Celebrity Advertising: An Assessment of Its Relative Effectiveness
Mohan K. Menon, University of South Alabama Louis E. Boone, University of South Alabama Hudson P. Rogers, Florida Gulf Coast
University
With escalating endorser fees, it is imperative to study the value added to the selling proposition by the celebrity. The study examines differences between advertisements with
celebrities and corresponding advertisements with non-celebrities. Overall findings do not reinforce the continued use of celebrities for certain types of products.
Introduction
Savvy marketers are discovering that both male and female consumers identify more with a "chunky Joe Six-pack than a buff Fabio. Regular guy marketing" is replacing the
age old notion that supermodels and beefy athletes are a necessary part of the advertising phenomena. This might "reflect demographic shifts, such as maturing Gen Xers and aging
Baby Boomers, who no longer identify with looking "buff' (USA Today, 1998). If this were true, it represents a sea change from the nineteen sixties, seventies, and the eighties
when celebrity spokespersons were ubiquitous. Though advertisers continue to use celebrities in advertising there is a realization that they are no panacea for achieving greater
attention and higher sales.
The general belief among advertisers is that advertising messages delivered by celebrities provide a higher degree of appeal, attention and possibly message recall than those
delivered by non-celebrities. Marketers also claim that celebrities affect the credibility of the claims made, increase the memorability of the message, and may provide a
positive effect that could be generalized to the brand (Cooper, 1984). Despite the potential benefits they can provide, celebrity advertising increases the marketers' financial
risk. Indeed, it is believed that using celebrities are an unnecessary risk unless they are very logically related to the product (Beverage Industry 1989; USA Today 1995).
Few studies have actually compared celebrity advertising to non-celebrity advertising to determine their relative effectiveness. For example, using advertisements with
celebrities and non-celebrities for a fictitious brand of sangria wine Friedman, Termini, and Washington (1977) found that the celebrity version of the advertisement had higher
scores on probable taste, advertising believability, and purchase intention, the three dependent variables. Similarly, Atkin and Block (1983) found that advertisements with
celebrity spokespersons had more favorable effect on consumers than those with non-celebrities. Therefore, based on previous research, it is hypothesized that advertisements
with celebrity spokespersons will have greater effectiveness as measured by higher scores on advertising believability and purchase intentions than those with
non-celebrities.
Literature Review
Researchers into the effectiveness of celebrity-centered advertising have appeared in both academic and trade publications (McCracken, 1989; Misra, 1986; DeSarbo and
Harshman, 1985; USA Today, 1998). At best the results of their effectiveness have been mixed. Most of the studies tend to highlight the efficacy of the phenomenon with
little attention given to why celebrity advertising work in some instances and not in others (Misra and Beatty 1990).
Assael (1984) suggests that celebrity advertising is effective because of their ability to tap into consumers' symbolic association to aspirational reference groups. Such reference
groups provide points of comparison through which the consumer may evaluate attitudes and behavior (Kamins 1990). In advancing reasons why celebrity advertising may be
influential, Atkins and Block (1983) assert that celebrity advertising may be influential because celebrities are viewed as dynamic, with both attractive and likable qualities.
Additionally, their fame is thought to attract attention to the product or service. However, in a study involving Edge disposable razor advertisements, Petty, Cacioppo, and
Schumann (1983) found that under high involvement conditions, arguments but not celebrities influenced attitudes, whereas under low involvement conditions, celebrities but
not arguments influenced attitudes. This suggests that celebrity influence may be related to the nature of the product rather than the person. Despite mixed findings, three
factors seem to be associated with the degree to which celebrity advertising is effective: source credibility, celebrity knowledge and trustworthiness, and celebrity appearance.
Source Credibility
Source credibility suggests that the effectiveness of a message depends on the "expertness" and "trustworthiness" of the source (Hovland, Janis, and Kelley, 1953; Sternthal,
Dholakia, and Leavitt 1978). In general, a message source with higher credibility tends to be more effective than one with less credibility (Sternthal, Phillips, and Dholakia
1978). Since higher levels of source credibility tend to be associated with more positive attitudes toward the message and lead to behavioral changes (Craig and McCann, 1978;
Woodside and Davenport, 1974), advertisers will opt to use celebrities if they think that they have a high level of credibility.
Researchers have identified three components as making up the credibility construct: knowledge or expertise, trustworthiness, and appearance or attractiveness (Baker and
Churchill, 1977; Joseph, 1982; Kahle and Homer, 1985; Maddox and Rogers, 1980). Attempts to measure the impact credibility on consumers' intentions to purchase indicate
that only "expertise" had any significant influence on intentions to purchase. There also seems to be a direct correlation between believability and overall advertisement
effectiveness, as measured by purchase intentions (Kamins, Brand, Hoeke, and Moe 1989).
Celebrity Knowledge and Trustworthiness
Celebrity knowledge or expertise is defined as the perceived ability of the spokesperson to make valid assertions. The expert spokesperson seems most appropriate when
advertising products and services that carry higher financial, performance, or physical risk while an ordinary consumer is considered best for low risk products or services
(Atkin and Block, 1983). When celebrity spokespersons were viewed as experts in the product category, they were more liked (Buhr, Simpson, and Pryor 1987). Further,
celebrity expertise tends to be highly correlated with believability and trustworthiness.
Celebrity Appearance
This has been taken to heart by advertisers. One only needs to watch television or print advertising to ascertain this fact. Physical appearance seems to induce positive feelings
toward the spokesperson and in some cases changes beliefs (Chaiken 1979). Spokespersons who are known to, liked by, and/or similar to consumers are attractive and to an
extent persuasive (McGuire, 1985). Numerous studies have indicated the link between celebrity attractiveness and attitude changes toward issues, product, and advertising
evaluations (Caballero and Pride, 1984; Chaiken 1979; Kahle and Homer, 1985). Others have suggested that when a celebrity's physical attractiveness "matches up" or is
congruent with the presence and degree to which the product or service advertised enhances attractiveness (i.e., attractive celebrity linked with an attractiveness-related
product) there would be a positive impact upon product/service and advertisement evaluations (Kahle and Homer 1985). On the other hand, if incongruence exists, then the
evaluations of both the product/service and the advertisement will be lowered (Kamis 1990).
However, not all the studies on physical attractiveness have found it to induce attitude changes. For example, Cooper, Darley, and Henderson (1974) found that a
deviant-appearing person, rather than an attractive person, was a more effective source of persuasion about income tax. Similarly, Maddox and Rogers (1980) found that "presence of
arguments" and "expertise, influenced consumer attitude ratings toward sleep while "physical attractiveness" did not. Further, likeability had a greater impact on attitude change
than other variables. Indeed, Freiden (1984) found that a celebrity spokesperson for a television set generated higher values for the likeability of the spokesperson when
compared to advertisements featuring an expert a typical consumer or a CEO. However, measures related to the knowledge and believability of the spokesperson, product
quality, trustworthiness of the advertiser and purchase intent did not provide favorable ratings.
Methodology
In order to assess the relative effectiveness of celebrity advertisements, the two types of advertising executions, celebrity versus non-celebrity, were tested across print
advertisements for six products: American Express (Arn), Apple computers (Ap), Avon cosmetics (Av), Milk (Mk), Pepsi Cola (Pc) and Ray Ban sunglasses (Rb). A celebrity
(C) advertisement and a non-celebrity (N-C) advertisement represented each brand. For this study, a celebrity spokesperson is described as one who is publicly recognized and
who use that recognition to further the goals of marketers by appearing in advertisements directed at consumers. Similarly, a non-celebrity is a person who, prior to placement
in the campaign, has no public notoriety but appears in an advertisement for the product.
Data was collected from students at an Southeastern university using a survey instrument that included two sixteenitem semantic differential scales. One scale is used to assess
spokesperson appearance, credibility, and knowledge (Ohanian, 1990) and the other scale is used to assess advertisement believability (Beltramini, 1982). The survey also
included questions related to familiarity with and liking for the spokespersons, likelihood of purchasing the product offered and demographic factors.
It is believed that this population was relevant because, unlike other studies that use students as surrogate consumers, the brands chosen in this study were being targeted at the
student segment. Eighty-three percent of respondents were between 1829 years, an equal percent of them were undergraduate and graduate students, and most (8
10/6)
were
unmarried business majors who were employed either full-time or part-time. Genderwise, there was an equal split between males and females and about 27 percent were foreign
students. Twenty-one percent of the students were Asians, 63 percent were Caucasians, and 16 percent were persons of African-American and Hispanic descent.
Each respondent was shown six of the twelve print advertisements at random and were asked to complete the questionnaire. In each case respondents were exposed to
advertisements for all six brands of products including three with celebrities (C), and three with non-celebrities (N-C). Data were collected from a total of 258 respondents
resulting in 1548 useable observations. Reliability assessment of the scales used in the study (Cronbach and Snow, 1977) indicates that all four had Cronbach's alpha greater
than .90: "Spokesperson Appearance (5 items, Alpha = .93); Spokesperson Credibility (4 items, Alpha= .95); Spokesperson Knowledge (6 items, Alpha =.96); and Advertising
Believability (10 items, Alpha =.96).
Results
Initially correlation analysis was conducted. In the case of the American Express advertisement, the appearance of the celebrity correlated highly with the person's credibility,
knowledge, liking for the person, believability of the advertisement, and purchase intentions. There were no significant correlations between any of the variables for the
advertisement
without a celebrity. With respect to the advertisements for Apple computers Computer, celebrity credibility and knowledge correlated with advertisement believability, which
in turn correlated with purchase intentions. Similar results were obtained for non-celebrity advertisement. The results indicate that credibility and knowledge correlated with
advertising believability. There were few significant correlations in the case of the Avon cosmetics advertisement and none of them were with advertising believability or
purchase intentions. For each execution of the advertisement, the significant correlations were between spokesperson knowledge and credibility.
In the advertisements for Milk, believability, knowledge, appearance and liking for the celebrity were highly correlated to each other and also with purchase intentions. For
advertising with non-celebrity spokesperson, credibility was highly correlated to advertising believability, which was in turn correlated to purchase intentions.
For advertisements about Pepsi-Cola significant correlations were identified between credibility and knowledge and between advertising believability and purchase intentions.
Celebrity appearance, knowledge, liking, and credibility of the celebrity were also highly correlated with advertising believability. At the same time, liking for the celebrity and
advertising believability were both correlated with purchase intentions. For advertising execution that involves non-celebrities, the analysis indicates that appearance,
credibility and knowledge were highly correlated with advertising believability.
When using celebrity-based advertisements involving Sunglasses the analysis indicate that there are high correlations between appearance, knowledge, liking, credibility,
advertising believability and purchase intentions. Advertisements using
non-celebrity spokes persons also yielded significant correlations between appearance, credibility, liking
and advertising believability, and purchase intentions.
As mentioned earlier, the intent of this article is to assess the effectiveness of celebrity advertising. In order to accomplish this, Multivariate Analyses of Variance between
the two advertising executions were conducted for each brand.
Hotelling's T2, the multivariate test of significance between the two groups, indicated that the null hypothesis of "no significant difference" could not be rejected in the
instance of American Express. In other words, one could not make an assertion that significant differences between the two (celebrity & non-celebrity) executions of the
advertisement exist. When the two executions were tested with respect to purchase intentions, the results were the same. The "null" could not be rejected.
In the case of Apple computers, the null hypothesis of no significant difference was rejected in the case of Apple computers computer (celebrity and non-celebrity)
advertising executions. In this case, spokesperson credibility, knowledge and advertising believability contributed to the overall difference. When the means were compared, it
was found that the three variables had lower values under the non-celebrity execution. ANOVA results for purchase intentions between the celebrity and non-celebrity
executions indicated that the hypothesis of no significant difference between them could not be rejected.
The Avon cosmetics advertisement had significantly different means between the two executions. In other words, the null hypothesis was rejected. Univariate F-tests revealed
that the overall differences were caused by three variables; spokesperson's appearance, knowledge, and believability of the advertisement. Examination of the means for these
variables revealed that they were higher for the non-celebrity executions. In the case of Avon cosmetics advertisements, ANOVA results for purchase intentions indicated that
the proposition of no significant differences were to be rejected at p<.05.
In the case of Milk advertising, the null hypothesis indicating a lack of significant difference between the three executions of the advertisement could not be rejected @ p<01
level. The result of ANOVA between the two executions with respect to purchase intentions was identical.
The difference between the two executions of the advertisement was also significant in the case of Pepsi advertising. According to the Univariate F-tests variables such as
spokesperson appearance and credibility contributed to the overall difference. Mean value for the appearance and credibility variables were higher in the case of the execution
with NC. In the case of purchase intentions, the null could not be rejected. The null hypothesis stating that there were no differences between the two executions was rejected
@ p<.05 for Ray Ban sunglasses. But when the Univariate F-tests were examined none of the variables were contributing significantly toward the overall difference at p<.05
level. Given these results, there was no reason to look at the individual mean values.
All the results of the analyses are exhibited in the following table. For each brand tested the table reveals whether or not there was an overall difference between the two
executions of the advertisement along with the listing of the variable(s) that contributed to the difference. In addition, the table displays relationships of the mean values of
the variables(s) that contributed to the overall difference variable(s) that contributed to the overall difference between the executions.
Table 5. Overall Results
| Product |
Overall Diff.b/w Executions |
Mean Values for Contributing Variables (C vs NC) |
| Am |
---- |
---- |
| Ap |
Credibility*
Knowledge*
Ad. Believability* |
C>NC |
| Av |
Appearance*
Knowledge*
Ad. Believability* |
C<NC |
| Mk |
--- |
--- |
| Pc |
Appearance*
Credibility* |
C<NC |
| Rb |
None |
---- |
*Significant at p<.05
Discussion
Milk, Pepsi, and Apple computers seem to be good examples of correlation between credibility of the spokesperson
and
advertising believability whether of not celebrities are used in the print advertising. Purchase intentions seem
to be correlated to advertising believability in the case of Milk
and Ray Ban sunglasses for both types of executions. Since advertising believability and purchase intentions are probably more important dependent variables from the
advertisers' perspective, it seems that for certain types of generic and branded products, celebrity advertising may not be any more effective than non-celebrity advertising.
With respect to the variance analysis, this study indicates that whether or not a celebrity spokesperson is used matters in the case of Apple computers, Avon cosmetics, Pepsi,
and Ray Ban sunglasses. At the same time, it does not seem to matter much in the case of American Express card and Milk. The differences might be in part due to the
conspicuous nature of the brands.
Since most of the celebrities in advertisements are attractive, consumers may use appearance as a differentiating variable between advertisements that they like and remember
and those that they do not like. However, this does not seem to have any influence on either the believability of the advertisements or purchase intentions.
Spokesperson credibility, knowledge, and advertising believability were contributing variables in the case of Apple computers and Avon cosmetics. Surprisingly non-celebrity
executions had higher mean values than celebrities in the case of Avon cosmetics but the opposite was true in the case of Apple computers. It could be hypothesized that in the
case of high-risk technical products, celebrities are perceived to be competent in the use of the product. The opposite situation occurred in the case of Avon cosmetics. Here
the non-celebrity executions score high. Meetings with the subjects revealed the general perception that the featured celebrities in Avon cosmetics advertisements are not
likely to use the brand. This finding provides support for the view that celebrities are willing to endorse almost any product or service if the fee is right. It seems that there is
no need for advertisers to rely exclusively on celebrities as they may exhibit low credibility & knowledge and advertisements featuring them may tend to be less believable. This
in turn may affect consumers purchase intentions. Earlier studies seem to indicate that product type has a mediating effect on the effectiveness of using celebrity spokespersons
(Atkin and Block, 1983). It is conceivable that the brands chosen for the study influenced the results, thus limiting the generalizability of the study. The case of Pepsi seems
similar with respect to appearance and credibility. Results for Ray Ban sunglasses show differences between executions but the data does not suggest any contributing factor.
Overall, the results from this study do not support the view that using celebrity advertising is more believable or effective than non-celebrity advertising for the brands tested in
this study. Consumers generally feel that celebrities are more attractive than non-celebrities, something that may draw initial attention to the advertisement. Beyond that, the
celebrities do not seem to make the advertising any more effective or believable. Further, purchase intentions did not vary between the executions for any of the brands tested.
The implication is that advertisers need to be cautious when using celebrity advertising as they are not believable in certain instances and hence may not deliver the
intended effect. This is certainly true of those celebrities who are seen as endorsing many types of products across a wide spectrum. It would seem that for some classes of
products, person-on-the-street type of advertising might be just as effective if not more so than those that use celebrities.
References
Advertising Age. Celebs Remain Entertaining, If Not Believable. (September 2, 1996): 18.
Assael, Henry. Consumer Behavior and Market Action. Boston, Massachusetts: Kent Publishing Company, 1984.
Atkin, Charles and Martin Block,. Effectiveness of Celebrity Endorsers. Journal ofAdvertising Research 23, 1 (February/March 1983): 57-62.
Baker, M and Gil Churchill, The Impact of Physically Attractive Models on Advertising Evaluation. Journal of Marketing Research 14,(1977): 538-555.
Basil, Michael, Identification as a Mediator of Celebrity Effects. Journal of Broadcasting and Electronic Media. (Fall 1996): 478-495.
Beltramini, Richard, Advertising Perceived Believability Scale. Proceedings of the Southwestern Marketing Association." D. R. Corrigan, F.B. KrA and R.H. Ross, Eds. Wichita,
Kansas: Southwestern Marketing Association, 1982. 1 - 3.
Beverage Industry. Simply Irresistible? Pepsi Learns There's a Downside to Signing Up Rock Stars. (May 1989): 1, 41.
Buhr, Thomas, Terry Simpson, and Burt Pryor, Celebrity Endorsers' Expertise and Perceptions of Attractiveness, Likability, and Familiarity. Psychological Reports 60,
(1987): 1307-1309.
Caballero, Marjorie and William Pride, Selected Effects of Salesperson Sex and Attractiveness in Direct Mail Advertisements. Journal of Marketing 48, (January 1984): 94100.
Chaiken, Shelley, Communicator Physical Attractiveness and Persuasion. Journal of Personality and Social Psychology 37, (1979): 1387-1397.
Cooper, Michael, Can Celebrities Really Sell Products? Marketing and Media Decisions 19, (September 1984): 6465.
Cooper, Joel, John Darley, and J. Henderson, On the Effectiveness of Deviant- and Conventional-Appearance Communications. Journal of Personality and Social Psychology
29, (May 1974): 752-757.
Craig, S and J. McCann, Assessing Communication Effects of Energy Conservation. Journal of Consumer Research 3,
(1978): 82-88.
Cronbach, L and R. Snow, Aptitudes and Instructional Methods: A Handbook for Research on Interactions. New York, New York: Irvington, 1977.
DeSarbo, Wayne and Richard Harshman, Celebrity-Brand Congruence Analysis. Current Issues and Research in Advertising. James Leigh and Claude Martin, eds. Ann Arbor,
Michigan: University of Michigan, 17-52,
Deutsch, Michael and Harland Gerard, A Study of Normative and Informational Social Influence Upon Individual Judgement. Journal ofAbnormal and Social Psychology 5 1,
(May 1955): 629-636.
Friedman, Hershey, Salvatore Termini, and Robert Washington, The Effectiveness of Advertisements Utilizing Four Types of Endorsers. Journal of Advertising 6, (1977): 2224.
Freidman, Jon, Advertising Spokesperson Effects: An Examination of Endorser Type and Gender on Two Audiences. Journal of Advertising Research 24, (October/November
1984): 33-41.
Hovland, Carl, Irving Janis, and Harold Kelley, Communication and Persuasion. New Haven, Connecticut: Yale University Press, 1953.
Joseph, W. The Credibility of Physically Attractive Communicators: A Review. Journal of Advertising 11, (1982): 15-24.
Kamins, Michael A, An Investigation into the 'Match Up' Hypothesis in Celebrity Advertising: When Beauty May be Only Skin Deep. Journal of Advertising 19, no. 1
(1990): 413.
Kamins, Michael A, Meribeth J. Brand, Stuart A. Hoeke, and John C. Moe, Two-Sided Versus One-Sided Celebrity Endorsements: The Impact on Advertising Effectiveness and
Credibility. Journal of Advertising 18, no. 2 (1989): 4-10.
Kahle, Lynn and Pamela Homer, Physical Attractiveness of the Celebrity Endorser: A Social Adaptation Perspective. Journal of Consumer Research 11, (1985): 954-961.
Kelman, Herbert, Processes of Opinion Change. Public Opinion Quarterly 33, (Spring 1961): 57-78.
Maddux, James and Ronald Rogers, Effects of Source Expertness, Physical Attractiveness and Supporting Arguments of Persuasion: A Case of Brains over Beauty. Journal of
Personality and Social Psychology 39, (1980): 235-244.
McCracken, Grant, Who is the Celebrity Endorser? Cultural Foundations of the Endorsement Process. Journal of Consumer Research 16, (December 1989): 310-32 1.
McGuire, William, Attitudes and Attitude Change. Handbook
of
Social Psychology, Vol. 2. Gardener Lindzey and Elliot Aronson, eds. New York: Random House, 1985. 233346.
Misra, Shekhar, The Role of Spokes Persons in Advertising: A Schematic Processing Perspective. Unpublished Doctoral Dissertation, University of Oregon, 1986.
and Sharon Beatty, Celebrity Spokesperson and Brand Congruence: An Assessment of Recall and Affect. Journal of Business Research 21, (1990): 159-173.
Ohanian, Roobina, Construction and Validation of a Scale to
Measure Celebrity Endorsers' Perceived Expertise,
Trustworthiness, and Attractiveness. Journal of Advertising
19, (1990): 39 - 52
-, The Impact of Celebrity Spokespersons' Perceived Image on Consumers' Intention to Purchase. Journal
of
Advertising Research. February/March (1991): 46-54.
Petty, Richard, John Cacioppo, and David Schumann, Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement. Journal of
Consumer Research 10, (September 1983): 135-146.
Sternthal, Brian, Ruby Roy Dholakia, and Clark Leavitt, The Persuasive Effect of Source Credibility: Tests of Cognitive Response. Journal of Consumer Research 4 (March
1978): 252-260.
L. Phillips, and Ruby Dholakia, The Persuasive Effect of Source Credibility: A Situational Analysis. Public Opinion Quarterly 3, (1978): 285-314.
USA Today. Ads with Sports Jocks Outscore Finger Pointers. November 20,1995. 5B.
__
. All's Not Well in Celebrity Pitchdom. November4, 1996. B I - B2.
. Being Incorrect, Politically and Otherwise, Sells. February 5, 1998. Al - A2.
Woodside, Arch and J. Davenport, The Effect of Salesman Similarity and Expertise on Consumer Purchasing Behavior. Journal of Marketing Research 11, (1974): 198-2
02.