
15 minute read
IRRATIONAL INFLUENCE IN THE INFORMATION AGE: AN EXAMINATION OF CELEBRITY ENDORSEMENTS IN POLITICS
Cory Millhouse
Abstract
Advertisement
This paper discusses the impact that celebrity endorsements can have on the public’s opinion of serious political matters. For the purposes of this research, celebrity endorsements are public statements of support for something or someone made by a person of significant influence. It attempts to identify whether celebrity endorsements of a candidate or political stance can have a substantial effect on a voter’s opinion, what kind of change can occur, and what kind of person that voter must be in order for change to happen. It examines empirical data recording voting habits before and after high profile political endorsements and research studying individuals’ opinions and voting intention following exposure to an endorsement. The research suggests that while celebrity endorsements do not always result in changes in voting intention, they usually have a significant impact on an individual’s perception of the candidate or party being enforced. It further suggests that such an impact on perception can indeed have a significant effect on the outcome of a given election.
Introduction
In 2007 the American democratic presidential nomination was hotly contested by both Barack Obama and Hillary Clinton. With the candidates neck and neck, campaign strategies that provided even the slightest edge were invaluable and necessary to claim victory for either candidate. Enter TV personality Oprah Winfrey. In May of 2007, Winfrey gave her official endorsement to Barack Obama, prompting speculation on whether this support would have any effect on Obama’s performance in the Democratic primary. Studies performed by the Pew Research Center (2007) at the time indicated that while almost 70% of the American public felt that the endorsement would have no impact on their individual vote, 60% believed that it would have a significant impact on the candidates’ success overall (para. 2). This poll reveals that many still put significant stock in endorsements made by their celebrity idols–after all, endorsements are only as impactful as the public thinks they are.
It is worth considering why people put so much stock in the statements and opinions of well-known figures, especially if they have nothing to do with the matter they are speaking on. Perhaps there is simply a wealth of willing ignorance among the public prompting people to side with the first opinion they hear from someone they consider nice. Maybe a longstanding cultural emphasis on celebrities has created a paradigm in which voter’s celebrity interests are seamlessly blended into their information, opinions, and rational thought. Shedding light on this mental process and the forces which shape public opinion is key to a robust understanding of the western democratic process. This study will examine the question: “What effect, if any, do celebrity endorsements have on individual voting habits?” If celebrity endorsements make a real difference in elections, their impact must be significant. Therefore, the following can be hypothesized:
Hypothesis 1: If celebrity endorsements can have a significant impact on elections, then a voter’s support for a candidate will change following exposure to an endorsement.
Hypothesis 2: Perceived impacts from endorsements should be explainable using existing theories on how individuals process and respond to celebrity endorsements.
Literature Review
Theoretical Models
There are two primary theoretical models that have been created to explain how consumers are influenced by external influences and why endorsements may
impact their voting habits: the Communication Process Theory and the Meaning Transfer Model.
A number of scholars have proposed a model that emphasizes overlap between audience’s and endorser’s experiences, opinions, and values as the factor that determines whether an endorsement accomplishes anything called the Communications Process Theory (Belch, 2012).
In this model the sender is the personality that has chosen to take a particular stance through an endorsement and the receiver is the audience being targeted by the sender’s particular message. Regardless of the channel used, all messages sent to the receiver must first undergo encoding under the sender’s field of experience (past experiences, opinions, values, etc.) and decoding based on the receiver’s field of experience. Encoding refers to the way a sender couches his or her message (i.e. the way they connect the message to their recipient). Similarly, decoding is the process by which the receiver translates a received message into their own thoughts. The way a particular message is received and decoded is heavily dependent on the receiver’s field of experience. Studies that have examined this theory concluded that communication is only effective when the two fields have significant overlap (Erdem & Swait, 2004). In other words, impactful communication is only likely when there is common ground between a celebrity and a voter.
The Meaning Transfer Model theorizes that the effectiveness of a celebrity endorsement stems from a celebrity’s ability to transfer his/her own significance to the product (McCracken, 1989). The theory is based around the concept of an invisible currency called “meaning.” The more significant an object or person, the more meaning it or they have. Therefore, the success of a celebrity endorsement is heavily dependent upon the process of a celebrity somehow transferring meaning to the person or object they are endorsing. Studies that have tested this theory break down the process into three stages: culture, endorsement, and consumption (Mitchell & Boustani, 1992; Langmeyer & and Langmeyer, 1993).
Figure 1: The Communication Process Model (Belch & Belch, 2012).

Figure 2: Meaning Transfer Model (McCracken, p. 315).
Stage 1 (culture) outlines how celebrities become charged with meaning. Their life experiences, accomplishments, etc. all contribute towards the creation of a persona with high societal meaning. Additionally, roles that a celebrity has assumed in their television, military, athletic, or other careers bring them into contact with objects and people that charge their meaning in a particular area of life. This grants the celebrity “specific meaning”:–increased influence in a certain area because of previous experiences that specifically relate to it. An example isa retired general having more influence in his endorsement of a new missile design than an actor. Stage 2 (endorsement) represents the actual advertisement of a product. The celebrity attempts to transfer their meaning to an item or person. Meaning transfer theory holds that this is most effective when the celebrity has specific meaning related to the product they are endorsing. If done correctly, the advertiser confers their meaning onto the product, increasing its value. Stage 3 (consumption) is the consumer’s response to the advertisement. Since meaning behaves like a pseudocurrency, once a product has received meaning from a celebrity, it is worth more to the consumer. Consumers who purchase this product now perceive a transfer of that meaning from the product to themselves. In elections consumers spend their votes on candidates they feel are worth enough to invest one in. Advocates of meaning transfer theory argue that an endorsement adds to the overall value of a would-bepolitician creating the perception that a vote for them is a better deal.
Comparison of the Two Models
Both models attempt to explain the source of an endorsement’s power. The Communication Process Model claims that endorsements are only effective when a celebrity shares characteristics with their target audience while the Meaning Transfer Model emphasizes a celebrity’s expertise related to the product being endorsed. As this paper attempts to determine the effects of endorsements on political campaigns,
these theories are useful for explaining the results of the raw data uncovered.
Laboratory Research
In order to fully test hypothesis 1, it is necessary to examine both qualitative and quantitative data that might indicate whether it is valid. Examining trends within a population can be risky when performing campaign analysis due to the plethora of factors that can act as intervening variables. Studies of polling trends and national popularity of candidates cannot adequately control for all the variables that may impact a voter’s habits. Therefore, it is necessary to first establish that individuals in a controlled environment can indeed be swayed by a celebrity endorsement. Once context from laboratory experiment data has been established, field data can be reviewed for a full understanding of the potential effects of celebrity endorsements.
Primary data was collected from a study performed by researchers at the University of Bath School of Management (Veer et al., 2010). The experiment sought to identify the impact ad endorsement could have on voter intentions and attitudes toward political parties.
Respondent Selection. 316 respondents were selected from citizens in Bath and Bristol in the United Kingdom. Adult respondents were selected using the following criteria: (1) respondents must be eligible to vote, and (2) respondents must have lived in the United Kingdom for more than two years. Respondents were gathered at public libraries, medical centers, and on public trains. In the final sample, 43% of respondents were male and 57% were female. 50% of all respondents included in the study were under the age of 28. Respondents were pre-screened using two five-point scales: the private self-consciousness scale (Fenigstein et. al., 1975), and a custom scale based on the voter engagement scale (Callero, 1985) that gauged their political involvement and the value they ascribed to their vote. The first scale asked respondents to rate how much attention they gave to politics with response options ranging from “1–Politics is something I rarely think about,” to “5–I am very attentive to politics and frequently reflect on my political beliefs.” The second scale asked respondents to indicate how important their vote was to them with response options ranging from “1–I would feel little to no loss if I were forced to give up voting at political elections,” to “5–Voting is an extremely important part of who I am.” Respondents were also asked to answer basic questions regarding their political persuasions and votes in previous elections. Researchers then used these responses to create a population with an even distribution of politically active and inactive conservatives and liberals.
Procedure. Respondents were told that researchers were attempting to determine the effect of literature on voting habits. They were then asked to read an article randomly selected by the researcher from two potential articles of similar length and readability. One article outlined the importance of voting and described poor voter turnout at the 2001 election while the other was a history of the City of Bath. Embedded in one-third of the articles was an ad containing an image of a well known British celebrity (ads included celebrities such as Helen Mirren, David Beckham, Kate Winslet, and Anthony Hopkins) with a slogan reading, “I vote Conservative, do you?” and embedded in another third was a similar ad featuring a random model instead of a celebrity. The remaining third contained no ad and served to establish a control group. Respondents were then issued a questionnaire relating to their voting intentions and their attitudes toward political parties. Attitudes toward the conservative party was determined through three five-point scales anchored by “like-dislike,” “unpleasant-pleasant,” and “bad-good,” respectively. Voting intentions were measured with a five-point response scale anchored by “strongly agree-strongly disagree” for the statement, “I am considering voting for the conservative party.” The researchers made the following prediction and attempted to test it as a hypothesis: “Celebrity endorsers of political parties will be seen as more familiar and likeable than non-celebrity endorsers, but may not necessarily impact voter intention.” (Ekant et al., 2010, p. 4)
Results Regarding Impact on Attitude. All respondents, regardless of their political activity, who were given articles that contained ads were on average much more positive in their ratings on the attitude-toward-the-conservative-party fivepoint scale. The total population of ad recipients saw a mean increase of about .5 points when compared to the control group. The respondents who specifically received celebrity endorsement ads saw an even greater increase in mean attitude score–a mean increase of .7 points over the control. However, the researchers noted that this trend can be somewhat misleading. Examining the mean scores of the total population ignores possible differences between politically active voters and inactive ones. When active and inactive populations are separated, slightly different trends appear. Politically salient individuals reacted identically. This is reflected in the increased scores to the model ad and the celebrity ad. The mean attitude scores for politically active respondents increased by .5 points over the control in response to both ads. Researchers concluded the endorsements had a baseline impact that could only be exceeded given certain conditions. In this case, respondents with little to no interest in politics gave endorsements additional impact. They concluded that the politically thoughtful individuals, while still swayed in attitude by the presence of an endorsement, were not as susceptible to the idea that celebrities’ opinions are somehow more valuable than someone else’s.
Results Regarding Impact on Voting Intentions. Researchers concluded from the experiment that the random advertisements using unknown models had no significant effect on voting intentions among either politically involved or uninvolved groups. However, they did see a one-point increase over the control group in mean voting support among politically uninvolved respondents.
The researchers concluded that lower levels of political salience empower celebrity endorsements and only celebrity endorsements— to create significantly higher voter intention. Researchers also noted that party affiliations remained almost entirely constant regardless of whether respondents received an endorsement or which endorsement they were exposed to. Respondents that had identified themselves as liberal voters maintained that they would not support conservative candidates with their votes and vice-versa. Researchers concluded that endorsements had no significant effect on allegiances to parties; while endorsements improved a party’s perception among many respondents, they did nothing to sway participants voting support away from parties they had already committed to in the past.
Figure 3: Mean Voter Intention by Political Involvement and Endorser Type (Veer, p. 12).

Limitations. One of the most glaring limitations of the research conducted is the geographical restriction of the respondents used. Participants were selected from random sites around a very small area of the U.K. True representation of the entire voter population would do a great deal to establish the external validity of this study and ensure that the data is valid. Another weakness of the study is its reliance on questionnaires and, more specifically, statements of intent rather than actual observation. Researchers merely inquired about voters’ intentions for support in future elections rather than actually examining how those respondents voted in later elections. While questionnaires can be useful in establishing the state of mind of a participant, self-evaluations, particularly when taken immediately following some sort of stimulus, can produce inaccuracies and may be affected by a number of invisible factors like internal bias or short-lived emotional drives that never manifest in action. If researchers had instead examined the future voting habits of the participants instead of simply asking them how they planned to vote, confidence in external validity of the data collected could be much higher. Finally, this experiment design has no ability to control for the respondents’ knowledge or opinions of specific celebrities. While random selection of respondents should help to account for this unknown variable, researchers had no way of knowing whether their respondents replied poorly to a celebrity endorsement because they were politically entrenched or because they had a personal dislike for the specific celebrity used in the ad. While not a colossal oversight by the experiment designers, small flaws in research design contribute towards unsurety in data gathered through experimentation. Future experiments can certainly aim to avoid these shortcomings.
Primary data was collected from a study performed by Dr. Anthony Nownes (2011) at the University of Tennessee Department of Political Science. The experiment sought to identify whether or not information on celebrities’ support for a specific party could create more positive evaluations of that party from respondents. It specifically exposed respondents to endorsements from celebrities that the individual respondents viewed positively.
Respondent Selection. Respondents for this study were drawn from a pool of 503 college students who agreed to participate for extra credit in Dr. Nownes’ classes. The students were randomly assigned to one of three groups – the control group who would receive no celebrity endorsement information, experimental group 1 (EG1) which received information about Jennifer Aniston’s support for the Democratic Party, and experimental group 2 (EG2) which received information regarding Peyton Manning’s support for the Republican Party. At the beginning of the experiment, all respondents (regardless of group) were issued a pretest questionnaire with 18 survey questions. Several questions gauged each respondents political attitudes and
affiliations (specifically asking about 2008 voting choices, political ideology, and the respondent’s party affiliation) while the remaining questions primarily identified respondent opinions on nine politically active celebrities, four current political figures, and four political parties. The survey also contained questions regarding respondents’ personal characteristics such as family income, academic major, personal income, age, home state, and ethnicity. Respondents were not informed of the true nature of the study in an attempt to limit the respondent expectancy effect. Participants were instead told that the study aimed to gauge the political attitudes of young adults and given the following project description:
This research is intended to continue the social scientific study of political attitudes and behavior in the United States. Specifically, it will seek to add to our growing knowledge of the development of attitudes in young people—that is, how young people form the opinions they have. Specifically, in this study we are seeking to determine how young people form their attitudes and opinions about political figures, UT personalities, and celebrities. (Data and Methods section, para. 483)
Dr. Nownes noted that the final experimental and control groups were statistically similar in all ways but one: he found that, for whatever reason, EG1 comprised substantially more women than the control group or EG2. To control for possibly skewed data, he performed a series of statistical tests (using the data collected from the experiment) which controlled for respondent sex in the multivariate models. He concludes that, given the similarity between results from the raw data and his later statistical tests, this imbalance poses no significant threat to external or internal validity.
Procedure. On day two of the study, respondents were treated differently based upon which group they had been assigned to. Respondents in the control group were given a survey identical to the one they had been given the day before to fill out again. Students in the experimental groups were given that same questionnaire containing the same questions but with a new cover sheet. Students in EG1 received a cover sheet that contained information detailing Jennifer Aniston’s financial contributions to the Democratic Party and Barack Obama’s 2008 campaign. Respondents in EG2 received a similar coversheet that instead outlined Peyton Manning’s financial support of Republican presidential and senatorial candidates. These cover sheets also included a story about John McCain and Barack Obama’s mutual love of animals in order to obscure the true purpose of the experiment from the participants. Survey results from each group were cross referenced with pre-test