Skin Color in Advertising
Natascha Heidner, 11086041
Master’s Thesis
MSc Business Administration, Marketing Track
University of Amsterdam – Faculty of Economics and Business
Supervisor: Anouar El Haji
Statement of Originality
This document is written by Natascha Heidner who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used creating it.
The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
ABSTRACT ... 4
1.0 INTRODUCTION ... 5
2.0 LITERATURE REVIEW ... 7
2.1SKIN COLOR AND DIFFERENT RACES ... 7
2.2BRAND ENDORSEMENT ... 8
2.3SOCIAL IDENTITY AND CONSUMER BEHAVIOR ... 9
2.4THE SIMILARITY HYPOTHESIS OF BYRNE ... 10
2.5SELF-REFERENCING AND ENDORSER SIMILARITY ... 10
2.6THE EFFECT OF PRODUCTS IN ADVERTISEMENTS ... 11
2.7EXPERT ... 12
2.8RESEARCH QUESTION ... 12
2.9THEORETICAL FRAMEWORK ... 13
3.0 METHODOLOGY ... 14
3.1RESEARCH DESIGN ... 14
3.2FOUR AUCTION TREATMENTS ... 15
3.4SAMPLE FRAMEWORK ... 17
3.5DEPENDENT VARIABLE ... 19
3.6INDEPENDENT VARIABLES ... 19
3.7MODERATING VARIABLE ... 20
4.0 RESULTS ... 20
4.1TEST OF H1:SIMILARITY BETWEEN SKIN COLORS ... 20
4.2TEST OF H2:ENDORSER WITH GREATER EXPERTISE ... 21
4.3TEST OF H3:SIMILARITY BETWEEN THE ENDORSER AND THE CONSUMER ... 22
4.4ADDITIONAL ANALYSES ... 23
5.0 DISCUSSION ... 24
5.1SIMILARITY BETWEEN SKIN COLORS ... 24
5.2ENDORSER WITH GREATER EXPERTISE ... 26
5.3SIMILARITY BETWEEN THE ENDORSER AND THE CONSUMER ... 28
5.4THEORETICAL IMPLICATIONS ... 29
5.5MANAGERIAL IMPLICATIONS ... 30
5.6ETHICAL ISSUES AND LIMITATIONS ... 30
5.7FUTURE RESEARCH ... 31
6.0 CONCLUSION ... 32
7.0 REFERENCES ... 34
8.0 APPENDICES ... 42
8.1TREATMENT 1:CAUCASIAN ENDORSER WITH EXPERTISE AND SIMILARITY ... 42
8.2TREATMENT 2:DARK ENDORSER WITH EXPERTISE AND SIMILARITY ... 43
8.3TREATMENT 3:CAUCASIAN ENDORSER WITHOUT EXPERTISE AND SIMILARITY ... 43
8.4TREATMENT 4:DARK ENDORSER WITHOUT EXPERTISE AND SIMILARITY ... 44
8.5TREATMENT 5:BASELINE WITHOUT ENDORSER AND WITHOUT EXPERTISE AND SIMILARITY ... 44
8.6QUESTIONS FOR THE RESPONDENTS ... 45
Abstract
The main purpose of this thesis is to investigate the effect of an endorser’s skin color on consumers’ willingness to pay. The research question of the thesis is therefore: "To what extent does the
endorser's skin color affect consumers’ willingness to pay?" The moderation effects of the level of
expertise and similarity of the endorser were investigated to see whether or not they had a positive influence on increasing the consumers’ willingness to pay. Data was collected from Dutch citizens (N = 1,116) using an online auction platform called Veylinx. Unexpectedly, and in contrast to previous studies, the results illustrate that the skin color, the similarity nor the expert-level of the endorser had a significant effect on consumers’ willingness to pay. The findings do no support the earlier studies that state that ‘skin color of an endorser should either increase or decrease the consumers’ willingness to pay’. This study makes a methodological contribution by examining consumers’ willingness to pay when purchasing a product in an online auction rather than as a survey in a hypothetical situation. Furthermore, it is contributing to research by shedding more light on the debate on whether brands should use endorsers with similar skin color as the majority of the consumers.
1.0 Introduction
A message that is being delivered by different endorsers can produce varied outcomes among consumers (Whittler and Spira, 2002). Different endorser features such as credibility, expertise and similarities can make an endorser more persuasive than others (Eagly and Chaiken, 1993). Hence, it is essential for companies to choose the right endorser to advertise their products, as the endorser can influence consumers’ buying intentions. Moreover, the skin color of an endorser may affect consumers’ evaluations of product advertisements and might either increase or decrease consumers’ willingness to pay. Skin color bias is where consumers tend to behave differently towards people with a different skin color than themselves (Maddox and Gray, 2002). Skin color bias has existed for a long time, yet there is still inconsistency in the research findings on whether or not it affects consumers’ buying behavior.
In the context of skin color, previous literature indicates different results when measuring the impact of an endorser’s skin color in advertisements (Petroshius & Crocker, 1989; Watson, Thornton & Engelland, 2010). The majority of the literature on how skin color affects consumers’ evaluations of advertisements report that consumers respond more favorably to advertisements featuring same race endorsers, meaning the endorser has the same skin color as the consumer (Appiah 2001; Choudhury & Schmid 1974; Green 1999; Pitts et al. 1989; Schlinger & Plummer 1972; Szybillo & Jacoby 1974; Whittler 1991; Williams, Qualls and Grier1995). Consumers prefer advertisements featuring same race endorsers when they share characteristics (Kareklas, 2010 and Woodside & Davenport, 1974). There are several facets of similar characteristics, including social class, level of education, gender, lifestyle and age (Kareklas, 2010).
Even though the majority of the studies report that consumers respond more favorably towards advertisements featuring the same race endorsers, there is an inconsistency in the literature. Various studies indicate that consumers have both negative and positive reactions towards endorsers having a darker skin color (Muse 1971; Solomon & Bush 1977; Bush et al. 1979). However, these previous
studies are not up to date and are therefore to be questioned, as they do not meet today’s standards of methodological rigor (Whittler, 1991). For instance, many of the studies failed to consider the issues of problems using real advertisements, carry-over effects and lack of control groups in the experiments (Whittler, 1991).
Social identity gives a basis for identification and belonging to people. The sense of identification can, among other things, be based on religion or race (Alekar, 2014). Together, social identity and the sense of identification can be reasons why the skin color of an endorser can have such a big impact. The sense of identification plays an important role in societies, and thus people might be more influenced by group norms (Alekar, 2014). When using brand endorsement, the objective is to match the right endorser with the right product and place them both in the right ad campaign.
This thesis is contributing to the extent of literature on the effect of skin color in advertising by exploring the effect of an endorser’s skin color on consumers’ willingness to pay in an online auction. We hypothesize that consumers have preferences towards endorsers that have a similar skin color. Furthermore, it is expected that the endorser’s skin color will have an impact on consumers’ willingness to pay (Woodside & Davenport, 1974 and Petroshius & Crocker, 1989; Watson, Thornton & Engelland, 2010; Kareklas, 2010). Hence the following research question is:
"To what extent does the endorser's skin color affect consumers’ willingness to pay?".
This thesis is structured as follows; the first part covers the literature review, the research question and theoretical framework; subsequently, the methodology and research design are covered. The results of the experiment are then analyzed and interpreted. Lastly, the discussion, implications, conclusion and recommendation for future research are given.
2.0 Literature Review
2.1 Skin Color and Different Races
Skin color is a social construction in which people attribute meaning to the shades of others’ complexions (Harvey et al., 2005; Hochshild and Weaver, 2007; Watson et al., 2010). Nevertheless, researchers often overlook that the skin color of an endorser may affect consumers’ evaluations of advertisements (Watson et al., 2010).
Since Barban and Cundiff (1964) started studying the racial effects of dark and Caucasian brand endorsers in advertisements, the number of studies has increased and has indicated different effects of the brand endorser’s skin color (Petroshius & Crocker, 1989; Watson, Thornton & Engelland, 2010). Barban and Cundiff’s (1964) results showed that Caucasian consumers had a neutral or positive reaction towards advertisements with dark people (Whittler, 1991). Later, Barban’s Chicago-based study (Bush et al. 1979) showed that Caucasian consumers prefer integrated ads as much or even more than the ones only including Caucasian brand endorsers. Guest (1970) did not find any difference in using dark and Caucasian endorsers the reactions among the consumers were the same. Tolley and Goett (1972) did not find negative reactions towards darker brand endorsers (Muse, 1971; Tolley and Goett, 1972) whereas in contrast Muse (1971) found negative reactions towards dark endorsers. Whittler (1991) found that even though Caucasian consumers do not react extremely negatively to darker endorsers, the reactions are not as positive as they are to Caucasian endorsers. Vice-versa, darker people respond more positively to darker endorsers.
More recent findings including: Williams et al. (1975), Watson et al. (2010), Kareklas (2010) and Alekar (2014), find that in the majority of the studies people are more attracted to endorsers having the same skin color as themselves. Nevertheless, the results are yet inconsistent where both Caucasian and dark consumers have responded positively, neutrally and even negatively towards brand endorsers of another race. However, though some studies show these differences, the greater part find that Caucasian consumers does not respond negatively towards dark endorsers, they just
respond more positively towards same race endorsers. Likewise, the dark consumers respond more positively towards endorsers who also have dark skin color.
Table 1 below gives an overview of the cited articles and their results on how consumers react towards the skin color of an endorser.
Table 1: Overview of Articles Concerning Skin Color Preferences
2.2 Brand Endorsement
Brand endorsement helps make a product or service even more attractive, noticeable and gets the attention of the consumer (Khongkok & YouLi, 2013). The brand endorser can furthermore help create and differentiate product images and generate sales, which all has proven to be a valuable and popular strategy to promote a company’s products (Dekker & Van Reijmersdal, 2013; Choi, Lee, & Kim, 2005).
In order to be attracted and interested in buying a product, it is important to have a positive association between the endorser and the product. Whether an endorser is attractive or credible
endorser to be credible in the eye of the consumer, he or she must be trustworthy, experienced, likeable and familiar to the consumer (Miciak & Shankklin, 1994). Hence, the more attractive and credible the endorser is, the more persuasive he or she will be as a brand endorser (Miciak & Shankklin, 1994). Previous studies support the theory that brand endorsers who are likable and familiar to the buyer are also more effective as endorsers.
2.3 Social Identity and Consumer Behavior
Social identity is an aspect of self-concept derived from an actual or perceived membership in a social group (White and Argo, 2009). White et al. (2012) argue that consumers may react to a given situation in ways that are consistent with their personal or social identity by responding either more positively or negatively to that given situation. Research also show that consumers engage in identity-congruent behaviors and evaluate products more positively when the product is linked with an aspect of social identity that is viewed as being important (Alekar, 2014).
Endorsers can express the values and ideals of a particular culture (Hung et al., 2011). Hence, consumers can identify with the common cultural symbolic element, which the endorser and the consumer share. This commonality forms the basis of social identity for the consumer. If the consumer identifies with the endorser, the consumer may want to mimic the attitude, belief and behavior of the endorser and thereby also strive to be more like him or her (Alekar, 2014).
People tend to be influenced by group norms. A variety of studies show that similar and dissimilar endorsers are more influential in changing a person’s attitude (Woodside & Davenport, 1974) – this is also known as ingroup favoritism (Tajfel et al., 1971). Ingroup favoritism is a tendency to have a more positive attitude towards people in your own group (e.g., you all have the same skin color) than to those belonging to another group (Spira & Whittler, 2004). For instance, when seeing an advertisement featuring an endorser with a different skin color, consumers may think “he is not like me so it’s hard to agree with him” (Kelman, 1961). Whittler (1991) found that Caucasian
consumers perceive themselves less similar to endorsers having a dark skin color. Vice-versa dark consumers perceived themselves less similar to endorsers being Caucasian. This is consistent with previous studies showing that both people with Caucasian and dark skin color prefer advertisements featuring same race endorsers (Kareklas & Polonsky, 2010).
2.4 The Similarity Hypothesis of Byrne
The similarity hypothesis of Byrne (1971) concerns interpersonal attraction, where the more similar a person is to another person, the more positive that other person will be assessed. When seeing a person with similar characteristics as oneself, it can give a certain satisfaction, which comes from actual and expected equality (Klein, 2013). Examples of similarities can, among other things, be same demographics (e.g., you come from the same region) or social characteristics (e.g., you both come from a poor family) (Reagans, 2011). The study of Tsui and O’Reilly (1989) found that some demographics such as gender, age and education play an important role for being perceived as being similar to others, and thereby also to be more positively assessed by others.
The study of Franke et al. (2006) supports the similarity hypothesis of Byrne. Franke et al. (2006) found, that the same kind of education or employment background when searching for a job had a significant influence on the selection process. The ones being similar to the employer was rated higher (Franke et al., 2006).
2.5 Self-referencing and Endorser Similarity
Self-referencing refers to the ability of referring to oneself. It is a reference to one’s own character or to a group with which one identifies (Roy, 2014). The more a consumer considers being similar to an endorser; the more the consumer can relate to him or her and thereby engage in self-referencing (Lee, Fernandez and Martin, 2002). Once a consumer perceives these similarities and self-referencing, the endorser is seen as being more trustworthy (Deshpandé and Stayman, 1994).
involves in self-referencing, which will lead to more positive attitudes and buying behavior (Zinkhan & Hong, 1991).
As already mentioned above, the right endorser of a brand is very important when contributing to the effectiveness of advertisement. Previous studies show that when consumers self-reference information by observing similarity when viewing an ad, the advertisement becomes even more efficient (Roy, 2014). Consumers can self-reference to endorsers or advertisements when they refer to a salient dimension of the consumer’s self, like ethnicity (Lee, Fernandez and Martin, 2002). If the consumer can make a self-reference and see similarities with the endorser solely by observing him or her without any further information given, then the advertisement is more likely to succeed. This will in turn lead to a more positive attitude and purchase intension (Martin and Kennedy, 1994). Therefore, it is important to include observable self-reference indicators in an advertisement as it can help the consumer to relate to it more easily. Examples of observable self-reference indictors can be familiar settings, common activities or similar attributes of the model (Roy, 2014).
2.6 The Effect of Products in Advertisements
Muse (1971), Solomon and Bush (1977) found that the type of product used in brand endorsements advertisements could have a negative effect on consumers purchase decision. Solomon and Bush (1977) found that if a product is advertised as a “socially consumed” product then the reactions towards darker endorsers might be more negative than if it had been a non-socially consumed product. However, Solomon and Bush did not give an explanation for this finding. Also, other product categories, such as cars, might stimulate a stereotyped criticism: “Dark people spend too much on cars” (Bush, Hair & Solomon, 1979). Muse (1971) found that “personal products,” such as feminine napkins, resulted in Caucasian consumers having negative responses towards the darker brand endorser. It appears the more personal or involving the product is, the more sensitive the Caucasian consumers are to dark endorsers (Muse, 1971). However, there is inconsistency across
studies, as Stafford and Birdwell did not find any significant differences between the reactions of Caucasian and dark endorsers for a “personally related” product such as lipstick (Muse, 1971). Nevertheless, marketers should be aware that negative reactions towards an endorser might depend on the type of product and the situational context of the advertisement.
2.7 Expert
How effective an advertised message is depends on the expertness and trustworthiness of the endorser (McCracken, 1989). Expertness is defined as the perceived ability of the endorser to make valid assertions (McCracken, 1989). Thus, an expert is an individual with superior knowledge of the product that is being endorsed. The superior knowledge is a result of experience, study and training (Friedman and Friedman, 1979). When endorsers show attributes such as expertise, trustworthiness, similarity and likableness, they will be perceived as being more credible and persuasive in consumers’ minds (McCracken, 1989; Friedman and Friedman, 1979; Ohanian, 1990). Expertise may therefore be seen as a logical connection with the products and thereby make the endorser more believable to consumers (Till and Busler, 2000; Choi et al., 2005). Overall, the perceived expertness of an endorser may therefore give consumers a more positive attitude towards a product. However, a study of Silvera and Austad (2001) shows that if there is an incongruity between the endorser and the endorsed product, the expert would be considered a non-expert and consumers would not be affected in the same way.
2.8 Research Question
As mentioned above the research question is as follows:
2.9 Theoretical Framework
Based on the hypothesis described above, the following theoretical framework has been established. The framework illustrates that a positive relation between an endorser’s skin color and the consumers’ willingness to pay is expected. In other words, if the consumer perceives the endorser to have the same skin color as the consumer him- or herself, then we would expect the consumer to be willing to pay more for the product; vice-versa if the consumer does not perceive to have a similar skin color to the endorser, then this should have a negative effect and thereby also decrease the consumer’s willingness to pay. In general, previous studies, have found that consumers prefer ads with similar race endorsers as themselves (Petroshius & Crocker, 1989; Watson, Thornton & Engelland, 2010); I therefore expect the following hypotheses (H1):
H1: Consumers will generally have a preference for product advertisements featuring
endorsers with the same skin color as themselves.
The first moderator, level of expertise similarly shows an expected positive relation. According to previous research and literature, the use of an expert can have a positive effect on consumers’ attitudes towards the product and may increase sales (Woodside & Davenport, 1974). The higher the level of an endorser’s expertise, the higher the ‘consumer’s willingness to pay’ is expected to be. This is expressing the second hypothesis (H2):
H2: Consumers are more likely to place a bid when the sales endorser is perceived as
having greater expertise.
Research have found that an endorser with similarities to oneself is more influential in changing the attitude and hence one will most likely respond more positively to advertisements by making a purchase (Woodside & Davenport, 1974; Roy, 2014). In the experiment using a written statement from the endorser, we will test the level of expertise. The endorser recommends the product to the consumer; this is expected to have an increasing effect on consumers’ willingness to pay. Thus, by
using a statement of level of expertise as a moderator, the effect of expertise between the endorser and the consumer can be analyzed. The same goes for the second moderator, level of similarity. By using a written statement, saying the endorser is from Amsterdam (all the participants in the experiment are based in the Netherlands), the impact of the level of similarity can be measured; here, it is therefore expected to have a positive relation and increase the consumers’ willingness to pay. Vice-versa, in the ads where there is not a written statement of similarity between the endorser and the consumer, it is expected to decrease the consumer’s willingness to pay. This represents hypothesis 3 (H3):
H3: The stronger the perceived similarity for the endorser, the bigger the probability of
bidding by the consumer.
Figure 1: Theoretical Framework
3.0 Methodology 3.1 Research Design
The research will be conducted by making use of an online auction called “Veylinx,” which is an online Vickrey auction platform. Veylinx is a sealed bid auction, where the bid of other participants is unknown, however the participant with the highest bid will only pay the second highest price.
The Vickrey auction reveals consumers’ maximum willingness to pay for a product or service. Veylinx gives the participants an incentive to bid their true willingness to pay, as consumers only have to pay the second highest price.
The reason for choosing Veylinx as a method over for example a survey is that in the survey participants might not reveal their true willingness to pay. Hence there might be a bias in the sample because they do not actually have to pay for the product; they are in a hypothetical situation. In the Veylinx experiment on the other hand, it has the advantage of letting consumers reveal their true value by making use of real auctions where consumers are placing their bid for a product in a real situation.
The experiment will be conducted by using Veylinx’s representative pool of 10,000+ consumers located in the Netherlands. The auction is an undisclosed bid auction, meaning that participants will not be able to see what other participants are bidding, and therefore it gives consumers the incentive to reveal their true willingness to pay for the product as they are not being biased by other people’s willingness to pay.
In the auction, participants see an ad for the given product and three questions they are asked to answer. The three questions are the three hypotheses mentioned previously. After the participant has seen the ad and answered the questions, he or she is able to make a bid for the product.
3.2 Four Auction Treatments
There are four treatments in the online auction: two where the endorser is Caucasian and two where the endorser has a darker skin color. In order to measure the impact of an endorser’s skin color the involvement dimensions will be measured on expert- and non-expert endorsers for a product. Since electronic product categories are easy to relate to and is something a lot of people consume, a high-involvement product such as a tablet has been chosen for this experiment.
The reason for using an expert and non-expert setting is that it will be possible to test the credibility of the endorser, which can help to determine how effective the person is as a salesperson.
Therefore, the more credible consumers find the salesperson, the more persuasive he or she will be as an endorser. For the treatments with an indication of expertise, there will be a small text saying: “Beyond all expectations, I recommend this iPad”. In the non-expert treatments there will not be any text phrasing the product. A figure for the four treatments will be as follows:
Caucasian Dark Skin Color
Expert Non-expert
Table 2: Matrix of Treatments 1 to 4
In the four experiments, participants will see a digital color ad presenting an endorser and a tablet. With the use of Adobe Photoshop, a photograph of a Caucasian female endorser will be digitally manipulated into having a darker skin color. It will therefore be the same endorser in all four treatments just with two different skin colors. Two of the ads will have a short text saying: “Technology expert from Amsterdam”. This shows similarity between the endorser and the respondents as the majority of the respondents are from Amsterdam. The similarity also functions as a moderator.
To sum up, the experimental ads differ in the endorser’s skin color (Caucasian or dark skin), the level or expertise (expert or non-expert) and the level of similarity between consumers and the endorser (similar or dissimilar). Also, a control ad will be made. Here there will not be any expert/non-expert and similarity/dissimilarity statements or a picture of the endorser, but the ad will only show the product and a text with information about the product.
After the participant makes a bid on the auction he or she will be asked to answer the following questions that are linked to the hypotheses mentioned above:
• How would you describe your skin color?
• To what extent do you consider yourself to be similar to the person shown in the ad?
3.4 Sample Framework
The total number of participants (N) was 1,116. According to the participants themselves 88.2% thought they were Caucasian, 5.9% thought they had a dark skin color and 5.9% did not want to reveal their skin color. When looking at the birth year of the participants, it is noticeable that not all participants have indicated their real age. The participants’ ages vary from 15 years old to 116 years old. Even though it is evident that not all participants stated their real age it is difficult to distinguish between the real ones and the fake ones. Therefore it was decided to keep and use all the data in the following analyses. Furthermore, it is unclear whether the participants have written other false information and it is therefore not possible to distinguish between the true and false ones in a clear way without it being subjective. Another reason for keeping all the data that has been gathered, is that even though numerous participants may have written the false birth year, they have still given a bid on the auction, which is valid and therefore not necessary to exclude. People are becoming more careful sharing personal information on the Internet (TRUSTe, 2013), and just because some do not want to express their real age it does not mean that their bid on the auction is also false. In this study, the age is not an important factor when measuring consumers’ willingness to pay for an online product and so, again, the data will not be excluded.
The distribution between men and women is 49.0% men and 51.0% women. Comparing this to the Dutch population, the distribution of men and women are equally represented as in November 2015, 49.5% of the Dutch population was men and 50.5% were women (see Appendix 8.7).
When looking at the participants’ willingness to pay (auction bid), note that the bid amounts visible are given in eurocents. Thus, consumers’ average willingness to pay (bid amount) of 8250,99 is equivalent to €82.51. The maximum bid amount was €400. When comparing consumers’ bid amounts with the official price of the exact same product on Apple’s own homepage, the price for an iPad mini 4 Wi-Fi 16GB is €409.
Table 3: Mean Bids for Treatments 1 to 5
Figure 2: Mean Bids for Treatments 1 to 5
The mean bids per treatment are shown in the figure above. When looking at the figure, it shows that the bid amounts are close to being similar across the different treatments. Firstly, when looking closer at the Caucasian endorser, the bid amounts are fairly alike, however, there is a higher mean for the expert-endorser. This is consistent with the stated hypothesis saying consumers are willing to place a higher bid when the endorser is perceived as having a higher level of expertise. Nevertheless, the differences between the two bids are only minor with a difference of €8. When comparing the two treatments with the dark endorser, there is a bigger difference between the mean bids. The dark non-expert endorser has a mean bid of €90 whereas the dark expert endorser has a mean bid of €75. This difference is more significant however it is of the contrary of the hypothesis stating consumers’ will place a higher bid amount when the endorser is perceived as having a
higher level of expertise. The baseline experiment where no endorser was presented has an almost equal bid amount to the dark expert-endorser, which makes sense, as a treatment without an endorser should decrease consumers’ willingness to pay and hence have a low mean bid. Nevertheless, in general the data is not consistent with the expected hypotheses, however this will be discussed and examined further below in this thesis.
3.5 Dependent Variable
The dependent variable in this experiment is consumers’ willingness to pay for a tablet. The willingness to pay reflects the extent to which consumers are willing to pay their maximum amount of money in order to get a desired product or service. The willingness to pay is measured by making use of the online auction platform Veylinx. The system will collect the different bid amounts from the respondents and thereby it will be possible to interpret and measure the effect the experiment had on the consumers’ willingness to pay. The bid amount of a Vickrey auction is chosen because it gives consumers the incentive to indicate their true willingness to pay. In other words, the consumer will not benefit from bidding a lower or a higher price than what he or she is willing to pay. The rule of a Vickrey auction is that the consumer with the highest bid will only pay the second highest bid. Due to this, consumers should reveal their true willingness to pay as they will always end up paying less than what they actually bid. Another benefit from using the Vickrey auction as a dependent variable, is that it diminishes the risk of people saying they are willing to pay one thing, but in reality they bid something else. This method places consumers in a real life situation and not only a hypothetical situation. All the bids in the auction are given in the currency of Euros.
3.6 Independent Variables
The independent variables are the skin color of an endorser and the level of expertise. The independent variables are the treatments taking place in the online auction. The skin color of the
two: expert and non-expert. The skin color of the endorser will be measured by directly asking the respondents, which skin color the endorser had. The second independent variable will be measured by comparing the means of the consumers’ bid for the product and thereby see whether or not the expertise level made a difference for the consumers’ buying intensions. The two independent variables are chosen because they will provide good insight and help the managers exploit the effectiveness of the endorsers more effectively. By knowing which attributes the consumers’ favor, the managers will have a better understanding of how to target the consumers in the best possible way.
3.7 Moderating Variable
The moderator used in this study is the level of similarity between the endorser and the consumers. As with the independent variables the moderator has two levels of similarity: similarity and non-similarity. The similarity is measured by directly asking the respondents how similar they perceive the endorser to be with themselves on scale from 1-5. To make consumers feel similar to the endorser a text was written in the ad (see appendices 8.1 and 8.2) stating the endorser was from Amsterdam. By knowing whether or not similarities between the endorser and the consumer will positively affect consumers’ buying behavior, managers can, as with the independent variables, gain a better understanding of how to target consumers.
4.0 Results
4.1 Test of H1: Similarity Between Skin Colors
The skin color of the participants as well as for the endorser have been recoded into new variables as a binary distribution. Not every participant wanted to disclose their skin color and not every participant felt like they had a similar skin color to the advertised person. Therefore the number of participants decreased to a total number of 663 participants.
A one-way between subjects ANOVA was conducted to compare the effect of perceived similarity between an endorser and the consumer and consumers’ willingness to pay. The results show there is no significant effect of perceived similarity with an endorser F(1,66) = 2.24, p = .135.
In other words, there is no significant difference between the mean of the auction and the participants believing they had a similar skin color to the endorser.
Nevertheless, when looking at the means, Caucasian participants have a mean of €87.34 whereas it for the participants with a dark skin color has a mean of €77.56. The overall mean was €82.80. Comparing these two numbers one see that the mean for Caucasian participants are on average willing to pay approximately €10 more than the participants with a dark skin color, which is 12.6% higher. Nevertheless, an interesting note is that the highest bid for dark skin colored participants are with its €400, €50 higher than what it is from the Caucasian participants. When comparing the means of the skin color with the total mean there is not a big difference. The Caucasian consumers believing their skin color is similar to the endorser is 5.5% higher whereas the dark consumers believing their skin color is similar to the endorser is 6.3% lower.
4.2 Test of H2: Endorser with Greater Expertise
The expertise of the endorser has been recoded into new variables as a binary distribution. The baseline treatment where no brand endoser or expertice statement was presented has been excluded from this analysis. The reason for this is that it will not serve its purpose as it is only the expert and non-expert that is of relevance. The total number of participants making a bid for either an expert or non-expert treatment was 906 participants.
A one-way between subjects ANOVA was conducted to compare the effect of an endorser’s perceived level of expertise and consumers’ willingness to pay. The results show there is no significant effect of perceived expertise of an endorser F(1,90) = .65, p = .421. In other words, it
can be concluded that there is not a significant difference in the means of the expert and non-expert and the bid amount.
For participants bidding on a product with an expertise statement the mean is €81.66 whereas for participants bidding on a product with no expertise it has a mean of €86.17. The overall mean was €83.84. Comparing these two numbers one see that they are close to being similar and does only have a difference of €5. For both the expert and the non-expert the highest bid was €400.
4.3 Test of H3: Similarity Between the Endorser and the Consumer
The data for this hypothesis is gathered by asking participants to what extent they find themselves to be similar with the endorser. The participants ranked their similarity with the endorser on a scale from 1 to 5, where 1 was incomparable and 5 was very similar. Because the data is presented on a scale from 1 to 5 it has not been recoded into new variables as a binary distribution. Several participants did not want to answer the question and hence the total number of participants making a bid was 899 participants (N).
A one-way between subjects ANOVA was conducted to compare the effect of similarity between an endorser and consumers and the consumers’ willingness to pay. The results show there is no significant effect of similarity with an endorser F(4,89) = .70, p = .589. In other words, it can be concluded that there is not a significant difference in the means of the perceived similarity between the participant and the endorser and the bid amount.
For participants bidding on a product where the similarity with the endorser was incomparable the mean is €77.41. For participants bidding on a product where the similarity with the endorser was perceived as being very similar the mean was €92.26. The overall mean was €83.67. Comparing these five groups one see that the mean of the bids is increasing the more the participant perceived to be similar to the endorser (level 2 = mean €84.52; level 3 = mean €87.20; level 4 = mean
€85.23). From the very two ends of the scale (incomparable and very similar) there is a large difference of €14.85. The maximum bids varied from €340 to €400.
4.4 Additional Analyses
A chi-square test has been run on gender and three different variables: Level of expertise, similarity in skin color of the endorser and level of similarity. However, the results for the level of expertise were insignificant, X2 (1, N = 906) = .01, p >.05 and are therefore not included in this assignment. The analyses of the similarity in skin color and level of similarity with the endorser will be discussed now.
A chi-square test was performed in order to examine the relation between gender and similarity with the endorser’s skin color. Since not all participants wanted to state the perceived similarity with the endorser’s skin color the number of participants are 663 (N). The results showed that the relation between these variables were statistically significant, X2 (1, N = 663) = 5.01, p<.05. We can therefore accept our assumption saying there is a significant association between gender and perceived similarity between the participant and the skin color of the endorser.
Another chi-square test was performed in order to examine the relation between gender and similarity with the endorser. Participants rated the similarity with the endorser on a scale from 1 (incomparable) to 5 (very similar). The total number of participants stating their perceived similarity with the endorser is 899 (N). The results showed that the relation between these variables were statistically significant, X2 (4, N = 899) = 16.68, p<.05. We can therefore accept our assumption saying there is a significant association between gender and perceived similarity between the participant and the endorser.
5.0 Discussion
The results obtained from each hypothesis will be discussed separately. Therefore, first a discussion of the consumers’ preference for product advertisements featuring ‘endorsers who have the same skin color as the consumer themselves’ will be done. Secondly, it will be discussed whether or not consumers are more likely to make a purchase when the sales endorser is perceived as having greater expertise. Lastly, it is discussed whether or not similarity between the endorser and the consumer increases the probability of consumers’ buying behavior.
5.1 Similarity Between Skin Colors
Previous research have shown that consumers would generally have a preference for products being advertised by an endorser having the same skin color as the consumer themselves (Appiah 2001; Choudhury & Schmid 1974; Green 1999; Pitts et al. 1989; Schlinger & Plummer 1972; Szybillo & Jacoby 1974; Whittler 1991; Williams, Qualls and Grier 1995). However, the results from this online experiment showed something else and did not support this hypothesis. The use of an endorser having the same skin color as the consumer was hypothesized to have a positive effect on the consumers’ willingness to pay; nevertheless, the data showed that this was not the case. When looking at the means of the consumers who thought they had a similar skin color to the advertised person, it was respectively €87.34 for Caucasian participants and €77.56 for darker participants. The overall average mean for all bids was €82.51. Nevertheless, according to the ANOVA-analysis, the increase was not large enough to be significant. The presence of an endorser with a similar skin color, therefore, did not have the expected effect of increasing the consumers’ willingness to pay. There can be several reasons for this unexpected turn of events. First the Caucasian endorser was not European white and hence the consumers might not have thought of them as having the same skin color as themselves. Perhaps if the study would be replicated in Asia with the Caucasian endorser looking like an Asian woman, it would have the expected positive effect on the consumers’ willingness to pay. Like this, the consumers would more likely see a greater similarity
between their own skin color and the one of the endorser. A reason why the mean for the darker participant was marginally lower than both the Caucasian and the overall average can be due to the fact that Caucasians are most commonly found amongst the native populations of Europe and therefore the consumers should, theoretically, according to the hypothesis, want to pay more for a Caucasian endorser, as that skin color is more like the consumer’s own.
Another reason for the unexpected results can be due to the conducting method that has been applied. In the previous studies, measuring the effect of an endorser’s skin color, they have all been conducted as a survey. This experiment stands out from what has been done to date by conducting an online auction that is putting the consumers in a real-life situation where they actually have to purchase the product. In the previous studies, the participants solely had to indicate on a piece of paper what they would prefer if they were in a hypothetical purchase situation. Therefore, the results of this experiment might have a different outcome, because the participants up until now did not have to act on their saying, but when standing in a real life-situation they might be acting differently from what they said they would. Also, since this was an online auction experiment, the consumers were only showed one of the four treatments whereas in the previous studies the participants were shown two treatments (one with a Caucasian endorser and one with a dark endorser). Therefore, in the previous studies, the participants were able to compare the two endorsers and their skin color and to determine their preference, whereas in this experiment they were only able to see one endorser with one skin color.
Finally, the inconsistency of the results could be due to the product used in the advertisement. Muse (1971), Solomon and Bush (1977) found that the type of product could have an effect on the consumers’ purchase decision. Even though it has not been researched whether or not a technological product like an iPad affects the consumers’ purchase decision, it may be that the cheaper a product is, the less the endorsers’ skin color and gender matters. The participants could have seen the online auction as a way to get a product cheaper and as such, might not have been concerned about the skin color or the gender of the endorser.
5.2 Endorser with Greater Expertise
Previous literature has shown that consumers are more likely to make a purchase when the sales endorser is perceived as having greater expertise (McCracken, 1989; Friedman and Friedman, 1979; Ohanian, 1991). However, the results from this online experiment did not support these previous findings, as they did not show any positive sign of likelihood for making a purchase when the endorser was perceived as having greater expertise. The use of an endorser having greater expertise was hypothesized to have a positive effect on the consumers’ willingness to pay; nevertheless, the data showed that this did not seem to be the case. When looking at the means of the consumers who saw a treatment where the endorser either had or did not have an expertise sentence on the ad, it was respectively €81.66 for endorsers with a high level of expertise and €86.17 for endorsers with no level of expertise. The overall average mean for all bids was €82.51; hence, there is not a significant difference in the results for using an endorser having greater expertise in the field of the product. Consequently, the presence of an endorser with greater expertise did not have the expected effect of increasing the consumers’ willingness to pay.
There can be numerous reasons for this unexpected turn of events. One reason why this research does not support the previous results from the literature can be that the consumers in the experiment did not see the endorser as an expert, resulting in a lack of trustworthiness. Trustworthiness is important for source credibility, and without it, any other attribute such as expertness is not effective in producing attitude change (Friedman and Friedman, 1979). Experts tend to look like people who have authority by wearing a suit or perhaps a white coat like doctors do. This gives the consumers the impression that the person is someone with a higher status in the society and someone who has expertise within a field. The woman used in the advertisement was wearing a white shirt, which might not resemble an outfit an expert would normally wear. Hence, the first reason could be that consumers simply did not see the endorser as an expert due to her look and outfit. Another reason why this research does not support the previous findings in the literature could be due to the text used in the advertisement to show that the endorser was an expert. It is
difficult with only a short sentence to write a convincing text telling someone that a person on a picture is an expert and make him or her believe it. The text used in the advertisement was “Beyond
all expectations. I highly recommend this iPad! Technology expert from Amsterdam.” Maybe the
consumers did not find this text convincing and hence, it did not serve the purpose it was supposed to. It could also be that the consumers did not believe the text stating that the endorser was an expert from Amsterdam, since, as mentioned above, the endorser does not look like a native Western European woman. People may have prejudices towards experts from other countries and may trust experts from their own country more, and even though the text said the endorser was from Amsterdam, perhaps the consumers did not believe that, as her skin color is not European white.
Another reason why the result of this study was not significant may be due to the endorser’s gender. For advertisers, it is important to not only consider whether or not the race of an endorser has an effect on the consumers’ buying intension, but also whether or not the gender of the endorser plays a role. Various studies have found that gender plays a significant role on consumers’ attitudes towards an advertised product (Debevec and Iyer, 1988; Bellizzi and Milner, 1991; Peirce and McBride, 1999). Since the 1980’s, the gender role stereotyping has changed a lot, where the images of women reflected in advertisements have gone from being few to almost equal to men (Wolin, 2003). Siu (1996) found that men are more likely to be portrayed as product authorities, whereas women are more likely to be portrayed as product users. Peirce and McBride (1999) also found that male endorsers are better remembered than female endorsers. Therefore, it might have had an effect on the consumers seeing that the endorser was a woman and they might have seen her as being a product user rather than a product authority. Lastly, another reason for this surprising outcome can be due to the credibility of the endorser. Ohanian (1991) argued that the credibility of an endorser has an effect on the consumers’ attitude towards the endorsed product. The credibility of the endorser in this experiment was not tested and analyzed, but for future studies this could be an opportunity to research this hypothesis even further.
5.3 Similarity Between the Endorser and the Consumer
The third hypothesis for this thesis was also not supported. According to previous literature, the stronger the perceived similarity between an endorser and a consumer, the bigger the probability of purchase by the consumer (Zinkhan & Hong, 1991; Martin and Kennedy, 1994; Roy, 2014; White et al., 2012; Alekar, 2014). Nevertheless, the results from this auction experiment were not significant and did not support these previous findings. The use of an endorser having a stronger perceived similarity with the consumer was hypothesized to have a positive effect on the consumers’ willingness to pay, meaning that the consumer would theoretically bid more money on the product. However, the data showed a neutral effect that was neither positive nor negative. When looking at the means of the consumers who either thought or did not think they had a similarity with the endorser in the advertisement, it was respectively €92.26 for a perceived high level of similarity and €77.41 for no perceived similarity between the endorser and the consumer. The overall average mean for all bids was €82.51; as a result, there is not a significant difference in the results for using an endorser having a stronger perceived similarity with the consumers. The presence of an endorser with stronger perceived similarity did therefore not have the expected effect of increasing the consumers’ willingness to pay.
There can be different reasons for these opposing results compared to the previous studies that have been made. One reason could be that only the consumers from Amsterdam who were participating in the online auction would see a similarity between themselves and the endorser, whereas consumers from other cities in the Netherlands would not see a similarity. The number of consumers indicating a similarity would therefore be small and not significant enough to support the hypothesis. Since the online auction is only available in Dutch for people living in the Netherlands, it would not have made sense to use “The Netherlands” as a similarity instead of “Amsterdam”. Even though more people could relate to “The Netherlands” and see a similarity the participants already know that only Dutch people will see the ad and therefore it would just be strange and obvious that the endorser would be from the Netherlands. Another thing that might
have triggered this difference in the results is the phrasing of the supposed similarity. Perhaps it would have been easier for the consumers to see a greater similarity between themselves and the endorser if the text provided would have been longer and indicated more things that could be of similarity. It is difficult to create a similarity with consumers by just using one word, such as “Amsterdam.” If the educational institution, for instance, would have been mentioned in combination with Amsterdam, then perhaps the consumers could relate more to the person and thereby also see a greater similarity.
A fourth reason why the results are different from what have been found previously might be that the consumers assess similarity differently. One consumer might assess it by looking at the gender and skin color of the endorser, whereas another consumer might assess it by looking at the text “Amsterdam” or the appearance of the endorser. Hence, there might be different variables that the consumer could have considered when indicating whether or not he or she had a similarity with the endorser. It is unknown whether the consumer assessed the similarity in terms of the provided text “From Amsterdam” or other unknown variables.
5.4 Theoretical Implications
The theoretical contribution of this research is that it contributes to analyzing the effect of an endorser’s skin color, where there have been inconsistent findings in the literature to date. The method being used in this research is an online auction platform that measured the consumers’ willingness to pay for a product. The outcome of the research adds significant value to the existing literature by examining the effect on the endorser’s skin color on the sales of a product. Several of the previous studies have been conducted before the Internet, where commercials were more local and country oriented; this research also contributes with a newer perspective as marketing today is another industry than it was 20-30 years ago. Furthermore, the research deepens and provides new insight into the understanding of the impact of skin color differences and preferences for willingness to pay when seeing an ad.
5.5 Managerial Implications
The managerial contribution of this research is that the outcome brings results for marketers to consider when using brand endorsement. On the basis of results obtained through the analysis, knowing whether or not the skin color of the endorser affects willingness to pay, marketers can better predict purchase intentions and choose how to perfectly use the right endorser in the right context in order to increase the consumers’ willingness to pay for a product.
It is furthermore worth noticing that since some of the previous studies have been found, society has changed significantly and people now have better opportunities for interacting across different races. Therefore, today’s population may also be more willing to accept endorsers of different races (Whittler & Scattone, 1991). This is consistent with the more recent studies showing that there appears to be little difference in how consumers evaluate endorsers with different skin color in advertisements before and after the Internet.
5.6 Ethical Issues and Limitations
Regarding ethical issues, it is made clear to the consumers that the data gathered from Veylinx will be processed anonymously and will not be shared with third parties (Veylinx).
Concerning limitations, this experiment will only be conducted on Dutch consumers due to restrictions on the Veylinx website; however, these results are of external validity where the assumption is that it represents the whole population and can be generated to other nationalities as well. Another limitation of the experiment is that in order to make the skin color of the endorser look credible, it was not possible to use a Caucasian European woman, as making her skin color dark would not look natural. Instead, an Asian looking woman was chosen in order to make the darker skin color look more reliable. Hence, the distinction between a Caucasian and dark person lies in the difference between a Caucasian Asian person and a dark Asian person. Another limitation to this study is that people from Europe and Asia have different face and hair features, however, this was not possible to edit in Photoshop. Therefore, it is to a small extent visible that the
dark version of the Asian woman does not have a dark skin color in reality as her face and hair features would have looked differently.
5.7 Future Research
As mentioned previously, people from Europe and Asia have different face and hair features, which were not possible to edit in Photoshop in a way that would look natural and credible. A future experiment could therefore be to run the same experiment but with two different endorsers, one being from Europe and one from Asia for instance. Like this, the endorser’s appearance would, without a doubt, be seen as being credible and there would be no questioning whether or not the person had the actual skin color that was shown. In an experiment like this, where two different endorsers would be used, it would however be necessary to measure the attractiveness of the models beforehand in order to make sure this would not affect the outcome of the respondents’ preference.
Another recommendation for future research is that the experiment could also be done with different nationalities to make sure the results would be valid for everyone and not only the Dutch population. However, when conducting the research with different nationalities, the best outcome would be achieved if an online auction method was also used, since this method gave different results than doing the experiment as a survey.
An implication of the research is that the endorser was considered either as an expert or non-expert based on the job title “technology expert” and a recommendation of the product. Thus, it is unclear whether the respondent actually considered the endorser to be an expert and whether or not this would have any influence on the respondent’s buying intensions. A final recommendation for future research is to take this into account and test whether it would make a difference or not for the consumer, as this could create a new perspective for the research.
6.0 Conclusion
By doing an online experiment, the aim of this study is to examine whether or not the skin color of an endorser has a positive effect on the consumers’ willingness to pay. In other words, the aim was to see if the consumers would be willing to pay more for a product if the endorser advertising the product had the same skin color as the consumer him- or herself. Furthermore, the effect of the endorser’s level of expertise and level of similarity on the relation between the endorser and consumers’ willingness to pay was examined.
The experiment was conducted on an online auction platform called Veylinx, which is interconnected with the Vickrey auction. Consumers were bidding online on the product and the one(s) with the highest bid won the auction, but only had to pay the second highest bid, which is also called a second-bid auction. For the experiment, four treatments were done; one treatment with a white endorser with a high level of expertise and similarity with the consumers, one treatment with a dark endorser with a high level of expertise and similarity with the consumers, one treatment with a white endorser with no level of expertise and similarity with the consumers, one treatment with a dark endorser with no level of expertise and similarity with the consumers and lastly one treatment as a baseline without any endorser being advertised.
This study has answered the research question, arguing that the use of an endorser with the same or a different skin color than the consumer does not have a significant effect on consumers’ willingness to pay. This means a consumer is not willing to pay more for a product being advertised by an endorser having the same skin color as him- or herself. Also, by not adding moderators such as level of expertise and similarity made a significant difference on the consumers’ willingness to pay. None of the treatments supported the hypothesis.
This research has contributed to the existing theory by adding a new dimension, by focusing on the consumer behavior and the act of buying an endorsed product. Previous studies had a hypothetical buying situation where the consumers were asked about the endorser and their intention to buy a
product, where this study takes it one step further by putting the consumers in a real life situation where they actually have to make a bid and purchase the product. It is, therefore, a great contribution to the methodology aspect of understanding the consumers and what affects their buying behavior. The results of this study also found that what appears to be a good strategy on paper is not always a good strategy when executed. What the consumers say they might prefer and what they actually prefer when purchasing a product can be two very different things, that managers should also keep in mind when using endorsers to advertise products.
7.0 References
Alekar, A., (2014), The relationship between the race of a celebrity endorser and consumer purchase behavior in multi-racial societies, University of Pretoria
Appiah, O., 2001, Ethnic Identification on Adolescents’ Evaluations of Advertisements, Journal of Advertising Research, Vol. 40, NO. 5, pp. 7-22
Arshad, S., (2014), Nike’s Top 10 Highest Paid Endorsement Deals to Sport Players, TSM Plug, Available from: http://www.tsmplug.com/richlist/nike-highest-paid-endorsement-deals/
Barban, A.M., and Cundiff, E.W., (1964), Negro and White Response to Advertising Stimuli, Journal of Marketing Research, Vol. 1, NO.4, pp. 53-56
Barrett Sales Blog, Helping People and Business Sell Better, Available from:
http://www.barrett.com.au/about-us/everybody-lives-by-selling-something.html
Bellizzi, J. A., and L. Milner, (1991), Gender Positioning of a Traditionally Male-Dominant Product, Journal of Advertising Research, Vol.31, NO. 3, pp. 72–79
Byrne, D.E., (1971), The Attraction Paradigm, Academic PR., Vol.11
Choi, S.M., Lee, W.N., and Kim, H.J., (2005), Lessons from the rich and famous: A Cross Cultural Comparison of Celebrity Endorsement in Advertising, Journal of Advertising, Vol.34:2. pp. 85-98.
Choudhury, P.K. and Schmid, L.S., 1974, African American Models in Advertising to African Americans, Journal of Advertising Research, NO. 14, NO. 3, pp. 19-22
Creswell, J., (2008), Nothing Sells Like Celebrity, The New York Times, Available from:
http://www.nytimes.com/2008/06/22/business/media/22celeb.html?_r=1
Dekker, K., and Van Reijmersdal, E.A., (2013), Disclosing Celebrity Endorsement in a Television Program to Mitigate Persuasion: How Disclosure Type and Celebrity Credibility Interact, Journal of Promotion Management, Vol. 19, NO. 2, pp. 224-240
Deshpandé, R., and Stayman, D.M., (1994), A Tale of Two Cities: Distinctiveness Theory and Advertising Effectiveness, Journal of Marketing Research, Vol. 31, NO.1, pp. 57-64.
Debevec, K., and Iyer, E., (1988), Self‐referencing as a Mediator of the Effectiveness of Sex‐role Portrayals in Advertising, Psychology & Marketing, Vol.5, NO.1, pp. 71-84
Eagly, A.H., and Chaiken, S., (1993), The Psychology of Attitudes, Forth Worth, TX Harcourt Brace Jovanovich College Publishers, pp. 794
Ecosystem Valuation, Basic Concepts of Economic Value, Available from:
http://www.ecosystemvaluation.org/1-01.htm
Franke, N., Gruber, M., Harhoff, D. and Henkel, J., (2006), What You Are is What You Like – Similarity Bias in Venture Capitalists’ Evaluations of Start-up Teams, Journal of Business Venturing, Vol.21, pp. 802-826
Friedman, H.H., Friedman, L., (1979), Endorser Effectiveness by Product Type, Journal of Advertising Research, Vol. 19. NO.5 pp. 67-71
Green, C.L., 1991, Ethnic Evaluations of Advertising: Interaction Effects of Strength of Ethnic Identification, Media Placement, and Degree of Racial Composition, Journal of Advertising, Vol. 28, NO. 1, pp. 49-64
Guest, L., (1970), How Negro Models Affect Company Image, Journal of Advertising Research, pp. 29-33
Gustafson, K., (2013), Gift card spending to reach an all-time high: NRF, CNBC, Available from:
http://www.cnbc.com/2013/11/21/gift-card-spending-to-reach-an-all-time-high-nrf.html
Hochschild, J.L., Weaver, V.W., and Burch, T.R., (2007), The Skin color Paradox and the American Racial Order, Social Forces, Vol. 86, NO. 2, pp. 643-670
Hung, K., Chan, K.W., and Tse, C.H., (2011), Assessing Celebrity Endorsement Effects in China, A Consumer-Celebrity Relational Approach, Journal of Advertising Research, Vol.51, NO. 4, pp. 608-623
Kareklas, I., and Polonsky, M. (2010), A Meta-Analytic Review of Racial Similarity Effects in Advertising, Association for Consumer Research, Vol. 37 pp. 829-832
Kelman, H.C., (1961), Processes of Opinion Change, Public Opinion Quarterly, Vol.25, NO.1, pp. 57-78
Klein, N.M.A.I., (2013), We are not biased! A research of the relationship between similarity, selection and winning, University of Amsterdam
Khongkok, W., and YouLi, w., (2013), Measuring the Impact of Celebrity Endorsement on Consumer Behavioral Intensions, A Study of Malaysian Consumers, International Journal of Sports Marketing and Sponsorship, Vol. 14, NO.3, pp. 157-177
Lee, C.K.C., Fernandez, N., and Martin, B.A.S., (2002), Using Self-Referencing to Explain the Effectiveness of Ethnic Minority Models in Advertising, International Journal of Advertising, The Review of Marketing Communications, Vol. 21, NO.3, pp. 367-379
Maddox, K.B., and Gray, S.A., 2002, Cognitive Representations of Black Americans: Reexploring the Role of Skin Tone, Personality and Social Psychology, Vol. 28 NO. 2 pp. 250-259
Marketing-Schools (2012), Celebrity Marketing, Available from: http://www.marketing-schools.org/types-of-marketing/celebrity-marketing.html
McCracken, G., (1989), Who is the celebrity Endorser? Cultural Foundations of the Endorsement Process, Journal of Consumer Research, Vol.16, NO 3. pp. 310-321.
Miciak, A.R., and Shanklin, W.L., (1994), Choosing celebrity endorsers, Marketing Management, Vol. 3 NP. 3, pp. 50-60.
Muruganantham, G., and Kaliyamoorthy, S., (2009) "Celebrity Effect on Brand Positioning: A Study With Reference to Female Personal Care Products, "International Journal of Applied Management and Technology, Vol. 7 NO.1.
Muse, W.V., (1971), Product-Related Response to Use of Black Models in Advertising, Journal of Marketing Research, Vol. 8, pp. 107-109
Ohanian, R., (1991), The Impact of Celebrity Spokespersons' Perceived Image on Consumers' Intention to Purchase, Journal of Advertising Research, pp. 46-54
Peirce, K., and M. McBride, (1999), Aunt Jemima Isn’t Keeping Up with the Energizer Bunny: Stereotyping of Animated Spokes-Characters in Advertising, Sex Roles Vol. 40 NO. 11/12, pp. 959–66
Petroshius, S.M., and Crocker, K.E., (1989) An Empirical Analysis of Spokesperson Characteristics on Advertisement and Product Evaluations, Journal of the Academy of Marketing Science, Vol. 17, NO. 3 pp. 217-225
Pitts, R.E., et. Al., 1989, Black and White Response to Culturally Targeted Television Commercials: A Values-Based Approach, Psychology and Marketing, Vol. 6, NO. 4, pp. 311-328
Reagans, R., (2011), Close Encounters: Analyzing How Social Similarity and Propinquity Contribute to Strong Network Connections, Organization Science, Vol.22, NO.4, pp. 835-849
Roy, S.A. (2014), The Effectiveness of Attractive Endorsers in Fashion Advertising: Assessing the roles of Viewer-Source Similarity and Self-Referencing, University of Amsterdam
Schlinger, M.J., and Plummer, J.T., 1972, Advertising in Black and White, Journal of Marketing Research, Vol. 9, NO. 2, pp. 149-153
Shogren, J., Shin, S., Hayes, D. & Kliebenstein, J. (1994), Resolving Differences in Willingness to Pay and Willingness to Accept, The American Economic Review. Vol. 89 NO. 1 PP. 255-270
