THE EFFECT OF CELEBRITY ENDORSEMENT VIA SOCIAL MEDIA ON BRAND IMAGE
by
LINDA ZUIDEMA
MASTER THESIS
THE EFFECT OF CELEBRITY ENDORSEMENT VIA SOCIAL MEDIA ON BRAND IMAGE
by LINDA ZUIDEMA Vrydemalaan 198 9713 WT Groningen 06-30531105 lindazuidema@gmail.com Student number 1997998 Supervisor: Dr. K.J. Alsem Second Supervisor: S. Beckers University of Groningen Faculty of Economics and Business
Msc Business Administration – Marketing Management
MANAGEMENT SUMMARY
Social media networks are upcoming and gaining importance. These networks enable customers to reach and be reached by almost everyone, anywhere and anytime (Hennig-Thurau et al, 2010). For marketing professionals, social media networks are a new outlet that can be used to help increase the interest in a product or service (Pooja et al, 2012). Celebrity endorsement via social media is not uncommon. One can find product recommendations from celebrities on networks like Facebook and Twitter. These recommendations might influence brand image largely and can save costs that are involved with the traditional celebrity endorsement.
Although literature about celebrity endorsement via social media hardly can be found, literature about the traditional offline celebrity endorsement is rich. Celebrity endorsement can be defined as a type of communication in which an individual who enjoys public recognition is the spokesperson of a certain brand and by using their recognition and popularity and personality, they certify the brand’s position and claim (Sonwalkar et al, 2011 and Biswas et al, 2006). The basic assumption of celebrity endorsement is that the value associated with the celebrity is transferred to the brand the celebrity is endorsing (Sonwalkar et al, 2011). Celebrity endorsement has proven to be valuable, however the fees the celebrities require have increased substantially (Jain et al, 2009).
The underlying phenomenon of celebrity endorsement is parasocial interaction. Some people might feel an imagined intimacy or interpersonal involvement with celebrities and are therefore open to their recommendations (Greenwood and Long, 2011). In comparison with other types of endorsements, like endorsement by professional experts, company president or typical consumers (Friedman & Friedman, 1979), celebrity endorsements are more effective in dimensions such as trustworthiness, believability, persuasiveness and likability (Biswas et al, 2006).
that gender and likeability of the celebrity endorser would influence this effect. To find out whether this was the case, this study has been conducted. A total of two experiments have been conducted. In study 1, respondents were confronted with a Tweet of a celebrity about Nespresso, containing just text. In study 2, respondents were confronted with a Tweet of a celebrity about Ray Ban, containing text and a picture of the celebrity wearing Ray Ban sunglasses. Brand image was measured with the help of the Young & Rubicam Brand Asset Valuator. This model consists out of five pillars, which are differentiation, relevance, esteem, knowledge and energy (Mizik & Jacobson, 2008). For each study, three questionnaires were developed; one without a celebrity endorser, one with a likeable endorser and one with a non-likeable endorser.
PREFACE
This thesis has been written for the master Business Administration – Marketing Management. It was final step in my academic study career. Since I wrote my bachelor thesis about social media, and found this topic very interesting, I wanted to write my master thesis about this relatively new subject as well. However, I also liked the subject of celebrity endorsement a lot, but I did not know what to research on this topic, since it already has been studied extensively in the past. The topic celebrity endorsement via social media, a combination of two topics I find very interesting, came to my mind when I heard about a scandal about a celebrity tweeting about a product that has false claims. Kim Kardashian was the endorser of Quicktrim, which is a supplement for losing weight and regularly tweeted about the product. Later on, it appeared that this supplement does not work, and Kim Kardashian got sued, since she promoted the product amongst her large group of social media followers. I thought about the influence celebrities might have when sharing their product experiences with their social media followers and found this a very interesting topic.
Doing this research was very interesting, although it was sometimes a challenge. Happily, I got enough help. For helping me with this thesis, I want to thank all my friends, who I annoyed with some pre-tests, the final questionnaires, and questions about my thesis in general. They also helped me with finding enough respondents. Furthermore, I want to thank all the people who filled in my questionnaires. I also want to thank dr. Karel-Jan Alsem for supervising me and giving me feedback on my proposals.
Linda Zuidema
TABLE OF CONTENTS
1. INTRODUCTION ...7
2. THEORETICAL FRAMEWORK ...10
2.1 Traditional celebrity endorsement...10
2.2 Celebrity endorsement and parasocial interaction ...11
2.3 The effects of celebrity endorsement ...11
2.4 Social media ...13
2.5 Celebrity endorsement via social media...14
3. RESEARCH DESIGN ...19 3.1 Design...19 3.2 Sample study 1 ...19 3.3 Stimuli ...20 3.4 Measurement ...24 3.5 Procedure study 1 ...25 3.6 Procedure study 2 ...26 3.7 Plan of analysis...27 4. RESULTS ...28
4.1 Reliability analysis study 1 ...28
4.2 Testing the hypotheses with study 1 ...29
4.3 Reliability analysis study 2 ...31
4.4 Testing the hypotheses with study 2 ...32
5. CONCLUSION...35
6. MANAGERIAL RECOMMENDATIONS ...38
7. LIMITATIONS AND FURTHER RESEARCH ...39
REFERENCES ...41
APPENDIX 1; QUESTIONNAIRE ...45
APPENDIX 2; TIMELINE PARIS HILTON AND NESPRESSO PICTURE ...47
APPENDIX 3; QUESTIONNAIRE STUDY 2 ...48
1. INTRODUCTION
Celebrities are commonly used as endorsers in advertisements for firms in order to support corporate or brand image (Biswas et al, 2006, Erdogan, 1999). About 25% of American advertisements employ celebrity endorsers (Amos et al, 2008). Celebrity endorsement can be seen as a type of communication where an individual who enjoys public recognition is the spokesperson of a certain brand and by using their recognition and popularity and personality, the celebrities certify the brand’s position and claim (Sonwalkar et al, 2011 and Biswas et al, 2006). Well-known examples of celebrity endorsers are the famous actor George Clooney, endorsing the Nespresso coffee or singer Beyoncé endorsing different products of L’Oréal. Celebrity endorsement is not a new phenomenon, on the contrary. Already since the late nineteenth century, celebrities have been endorsing products (Erdogan, 1999). Celebrity endorsement has proven to be valuable, however the fees the celebrities require have increased substantially (Jain et al, 2009).
Currently, the marketing environment is changing. Advertisements go further than the regularly used television commercials or ads in newspapers and magazines. The rise of Internet has contributed to this phenomenon. Nowadays Internet is part of our daily lives, and social media networks such as Facebook, YouTube and Twitter have gained importance (Hennig-Thurau et al, 2010). Social media networks enable customers to reach and be reached by almost everyone, anywhere and anytime (Hennig-Thurau et al, 2010). These social networks are used often; according to Trusov et al (2010) 90 percent of all American teenagers and young adults are active on a social network platform. Advertising via social media networks is not uncommon. For marketing professionals, social media networks are a new outlet that can be used to help increase the interest in a product or service (Pooja et al, 2012). This new way of advertising can be very effective since typically social media are highly accessible and can be used to reach large numbers (Berthon et al, 2012).
send. However, literature about this type of celebrity endorsement via social media is rare.
As Pooja et al (2012) describe, through their connections to others, consumers can wield a lot of power, engaging in activities that can enhance or reduce a brand's profitability. This is certainly the case for celebrities, since they often have a lot of connections via their social network. The power of celebrity endorsement via social media might be larger than the traditional endorsement, since it seem to be more honest; the celebrity shares recommendations via a personally social media account and not via (paid) mass media. It can also save costs, since traditional celebrity endorsement costs companies a lot of money (Jain, 2009).
But what is the actual effect of this online celebrity endorsement? And what are the influences of using a certain celebrity versus using another celebrity? Fit between the celebrity and the product is important in the offline environment (Kamins, 1990), and might also play an important role in the online environment. Likeability of the celebrity might also have an influence. And do characteristics of the consumers affect the influence of celebrity endorsers? Premeaux (2009) showed that gender and social class might have an influence. Debevec and KerTian (1984) also showed the influence of gender; they found that attractive female models generated more enhanced attitudes than attractive male models across both genders but particularly among males. These results are for offline endorsement, so whether this is true for social media endorsement is not known. Literature about this topic is rare. Therefore, two studies are conducted. The research question for these studies will be: Which factors influence the effect of celebrity endorsement via social media on brand image? The results of the studies might give companies a clear indication of how celebrity endorsement via social media networks can be used to enhance brand image and will indicate the importance of celebrity endorsement via social media. The difference of this effect for males and females will be discussed as well as the differences of a likeable person versus a less likeable person. It can help managers in determine their social media marketing strategy.
2. THEORETICAL FRAMEWORK
In this chapter, the phenomenon of celebrity endorsement will be discussed. Furthermore, some information about social media will be shared. In paragraph 2.5, celebrity endorsement via social media will be explained. This topic has not been studied a lot before in literature. In this paragraph, the hypotheses are also proposed. At the end of this chapter, the conceptual model will be presented.
2.1 Traditional celebrity endorsement
The definition of celebrity endorsement used in this paper is as follows: a type of communication in which an individual who enjoys public recognition is the spokesperson of a certain brand and by using their recognition and popularity and personality, they certify the brand’s position and claim. This definition is based on the definitions of Sonwalkar et al (2011) and Biswas et al (2006). However, Sonwalkar et al (2011) talk about celebrity endorsement as some type of communication channel. It can be discussed whether a person in itself is a communication channel and therefore, this part has been left out of the definition used for this study.
Celebrity endorsement is used a lot these days. Just open a magazine and one will see a range of celebrities promoting facial creams, perfumes and watches. According to Amos et al (2008) about 25% of American advertisements employ celebrity endorsers. Celebrity endorsement is especially typical for nationally marketed products rather than for local or niche products (Sonwalkar et al, 2011).
Celebrity endorsers can be all kind of well-known people, ranging from singers to sportspeople. Consumers rate films stars as the most preferred endorsers, followed by models and sports stars (Sonwalkar et al, 2011).
2.2 Celebrity endorsement and parasocial interaction
With celebrity endorsement, the popularity and recognition of the celebrity is transferred to the product. People want to use the same products as their idols and feel connected to them by doing this. This underlying phenomenon of this is parasocial interaction. This concept has been introduced by Horton and Wohl (1956). They researched this topic due to the upcoming success of the new mass media in those years, like radio, television and the movies. These media give people the illusion of face-to-face relationship with the performer (Horton and Wohl, 1956). This will lead to the fact that the viewers may consider this celebrity as for example a friend, and this will create an illusion of intimacy (Horton and Wohl, 1956). Perceived attractiveness of the celebrity intensifies the parasocial interaction (Hartmann and Goldhoorn, 2011). However, where Horton and Wohl (1956) as well as Hartmann and Goldhoorn (2011) see parasocial interaction as regular people that are having a feeling of interacting with people on television or the radio, other see the phenomena more as an enduring relationship between viewers and celebrities (Hartmann and Goldhoorn, 2011). Thorson and Rodgers (2006) for example, define parasocial interaction as imagined intimacy of the media user with what he or she consumes. So instead of only an interaction between celebrity and people, they see it more as an enduring relationship between the two. This last mentioned definition is mostly used nowadays and explains the best why celebrity endorsement works. Some people might feel an imagined intimacy or interpersonal involvement with celebrities (Greenwood and Long, 2011). Therefore, they are open to their recommendations (Greenwood and Long, 2011).
2.3 The effects of celebrity endorsement
However, there are also people who believe celebrity endorsement is not that effective. According to Mehta (1994) there are no significant differences between celebrity and non-celebrity endorsement situations with regard to attitudes towards advertising, attitudes towards the brand and intentions to purchase endorsed brands. The effects of celebrity endorsement are thus not straightforward.
Companies should carefully select a celebrity for representing their brand. Amos et al (2008), Bailey (2007), and Edwards and La Ferle (2009) explain that negative celebrity information could harm an advertising campaign a lot. Due to the paparazzi-interested media, information about celebrities is easy to find and often hard to ignore or escape (Edwards and La Ferle, 2009). The upcoming availability of blogs and the interest in them, make collecting information about celebrities a sport and has led to a peak in celebrity gossip (Edwards and La Ferle, 2009). Negative news about a celebrity is spread quickly and this negativity transfers to the image of the brand the celebrity represents. So when a celebrity is involved in a scandal, this can have a negative effect on the brand he or she represents as well.
However, celebrity endorsement, when carefully selected, can be very valuable to a brand. Amos et al (2008) found out that celebrity trustworthiness, celebrity expertise and celebrity attractiveness are of high influence on purchase intentions, brand attitudes and attitudes towards the advertisement. When a celebrity is very popular, this can help to improve the image of the brand as well. The effect also depends on how famous the celebrity is; there is a positive relation between knowledge of the celebrity’s name and endorsement recall (Huston et al, 2003).
However, the success of the celebrity endorsement also depends on other factures, such as on fit (Friedman & Friedman, 1979, Kamins, 1990). Kamins (1990) developed the match-up hypothesis based on findings of other researchers. This hypothesis suggests that the message conveyed by the image of the celebrity and the image of the product should be congruent in advertisements in order to be effective. In other words, there should be a fit between the celebrity and the product. A good example of this is tennis player Andre Agassi, endorsing products for the sports brand Niké. However, it still remains unclear between which dimensions the celebrity/product fit exactly should be (Amos et al, 2008).
attitudes. Debevec and KerTian (1984) found that attractive female models generated more enhanced attitudes than attractive male models across both genders but particularly among males. Edwards and La Ferle (2009) found in their research that women rated a female endorser more trustworthy and men rated a male endorser as more trustworthy. They also found that women rated celebrity endorsements slightly more positive than men did. In his research of 2009, Premeax found that middle and upper class men and upper class women were more influenced by celebrity endorsers appearing in commercials than were middle class women. It can be concluded that there are many results, which are contradicting each other sometimes.
2.4 Social media
Internet has changed the world a lot. The introduction of web 2.0 and upcoming social media have led to the fact that the world has become smaller and people are more connected with each other (Pooja et al, 2012). Since 2009, social networks have emerged as the primary way active Internet users stay in touch with one another (Hutton & Fosdick, 2011). Facebook, Twitter, Myspace, LinkedIn, YouTube; (almost) everyone has heard of these new social media networks. Pooja et al (2012) state that more than half of the U.S. population has some social-media profiling. Trusov et al (2010) even state that 90 percent of al American teenagers and young adults are active on a social media network.
The idea behind all social media networks is basically the same. All social media networks are collections of user profiles, where people can subscribe themselves for and then can place information on which they want to share with others (Trusov et al, 2010). People can share pictures, upload music, write messages, but on the other hand also can consume content that others have shared with them (Trusov et al, 2010). Social media has a private character; many people share their personal things with their friends via these platforms. Celebrities also do this; they give you the feeling that you are getting to know them on a more personal level when they share their status update or a photo of their child with you.
quickly that it goes viral. One could say that social media can lead to an extreme form of word-of-mouth.
There is no doubt that social media networks are gaining popularity: Facebook has more than 400 million members worldwide and is now the most frequent visited website in the United States, surpassing Google (Katona et al, 2011). Enormous amounts thus and it is only becoming more and more.
All these social network sites are valued at billions of dollars and companies are interested in advertising on these networks (Trusov et al. 2010). However, companies also see that advertising is not the only way of getting attention on social media networks. Consumers can share information and opinions quickly with each other and this can make or break a brand. Due to the upcoming popularity of these networks, consumers are nowadays playing a more important role in defining brands and influencing the reputation of organizations (Pooja et al, 2012). Katona et al (2011) identified network marketing as a new opportunity for companies. The idea is that analyzing the network of members’ connections can help implement effective viral marketing strategies, by identifying influences and predicting consumers’ adoption probabilities (Katona et al, 2011). Targeting the right people might lead to a high word of mouth probability and a large reach. The research of Katona et al (2011) shows that the position a person has within the social network, is a good predictor of influential power. The number of connections and how connected these friends are determine the position largely (Katona et al, 2011). In the case of high-involvement products, companies might want to focus on people who have fewer connections, since they have a higher average influential power, but with low-involvement products, high-connected customers might be more suitable, since they reach more people (Katona et al, 2011).
However, Trusov et al (2011) state that a link between two profiles on a social media platform does not necessary mean influence.
2.5 Celebrity endorsement via social media
social media platforms. These celebrities might share experiences with brands and products. Although research on this influence does not exist, one might expect that this can lead to that ‘regular’ people rely on recommendations that celebrities share with their online friends.
This type of endorsement via social media can have a much larger power than endorsement via the traditional offline channels like print ads and television commercials. Social media are typically highly accessible and can be used to reach large numbers against relatively low costs (Berthon et al, 2012). Jain et al (2009) state that the effect of the message depends mostly on the source level of expertise (honesty, integrity and believability) and trustworthiness. Celebrity messages via social media seem to be more reliable than offline celebrity endorsement; celebrities recommend products themselves via their personal profile and it looks like there are not getting paid for recommending the product. Maybe they are getting paid, but they also might actually like a product. Therefore, it is expected that these recommendations are seen as honest, trustworthy and believable. Endorsement via offline channels, on the contrary, is paid and everyone knows that. Therefore consumers might believe that celebrities recommend products that they do not actually use or like, but just because they receive money for recommending it. It must be noted that this topic has not been studied and therefore this reasoning might not be true.
Parasocial interaction is also relevant with social media. As mentioned in chapter 2.2, parasocial interaction can be defined as an imagined intimacy between people and celebrities (Greenwood and Long, 2011). When people are ‘friends’ with a celebrity on a social media platform, they might have a feeling being connected with the celebrity and feel like they know them well. When a celebrity shares opinions about products and brands, this might influence their social media friends, since interpersonal networks are sometimes powerful sources of influence (Thorson and Rodgers, 2006).
When social media users have this emotional bond with the celebrity, they are open to social influence of these social media connections. Social influence occurs when a person adapts his behavior or attitudes to the behavior or attitudes of others (Leenders, 2002), which are in this case the celebrities. Since parasocial interaction is indicative of active, involved (social) media use, it can affect attitudes and behaviors (Thorson and Rodgers, 2006). Celebrities are often having a lot of connections on social media networks and therefore might exert influence on many people. A positive Tweet or Facebook update about a certain product might therefore influence the image of a brand. Besides that, Thorson and Rodgers (2006) say that individuals who collected product information from consumer-generated (eWOM) forums expressed higher product category interest than individuals who collected product information from traditional, marketer-generated forms. And eWOM can also be generated by celebrity endorsement via social media networks. Based on the arguments described above, the following hypothesis has been formulated.
H1: Celebrity endorsement via social media has a positive influence on brand image.
From research on the moderating role of gender on celebrity endorsement in general no clear conclusions are drawn. Some say that the effect is significantly larger for men (Premeaux, 2009, Debevec and KerTian, 1984), but no clear differences between men and women were found.
H2: Celebrity endorsement via social media has a larger effect on brand image for women than for men.
Whether celebrity endorsement is effective might not only depends on gender of the recipient, but also on the likeability of the celebrity. Biswas et al (2006) define likeability as affection for the source as a result of the source’s physical attractiveness, behavior, or credentials. According to McCracken (1989) likeability indicates affection toward a celebrity as a result of his or her physical appearance and the celebrities’ behavior.
In the online world liking a celebrity might even have a higher influence on the effects of celebrity endorsement than in the offline world. When liking a celebrity, the chance of following the activities of celebrities via social media becomes higher. Everyone can choose him- or herself which persons to follow online and whether to follow the activities of celebrities or not. When following the activities of a celebrity the chance that one will read about a product the celebrity recommends becomes larger, since these messages are shared on your personal wall (Facebook) or timeline (Twitter). However, one can also be confronted with messages from people they do not follow. This can happen because friends or relatives do follow these people and share the message the celebrity has posted. In this way, one can be confronted with a message from a celebrity he or she does not like.
Silvera and Austad (2004) also showed the importance of positive attitudes toward the endorser in their research. They found that liking an endorser also resulted in a higher liking of the product endorsed. Amos et al (2008) also state that likeability influences the attitude towards the brand. The higher the likability, the more the brand benefits from this. However, it might also work the other way around; if a celebrity is not liked, this might harm the brand image. Based on this, the following hypothesis has been developed.
H3a: Likeability of the celebrity has a positive influence on the effect of celebrity endorsement via social media on brand image.
In figure 1, the conceptual model is presented. H2 +/- H1 + H3 +
Figure 1. Conceptual model
3. RESEARCH DESIGN
To test the hypothesis stated in chapter two, quantitative research has been done. Quantitative research can give a good insight in the opinion or attitudes of large amounts of people (Malhotra, 2009, p.171). There has been chosen to conduct two studies, since this can indicate the differences in influence of just a text message of a celebrity or a text message accompanied with a picture of the celebrity. In the first study, a text message from a celebrity endorser is shown to the respondents. In the second study, a text message accompanied by a picture from the celebrity endorser wearing the brand is shown. The message is also accompanied by less other messages in the second study. This is done in order to enhance the visibility of the message and to see whether a text message from a celebrity is enough or more is needed to change brand image. In the following chapter, the research design of the two studies conducted for this paper will be discussed.
3.1 Design
For both studies, a 3 (no celebrity, likeable celebrity, not likeable celebrity) x 2 (male, female) between subjects factorial experimental design was performed in order to test the hypothesis. This design resulted into six experimental groups, which are displayed in table 1 and 2. The people who do not see a celebrity during the experiment are the control groups (group 1 and 2).
For both studies, three questionnaires have been developed and distributed via social media networks as well as via direct mail. One questionnaire was developed for the control groups, one for the likeable celebrity and one for the less likeable celebrity. People were requested to send the questionnaire further towards their friends, colleagues and relatives. This distribution technique is called snowball sampling. Snowball sampling is a good method to reach a large population and is low in costs (Malhotra, 2009, p.381). The questionnaires were accessible online, via thesistools.com.
3.2 Sample study 1
The target population of this study consists of Dutch males and females who are active on a social media platform.
completely and therefore excluded from this research. From this group of respondents were 60 male (= 50%) and 60 were female (= 50%). The average age of the participants of this study was 26.82 years (SD= 6.941) with a range from 18 to 56 years old. The average age per experimental group is displayed in table 1.
1. No celebrity/Male N=20 Average age: 27.30 3. Likeable celebrity/Male N=20 Average age: 28
5. Not likeable celebrity/Male N=20 Average age: 28.9 2. No celebrity/Female N=20 Average age: 25.55 4. Likeable celebrity/Female N=20 Average age: 24.80
6. Not likeable celebrity/Female N=20
Average age: 26.40
Table 1: Experimental groups study 1
The questionnaires from study 2 were filled in by a total of 156 people, of which 129 questionnaires could be used. 22 Questionnaires were not filled in completely and therefore excluded from this research. An additional five questionnaires were excluded since people did not indicate correctly which celebrity they say on the Twitter timeline. From this group of respondents were 64 male (= 49,6%) and 65 were female (= 50,4%). The average age of the participants of this study was 26.62 years (SD= 8.790) with a range from 16 to 69 years old. The average age per group is displayed in table 2.
Table 2: Experimental groups study 2 3.3 Stimuli
3.3.1 Celebrity
To select a likeable celebrity and a celebrity who is not liked that much, a list of 24 celebrities has been send to a group of 20 people. These people were asked rank the celebrities in order of liking. Of these 20 people, 14 people responded, which results
1. No celebrity/Male N=22 Average age: 27.95 3. Likeable celebrity/Male N=21 Average age: 28.19
5. Not likeable celebrity/Male N=21 Average age: 26.24 2. No celebrity/Female N=20 Average age: 27.50 4. Likeable celebrity/Female N=23 Average age: 24.96
6. Not likeable celebrity/Female N=22
in a response rate of 70%. This list consisted of 24 celebrities, from whom 12 are expected to be likeable and 12 are expected to be unlikeable. The list was based upon an American list of most popular celebrities. This pre-test showed that Doutzen Kroes was chosen as most likeable celebrity and Paris Hilton was chosen as the least likeable celebrity (see table 3) from the send list of celebrities. Therefore, these celebrities are selected for both studies.
Expected to be liked Actual liking rank Expected to be disliked Actual liking rank
Doutzen Kroes 1 Kanye West 10
George Clooney 2 Christina Aguilera 13 (shared)
Cameron Diaz 3 Hilary Duff 13 (shared)
Oprah Winfrey 4 Britney Spears 16
Brad Pitt 5 Donald Trump 17
Angelina Jolie 6 Arnold Schwarzenegger 18
Johnny Depp 7 Charlie Sheen 19
Eva Longoria 8 Lindsay Lohan 20
Tom Cruise 9 Kim Kardashian 21
Jay-Z 11 Justin Bieber 22 (shared)
Lady Gaga 12 Tiger Woods 22 (shared)
Madonna 15 Paris Hilton 24
Table 3: List of celebrities 3.3.2 Brand study 1
3.3.3 Brand study 2
In this second study, the brand Ray Ban was used. Ray Ban is a well-known brand, with which many people have experience, or at least heard about. Furthermore, males as well as females purchase the sunglasses brand. Another reason why this brand has been chosen is because it was possible to find pictures of Doutzen Kroes and Paris Hilton wearing sunglasses of this brand. This was an important prerequisite for the chosen brand, since a picture of the celebrity wearing the brand was a very important element of the experiment.
3.3.4 Social media network
Image 1: Twitter timeline study 1
3.4 Measurement
Brand image will be measured with the help of fifteen questions based on the Young & Rubicam Brand Asset Valuator in study 1 as well as in study 2. This model consists out of five pillars, which are differentiation, relevance, esteem, knowledge and energy (Mizik & Jacobson, 2008). This model is based on the premise that brand is a multidimensional construct that can be assessed through customer perception measurements (Mizik & Jacobson, 2008). The original model monitors more than 50 aspects of brand perceptions, but for this research a shortened version will be used in order to enhance the willingness of filling in the questionnaire. The statements used in the first study are presented in table 4. The statements for the second study are basically the same; only the brand name of Nespresso is replaced by Ray Ban.
A short explanation of the different BAV pillars; Differentiation is about the perceived distinctiveness of the brand. It defines the brand and reflects its ability to stand out from competition (Mizik & Jacobson, 2008). Relevance is about the personal relevance, appropriateness and perceived importance of the brand (Mizik & Jacobson, 2008). Furthermore, esteem measures the level of regard that consumers hold for the brand, the valence of consumer attitude and it reflects how well the brand fulfills it promises (Mizik & Jacobson, 2008). Knowledge measures awareness and understanding of the brand identity and captures consumer intimacy with the brand (Mizik & Jacobson, 2008). The last pillar energy is a new addition to the model. It indicates future orientation and capabilities of the brand (Mizik & Jacobson, 2008).
BAV Pillar Statement
Differentiation 1. Nespresso is unique Differentiation 2. Nespresso is distinctive
Differentiation 3. Nespresso coffee are very different from other brands Relevance 4. I purchase Nespresso coffee
Relevance 5. Nespresso is relevant to me
Esteem 6. Nespresso is the leading brand in the coffee industry Esteem 7. Nespresso is important for me
Esteem 8. Nespresso is reliable
Esteem 9. Nespresso is a high quality brand
Knowledge 11. I have tried Nespresso coffee
Knowledge 12. I am familiar with the brand Nespresso
Energy 13. Nespresso can meet consumers’ needs in the future Energy 14. Nespresso is innovative
Energy 15. Nespresso is able to adapt to changing tastes and needs
Table 4: Statements to measure brand image
3.5 Procedure study 1
Before the final questionnaires were distributed, a group of five people pretested the questionnaires. After this pretest, some small adoptions were made and the final questionnaires were placed online.
People are invited via email or via social media to participate in the research. They receive a link to an online questionnaire. There are three different versions of the questionnaire: one with no celebrity endorser, one with a likeable celebrity endorser and one with a non-likeable celebrity endorser. The website thesistools.com randomly selects which respondent gets to fill in which questionnaire.
At the beginning of the questionnaire, respondents are asked to fill in their age and gender. After this, there are some small differences between the three questionnaires.
Four out of the six groups of respondents are asked to take a good look to screenshots of a Twitter timeline. On this timeline, different people have shared their activities and thoughts. On this timeline, there is also a post from Doutzen Kroes or Paris Hilton (see appendix 2), which says “Just had a delicious cup of Nespresso..love this coffee!” The control groups are just confronted with a picture of a cup of Nespresso coffee with the logo of Nespresso next to it (see appendix 2).
likeable celebrity really like the celebrity and the people who see an unlikeable celebrity really dislike the celebrity.
In appendix 1, the complete questionnaire for the likeable celebrity can be found.
3.6 Procedure study 2
Before the final questionnaires were distributed, a group of five people pretested the questionnaires. After this pretest, some small adoptions were made and the final questionnaires were placed online.
Just like the first study, people are invited via email or via social media to participate in the research. They receive a link to an online questionnaire. There are again three different versions of the questionnaire: one with no celebrity endorser, one with a likeable celebrity endorser and one with a non-likeable celebrity endorser. The website thesistools.com randomly selects which respondent gets to fill in which questionnaire.
The structure of these questionnaires is almost the same as the structure of the questionnaires from study 1. At the beginning of the questionnaire, respondents are asked to fill in their age and gender. They are also asked to indicate whether they have participated in study 1 as well. After this, there are some small differences between the three questionnaires.
Four out of the six groups of respondents are asked to take a good look to screenshots of a Twitter timeline. On this timeline, they do not only see text messages from people, but also a picture of Doutzen Kroes (or Paris Hilton, see appendix 4) wearing Ray Ban sunglasses, accompanied with the text “So happy with my new sunglasses, Ray Ban is great!” The control groups are just confronted with a picture of different Ray Ban sunglasses (see appendix 4).
In appendix 3, the complete questionnaire for the likeable celebrity can be found.
3.7 Plan of analysis
To analyze the results of the experiments, some statistical tests need to be done. Brand image is measured with the help of the five pillars of the Young & Rubicam Brand Asset Valuator. The questions belonging to each pillar are combined and form the score of the pillar. To test whether the questions can be combined into one pillar, a Cronbach's Alpha reliability test will be done.
To test hypothesis 1 (Celebrity endorsement via social media has a positive
influence on brand image), experiment group 1 and 2 (no celebrity endorser) are
compared with group 3,4,5 and 6 (likeable and unlikeable endorser). The groups are recoded first, in order to perform an independent samples t-test later on.
To test hypothesis 2 (Celebrity endorsement via social media has a larger effect on
brand image for women than for men) the difference between experiment groups 3
and 5 (male-likeable and unlikeable celebrity) and experimental group 1 (no endorser-male) is compared with the difference between groups 4 and 6 (female-likeable and unlikeable celebrity) and group 2 (no endorser-female). The difference is measured by subtracting the average score of the no endorser groups from the scores of the groups who are confronted with a likeable or unlikeable celebrity endorser. The differences between the genders are compared with the help of an independent samples t-test.
To test H3a (Likeability of the celebrity has a positive influence on the effect of
celebrity endorsement via social media on brand image.) group 3 and 4 (male and
female-likeable celebrity) are compared with group 1 and 2 (male and female-no celebrity endorser) with the help of an independent samples t-test. To test hypothesis 3b (Not liking a celebrity has a negative influence on the effect of celebrity
endorsement via social media on brand image) group 5 and 6 (male and
4. RESULTS
4.1 Reliability analysis study 1
In order to assess the reliability of a summated scale where several items are summed to form a total score, a Cronbach’s alpha test is needed (Malhotra, 2009, p.319). Cronbach’s alpha is a measure of internal consistency reliability that is the average of all possible split-half coefficients resulting from different splitting of the scale items (Malhotra, 2009, p.319). This test has been performed for each of the five pillars of the Young & Rubican model. The score of the Cronbach’s alpha test can range from 0 to 1 and a value of 0.6 or less generally indicates unsatisfactory internal consistency reliability (Malhotra, 2009, p.319).
From table 5, it can be concluded that the Cronbach’s alpha is sufficient for all five pillars. Therefore, the original variables for the questions have been recoded into five new variables, corresponding with the five BAV pillars differentiation, knowledge, relevance, energy and esteem. This is done by summing up the mean scores of the different questions belonging to a certain pillar and then dividing this score by the amount of items belonging to the pillar. There has been decided to measure brand image by each of the five pillars instead of combining them into one variable, since the pillars represent complete different aspects and there is a lot of difference in scores among the pillars.
BAV Pillar Cronbach's alpha score Reliable?
Differentiation 0.812 Yes
Relevance 0.886 Yes
Esteem 0.799 Yes
Knowledge 0.619 Yes
Energy 0.873 Yes
Table 5: Cronbach's alpha results study 1
Personality trait Doutzen Kroes Paris Hilton
Awful-nice 4.28 2.48
Unfriendly-friendly 4.15 2.80
Unkind-kind 4.13 2.58
Unpleasant-pleasant 4.22 2.50
Table 6: Scores personality traits study 1
4.2 Testing the hypotheses with study 1
To test hypothesis 1 (Celebrity endorsement via social media has a positive influence
on brand image), experiment group 1 and 2 (no celebrity endorser) are compared with
group 3, 4, 5 and 6 (likeable and unlikeable endorser) with the help of an independent sample t-test. Group 1 and 2 are first recoded into group 1, and group 3 till 6 are recoded into group 2. The confidence interval was set at 5%.
The p-values can be found in table 7. The p-values are divided by two, since we only want to know the probability that celebrity endorsement has a positive effect and not whether it has a positive or negative effect (two-sided). However, as can be seen in table 7, the averages of the different aspects of brand image are lower for the respondents who were confronted with a celebrity endorser (either a likeable one or an unlikeable one), which is in contradiction with the hypothesis. This is unexpected, but the differences are not significant.
BAV Pillar Celebrity No
celebrity P-value Differentiation 3.0208 3.2833 .118/2 =.0.059 Relevance 1.7188 1.8750 .455/2 =.228 Esteem 2.6250 2.7000 .595/2 =.298 Knowledge 3.0062 3.1500 .576/2 =.288 Energy 3.2333 3.3250 .574/2 =.287
Table 7: Results hypothesis 1 study 1
To test hypothesis 2 (Celebrity endorsement via social media has a larger effect on
brand image for women than for men) the difference between experiment group 3 and
(female-likeable and un(female-likeable celebrity) and group 2 (female-no endorser). The results of the independent samples t-test (see table 8) show that the differences of esteem, knowledge and energy of brand image are larger for women than for men. However, not all differences are significant on a 5% confidence interval level, and some scores are in contradiction with the hypothesis, as can be seen in table 8. Therefore, we can only accept H2 for the aspect energy.
BAV Pillar Difference
male Difference Female P-values Differentiation -.3348 -.1925 .480/2 =.240 Relevance -.4500 .1375 .012/2 =.006* Esteem -.0700 -.0800 .950/2 =.475 Knowledge .0500 -.3375 .202/2 =.101 Energy .1423 -.3250 .012/2 =.006*
Table 8: Results hypothesis 2 study 1
To test hypothesis 3a (Likeability of the celebrity has a positive influence on the effect
of celebrity endorsement via social media on brand image.) group 3 and 4 (male and
female-likeable celebrity) are compared with group 1 and 2 (male and female-no celebrity endorser) with the help of an independent samples t-test as well. Group 3 and 4 are combined into group 1 and group 1 and 2 are combined into group 2. The results of the test, which are presented in table 9, show that the five aspects of brand image are indeed graded higher after seeing a likeable celebrity. However, on a 5 percent confidence interval, no aspects of brand image differ significantly. Therefore, we cannot accept H3a.
Table 9: Results hypothesis 3a study 1
BAV Pillar No celebrity Likeable
To test hypothesis 3b (Not liking a celebrity has a negative influence on the effect of
celebrity endorsement via social media on brand image) group 5 and 6 (male and
female-unlikeable celebrity endorser will be compared with group 1 and 2 (male and female-no celebrity endorser). First, group 5 and 6 are combined into group 1 and group 1 and 2 are combined into group 2. The results of the independent samples t-test, presented in table 10, show that the aspects of brand image are indeed graded lower after seeing an unlikeable celebrity. On a 5 percent confidence interval level, the aspects differentiation, knowledge and energy differ significantly. Therefore, we can accept H3b partially.
Table 10: Results hypothesis 3b study 1
4.3 Reliability analysis study 2
To test whether the reliability of the summated scale of the second study is sufficient as well, the Cronbach’s alpha test has been performed again. This test has been done for each of the five pillars of the Young & Rubican model. Table 11 shows the results of this reliability analysis.
BAV Pillar Cronbach's alpha score Reliable?
Differentiation 0.864 Yes
Relevance 0.863 Yes
Esteem 0.840 Yes
Knowledge 0.426 No
Energy 0.823 Yes
Table 11: Cronbach's alpha results study 2
BAV Pillar No celebrity Unlikeable
From table 11, it can be concluded that the Cronbach’s alpha is sufficient for four out of the five pillars. Therefore, the original variables for the questions belonging to the pillars differentiation, relevance, esteem and energy have been recoded into four new variables, corresponding with the four BAV pillars. This is done by summing up the mean scores of the different questions belonging to a certain pillar and then dividing this score by the amount of items belonging to the pillar. The pillar knowledge stays represented by two separate variables, which are “familiarity with” and “have had”. To test whether Doutzen Kroes was actually seen as a likeable celebrity and Paris Hilton as a non-likeable celebrity, some control questions were included in the questionnaires. These were about four personality traits, which are represented in a semantic differential scale. The answers could range from 1 (e.g. awful) to 5 (e.g. nice). The results in table 12 show that Doutzen Kroes was indeed ranked as more likeable than Paris Hilton. There is a significant difference between the averages (p=.00 for all four aspects).
Personality trait Doutzen Kroes Paris Hilton
Awful-nice 4.11 2.23
Unfriendly-friendly 4.30 2.51
Unkind-kind 4.16 2.42
Unpleasant-pleasant 4.27 2.23
Table 12: Scores personality traits study 2
4.4 Testing the hypotheses with study 2
To test hypothesis 1 (Celebrity endorsement via social media has a positive influence
on brand image), experiment group 1 and 2 (no celebrity endorser) are compared with
probability of a positive influence. As can be seen in table 13, H1 can only be accepted for the aspect familiarity with.
BAV Pillar Celebrity No celebrity P-values
Differentiation 3.0613 3.1984 .386/2 =.193 Relevance 2.5805 2.2976 .136/2 =.068 Esteem 3.2621 3.1286 .319/2 =.160 Energy 3.2031 3.1508 .680/2 =.340 Familiarity with 3.70 3.31 .075/2 =.0375* Have had 2.57 2.24 .185/2 =.0925
Table 13: Results hypothesis 1 study 2
To test hypothesis 2 (Celebrity endorsement via social media has a larger effect on
brand image for women than for men) the difference between experiment group 3 and
5 (male-likeable and unlikeable celebrity) and experiment group 1 (no endorser-male) are compared with the differences between experimental group 4 and 6 (female-likeable and un(female-likeable celebrity) and group 2(female-no endorser). The differences are indeed larger for female respondents than for male respondents as can be seen in table 14, except for the aspect differentiation. As can be concluded from table 14, esteem is the only aspect of brand image above the confidence interval level of 5%, and therefore H2 can only be partially accepted.
BAV Pillar Differences Male Differences
Female P-values Differentiation -.2549 -.0335 .306/2 =.153 Relevance .1092 .4444 .239/2 =.120 Esteem -.0874 .3456 .020/2 =.010* Energy -.0276 .1178 .421/2 =.211 Familiarity with .3031 .4833 .474/2 =.237 Have had .0909 .5944 .145/2 =.073
Table 14: Results hypothesis 2 study 2
To test hypothesis 3a (Likeability of the celebrity has a positive influence on the effect
of celebrity endorsement via social media on brand image.) group 3 and 4 (male and
celebrity endorser) with the help of an independent samples t-test as well. Group 3 and 4 are combined into group 1 and group 1 and 2 are combined into group 2. The results of the test show that on a 5 percent confidence interval, four out of the six aspects of brand image differ significantly, namely relevance, esteem, familiarity with and have had (see table 15). This means that H3a can be partially accepted.
BAV Pillar No celebrity Likeable celebrity P-value
Differentiation 3.1984 3.2424 .819/2 =.420 Relevance 2.2976 2.7386 .077/2 =.039* Esteem 3.1286 3.5455 .012/2 =.006* Energy 3.1508 3.3864 .153/2 =.077 Familiarity with 3.31 3.84 .040/2 =.020* Have had 2.24 2.80 .080/2 =.040*
Table 15: Results hypothesis 3a study 2
To test hypothesis 3b (Not liking a celebrity has a negative influence on the effect of
celebrity endorsement via social media on brand image) group 5 and 6 (male and
female-unlikeable celebrity endorser will be compared with group 1 and 2 (male and female-no celebrity endorser). First, group 5 and 6 are combined into group 1 and group 1 and 2 are combined into group 2. The average scores on the aspects of brand image indeed are higher for the respondents who were confronted with just a picture of Ray Ban instead of a Tweet of Paris Hilton (see table 16). However, not all differences are significant. Therefore, H3b can only be accepted for the aspect differentiation.
BAV Pillar No celebrity Unlikeable celebrity P-values
Differentiation 3.1984 2.8760 .090/2 =.045* Relevance 2.2976 2.4186 .591/2 =.296 Esteem 3.1286 2.9721 .305/2 =.153 Energy 3.1508 3.0155 .345/2 =.173 Familiarity with 3.31 3.56 .317/2 =.159 Have had 2.24 2.35 .706/2 =.353
5. CONCLUSION
Celebrity endorsement is a well-studied topic. However, social media celebrity endorsement is relatively new and has not been studied a lot yet. Since social media are gaining more importance nowadays and has many users, this new type of celebrity endorsement might influence the success of brands a lot. This paper has tried to evaluate the effect of this new phenomenon.
To test all hypotheses, two studies were conducted. In study 1, respondents were confronted with a Tweet of a celebrity about Nespresso, containing just text. In study 2, respondents were confronted with a Tweet of a celebrity about Ray Ban, containing text and a picture of the celebrity wearing Ray Ban sunglasses. The main conclusion is that celebrity endorsement via social media in general has no significant effect on brand image.
It was expected that celebrity endorsement via social media has a positive influence on brand image. Brand image was measured with the help of the Young & Rubicam Brand Asset Valuator. This model consists out of five pillars, which are differentiation, relevance, esteem, knowledge and energy (Mizik & Jacobson, 2008). In study 1, these pillars are used. In study 2, these pillars are also used, but since the Cronbach’s alpha of knowledge was not reliable, this pillar was split into “familiarity with” and “have had”.
It was expected that celebrity endorsement via social media influences brand image positively (H1), since people find these online recommendations more reliable than offline recommendations. The results of study 1 show that in this case, celebrity endorsement via social media had little influence on brand image. Being confronted with a picture of the brand Nespresso and being confronted with a Tweet of Doutzen Kroes or Paris Hilton did not result into significant differences in brand image. The results of study 2 show that most of the averages of the different aspects of brand image (except differentiation) are indeed higher when respondents were confronted with a celebrity endorser. However, only the aspect familiarity with differs significantly.
product recommendations than men.
Both studies do not show large differences between men and women. The results of the first study show that women are more influenced by celebrity endorsement than man for three out of the five aspects of brand image. However, brand image is also sometimes more negatively influenced. The only aspect that differs significantly is energy, which leads to only partial support for the suggestion hat women are more influenced than men by celebrity endorsement via social media. The results of study 2 also show little significant differences; only esteem is influenced more for women than for men.
A reasonable explanation of the fact that almost no aspects of brand image differ significantly between celebrity endorsement and no celebrity endorsement is because we used two celebrities; a likeable one and an unlikeable one. A likeable celebrity might result into higher averages, but a less likeable celebrity into lower averages. To find out whether this is true, some other tests have been performed. It was expected that the likeability of the celebrity has a positive influence on brand image (H3a). Study 1 showed that using a likeable celebrity versus just showing a picture of the brand has indeed a positive influence on brand image. However the differences are not significant. Study 2 showed that four out of the six aspects of brand image differ significantly, namely relevance, esteem, familiarity with and have had. This means that after being confronted with a Tweet including a text message and a picture of Doutzen Kroes wearing Ray Ban sunglasses, the respondents found Ray Ban more relevant to them (relevance) and see Ray Ban as more reliable, very popular and of high quality (esteem). They also indicated that they are more familiar with the brand Ray Ban and have had more Ray Ban products (knowledge).
The results of the second study are different; only differentiation differed significantly. This means that after seeing a Tweet of Paris Hilton including text and a picture of her wearing Ray Ban sunglasses, respondents see Ray Ban as less distinctive from other brands then when being confronted with a picture of Ray Ban sunglasses.
6. MANAGERIAL RECOMMENDATIONS
This study has made a start in studying the effects of celebrity endorsement via social media. Even though it is just a start, some managerial recommendations can be drawn from this research.
First of al, it is important that companies should be careful in selecting a celebrity. Celebrity endorsement in general might not be effective. As both studies have shown, using an unlikeable celebrity might result in a less favorable brand image. Companies should focus on selecting a likeable spokesperson for their brand. Market research is necessary for this. It is advised to select a celebrity who has a good image, so that the possibility on scandals is low, since this can also influence the image of the brand.
7. LIMITATIONS AND FURTHER RESEARCH
In this chapter, the limitations of the two conducted studies will be discussed. First of all, the sample is quite small. Each of the six experimental groups consists of only (around the) twenty people and this can lead to insignificant results. These results might not be representative for the population. The sample mainly consists out of young people, which might influence the results as well. Further research might be held under different age categories.
Furthermore, the research setting may also have leaded to the insignificant results. First of all, there has been chosen to use an image of a Twitter timeline. In study 1, a tweet of Doutzen Kroes (or Paris Hilton) was showed on a Twitter timeline, but also nine other Tweets were displayed. This may have caused that some people did not saw the Tweet, even though the respondents were asked to read all Tweets carefully. However, when the Tweet of Doutzen Kroes (or Paris Hilton) was shown solely, this would have not been realistic, as people always see the complete timeline, and not just one Tweet. It also would not have been realistic when the Tweet was displayed several times on the timeline, or several celebrities Tweeted about the product. To overcome this problem of possible low visibility, the Tweet of the celebrity was made more visible in the second study; instead of nine other Tweets, only four other Tweets are presented. The Tweet of the celebrity also included a picture, so that it could not be missed. Furthermore, respondents were asked to indicate from which celebrity they saw a Tweet. Respondents who did not know the correct answer were excluded from the experiment.
Testing the hypothesis can also be done with the help of other social media networks. In future research other social networks can be used, which might lead to other results. For this study only Twitter was used, due to the relatively small possible scope of the research.
Other limitations of this study are the selection of the celebrity. Although these celebrities were shown to be most likeable and least likeable in the pre-test, others also might have been good to use. The chosen celebrities were both females, and this might influenced the outcome. Gender of the celebrity endorser is a good point for further research. Furthermore, there might be a difference between type of endorsers (e.g. sportspeople, actress, model).
REFERENCES
Amos, C., Holmes, G. and Strutton, D., 2008, Exploring the relationship between celebrity endorser effects and advertising effectiveness: A quantitative synthesis of effect size, International Journal of Advertising, Vol. 27(2): p209-234
Berthon, P.R., Pitt, L.F., Plangger, K., Shapiro, D., Marketing meets Web 2.0, social media and creative consumers; implications for international marketing strategy,
Business Horizons, Vol. 55 (3): p. 261-271
Biswas, D., Biswas, A., Das, N., 2006, The differential effects of celebrity and expert endorsement on consumer risk perceptions: The Role of Consumer Knowledge, Perceived Congruency, and Product Technology Orientation, Journal of Advertising, Vol. 35 (2): p.17–31
Biswas, S., Hussain, M., O’Donnellm, K., 2009, Celebrity Endorsements in Advertisements and Consumer Perceptions: A Cross-Cultural Study, Journal of
Global Marketing, Vol. 22 (2): p.121-137
Colliander, J., Dahlén, M., 2011, Following the Fashionable Friend: The Power of Social Media, Journal of Advertising Research, Vol. 51 (1): p.313-320
Debevec, K. Kernan, J.B., 1984, More evidence on the effects of a presenter’s attractiveness: some cognitive, affective and behavioral consequences, Advances in
Consumer Research, Vol. 11: p.127-132.
Erdogan, B.Z., 1999, Celebrity Endorsement: A Literature Review, Journal of
Marketing Management, Vol. 15(4): p291-314
Friedman, H.H., Friedman, L., 1979, Endorser effectiveness by product type, Journal
Greenwood, D.N., Long, C.R., 2011, Attachment, Belongingness Needs, and Relationship Status Predict Imagined Intimacy with Media Figures, Communication
Research, Vol. 38(2): p. 278-297
Hartmann, T., Goldhoorn, C., 2011, Horton and Wohl revisited; exploring vieuwers’ experience of parasocial interaction, Journal of communication, Vol. 61(6):p.1104-1121
Hennig-Thurau, T., Malthouse, E.C., Friege, C., Gensler, S., Lobschat, L., Rangaswamy, A., Skiera, N., 2010, The Impact of New Media on Customer Relationships, Journal of Service Research, Vol. 13 (3): p.311-330
Horton, D., Wohl, R., 1956, Mass communication and para-social interaction: Observations on intimacy at a distance, Psychiatry, Vol. 19: p.215 – 229
Hutton, G., Fosdick, M., 2011, The Globalization of Social Media, Journal of
Advertising Research, Vol. 51(4): p564-570
Jain, V., Sudha, M., Daswani, A., 2009, Customer Perception about celebrity endorsement in television advertising for retail brands, Journal of brand management, Vol. 6 (3): 7-25
Kamins, M.A., 1990, An investigation into the ‘match-up’ hypothesis in celebrity advertising: when beauty may be only skin deep, Journal of Advertising, Vol.19(1): p. 4–13
Katona, Z., Zubcek, P.P., Sarvary, M., 2011, Network effects and personal influences; the diffusion of an online social network, Journal of Marketing Research, Vol. 48(3): p.425-443
Leenders, R.A.J., 2002, Modeling Social Influence Through Network
Malhotra, N., 2009, Marketing research: an applied orientation, Pearson education, 6th edition
McCracken, G., 1989, Who is the celebrity endorser? Cultural foundations of the endorsement process, Journal of Consumer Research, Vol.16(3):p.310-321 Mehta, A., 1994, How advertising response modeling (ARM) can increase ad effectiveness, Journal of advertising Research, Vol. 34(3): p.62-74
Mizik, N., Jacobson, R., 2008, The Financial Value Impact of Perceptual Brand Attributes, Journal of Marketing Research, Vol. 45(1): p. 15-32
Monahan, J.L., 1998, I don’t know it but I like you: the influence of non-conscious affact on person perception, Human communication research, Vol. 24(4): p.480-500 Petroshius, S.M., Crocker, K.E., 1989, An Empirical Analysis of Spokesperson Characteristics on Advertisement and Product Evaluations, Journal of the Academy of
Marketing Science, Vol. 17 (3):p.217-226
Pfeiffer, M., Zinnbauer, M., 2010, Can Old Media Enhance New Media? How Traditional Advertising Pays off for an Online Social Network, Journal of Advertising
Research, Vol. 50 (1): p42-49
Pooja, M., Black., J.E., Jianmei, C., Berger, P.D., Weinberg, B.D., 2012, The Impact of Social Media Usage on Consumer Buying Behavior, Advances in Management, Vol. 5(1):p14-22
Premeaux, S.R., 2009, The Attitudes of Middle Class versus Upper Class Male and Female Consumers Regarding the Effectiveness of Celebrity Endorsers, Journal of
Promotion Management, Vol. 15 (1/2): p.2-21
Thorson, K.S., Rodgers, S., 2006, Relationships between blogs as eWOM and Interactivity, Perceived Interactivity, and Parasocial Interaction, Journal of Interactive
Advertising, Vol. 6(2): p.39-50
Till, B.D., Busler, M., 2000, The match-up hypothesis: physical attractiveness, expertise, and the role of fit on brand attitude, purchase intentions, and brand beliefs,
Journal of Advertising, Vol. 29(3): p.1–13
APPENDIX 1; QUESTIONNAIRE
Hello, My name is Linda Zuidema and I am a Master student at the University of Groningen. At the moment I am conducting a study on celebrity endorsement via social media. I have developed the following questionnaire in order to answer the research questions. By filling in the questionnaire you participate in my research and help me finish my thesis. In order to participate in this study, you should be active on social media, for example Facebook or Twitter. Filling in the questionnaire will take approximately 7-10 minutes. I would really appreciate if you dedicate this time and try to answer the following questions precisely. All answers to the questions are anonymous and confidential. Thank you in advance.
1. Age: .. 2. Gender: Male
Female
Please indicate to which extent you agree with the following statements on a scale from 1-5 (in which 1=totally disagree, 5=totally agree).
1. Nespresso is unique 2. Nespresso is distinctive
3. Nespresso coffee are very different from other brands 4. I purchase Nespresso coffee
5. Nespresso is relevant to me
6. Nespresso is the leading brand in the coffee industry 7. Nespresso is important for me
8. Nespresso is reliable
9. Nespresso is a high quality brand
10. Nespresso is the most popular coffee brand 11. I have tried Nespresso coffee
12. I am familiar with the brand Nespresso
13. Nespresso can meet consumers’ needs in the future 14. Nespresso is innovative
15. Nespresso is able to adapt to changing tastes and needs
Please indicate which of the two extremes fits Doutzen Kroes best. 1. Awful 0 0 0 0 0 Awful
2. Unfriendly 0 0 0 0 0 Friendly 3. Unkind 0 0 0 0 0 Kind
4. Unpleasant 0 0 0 0 0 Pleasant.
APPENDIX 3; QUESTIONNAIRE STUDY 2
Hello,
My name is Linda Zuidema and I am a Master student at the University of Groningen. At the moment I am conducting a second study on celebrity endorsement via social media. This study is an extension to my previous study on the same topic. I have developed the following questionnaire in order to answer the research questions. By filling in the questionnaire you participate in my research and help me finish my thesis. In order to participate in this study, you should be active on social media, for example Facebook or Twitter. Filling in the questionnaire will take approximately 3-4 minutes. I would really appreciate if you dedicate this time and try to answer the following questions precisely. All answers to the questions are anonymous and confidential. Thank you in advance.
1. Age: .. 2. Gender: Male
Female
3. Have you participated in my previous study? (This study concerned some questions about the brand image of Nespresso). Yes/No
4. The following question is about the Twitter timeline you just saw. Please indicate who posted the Tweet (including a picture) about the Ray Ban sunglasses.
5. Please indicate to which extent you agree with the following statements on a scale from 1-5 (in which 1=totally disagree, 5=totally agree).
1. Ray Ban is unique 2. Ray Ban is distinctive
3. Ray Ban sunglasses are very different from other brands 4. I purchase Ray Ban sunglasses
5. Ray Ban is relevant to me
6. Ray Ban is the leading brand in the sunglasses industry 7. Ray Ban is important for me
8. Ray Ban is reliable
9. Ray Ban is a high quality brand
10. Ray Ban is the most popular sunglasses brand 11. I have tried Ray Ban sunglasses
12. I am familiar with the brand Ray Ban
13. Ray Ban can meet consumers’ needs in the future 14. Ray Ban is innovative
15. Ray Ban is able to adapt to changing tastes and needs
6. Please indicate which of the two extremes fits Doutzen Kroes best. 1. Awful 0 0 0 0 0 Awful
2. Unfriendly 0 0 0 0 0 Friendly 3. Unkind 0 0 0 0 0 Kind
4. Unpleasant 0 0 0 0 0 Pleasant.
