(32)

32 The first part was designed to check if the attitude towards Federer and the attitude towards the brand were positive. Because if not, it would be difficult to measure any changes in attitude towards the brand. The second part checked if the brands were seen as either a sports brand (Puma) or a family brand (Nutella), and if the scandals were seen as a sport scandal (doping) or a family scandal (cheating). It was asked if the core values are consistent with the brands, because the core values are crucial for measuring brand-scandal fit in this research. Finally, the news feed was shown, and questions were asked about the realism of this news feed. Al the items were measured on a scale from 1 (“strongly disagree”) to 7(“strongly agree”). The measurement methods of the final questionnaire will be further discussed in the later paragraphs.

Procedure

After the pre-test the final questionnaire was distributed. It was online for 2 weeks. The dispersion of the survey was mainly through modern social media such as e-mail and Facebook. Personal messages with the survey link were sent out and additionally went to closer friends and relatives it was asked to send the link through to their friends or relatives. There are no limits set in participating in this research. This way the response rate was highly optimized. However, preferably, the targeted ages are between 18 and 65 years old.

The respondents were randomly assigned to one of the four scenarios, see figure 9. They were not aware of the other treatments. In order to make sure everyone was familiar with the brand’s core values, a small introduction was given at the beginning of the survey, shortly mentioning the brand and it core values. The first questions were concerning the attitude towards Federer and also towards the brand, as well as about the core values of the brand. In the introduction to the news feed, instructions were given to read this news feed carefully. After showing this news feed about the endorser being involved in a scandal, questions were asked to measure the dependent variable and. Furthermore, respondents were asked again to rate the core values. By rating the core values before and after the scandal, it can be checked whether the brand-scandal manipulation was perceived as intended. Then another small introduction was given to read the news feed about the brand’s decision carefully. Again, questions were asked

(33)

33 after showing this news feed to measure the dependent variable and to check the

manipulation for brand-response fit. To check this, it was asked if participants agreed that the brand made the right decision concerning the scandal. Additionally, a question to check the realism of the scenario was incorporated after each newsfeed. The survey ended with questions about message involvement, age, gender and nationality. The final question of the survey was an open question to see if anything specific came to mind when filling in the questionnaire. This could be useful at the end of the research to see if there were any implications. An example of one of the four questionnaire versions can be found in appendix 1.

Figure 9- Different survey versions

Endorser-Brand-Scandal-Response combination Brand- Scandal Fit Condition Brand-Response Fit Condition 1 Federer-Puma-doping-discontinue contract HIGH HIGH 2 Federer-Puma-extra-marital affairs-continue contract LOW HIGH 3 Federer-Nutella-doping-discontinue contract LOW LOW

4 Federer-Nutella- extra-marital affairs-continue contract

(34)

34

Measurements

Attitude towards the brand

The dependent variable that relates to consumer behaviour is attitude towards the brand. Attitudes toward the brand where asked in the beginning of the questionnaire, after the news feed about the scandal, and after the news feed about the brand’s response to the scandal. The items included: “I think Puma is a good brand”, “I am

positive toward Puma”, I like Puma”, I think Puma is a high quality brand”, I think Puma is appealing”. On a 7-point Likert scale, the answers varied from “strongly disagree” to

“strongly agree”. These items were taken from Spears and Sing’s (2004) study about measuring attitude towards the brand. Additionally, one item was added from a

research from Fong and Wyer (2012) about consumer’s reaction to a celebrity endorser scandal. The question was: “How did your attitude toward Puma change after reading the news feed?” It was asked on a 7-point Likert scale, varying from “much more negative” to “much more positive”.

Brand-Scandal fit

Brand-Scandal fit is the independent variable. This is whether the core values

correspond or conflict with the scandal; high brand-scandal fit is when the scandal is goes against the core values of the company for instance Puma-doping. It can be high or low and this variable is expected to have an effect on attitude towards the brand.

Previous literature from Greyser (2010) showed that a scandal could become most problematic to a brand when the ‘essence’, or their core values are being inflicted. Brand-Scandal fit is high when the core values are damaged. Therefore the

questionnaire asks about the core values before and after the news feed about the scandal. When taking Puma as an example, the items included: “Puma stands for

performance”, “Puma stands for honesty in the game”, and “Puma stands for being inspirational”. The answers varied from “strongly disagree” to “strongly agree” on a

7-point Likert scale. After showing the newsfeed, two extra items were added, taken from a research from Fong and Wyer (2012) about consumers’ reactions to a celebrity

endorser scandal. The items include: “I think this scandal has a negative impact on Puma”, and “This scandal has impacted my view on Puma”. The answers varied again on a 7-point Likert scale from “strongly disagree” to “strongly agree”.

(35)

35

Brand Response to scandal and Brand-Response fit

In this research, it is expected that if the brand takes the right decision after being involved in a scandal, the damage on attitude towards the brand could be reduced. Previous research by Griffith and Ryans (1997) showed that a brand’s corporate

communications could be a big threat for brand equity. Keeping this in mind it could be said it is important for a company to react to a scandal. However, little to no academic research has been done on what consumer’s think is the best decision after the scandal; to end or to continue the contract with their endorser? Building on research from Louis and Obermiller (2002) stating that a brand is better off disassociating with the endorser when there is a high level of blame for the endorser, it could be expected that it will also be better for a brand to end the contract when the core values are inflicted.

The items that will try to answer this question include: “Puma made the right decision”,

“Puma’s decision has impacted my view on them”, “I would have done the same in Puma’s situation”, “I think Puma is responsible for Federer’s behaviour”, “Puma made an effective decision”. Answers vary from “strongly disagree” to “strongly agree” on a 7-point Likert

scale.

The control variables that are used in the survey are attitude towards Federer and message involvement. These variables are used because the impact that is measured in this questionnaire could be affected by the characteristics of the respondents and stimulus.

Attitude towards the Celebrity Endorser (Roger Federer)

It is important for this research that all the participants think (slightly) positive about Federer and that there is no difference in liking between the different conditions. Therefore 2 items are asked, including: “I like Roger Federer”, and “I am positive toward

Federer”. The items will be measured on a 7-point Likert scale, varying from “strongly

disagree” to “strongly agree”.

Message involvement

The Elaboration Likelihood Model from Petty and Cacioppo (1983) shows that

involvement in the ad or message can influence results. For example, a highly-involved person will perceive the message with more intention and care than a low-involved

(36)

36 person. In this research it could well be that a low involved person does not care at all about what happens, so it could be important to keep this in mind. The items to measure this include: “I was interested in the contents of the newsfeed”, “I perceived the

information with attention”. A 7-point Likert scale was used, with answers varying between “strongly disagree” and “strongly agree”.

Additionally, at the end of the questionnaire, the gender was asked, as well as the age and the nationality. In order to give their age the participant had to fill in a number. The final question of the questionnaire was an open question, stating: “Did anything specific come to mind when filling in this questionnaire”? This feedback question could help in explaining limitations of the design or research.

(37)

37

5. RESULTS

The questionnaire was online for 11 days and this resulted in a total of 221 respondents. 50 of these did not fill out more than 75% of the survey and were therefore deleted. 2 participants were identified that filled in the same scale number throughout all the questions. These were also deleted, and this resulted in a final data set of 169 respondents.

Data preparation

After eliminating the unusable cases from the dataset, as said in the previous paragraph, it was identified that some values were missing. These values were coded as -99 and put on missing. A control check was carried out to make sure this was all executed correctly throughout the entire dataset. The next step that needed to be performed in oder to conduct analysis was to make sure the answers of the different conditions were put together into one column per variable. This was necessary because in SPSS, all the variables of the different treatments were in a separated column. The new variables were computed by calculating the mean per variable of the four different treatments.

Sample profile

The final data set consisted of 169 respondents. Each one of these was randomly assigned to either one of the four questionnaire versions. The participants were not aware of the fact that there were different treatments and therefore different versions of the survey. Version one consisted of 37 respondents, version two of 46 respondents, version three of 39 respondents and version four of 47 respondents (N total = 169). These subjects were asked for their age and gender. The distribution of these descriptive

(38)

38

Figure 10- Gender distribution

Looking at the descriptive statistics of the participants 41.4% (N=70) were male and 58.6%(N=99) female as shown in figure 10. There was no significant difference in the distribution of gender among the different surveys (x=3.487, p=0.324). With an age range from 18 to 67, the average age was 35.1 (SD = 15.2). Figure 11 shows the distribution of age of the participants. A One-Way ANOVA showed there were no significant age differences (F=0.675, p=0.568).

Figure 11- Age distribution

An overview of the Descriptive Statistics can be found in Appendix 2.

Male 41% Female 59%

Gender

<20 11% 21-30 49% 31-40 3% 41-50 12% >50 25%

Age

(39)

39

Reliability Analysis

As said in the data preparation paragraph, in order to conduct analysis new variables had to be computed. However, before conducting analyses with these new variables, a reliability analysis had to be carried out to prove that the items measured the same underlying attribute. This analysis is conducted by checking the Cronbach’s Alpha score of all the variables. The general rule is that in order to be labeled reliable, a scale needs a

Cronbach’s Alpha score of above 0.7. The three different brand attitude scales, consisting of five items, all showed a

Cronbach’s Alpha score of above 0.7. Also the core values scales, which consist of three items, can be labeled as reliable. Furthermore, both consisting of two items, the attitude towards Federer and message involvement scale all scored a Cronbach’s Alpha above 0.7. However, only the brand response scale showed a score of 0.614, which means it could not be labeled as reliable. However, if one item were to be deleted the Cronbach’s Alpha would go up to 0.726. The decision was therefore made to delete one item on this scale. An overview of all the reliability analyses is shown in figure 12.

Figure 12 - Results of Reliability Analysis tests

Cronbach’s

Alpha Items in scale Items eliminated

Brand attitude before 0.924 5 0

Brand attitude after scandal 0.938 5 0

Brand attitude after response 0.937 5 0

Core values before 0.864 3 0

Core values after 0.756 3 0

Brand response 0.726 4 1

Attitude toward Federer 0.934 2 0

(40)

40

Control variable checks

The control variables were checked for differences between groups. A One-Way ANOVA was performed for these variables. It shows that the respondents are generally positive toward Roger Federer (M= 5.61, SD= 1.29). Between the groups there is no significant difference between how positive they are towards Roger Federer (F= 1.560, p=.201). With a mean of 4.83 the respondents are slightly involved in the messages (M=4.83, SD=1.50). The One-Way ANOVA test (F=2.004, p=.115) shows there is no significant difference between the levels of message involvement between the manipulation groups. There have been no selection bias. More detailed information can be found in appendix 3.

Manipulation checks

In order to assess the success of the manipulations, manipulation checks needed to be performed.

The independent variable brand-scandal fit was created by manipulating brand and scandal. As stated earlier, there is a high brand-scandal fit when the core values of the brand are inflicted. It is therefore expected that the core values of the brand will be rated lower for the manipulations where there is a high brand-scandal fit than where there is a low brand-scandal fit. A mixed-ANOVA test was computed to check the manipulation.

Figure 13- Results Mixed-Anova

Descriptive Statistics

Mean SD N

Core Value Rating Before scandal

4,71 1,14 166

Core Value Rating After scandal

Perception becomes reality : the impact of a celebrity endorser scandal and the effect of a powerful response on consumer attitudes towards the brand

Faculty Economics and Business