On Explaining and Predicting the Effectiveness of Celebrity Endorsers

John C. Mowen, Oklahoma State University
Stephen W. Brown, Arizona State University
ABSTRACT - Consumer behavior scholars and advertising practitioners are devoting increased attention to promotional endorsements. Building upon two theories from social psychology, balance theory and attribution theory, this paper examines a new approach to explaining and predicting endorser effectiveness. More specifically, the results are presented of an experiment investigating consumers' perceptions of endorsers who promote multiple versus single products and of products promoted by multiple versus single endorsers.
[ to cite ]:
John C. Mowen and Stephen W. Brown (1981) ,"On Explaining and Predicting the Effectiveness of Celebrity Endorsers", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 437-441.

Advances in Consumer Research Volume 8, 1981      Pages 437-441


John C. Mowen, Oklahoma State University

Stephen W. Brown, Arizona State University


Consumer behavior scholars and advertising practitioners are devoting increased attention to promotional endorsements. Building upon two theories from social psychology, balance theory and attribution theory, this paper examines a new approach to explaining and predicting endorser effectiveness. More specifically, the results are presented of an experiment investigating consumers' perceptions of endorsers who promote multiple versus single products and of products promoted by multiple versus single endorsers.


Celebrity endorsers are regularly used to promote an endless list of products, including credit cards, automobiles, headache remedies, beer, and diet program. A number of factors have bean mentioned as important in using endorsers, such as the endorser's personal qualities--reputation, believability, likeability, etc. (Nelson 1974). With the Large number of products endorsed by celebrities, a natural outcome is that some celebrities endorse a number of products, e.g., Jack Nicklaus, Joe Namath. However, as noted in a monograph published by the advertising firm, Foots, Cone, and Balding (1978), care should be exercised to select an endorser not tarnished by the association with other products. Of course, "virgin" endorsers tend to be highly expensive, leading to one practical question. Does the endorsement of multiple products indeed tarnish a celebrity's effectiveness? A second practical consideration involves whether or not to use multiple endorsers, such as in advertisements for Miller's Lite Beer and American Express Credit Cards. Are there advantages in having a number of different endorsers recommend a product to offset the cost of such an approach?

As proposed previously by Mowen, Brown, and Schulman (1979) a general communications model, based upon the integration of balance theory (Heider 1958) and attribution theory (Kelley 1967) can be used to analyze factors affecting the effectiveness of product endorsers. In the Novas, et al. study, balance theory was used to define the relationships among the triad of cognitive almanacs of consumer (C), endorser (E), and product (P). According to the analysis, the consumer's perceptions of an endorser and a product represent sentiment or effective relations. The relationship between the endorser and the product represents a unit connection, or a perception by the consumer of the extent with which the endorser is associated or bonded with the product. Based upon the cognitive consistency analysis, an endorser will be maximally effective when both a strong sentiment relationship exists between the consumer and the endorser, and a strong unit relationship exists between the endorser and the product. With the development of strong positive sentiment and unit relations, the consistency forces are hypothesized to cause the consumer to cognitively reorganize the weaker consumer-product (C-P) relation so as to perceive the product more favorably (Abelson and Rosenberg 1958).

The extensive literature on communicator effects (Cohen 1964, Simons, Berkowitz, and Moyer 1979, Sternthal,

Philips, and Dholakia 1978) adequately details the factors influencing the consumer-endorser (C-E) sentiment relation. However, surprisingly little attention has been devoted to delineating the nature of unit relations in general and the relation Between an endorser and a product in particular. Following the suggestions of Mowen (In Press), Mowen, et al. (1979) investigated the application of attribution theory (Kelley, 1967) as a method of analyzing the unit connection between endorser and product.

From an attribution theory perspective, consumers may be conceptualized as seeking to determine the causal reasons for a celebrity endorsing a product. The question is, does the celebrity endorse the product because of his or her belief in its positive characteristics or because of external factors, such as monetary incentives? To the extent that consumers perceive that the endorser really believes in the product, the unit connection between endorser and product (the E-P relation) is hypothesized to be strengthened.

The use of attribution theory as an approach for understanding the nature of the unit connection is new. Despite the proposition's novelty, however, several factors suggest its application. First, the concept of association, delineated by Heider (1958) to describe the nature of the unit connection, is generally considered to be one of the fundamental elements in determining causality. Attribution theory was developed to predict individuals' perceptions of social causality. Therefore, it makes conceptual sense to use attribution theory as an approach to specify the nature of the unit connection. Second, that attribution theory and balance theory may be related should not be surprising for both were analyzed first by Heider in his seminal (1958) work.

To analyze the factors potentially influencing consumer perceptions of the E-P relation, one may utilize Kelley's (1967) attributional approach for individuals having multiple observations of an endorser's actions. Kelley argued that in circumstances in which an observer has multiple observations, three sets of attributional cues are utilized to establish causality--distinctiveness, consistency, and consensus. In an advertising context, distinctiveness refers to the extent that the endorsement occurs uniquely in the presence of the product. If a celebrity endorses several products, the relation between himself and a particular product is not distinctive, leading to an inference that the nature of the particular product was not the reason for endorsement. Thus, attribution theorists tend to agree with some professional advertisers, i.e., that endorsers can become tarnished by endorsing multiple products (Foots, Cone, Balding 1978). Consistency deals with the relationship between endorser and product over time and modality. High consistency will strengthen the perceived E-P bond. For this to occur, the endorser should advocate the product over a variety of media in a long running campaign. The third criteria, consensus, concerns the consumer's perception of whether other individuals view the product similarly to the endorser. To the extent which friends, reference others, or possibly other endorsers support the product, the consumer will perceive the message to result from the nature of the product rather than from situational factors such as a payment for services. Thus, by using multiple endorsers, advertisers may be effectively utilizing the concept of consensus.

Utilizing George C. Scott, Steve McQueen, and Paul Newman as endorsers of a fictitious television set, Mowen et al. (1979) manipulated distinctiveness and consensus information available to subjects. Distinctiveness was varied by informing subjects that the endorsers had signed contracts to endorse only one product--a new television set (high distinctiveness) or the television and four additional products (low distinctiveness). Consensus was varied by depicting the fictitious company as signing only one individual to endorse the product (low consensus) or signing six celebrities to endorse the television set (high consensus ).

The results of the study supported neither the balance nor the attribution theory analyses. First, the manipulation of consensus and distinctiveness information failed to influence consumer perceptions of the causality for the endorsement (the endorser-product connection), thereby, not supporting the attributional analysis. With the lack of variations in the unit connections, according to the balance analysts, one would not expect the manipulations to influence the perception of the product. In addition, several complex interactions were found involving consensus information, distinctiveness information as well as the age of the consumer. Such interactions were also not supportive of the balance analysis.

The research presented in this paper conceptually replicates the Mowen et al. (1979) research. However, two important modifications were made. First, a new product was selected for endorsement. Consumers saw a story-board of an advertisement for a fictitious new ball point pen--the "Easy Writer." Utilizing a television set as the product in the previous experiment could have resulted in consumers perceiving risk in allowing an advertisement to influence their perceptions of a relatively costly consumer good. Furthermore, in the earlier study, the actors (Newman, McQueen, and Scott) may not have been perceived by consumers to have much relevancy to a television set. By developing advertising copy in which the endorser discussed using the pen to sign autographs, it was hoped that greater relevancy would be established. A second modification in the present study was the addition of conditions in which the endorser was said to have signed contracts to endorse products normally used by higher social status individuals. In the earlier study, the products were relatively low status--motorcycles, cigars, cologne, and rums. In the present study, higher status products were utilized (i.e., Hilton Hotels, Chivas Regal Scotch, and Mercedes Benz Automobiles) as well as the lower social status products.

The experiment snagging from these considerations was developed as a 313, full factorial, between subjects design. The three levels of distinctiveness were: high distinctiveness (the celebrity endorsed only the pen), Low distinctiveness-low social status product, and low distinctiveness-high social status product. The three levels of consensus information was: low (no others endorsed the pen), high consensus-celebrity (Steve McQueen also endorsed the pen), and high consensus-student (a student also endorsed the pen).



Ninety-nine individuals of both sexes served as subjects. Subjects were students interviewed in their dorm rooms at a large midwestern university. Students were selected as subjects for two reasons. First, students are a primary target market for producers of ball point pens. Second, the use of a relatively young, homogeneous population allowed a more precise test of our hypotheses. In previous research, Mowen et al. (1979) found interactions between age and the variables of consensus and distinctiveness. In order to test for other possible explanations for the results of the Mowen et al. (1979) study, it was important to hold age relatively constant in order to avoid a larger, more cumbersome experimental design.

Subjects were approached in their dorm rooms by a same sexed experimenter. The experimenter would briefly state that a new advertising campaign was being tested, and ask if they could help out. All subjects agreed to help and were then handed an experimental booklet. On the first page subjects read:

We are interested in obtaining your opinions concerning a particular test advertisement. You will be shown an advertisement and then asked several questions concerning your reaction to the ad and to the particular product described in the ad.

One important point to remember: since this is a test advertisement, it is presently in rough draft form. We ask that you keep in mind the fact that if this ad were to appear in a magazine, it would be in color and professionally laid out. Everything else, however, would remain exactly as you see it. Do you have any questions?

On the second page of the experimental booklet, subjects read material which was labeled background information. The experimenter explained that since the subject was receiving only a draft of the proposed magazine ad, he or she should get additional information. All subjects read the same first paragraph:

This advertising campaign has a total budget of $1,200,000. The "Easy Writer" will be distributed in all 50 states and in ten foreign countries through better stationery and department stores.

Subjects then read that Paul Newman had signed a two-year contract to endorse the pen. Next, the manipulation of distinctiveness was made. In the high distinctiveness condition subjects read that "Paul Newman has refused to endorse any products other than the "Easy Writer". In the low distinctiveness, low status conditions, subjects read: "Paul Newman has also signed contracts to endorse Puerto Rican Rums, Grendiers Cigars, Yamaha X5650 motorcycles, and Brut Aftershave Cologne." In the low distinctiveness, high status conditions, subjects read: "Paul Newman has also signed contracts to endorse Hilton Hotels, Mercedes Benz Automobiles, Chivas Regal Scotch, and Botany 500 suits."

Subjects were next shown the 11 x 14 inch storyboards. The manner in which the storyboards were shown constituted the manipulation of consumer information. In low consensus conditions, subjects saw only one storyboard having a large photograph of Paul Newman on it. To the side of the photograph, the below standard endorsement was printed:

PAUL NEWMAN does. Why?

"I've signed a lot of autographs during my career. But whether I'm signing autographs or writing a check or letter, I require a superior writing instrument. And Easy Writer is very definitely a superior writing instrument."

Your Easy Writer will never skip, never stick. It is so well engineered, that it can't leak or blot, either on your fingers or on your paper. The Easy Writer flows with such ease and comfort that it actually lessens writing fatigue.

And since it's refillable, you can expect to sign your name again and again and again.

Ask for the Easy Writer in any good art, stationery or college bookstore.

In the "high consensus-celebrity" conditions, subjects saw the Paul Newman storyboard and in addition saw storyboards with pictures of Steve McQueen and George C. Scott. The same standard endorsement was printed for each of the celebrities. In the high consensus, average person conditions, subjects saw the Paul Newman storyboard plus two story-boards, each having a different picture of two male college students. To the right of the photo of the student with pen in his hand, the copy was placed.


"Being a college student requires a lot of writing everyday. But whether I'm taking notes, an exam, or writing a check, I require a superior writing instrument. And Easy Writer is very definitely a superior writing instrument."

Your Easy Writer will never skip, never stick. It is so well engineered, that it can't leak or blot, either on your fingers or on your paper. The Easy Writer flows with such ease and comfort that it actually lessens writing fatigue.

And since it's refillable, you can expect to sign your name again and again and again.

Ask for the Easy Writer in any good art, stationery or college bookstore.

Dependent Variables

Eleven dependent variables were taken assessing subjects' impressions of the product, the advertisement, and the major celebrity endorser--Paul Newman. Concerning the product, subjects were asked: "How favorable an opinion do you have of the Easy Writer?", and "If you were in the market for a pen, would you buy the Easy Writer?" Concerning the advertisement, subjects were asked, "What is your overall reaction to the advertisement?" In addition, subjects rated the advertisement on a series of five seem-tic differential scales: interesting-dull, appealing-unappealing, believable-unbelievable, informative-uninformative, and eye-catching-not eye-catching. Concerning their impression of Paul Newman, subjects were asked: "Is Paul Newman doing the advertisement solely for the money or because he really believes in the Easy Writer?", "To what extent does Paul Newman believe in the Easy Writer?", "How likeable is Paul Newman?" and "How much do you trust Paul Newman?" All dependent measures were assessed on seven point scales anchored with appropriate descriptors. (Lower scale values indicate a more positive impression.)


Subjects' perceptions of Paul Newman were analyzed first. For the question asking subjects to indicate their liking for the actor, one significant main effect was found for the consensus variable, F(2,90) = 3.03, P<.055. The pattern of means revealed that subjects liked Newman most when the student endorsers were also shown (M = 1.45) and liked Newman less when his endorsement appeared alone or in conjunction with Scott and McQueen (M, alone = 2.45; M, celebrity endorsers = 2.64).

On the two questions asking for subjects' reactions to the "Easy Writer Pen", similar effects were found. When asked for their opinion of the pen, F(2,90) = 4.3, p<.02 and whether they would buy the pen, F(2,90) = 3.7, p<.05, a main effect for distinctiveness occurred. In each case, one finds that subjects reacted more favorably to the pen in high distinctiveness (M = 2.09 for opinion and M = 2.18 for buying likelihood) than in the two low distinctiveness conditions (Opinion-HSS M = 2.75, Opinion-LSS = 2.56; Buy-HSS = 2.85, Buy-LSS = 2.99).

Of the six questions asking subjects to give their impressions of the advertisement, two significant effects were found, each paralleling those found directly above. On the question asking for the subjects' overall reaction to the advertisement, a main effect was found for distinctiveness F(2,90) = 4.12, p<.02, revealing that impressions were most favorable in the high distinctiveness conditions. (M high distinctiveness = 2.42, M HSS = 3.12, M LSS = 3.03). Similarly, subjects found the advertisement more interesting in the high distinctiveness conditions (M high distinctiveness = 2.60, M HSS = 3.15, M LSS = 3.15).

On the question asking subjects to indicate the extent which Newman believes in the product, a significant consensus by distinctiveness interaction occurred, F(4,90) = 2.87, p<.03. Figure 1 reveals the nature of the interaction. (Table 1 presents the means for all of the dependent variables.) As shown by Figure 1, the interaction resulted from the pattern of the "low distinctiveness-high SS product" conditions diverging substantially from the two remaining distinctiveness conditions. Thus, one finds that in the low distinctiveness-high SS conditions, subjects viewed Paul Newman as believing in the pen in "low consensus" conditions. However, in the "high consensus-celebrity'' condition, belief decreases and in the "high consensus-student" conditions, perceptions of his belief drop substantially. This pattern diverges from that of the "low distinctiveness" conditions and "low distinctiveness-low SS" conditions, in which the perceptions of belief increased in the "high consensus-student" conditions. No significant effects were found for the variables assessing subjects trust in Paul Newman. on the question asking whether Newman was doing the advertisement for the money or for the product, a marginally significant interaction occurred, F(4,90) = 2.13, p<.10, which generally followed the same pattern as the belief variable.


The results of the experiment generally supported the predictions for the effects of distinctiveness information, as derived from attribution theory. When the celebrity endorsed only one product, the product was seen more favorably, individuals indicated a greater interest in buying the product, subjects viewed the advertisement more favorably, and subjects responded more positively to the promotional massage. These important distinctiveness effects show the value to an advertiser of establishing an exclusive agreement with a celebrity.

While generally supporting attribution theory, the results for distinctiveness information failed to support the balance theory analysts. According to the analysis, perceptions of the product are hypothesized to change as a result of a highly positive endorser being placed in a unit connection with the product. As such, high distinctiveness information should first influence subjects' perceptions of the connection between endorser and product and their view of the product. (Perception of the unit connection were operationalized as the perception that the endorser believes in the product and that the endorser is doing the endorsement for the product, not for the money). However, those perceptions were not influenced in the same manner by variations in distinctiveness information as were variables assessing perceptions of the product. Therefore, as with the Mowen et al. (1979) study, the balance theory analysis was not supported.

Turning to the results for consensus, significant main effect was found in which subjects liked Newman most when a student endorser also recommended the product.

The student subjects could apparently relate more to Newman when people like themselves (i.e. student actors) were also favorable toward the same product he was endorsing.

A complex pattern emerged in the consensus by distinctiveness interaction for the dependent variable of Newman's belief in the pen. (See Figure 1. ) For high status products, Newman's association with the student endorser resulted in lower believability than in other conditions. This result suggests that advertisers should be careful not to associate a celebrity, who typically endorses high status products, in the same advertising campaign in which the "masses" also endorse the same product. In the high distinctiveness condition, a similar pattern of results occurred for the believability measure in consensus and in high consensus, celebrity endorser conditions. Higher believability resulted when Newman endorsed only the pen, and the pen was also endorsed by students. This finding supports both the distinctiveness and consensus predictions and suggests that an "exclusive" celebrity can be effectively used with lay endorsers whose characteristics (e.g. age, occupation) are similar to the target audience.

In the low distinctiveness-low social status conditions, the association of Newman and the students with the pen produced more believability for the celebrity, even when Newman was also endorsing other products. This suggests that even multiple product celebrity endorsers associated with low status products can have relatively high believability if linked with "lay endorsers" similar to the target audience.

The results from the present research differ from the earlier Mowen, et. al. (1979) study. The 1979 study supported neither the balance nor attribution theory analyses. The study reported in this paper, however, provides evidence that distinctiveness information does provide relevant input which consumers utilize in forming or not forming unit connections between the endorser and the product. Nevertheless, the complex interactions involving the types of consensus and distinctiveness information were found. In general, these interactions were not supportive of the theoretical propositions.


The results of this study provide some guidance for marketers, advertisers, and public policy makers. Each group should conduct careful pretesting of promotional messages to assess, a) whether the audience perceives the product more favorable when one versus multiple endorsers are used, b) audience perceptions of both the endorser and of the other products he or she endorsee, and c) whether alternative market segments perceive distinctiveness and consensus information differently.

This study's findings provide a clearer answer to the question of whether an integration of balance theory and attribution theory adequately explain the effects of celebrity endorsers than the earlier study by Mown, Brown, and Schulman (1979). This is not startling since the more recent study benefited from the earlier experience. At this point in time, little evidence exists that balance theory contributes significantly to our understanding of celebrity endorsers. According to the balance analysis, distinctiveness and consensus information are hypothesized to affect first the unit connection between endorser and product. Secondarily, the distinctiveness and consensus information affect perceptions of the product. The results in the present study and in the Mowen at al. (1979) research reveal, however, that one consistently finds that distinctiveness and consensus interact to influence the unit connection measure but influence the product ratings in a non-interactive manner. Such results cannot be predicted from the balance theory approach.

The research does indicate greater applicability of concepts derived from attribution theory and in particular the distinctiveness concept. A lack of support exists, however, for predictions based upon the consensus formulation. It may be that in advertising contexts, attempts to create the impression of consensus are doomed to failure. In order for consumers to gain an impression of consensus, it is probably necessary that endorsements appear independent to consumers. That is, the endorsers need to have the appearance that each acted independently of the other and were not "bought off" by the company. However, with consumers increasingly becoming wary of advertisements, the ability to create the impression of the independence of endorsers may be impossible, even if it in fact exists.

Despite the completion of two related studies, additional work is certainly needed on the topic of endorser effectiveness. Such studies could introduce additional product types, new endorsers, non-student subjects, and different operationalizations of distinctiveness, consensus, and consistency information. Further research, including replications with extensions, will help consumer research scholars and communication practitioners better understand and measure endorser effectiveness.






Abelson, R. P. and Rosenberg, M. J. (1958), "Symbolic Psychologic: A Model of Attitudinal Cognition," Behavioral Science, 3, 1-13.

Cohen, A. R. (1964), Attitude Change and Social Influence, New York: Basic Books.

Fireworker, R. B. and Friedman, H. H. (1977), "The Effects of Endorsements on Product Evaluation," Decision Sciences, 8 (July), 576-583.

Foote, Cone, and Belding (1978); Profiting from Celebrity Endorsers, New York: Foote, Cone, and Belding Communications, Inc.

Heider, F. (1958), The Psychology of Interpersonal Relations, New York: John Wiley and Sons, Inc.

Kaplan, G. (1978), Profiting from Celebrity Endorsers, New York: Foote, Cone, and Belding Communications, Inc.

Kelley, Harold H. (1967), "Attribution Theory in Social Psychology," in Nebraska Symposium on Motivation, David Levine, ed., Lincoln: University of Nebraska Preen, 192-238.

Mowen, John C. (In Press), "On Product Endorser Effectiveness: A Balance Model Approach," Current Issues and Research in Advertising--1980, James H. Leigh and Claude R. Herein, Jr. (eds.).

Mowen, John C., Brown, Stephen W., and Schulman, Meg (1979), "Theoretical and Empirical Extensions of Endorser Effectiveness,'' 1979 Educators' Conference Proceedings, Neil Beckwith et al., editors, Chicago: American Marketing Association, 258-262.

Nelson, Alan R. (1974), "Can The Glamour and Excitement of Sports Really Carry the Ball for Your Product?" Marketing Review, 29 (February), 21.

Simons, H., Berkowitz, N. and Moyer, R. (1970), "Similarity, Credibility, and Attitude Change: A Review and Theory," Psychological Bulletin, 73, 1-16.

Sternthal, B., Philips, L. W. and Dholakia, R. (1978), "The Persuasive Effect of Source Credibility: A Situational Analysis," Public Opinion Quarterly, (Fall), 285-314.

Winer, B. J. (1971), Statistical Principles in Experimental Design, New York: McGraw Hill Book Company.