Do Celebrities Affect Brands They Endorse?

Kathleen Pinter Ashley Stubbs Jessica Schwartz Christina Thomas Jessica Wesley

Introduction
According to a study done by Marina Choi and Nora Rifon, celebrities are individuals who are symbolic icons, popular in the culture, and transfer their symbolic meaning to the products they endorse in advertising. Advertisers are well aware of the positive influence that celebrities can bring to a persuasive message, because approximately 25% of all American television commercials feature celebrities (Choi, 2007). It is evident that advertising uses these celebrities to aid in their brand being purchased. Although some celebrities share common characteristics a celebrity for one brand may be completely wrong for the next based on the public’s perception of the celebrity. Through our research and survey we hope to determine if a consumer’s decision about a brand is affected by a celebrity endorsing the brand. We want to test if celebrities’ behaviors positively or negatively affect the brand image male and female consumers have for the products the celebrities endorse. The American public has been bombarded with celebrity mishaps ranging from stays in rehabilitation facilities and spending time in jail. Nowadays, celebrities are plastered all over newspapers, magazines, television, and advertisements for specific products. This current explosion of celebrity advertising causes us to look further into how celebrities affect the brand they advertise. We want to determine if actions of celebrities, whether good or bad, affect consumer thoughts when buying the product. For example, if a celebrity is endorsing Coca-Cola and is seen on the cover of People in rehabilitation will that affect the sale of the brand? Are male and female celebrities treated differently for their bad behaviors as far as consumers are concerned? Does the sex of the consumer affect his or her view of these celebrity endorsers? Additionally, how does one feel about celebrity endorsers in general?

2

Heider’s theory states that if a perceiver (consumer) likes a certain celebrity and that celebrity likes the product they endorse then in turn the perceiver will like the overall brand. If the consumer does like a certain brand, but does not like the celebrity that endorses it, according to Heider’s balance theory the consumer would have a negative image of the brand. Through our research we want to test if consumers’ responses match Heider’s balance theory. In a study by Joshua Stuart, his results showed that celebrities provide us with heroic role models in an age of conformity. These "role models" are used in a variety of product endorsements that affect consumers’ opinion of the brands. Nike alone spent over $1.4 billion on celebrity endorsements in 2003-2004 (Stuart, 2007). The effectiveness of endorsements can be categorized by perceived expertise, trustworthiness, and physical attractiveness of the celebrity. Stuart’s study showed that a combination of all three attributes is the most effective in selling a product. Furthermore, this study found that female celebrities were most influential for male test subjects, and vice versa for females, in terms of changing their perception about advertised products used in this study. The researchers found that celebrities who can also be characterized as heroes, being admired for their ability or bravery, are the most effective in selling a brand. According to the study, the celebrity who subjects most considered to be a hero is Michael Jordan throughout his Nike, Hanes, and Space Jam endorsing (Stuart, 2007). Although celebrities aid in sales for a particular product, they can likewise hurt sales. David Moore, president of Leo Burnett Canada, thinks that whenever a brand decides to have a celebrity endorsement there is always a risk because if the celebrity has negative exposure, that adversely affects the brand (Prashad, 2005). "When Lance Armstrong faced allegations this summer of performance-enhancing drug use in 1999, which the cyclist denied, organizations supported the athlete because fans associated Armstrong with powerful values and he was hugely

3

respected by fans" (Prashad, 1). While the people cited in this article hold extreme views of celebrity endorsers, we want to use our study to see if these views hold true across different celebrities and different product categories. The study by Sarah Roberto was held in regards to Martha Stewart’s insider-trading scandal. The purpose of the study was to see if Martha Stewart’s conviction affected the consumer’s view of Martha Stewart Living Omnimedia Corporation. The researcher tested the views of consumers about her credibility before and after her trial, and noted the differences. It was found that consumers’ view of her credibility before and after her trial differed between people who would purchase her products versus those who would not purchase. “A classic study conducted by Mizerski (1982) found respondents were more likely to process negative or unfavorable information as opposed to favorable information. Unfavorable information tended to trigger a stronger stimulus than favorable information” (Roberto, 3). Therefore, subjects who said they would not purchase Martha Stewart’s products were more likely to view her as being less credible following her trial because this negative information was more prominent to them (Roberto, 2006). In another study, an experiment was performed by Therese Louie to examine how a firm’s financial performance was influenced when celebrity endorsers become involved in undesirable events. An experiment was conducted that asked respondents to attach a level of blame to celebrities for each type of undesirable event they were associated with, such as drinking and driving. The effect on the brand in such events was then analyzed. According to this study, if, for example, a celebrity participates in illegal drug use, the target market for their endorsement will become more aware of this celebrity though negative attention. The experiment’s results suggested that undesirable actions by a celebrity endorser negatively

4

affected the brand image. Overall, the effectiveness of an endorser depends jointly on awareness and attitudes towards the celebrity. For example, if a consumer is more aware of a celebrity’s negative behavior it will negatively affect their attitude toward the product the celebrity endorses (Louie, 2001). We anticipate the results of our survey will confirm that negative celebrity behavior will negatively affect the way respondents view the brand the celebrity endorses, and, likewise, positive celebrity behavior will positively affect respondents’ views of brands. Using Heider’s Balance Theory we expect that cognitive dissonance or cognitive resonance will be created. By evaluating respondents’ opinions of celebrity endorsers, we plan to see how celebrity behavior affects consumers’ purchasing decisions of brands.

5

Study 1 Methods
Participants In this survey participants were part of a convenience sample. A convenience sample is a sample where participants are selected at the convenience of the researcher. With a convenience sample, the researcher makes no attempt, or only a limited attempt, to insure that this sample is an accurate representation of some larger group or population. Although a convenience sample was used for this survey it still provides useful information. However, for the most accurate results replication would be necessary. The participants chosen were 50 students from the University of Illinois Urbana-Champaign. These students were at least eighteen years or older and consisted of both males and females. Students were surveyed from sororities, classes, campus acquaintances, and other organizations, such as the National Society of Collegiate Scholars and the American Advertising Federation.

Procedure The complete survey was handed out in person to friends, roommates, and classmates. Every member of the group contacted members of various sororities and other campus organizations to get a wide-range, diverse response. Then, students were asked to voluntarily respond to the questionnaire and were not asked to write their name on the survey to make them feel comfortable about their answers as well as attempt to get accurate and honest responses. The participants were allowed as much time as they needed to complete the survey. After participants completed the survey they handed it back and their answers were recorded.

6

Materials A variety of questions were asked in the questionnaire to keep respondents interested and to ensure they responded accurately. All of the participants were asked close-ended questions that asked them to choose the answer closest to his/her viewpoint. These questions are specific and have limited responses, and therefore are more reliable, have less ambiguity, and are easier to analyze. The different types of questions used were likert scale, frequency and staple scales, as well as simple dichotomy questions. The survey consisted of thirteen questions that covered a variety of topics. One topic focused on male versus female celebrity endorsers as well as positive versus negative celebrity endorsers. Another topic focused on the amount of persuasion

celebrities have in advertising and how effective they are on consumers. Last, fictitious endorsements linking existing celebrities to very neutral products were created. Respondents were asked to rate how positively or negatively they felt about the endorser-brand relationship.

7

Study 1 Results
Question 2 asked respondents the difference in views on effectiveness of celebrity endorsement. The question we asked on the survey was a Likert Scale that said: I think celebrity endorsements are ineffective. We want to compare the views of males and females by frequency distribution. Strongly Disagree 3 7 Disagree 11 21 Uncertain 2 4 Agree 2 0 Strongly Agree 0 0

Males Females

Results show that the majority of male and female respondents disagreed that celebrity endorsers are ineffective. Therefore, the majority believes that celebrity endorsers are effective according to this question. We then reverse scored the results because the question is negative and the rest of the Likert Scale questions are positive. This helps us to see if respondents were paying attention when they filled out the survey. It also allows us to compare the result to this question to the rest of the survey. We then calculated the mean and standard deviation.
Mean 4.0 SD .70

Ineffective endorsements

Question 3 asks if celebrity endorsers affect the sale of the product with which they are associated and Question 9 asks if the respondent intentionally purchases certain products because of their association with a celebrity. For Question 3, we did not give respondents a specific product to keep in mind when answering the question to prevent the answers from being skewed due to preconceived notions the respondents may have had about a certain brand or product. We are looking to see if the responses to Questions 3 and 9 were similar because they ask similar

8

questions, so we thought the responses would be similar. Since we did not ask in Question 3 whether respondents thought celebrity endorsers positively or negatively affected sales, we used the responses to Question 9 to determine this.
Mean 4.0 2.22 SD .639 .887

Affect sales Intentionally purchase

These results show that our respondents believe celebrity endorsements affect sales of products, but they do not purchase products because of their association with celebrities. This leads us to believe that respondents feel they are not affected by endorsements, but the general public is. We then did a cross-tabulation of the results for Questions 3 and 9.

Question 11 on the survey asked respondents how certain celebrity incidents affects their decision to purchase a product that was being endorsed by that particular celebrity. Respondents gave their answers by ranking each situation with a number from one to five, one meaning “would stop using product” and five meaning “would greatly increase purchases of the product.” We compared the mean scores of all the incidents and compared male versus female as well. Standard deviations are given in parentheses next to the mean scores.
Drunk Driving Adopt Children Late Partying Tabloid coverage Eating Disorder Charitable donation Public Break Up Won Award an

Males Females Total SD

2.44 2.19 2.28 .61

3.00 3.25 3.16 .51

2.89 2.78 2.82 .44

3.06 3.16 3.12 .66

2.72 2.69 2.70 .61

3.61 3.81 3.74 .63

2.89 3.06 3.00 .45

3.39 3.50 3.46 .73

The results of these data show that drunk driving would most likely decrease purchases of a product. A large charitable donation would most likely increase purchases of a product. In general, males and females had similar views on how celebrity behavior affects their purchasing.

9

Most of the results indicate that celebrity behavior would either decrease purchasing or has no effect on purchasing.

Question 12 on the survey was a set of fictitious endorsements. The respondent was asked to rate their feelings toward the celebrity endorsements. We want to explore the differences between how males and females responded to celebrities of the same sex as well as the opposite. For each question, the dependent variables are the celebrity and the product they are endorsing, and the independent variable is the sex of the respondent. We chose the products based on what we perceive them as being gender neutral, that is, we thought male and female respondents would be equally likely to purchase these products. In choosing celebrities, we selected a mix of celebrities who we thought had either positive or negative public images. Lindsay Lohan, Tom Cruise, and Kevin Federline are celebrities we consider to have a negative image. Celebrities we see as having positive public images are Justin Timberlake, Jennifer Aniston, and Jessica Alba. Lindsay Lohan- Pepsi
-3 (Strongly Dislike) -2 -1 +1 +2 +3 (Strongly Like)

Males 2 4 4 4 2 2 Females 8 6 7 7 4 0 These data shows that more females dislike the celebrity endorsement than males. Females would be less likely to purchase Pepsi if she endorsed it.

Justin Timberlake- Apple
-3 (Strongly Dislike) -2 -1 +1 +2 +3 (Strongly Like)

Males 0 1 1 6 7 3 Females 2 0 0 8 13 9 These data shows that both males and females like this celebrity endorsement and would likely purchase Apple products.

10

Jennifer Aniston- Verizon Wireless
-3 (Strongly Dislike -2 -1 +1 +2 +3 (Strongly Like)

Males 0 4 5 3 2 4 Females 0 1 2 8 13 8 These data shows that no one strongly dislikes the endorsement, but the data was spread out among the choices. Tom Cruise- McDonald’s
-3 (Strongly Dislike -2 -1 +1 +2 +3 (Strongly Like)

Males 1 8 5 2 1 1 Females 8 11 8 4 1 0 These data tells us that both males and females dislike this endorsement. Even though we had more female respondents, a majority of the males we surveyed had comparable feelings of dislike for Tom Cruise endorsing McDonald’s as female respondents. Kevin Federline- Subway
-3 (Strongly Dislike -2 -1 +1 +2 +3 (Strongly Like)

Males 5 8 2 3 0 0 Females 15 9 7 0 0 1 Again, everyone dislikes this endorsement. Both genders would most likely not purchase the product. Jessica Alba- Evian
-3 (Strongly Dislike -2 -1 +1 +2 +3 (Strongly Like)

Males 0 0 2 6 6 4 Females 1 0 4 13 9 5 Both genders liked this endorsement. They would most likely purchase this product.

Study 1 Discussion

11

Heider’s balance theory can be described saying a Person who likes an Other person will be balanced by the same valence attitude on behalf of the other. Symbolically, P (+) > O and P < (+) O results in psychological balance. This can be extended to objects (X) as well, thus introducing triadic relationships. If person P likes object X but dislikes other person O, what does P feel upon learning that O created X? Therefore, we suspected through Heider’s balance theory that if respondents (P) like the celebrity (O) then they then will like the endorsed product (X). However, if respondents (P) dislike the celebrity (O) then they won’t like the endorsed product (X). Balanced on their attitude celebrities portrayed negatively would receive low scores for the product they endorse, and celebrities portrayed positively would receive high scores for the product they endorse. To test if Heider’s theory is true a variety of questions were asked to participants and their results were recorded. In question two we asked if celebrity endorsements were ineffective, our results showed us that they were effective because eleven males, which is over sixty percent of the males that we surveyed, said that they disagreed with this statement. Of female respondents, about sixty-five percent put that they also disagreed that celebrity endorsers are ineffective. This tells us that both males and females agree that celebrity endorsements are effective in advertising. Questions five and seven both dealt with a male endorser for a product, one which was considered a gender neutral product (cell phone) and the other which was considered a masculine product (car). After analyzing our results, we found that respondents feel that the endorser’s gender does not have to correlate to the gender association of the product. Questions six and eight both dealt with a female endorser for a product, one which was considered a gender neutral product (soft drink) and the other which was considered a feminine product (hair product). Once

12

again, we found that respondents feel that the endorser’s gender does not correlate to the gender association of the product. Question three asked respondents if in general they think that celebrity endorsers affect the sale of products that they are associated with. Question nine asked respondents if they personally purchased products because of their celebrity endorsers. We thought that these responses would be similar; however the mean for question three was 4.0, whereas the mean for question nine was 2.22. This suggests that the respondents believe that celebrity endorsers do affect the sales of the products that they endorse, but said they do not personally purchase products because of celebrity endorsements. This implies that people think they are immune to celebrity endorsements, whereas the general population is affected. We created question twelve because we wanted to provide respondents with examples of various celebrities and how they felt about hypothetical product endorsements. We suspected through Heider’s balance theory that celebrities that were portrayed negatively would produce negative opinions toward the brand they are associated with. For example, we thought that since Lindsay Lohan is portrayed negatively then respondents would automatically perceive Pepsi negatively because Lohan is associated with it. Through our results we found that to be true. Justin Timberlake and Jennifer Aniston received the highest scores for their endorsements, whereas Kevin Federline and Tom Cruise received the lowest. This may have occurred because Jennifer Aniston and Justin Timberlake are positively displayed through charitable functions and movies, and this reflected in the respondent’s positive answers. On the other hand, Kevin Federline and Tom Cruise, two celebrities displayed negatively in the media, received more negative responses from respondents.

13

A way to improve our survey if we were to perform this again would be to change how we had respondents answer questions eleven and twelve. In order to make a better comparison, we would compare certain celebrities with their specific behavior. For example, in question twelve we give an example of a fictitious Lindsay Lohan advertisement with Pepsi. We would have liked to compare this advertisement with what people rated for drunk driving and late night partying in question eleven, but were unable to compare these results. If we were able to provide this type of comparison we believe that the results would suggest that the scores for Lindsay Lohan would be closely related to the scores for late night partying and drunk driving. However, since we were unable to make this comparison we are unable to find out if our prediction is accurate for this study. We also would have been able to make comparisons between other celebrities in question twelve and actions they have committed in the past listed in question eleven. When we wrote the survey we were still unsure how we were going to analyze our results and therefore did not think to code these questions in such a manner to compare them. If we produced this survey again we would attempt to code questions eleven and twelve on the same scale; either five point or six point. Another way to improve our survey would be to have a more equal gender distribution among the respondents as well as a larger sample size approximately 80 to 100. Although we had a decent ratio of male to female respondents it was not as close to being half and half as we would have liked. This may have occurred because we are all females involved in sororities and it was more convenient for us to get female respondents rather than male respondents. If we were to give out the survey again we would attempt to find more males willing to take our survey to even out the male to female ratio.

14

The next step in finding support for our hypothesis was to conduct a controlled experiment to provide us with detailed results pertaining to students’ feelings towards celebrity endorsements. While our survey provided us with information regarding consumers’ general opinions about celebrity behavior and endorsements, our experiment will actually test the data collected in our initial surveys using a specific example of a celebrity endorsement. In the experiment we will be manipulating our variables to have two independent variables and numerous dependent variables to test if celebrities have an effect on brands they endorse.

15

Study 2 Methods
Participants In study 2 an experiment was conducted using participants that were part of a convenience sample. A convenience sample is a sample where participants are selected at the convenience of the researcher. With a convenience sample, the researcher makes no attempt, or only a limited attempt, to insure that this sample is an accurate representation of some larger group or population. Although a convenience sample was used for this experiment it still provides valid and useful information. However, for the most accurate results replication would be necessary. The participants sampled were eighty students from the University of Illinois Urbana-Champaign. These students were at least eighteen years or older and consisted of both males and females. The students were chosen randomly from various sororities, classes, campus acquaintances, and other organizations. Procedure The experiment consisted of three parts: a celebrity description, an advertisement, and a survey asking questions about the previous two parts. To control for gender differences, only male participants read about the male celebrity and female participants only answered questions about the female celebrity. Using this guideline, the convenience sample of participants saw either a “good” or “bad” celebrity description. With the exception of the celebrity’s name, the male and female descriptions in each scenario were identical. There was only one female and one male advertisement, so regardless of what description participants saw, everyone saw the same advertisement according to their own gender. So, for example, all female participants saw the same advertisement featuring the female “celebrity.” There was only one questionnaire; therefore all participants answered the same questions.

16

The questionnaire was handed out in person during classes to friends and roommates. Students were asked to voluntarily respond to the questionnaire. They were not asked to write their name on it to make them feel more comfortable about their answers as well as to ensure we received accurate and honest responses. Participants were allowed as much time as needed to complete the survey. Respondents first read the celebrity description which was randomly assigned to them, then looked at the celebrity advertisement, then responded to the questionnaire with their personal opinions. Materials The independent variables in this experiment are the sex of celebrities and participants as well as the type of scenario they received. The independent variables were manipulated based on the gender of the advertisement and respondent. For example, the female advertisement was seen by only females. The dependent variables in the experiment are the questions in the questionnaire. Fictitious celebrities were used to prevent respondents from having a biased opinion towards an existing celebrity. For the “good” celebrities we described the celebrity as being talented, polite, and personable. The “bad” celebrities were described as being egotistical, irresponsible, and party animals. Situations were exaggerated to ensure our respondents would have definite positive or negative opinions of the celebrity so we would have as few “uncertain” responses as possible in the questionnaire. Fictitious celebrities were created linking existing celebrities to what we consider to be a gender-neutral product, Coca-Cola. We believe that men and women are equally likely to purchase Coca-Cola products, which would help ensure against gender biases. Following the description, respondents saw a fake advertisement for Coca-Cola featuring the celebrity. We

17

created the advertisements using Photoshop from pictures we found online of a male or female and the Coca-Cola logo and can. We tried to make the male and female advertisements look as similar as possible as well as have both of them smiling in the advertisement. We did this to ensure as much congruity between advertisements as possible. We used a Likert scale and a semantic-differential scale for questions for the controlled experiment. We used this variety of questions to keep the respondents interested and to ensure they responded accurately. We asked all of our participants’ closed-ended questions and had them choose the one closest to his or her viewpoint. These questions are specific and have limited responses and therefore are more reliable, have less ambiguity, and are easier to analyze. The questionnaire consisted of twelve questions that covered consumer’s opinions about the brand and celebrity alone as well as the brand and celebrity together. Respondents were asked to rate how positively or negatively they felt about the endorser-brand relationship based on the description and advertisement, and we recorded their data.

18

Study 2 Results
Coke (Determine if participants like Coke as a brand)

There is a significant effect of the celebrity’s image, F(1, 76)=6.18, p < .05. The bad celebrity (M=4.22) effected slightly more favorable attitudes than the good celebrity (M=3.75). There is a significant effect of the celebrity’s gender, F(1,76)=4.69, p < .05. The female celebrity (M=4.19) effected slightly more favorable attitudes than the male celebrity (M=3.78). There is a marginally significant celebrity image by celebrity gender interaction effect, F(1,76)=.41, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image.

19

Ad (Determine if participant liked advertisement)

There is a significant effect of the celebrity image, F(1, 76)=11.95, p < .05. The good celebrity (M=3.11) effected slightly more favorable attitudes than the bad celebrity (M=2.44). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.007, p < .10. The male celebrity (M=2.78) effected slightly more favorable attitudes than the female celebrity (M=2.77). There is a marginally significant celebrity image by celebrity gender interaction, F(1,76)=.003 , p <.10 . As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image.

20

Celeb (Approve of celebrity behavior)

There is a significant effect of the celebrity image, F(1, 76)=46.577, p < .05. The good celebrity (M=4.048) effected slightly more favorable attitudes than the bad celebrity (M=2.7). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.869 p < .10. The female celebrity (M=3.47) effected slightly more favorable attitudes than the male celebrity (M=3.28). There is a slightly significant celebrity image by celebrity gender interaction, F(1,76)=.1.534, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image.

21

Positive (Advertisement affected respondents positively)

There is a significant effect of the celebrity image, F(1, 76)=7.68, p < .05. The good celebrity (M=3.11) effected slightly more favorable attitudes than the bad celebrity (M=2.59). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.05, p < .10. The male celebrity (M=2.87) effected slightly more favorable attitudes than the female celebrity (M=2.83). There is a slightly significant celebrity image by celebrity gender interaction, F(1,76)=4.31, p < .05. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image. Behavior (Like celebrity behavior)

22

There is a significant effect of the celebrity image, F(1, 76)=35.090, p < .05. The good celebrity (M=3.753) effected slightly more favorable attitudes than the bad celebrity (M=2.51). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.083 p < .10. The female celebrity (M=3.162) effected slightly more favorable attitudes than the male celebrity (M=3.101). There is not a significant celebrity image by celebrity gender interaction F(1,76)=1.129, p < .10. As can be seen in the above table, the effect of the celebrity’s image did not depend at all on whether or not the celebrity, and therefore the respondent, was male or female. Image (Negative behavior reflected negative brand image)

There is a significant effect of the celebrity image, F(1.76)=29.69, p < .05. The good celebrity (M=3.89) effected more favorable attitudes than the bad celebrity (M=2.70). There is a significant effect of the celebrity’s gender, F(1,76)=.634, p < .05. The female celebrity (M=3.38) effected slightly more favorable attitudes than the male celebrity (M=3.21). There is a slightly significant celebrity image by celebrity gender interaction, F(1,76)=.049, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on

23

whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image. Changed (Celebrity changed opinion of brand)

There is a significant effect of the celebrity image, F(1.76)=12.79, p < .05. The good celebrity (M=2.86) effected more favorable attitudes than the bad celebrity (M=2.16). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.11, p < .10. The female celebrity (M=2.55) effected slightly more favorable attitudes than the male celebrity (M=2.48). There is a marginally significant celebrity image by celebrity gender interaction, F(1,76)=.28, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image.

Sex (Sex of advertisement seen)

24

There is a significant effect of the celebrity image, F(1.76)=6.18, p < .05. The good celebrity (M=2.25) effected more favorable attitudes than the bad celebrity (M=1.79). There is a significant effect of the celebrity’s gender, F(1,76)=4.69, p < .05. The male celebrity (M=2.22) effected slightly more favorable attitudes than the female celebrity (M=1.82). There is a marginally significant celebrity image by celebrity gender interaction, F(1,76)=.41, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image. Trashy (Trashy vs. Sophisticated opinion of celebrity)

25

There is a significant effect of the celebrity image, F(1.76)=63.04, p < .05. The good celebrity (M=3.59) effected more favorable attitudes than the bad celebrity (M=2.16). There is a marginally significant effect of the celebrity’s gender, F(1,76)=2.87, p < .10. The male celebrity (M=3.03) effected slightly more favorable attitudes than the female celebrity (M=2.72). There is a significant celebrity image by celebrity gender interaction, F(1,76)=4.47, p < .05. As can be seen in the above table, the effect of the celebrity’s image depended on whether or not the celebrity, and therefore the respondent, was male or female. So when the celebrity was male there was a larger affect on the perceived trashiness of the celebrity than when the celebrity was female. Responsible (Responsible vs. Irresponsible opinion of celebrity)

There is a significant effect of the celebrity image, F(1.76)=65.01, p < .05. The good celebrity (M=3.87) effected more favorable attitudes than the bad celebrity (M=2.16). There is a significant effect of the celebrity’s gender, F(1,76)=4.25, p < .05. The male celebrity (M=3.23) effected slightly more favorable attitudes than the female celebrity (M=2.79). There is a marginally significant celebrity image by celebrity gender interaction, F(1,76)=.098, p < .10. As can be seen in the above table, the effect of the celebrity’s image depended slightly on

26

whether or not the celebrity, and therefore the respondent, was male or female. So whether the celebrity was male or female, there was only a slight effect of the celebrity’s image.

Friendly (Friendly vs. Rude opinion of celebrity)

There is a significant effect of the celebrity image, F(1.76)=49.03, p < .05. The good celebrity (M=4.24) effected more favorable attitudes than the bad celebrity (M=2.55). There is a marginally significant effect of the celebrity’s gender, F(1,76)=.326, p < .10. The male celebrity (M=3.47) effected slightly more favorable attitudes than the female celebrity (M=3.33). There is a not a significant celebrity image by celebrity gender interaction, F(1,76)=2.68, p=.105. As can be seen in the above table, the effect of the celebrity’s image did not depended on whether or not the celebrity, and therefore the respondent, was male or female.

27

Trust (Trust vs. Untrustworthy opinion of celebrity)

There is a significant effect of celebrity image, F(1.76)=23.15, p < .05. The good celebrity (M=3.38) effected more favorable attitudes than the bad celebrity (M=2.51). There is a slightly significant effect of the celebrity’s gender, F(1,76)=.043, p < .10. The male celebrity (M=2.96) effected slightly more favorable attitudes than the female celebrity (M=2.92). There is a significant celebrity image by celebrity gender interaction, F(1,76)=.5481, p < .05. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. Male celebrities had a larger effect on the trustworthiness of the celebrity than when the celebrity was female.

28

Model (Role Model vs. Bad influence opinion of celebrity)

There is a significant effect of the celebrity image, F(1.76)=102.49, p < .05. The good celebrity (M=3.92) effected more favorable attitudes than the bad celebrity (M=2.00). There is a significant effect of the celebrity’s gender, F(1,76)=.528, p < .05. The female celebrity (M=3.03) effected slightly more favorable attitudes than the male celebrity (M=2.89). There is a significant celebrity image by celebrity gender interaction, F(1,76)=.528, p < .05. As can be seen in the above table, the effect of the celebrity’s image depended slightly on whether or not the celebrity, and therefore the respondent, was male or female. Female celebrities had a lesser effect on whether the celebrity was perceived as being a role model than when the celebrity was male.

29

Study 2 Discussion
The first question on our survey simply had respondents rate how much they like CocaCola. Respondents with bad celebrities generally like Coca-Cola more than respondents with good celebrities. We think this could be due to the fact that the surveys were only given to college students, who probably had more favorable attitudes towards the bad celebrity because the description of the celebrity mentioned lots of partying. Obviously partying would not be considered a negative action by college students because they live a similar lifestyle to that of the celebrity described. Had we given the surveys to older people, for example, the results may have been more favorable for people who read about the good celebrity. For the same question female respondents had more favorable attitudes for Coca-Cola than male respondents. We think this may have happened because young women may drink Coca-Cola (especially Diet Coke) more than young men do. Product placement in shows like “American Idol” and “Sex and the City,” which are both targeted towards females rather than males, may have also contributed to our results. The results regarding respondents’ opinion of the advertisement is slightly in favor of the advertisement featuring the good celebrity. However, the difference is not as drastic as what we expected to see. This suggests that a celebrity’s public image may not affect their endorsements to a large degree, as we thought it would. There was very little difference between male and female respondents’ opinions of the advertisement, and the interaction effect is small as well. Since the interaction effect is so small, we think that the respondent’s gender does not play a role in how much he or she likes an advertisement. Respondents preferred the good celebrities over the bad celebrities, according to the results to Question 3. We expected this to happen because we thought the good celebrity seemed much more likeable than the bad celebrity. Respondents

30

approved of the good celebrity’s behavior more than that of the bad celebrity. Again, these results are what we expected. Regardless of the celebrity’s image, respondents said they would not like the advertisement more if a celebrity of the opposite sex was featured. Respondents who read about the good celebrity would like a celebrity of the opposite sex more than respondents who read about the bad celebrity. The difference between the two scenarios was significant, as we showed in our results. We think these results occurred because respondents did not want to admit that sex appeal in advertising makes them think more favorably of the product. The difference between male and female respondents was significant as well, with males having a greater preference to see a female celebrity in the advertisement. However, the interaction effect between the celebrity’s image and gender was only marginally significant for this question. We think there was little significance in the interaction effect because respondents may not care about a celebrity’s personality if the celebrity is of the opposite sex; respondents may care more about superficial aspects of celebrities of the opposite sex featured in advertisements. Respondents rated the good celebrities as being more sophisticated and responsible than the bad celebrities. We predicted this would happen because we wrote the descriptions to suggest that the good celebrities were more mature than the bad celebrities. Male celebrities were viewed as more sophisticated than female celebrities. There is a significant interaction between celebrity image and gender in terms of sophistication as well. Women our age have higher standards of sophistication than males our age, which is why we think the results are as shown. Male celebrities were viewed as more responsible than female celebrities. We think this is because women are more hesitant to say a celebrity is responsible, and perhaps they needed more information to judge the celebrity.

31

Good celebrities were viewed as more trustworthy than bad celebrities. Trustworthiness had a high interaction effect as well, which may be because women may value this attribute more than men. The question regarding whether or not the celebrity is a role model also had a high interaction effect. Since we think women read more gossip magazines, they may care more than men about a celebrity’s actions.

32

General Discussion
Overall, we reject our null hypothesis, which was the results of our survey will confirm that negative celebrity behavior will negatively affect the way respondents view the brand the celebrity endorses. We will accept the alternative hypothesis. In our first study, we found that males and females thought that celebrity endorsements were positively effective. In the second study, we found that celebrity endorsers are effective on audiences. However, contrary to our hypothesis, the “bad” celebrity scenarios had the most positive effect on audiences’ opinion of a brand. In our first study, respondents reacted more favorable towards “good” celebrities. These results contrast with our findings in study two. In study two, the brand was viewed more favorably when endorsed by “bad” celebrities. Our research has helped us to learn about advertising to the college market. Before conducting this study, we thought that “good” celebrities would have a more positive effect when endorsing a brand. However, our research showed that college-aged consumers prefer more controversial celebrity endorsers. We can apply this research to advertising practice by not assuming what a target market thinks. Positive behavior does not always affect a celebrity or his or her endorsements positively. If we were to re-do this study, we would make our first survey more focused. We searched for too much information and instead we should have focused on one specific variable. Study two proved to be effective in finding useful information because we knew exactly what results we wanted to find when we wrote the survey.

33

References
Choi, M. S., & Rifon, N. J. (2007). "Who is the Celebrity in Advertising?" Periodical Abstract, 304-324. Kampert, P. (2007). "The Right Way to Talk About Fallen Stars." Chicago Tribune. Academic Search Premier. Louie, T. (2001). "When Bad Things Happen to the Endorsers of Good Products." Marketing Letters, 13-23. Prashad, S. (2005). "Bad Branding: Here's Some Marketing That Can't Help." The Toronto Star, 1. Roberto, S. (2006). “Consumer Buying Behavior in Response to Corporate Scandal.” Consumer Interests Annual. Stuart, J. (2007). "Heroes in Sports: Assessing Celebrity Endorser Effectiveness" International Journal of Sports Marketing and Sports Marketing and Sponsorship.

34

Appendix: Advertising Research Survey
For the following questions 1-9 please circle the number which you think is the best answer. 1 Strongly Disagree 2 Disagree 3 Uncertain 4 Agree 5 Strongly Agree

1. Celebrity endorsers are the most persuasive advertising method. 1 2 3 4 5

2. I think celebrity endorsements are ineffective. 1 2 3 4 5

3. Celebrity endorsers affect the sale of the product with which they are associated. 1 2 3 4 5

4. A celebrity endorser’s behavior affects my purchasing decision. 1 2 3 4 5

5. A male celebrity endorser for a car will increase my likelihood to purchase the product. 1 2 3 4 5

6. A female celebrity endorser for a soft drink will increase my likelihood to purchase the product. 1 2 3 4 5

7. A male celebrity endorser for a cell phone will increase my likelihood to purchase the product. 1 2 3 4 5

8. A female celebrity endorser for a hair product will increase my likelihood to purchase the product. 1 2 3 4 5

9. I intentionally purchase certain products because of their association with the celebrity. 1 2 3 4 5

35

10. In general, do you think celebrity endorsement increases or decreases the sales of products? Increase Decrease Not Sure 11. Please rank using numbers 1-5 how the following celebrity incidences would affect your decision to purchase a product 1= Would stop using product 2= Would somewhat decrease purchases of the product 3= Would not affect purchasing decision 4= Would somewhat increase purchases of the product 5= Would greatly increase purchases of the product Drunk Driving___ Adopting children___ Late night partying___ Frequent tabloid coverage___ Eating disorder___ Large charitable donation___ Public Break-up___ Won an award (Emmy, Grammy, etc.)__ 12. Check under the number that represents your feelings toward the celebrity endorsement. (These are fictitious endorsements) (Strongly Dislike) -3 -2 -1 +1 +2 +3 (Strongly Like) Lindsay Lohan- Pepsi Justin Timberlake- Apple Jennifer Aniston- Verizon Tom Cruise- McDonalds Kevin Federline- Subway Jessica Alba- Evian __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __

13. Are you: Male or Female (circle one)

36

Experiment Survey Scenario 1
For this experiment we are going to show you an ad featuring a new celebrity and the product that they are currently endorsing. Based off this advertisement and the description below, answer the questions following the advertisement. John Anderson has just won American Idol and has just released his critically acclaimed first album. On American Idol viewers got to know his sparkling personality and were able to get a glimpse of his small town life. John was a favorite of fans from the beginning with his amazing vocal talent and stylish fashion sense. Even though he was the top pick throughout the competition, he remained modest and level headed. After his big win, he was just picked by Coca-Cola to be the new face of their brand. CocaCola is working closely with John and has also agreed to sponsor his upcoming tour. John has been frequently seen in magazines such as People at sporting events with his girlfriend and friends in New York, where he just finished recording his album.

37

Experiment Scenario 2
For this experiment we are going to show you an ad featuring a new celebrity and the product that they are currently endorsing. Based off this advertisement and the description below, answer the questions following the advertisement John Anderson has just won American Idol and has just released his album which received less than perfect reviews from critics. On American Idol viewers saw him as conceited and self-centered. Despite his attitude he remained in the competition due to his snarky comebacks to the judges, especially Simon. This combined with his voice, although not the best, won the votes he needed to become the newest American Idol. Recently, John has been caught hanging out with Lindsay Lohan in popular Los Angeles Night clubs, even though he is only twenty years old. He has frequently appeared on the cover of Us Weekly sporting a new woman each time. Trying to capitalize on his new popularity, he has just signed a contract with Coca-Cola to appear in their advertisements.

38

Experiment Scenario 3
For this experiment we are going to show you an ad featuring a new celebrity and the product that they are currently endorsing. Based off this advertisement and the description below, answer the questions following the advertisement. Jen Smith has just won American Idol and has just released her critically acclaimed first album. On American Idol viewers got to know her sparkling personality and were able to get a glimpse into her small town life. Jen was a favorite of fans from the beginning with her amazing vocal talent and stylish fashion sense. Even though she was the top pick throughout the competition, she remained modest and level headed. After her big win she was chosen by Coca-Cola to be the new face of their brand. Coca-Cola is working closely with Jen and has also agreed to sponsor her upcoming tour. Jen has frequently been seen in magazines such as People shopping with her mom and friends in New York, where she just finished recording her album.

39

Experiment Scenario 4
For this experiment we are going to show you an ad featuring a new celebrity and the product that they are currently endorsing. Based off this advertisement and the description below, answer the questions following the advertisement. Jen Smith has just won American Idol and has just released her album which received less than perfect reviews from critics. On American Idol viewers saw her as conceited and self-centered. Despite her attitude she remained in the competition due to her snarky comebacks to the judges, especially Simon. This combined with her voice, although not the best, won her the votes she needed to become the newest American Idol. Recently, Jen has been caught hanging out with Lindsay Lohan in popular Los Angeles Night clubs, even though she is only twenty years old. She has frequently appeared on the cover of Us Weekly sporting a new man each time. Trying to capitalize on her new popularity, she has just signed a contract with Coca-Cola to appear in their advertisements.

40

41

42

Study 2 Experiment
Please answer the following questions based of the information you have just seen. For the following questions 1-8 please circle the number that best represents your opinion using the scale below. 1 2 3 4 5 Strongly Agree Agree Uncertain Disagree Strongly Disagree 1) I like Coca-Cola as a brand.______

2) I like the ad that I just saw.______

3) I have a negative opinion of the celebrity I just read about._____

4) I have a positive opinion of Coca-Cola based on the ad I just saw._____

5) I approve of the celebrity’s behavior based on the description I just read.____ 6) Brand image is how consumers feel about a brand. I feel that the celebrity’s behavior negatively affects Coca-Cola’s brand image._____ 7) Based on the celebrity description I read, this ad has made my opinion of Coca-Cola more positive. ____ 8) I would have a better opinion of Coca-Cola as a brand if it featured a celebrity of the opposite sex. ____ Based on the description of the celebrity and the advertisement, for questions 9-13, please rate your overall opinion of the celebrity by checking the line which best reflects your feelings. 9) Trashy__: __: __: __: __Sophisticated 10) Responsible__: __: __: __: __Irresponsible 11) Friendly__: __: __: __: __Rude 12) Untrustworthy__: __: __: __: __Trustworthy 13) Bad Influence __: __: __: __: __Role model 14) Are you male or female? (circle one)

43

Codebook For Study 1 Survey

Column(s) 1

Variable Name ID

Description Questionnaire identification number 1=Strongly Disagree 2= Disagree 3= Uncertain 4= Agree 5= Strongly Agree 1= Not Sure 2= Decrease 3= Increase 1= Would Stop Using Product 2= Somewhat decrease purchase of product 3= Not affect purchasing decision 4= Somewhat increase purchase of product 5= Greatly increase purchase of

2-10

Agreement of celebrity endorsers

11

Willingness to purchase product because of celebrity Confidence in buying product

12-19

product 20-25 Likelihood of purchasing product 1=Strongly Dislike 2= Dislike 3= Somewhat Dislike 4= No preference 5= Like 6= Strongly Like 1=male 2=female

26

Gender

44

Study One SPSS Code Book Pertaining to Survey One

Questions 1=Association 2=General 3=Driving 4=Children 5=Partying 6=Tabloid 7=Eating 8=Donation 9=Breakup 10=Award 11=Lindsay 12=Justin 13=Jennifer 14=Tom 15=Kevin 16=Jessica 17=Gender

45

Codebook For Study 2 Survey

Column(s) 1 2-9

Variable Name ID Opinion of Coca-Cola and celebrity

Description Questionnaire Identification number 1=Strongly Agree 2= Agree 3= Uncertain 4= Disagree 5= Strongly Disagree 1=Very Negative 2=Negative 3=Uncertain/Neutral 4=Positive 5=Very Positive 1=Very Positive 2=Positive 3=Uncertain/Neutral 4=Negative 5=Very Negative 1=Very Negative 2=Negative 3=Uncertain/Neutral 4=Positive 5=Very Positive 1=male 2=female

10

Celebrity Personality Trait

11-12

Celebrity Personality Traits

13-14

Celebrity Personality Traits

15

Gender

46

Study Two SPSS Code Book Pertaining to Survey Two
Questions 1=Coke 2=Ad 3=Celeb 4=Positive 5=Behavior 6=Image 7=Changed 8=Sex 9=Trashy 10=Responsi 11=Friendly 12=Trust 13=Model 14=Gender

47

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.