Endorsing Products For the Money: the Role of the Correspondence Bias in Celebrity Advertising

Maria L. Cronley, University of Cincinnati
Frank R. Kardes, University of Cincinnati
Perilou Goddard, Northern Kentucky University
David C. Houghton, Northwest Nazarene College
ABSTRACT - Although there have been many studies regarding celebrity endorsed advertising, little work has been done examining why celebrity advertising is effective. The present study examines celebrity advertising in terms of the correspondence bias. This bias is the tendency to assume that a person’s behavior is a reflection of their true underlying dispositions when in fact, their behavior could be explained by situational factors. Results show that people exhibit correspondent inferences when evaluating celebrity endorsed ads, despite knowing the celebrity was paid a large endorsement fee. These inferences are subsequently related to brand evaluations.
[ to cite ]:
Maria L. Cronley, Frank R. Kardes, Perilou Goddard, and David C. Houghton (1999) ,"Endorsing Products For the Money: the Role of the Correspondence Bias in Celebrity Advertising", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 627-631.

Advances in Consumer Research Volume 26, 1999      Pages 627-631

ENDORSING PRODUCTS FOR THE MONEY: THE ROLE OF THE CORRESPONDENCE BIAS IN CELEBRITY ADVERTISING

Maria L. Cronley, University of Cincinnati

Frank R. Kardes, University of Cincinnati

Perilou Goddard, Northern Kentucky University

David C. Houghton, Northwest Nazarene College

ABSTRACT -

Although there have been many studies regarding celebrity endorsed advertising, little work has been done examining why celebrity advertising is effective. The present study examines celebrity advertising in terms of the correspondence bias. This bias is the tendency to assume that a person’s behavior is a reflection of their true underlying dispositions when in fact, their behavior could be explained by situational factors. Results show that people exhibit correspondent inferences when evaluating celebrity endorsed ads, despite knowing the celebrity was paid a large endorsement fee. These inferences are subsequently related to brand evaluations.

INTRODUCTION

A person takes a container of ice cream from their freezer and eats some. This is a rather common occurrence, of course, but why did the person exhibit this behavior? The first explanation that may come to mind may be that the person really likes ice cream. But, what if the person had just burned their tongue or the electricity controlling the freezer had gone out? In these cases, situational factors may be determining the person’s behavior rather than the person’s preference or liking for ice cream. This is the main idea behind the correspondence bias.

The correspondence bias is the tendency to assume that a person’s behavior is a true reflection of their beliefs or opinions, and thus, their underlying dispositions when in fact, their behavior could be explained entirely by situational factors (Jones 1979; 1986; Gilbert and Malone 1995). In other words, people make strong inferences from behavior and fail to adjust sufficiently for situational constraints. The correspondence bias is one type of inferential bias that arises when a person has to make a judgment, such as a prediction, causal attribution, or an attitude formation. Inferential biases result from the limited amount of cognitive capacity (i.e., limited attention and memory) people have to process information (Bettman 1979; Kahneman 1973) and the inappropriate use of relevant and irrelevant information (see Kardes, 1993, p. 174 for an extensive list of inferential biases).

Although social and consumer psychologists have long been aware of the correspondence bias, researchers still struggle with explaining its many causes and consequences (Gilbert and Malone 1995). And although correspondent inferences have been examined within some advertising contexts (e.g., Pechmann 1992), the correspondence bias has not been discussed extensively in the marketing literature, and the phenomenon has not been tested within a celebrity advertising context. The correspondence bias may have strong consequences for advertising, specifically celebrity endorsed advertising (Kardes 1993). Because most consumers today are quite savvy, they realize that celebrity endorsers are well-paid for their endorsements. And often, the media will disclose how much celebrities are paid to promote a company or product, such as the Forbes Magazine "Top Athletes" list (1997). Despite this strong situational factor, consumers may still make dispositional inferences with regard to the spokesperson’s attitude or preference for the product based on the advertisement, which may in turn affect the individual’s own brand evaluations. The present study examines if the correspondence bias does occur in this type of advertising.

The correspondence bias is not a new phenomenon. It has a rich and varied history in the field of social psychology, dating back to the 1960’s when Jones and Harris (1967) first investigated the phenomenon while doing work on correspondent and dispositional inferences. In a classic study, participants were shown essays that reflected either a pro- or anti- position on Cuba’s leader, Fidel Castro. Participants were told either that the essayist had been free to choose his written position or had been instructed by the debate coach to defend an assigned position. Participants then tried to infer the true attitudes of the essay writers. As expected, under the free choice condition, pro-Cstro essay writers were seen as very pro-Castro and anti-Castro essay writers were seen as very anti-Castro. But surprisingly, in the assigned position condition, test participants still ascribed pro- or anti- Castro attitudes to the essayists in a similar manner, even though they knew the position was assigned. Since this experiment, the correspondence bias has been investigated a number of times (e.g., Jones 1979; Fein, Hilton and Miller 1990; Allison, Mackie, Muller and Worth 1993; Lawson and Miller 1996), with results which have "...proven both robust and enigmatic..." (Gilbert and Malone 1995, pg. 24).

The correspondence bias has its roots in attribution theory, a family of diverse theories which attempt to explain how people gather and process information in order to formulate causal explanations (Fiske and Taylor 1991). In formulating causal explanations of other people’s behaviors, people ordinarily assume that someone’s behavior is the result of their natural, internal dispositions or is the result of the situation in which the behavior occurs. This idea is reflected in the law of noncommon effects (Jones and Davis 1965) which formalized the idea that when behavior occurs in the presence of strong situational factors, an observer should not assume the other’s behavior is the result of an internal disposition (Gilbert and Malone 1995). In other words, one should not discount the situation when attempting to explain another’s behavior. Kelley (1967) later extended this idea to form the discounting principle, and Ross (1977) labeled the tendency to discount the situation as the fundamental attribution error (Fiske and Taylor 1991; Myers 1994).

The reality that many people do not adhere to the directives given in the law of noncommon effects and the discounting principle is the correspondence bias. It can be shown that time and time again people make inferences about the dispositions of others when situational factors may explain the behavior completely (Gilbert and Malone 1995). For example, persons who are assigned a pro-choice position are viewed as more pro-choice than persons who are assigned pro-life positions (Jones and Harris 1967), and actors who are assigned to play a role are viewed as more inherently like that role than actors who are assigned a different role (Fleming and Darley 1993). These examples show just how pervasive this error in human judgment is within every day judgment.

As stated previously, the correspondence bias may have play a significant role in the way consumers evaluate celebrity endorsed advertisements. Although there is a broad set of studies regarding celebrity endorsed advertising, most of these examine characteristics about the endorser, such as physical attractiveness (e.g., Kamins 1990; Friedmen 1984), or examine if they are more effective than some other type of advertising (e.g., Friedman 1976; Atkin 1983). McCracken (1989) has examined why particular endorsers are effective, proposing that it is the cultural meanings ascribed to the endorser which makes him effective. If there is a good "match" between the endorser and the product so that the transfer of cultural meaning is possible, then the endorser is effective. This work is complementary to McCracken’s work on meaning transfer (McCracken 1986; 1988) in that it examines more specifically one process by which consumers may formulate judgments about celebrity endorsers. This study attempts to show how the correspondent inferences that people make when judging advertisements may contribute to the effectiveness of celebrity endorsed advertising.

This study addresses the basic research question: do individuals make correspondet inferences when evaluating a celebrity endorsed advertisement? In other words, when a consumer evaluates an advertisement with a celebrity endorser, does the consumer make correspondent inferences about the celebrity, even when it is clear that the endorser’s behavior is constrained by a lucrative contract?

Within the area of advertising, correspondent inferences have been studied on a limited basis. Pechmann (1992) used correspondent inference theory to explain how consumers view advertisers as inherently honest when presenting two-sided advertisements, which subsequently influences brand evaluations. These findings suggest that people can and do make some correspondent inferences when evaluating advertising claims. With regards to celebrity advertising, we suggest that consumers may make correspondent inferences about the celebrity’s attitudes and preferences. The following is hypothesized: when consumers are presented with a celebrity endorsing a product, they will attribute the celebrity’s appearance in the ad to a dispositional attitude of having a preference for the product and the brand, especially when it appears that the celebrity is endorsing the product freely. This in turn may influence consumers’ evaluations of the product.

Many studies have shown that the correspondence bias occurs even when the situational constraint is obviously known by the observer, although the inferences are weaker (e.g., Gilbert and Jones 1986; Gilbert and Malone 1995). Therefore, when evaluating a celebrity endorsed advertisement, the correspondence bias may cause people to make dispositional inferences regarding a celebrity endorser’s attitudes and preferences for a product, even when they know that the endorser is being highly paid for the ad and the advertisement is following a set script determined by the advertiser. In other words, the endorser is in a constrained situation where he is forced to state an opinion set forth by the sponsor. The following is hypothesized: even with the knowledge that the celebrity is under a highly constrained condition (i.e., a large endorsement contract), the correspondence bias will still occur, although the inferences will be lower in this constrained condition than in the free choice, unpaid condition. And again, this may influence consumers’ evaluations of the product.

METHOD

The general methodological organization of the experiment was based upon the Jones and Harris (1967) essay experiments. Participants were 100 undergraduate students at a midwest university. They were randomly assigned to one of four conditions and asked to complete a paper and pencil questionnaire. Participants were asked to examine a set of instructions, which contained one of the experimental manipulations, and a print advertisement, which showed Cindy Crawford endorsing a popular brand of orange juice.

A 2 X 2 between-subjects design was used, with a free choice versus no free choice (constrained) condition by an extreme versus moderate endorsement condition. The choice condition was presented within the context of a paid or a nonpaid celebrity endorsement. In the nonpaid condition, subjects were told that Cindy Crawford was "endorsing the product completely free of charge" as part of her support for organizations that promote healthy eating habits. In the paid condition, subjects were told that Cindy Crawford was endorsing the product as part of a paid endorsement contract and that she "is receiving $6 Million dollars" to endorse the product. The contract fee of $6 million dollars was disclosedin the constrained condition instructions in order to emphasize that Miss Crawford was extremely well compensated for her appearance. This endorsement amount, as part of a sponsorship contract, is typical for a supermodel of Cindy Crawford’s status and was verified through an informal investigation of celebrity management organizations and their fee schedules. The extreme versus moderate endorsement condition was manipulated through the advertising copy. In the extreme endorsement advertisement condition, five product benefits were presented in the ad, and Cindy Crawford’s name was repeated three times in the advertisement. In the moderate endorsement advertisement condition, the advertisement listed no product benefits and Cindy Crawford was mentioned only once as the endorser.

The survey included measures for correspondent inference, attitude toward the advertisement, attitude toward the brand, and attitude toward the celebrity endorser. Correspondent inference was measured with three items, using an 11-point scale. Subjects were asked to indicate their level of belief that Cindy Crawford likes the brand on a scale from 0 (Not at all likely) to 10 (Extremely likely). They also indicated their level of belief that Cindy Crawford frequently uses the brand on a scale from 0 (Not at all likely) to 10 (Extremely likely). And finally, subjects indicated their level of belief that Cindy Crawford views the brand as a good (healthy) product on a scale from 0 (Not at all likely) to 10 (Extremely likely).

The attitude scales were adapted from scales developed by Madden, Allen and Twible (1988). All of the scale items were a 7-point semantic differential scale with values ranging from 7 (the higher, positive rating) to 1 (the lower, negative rating). Attitude toward the ad was measured with four items: pleasant/unpleasant, likable/not likable, interesting/uninteresting, and good/bad. Attitude toward the brand was measured with four items: desirable/undesirable, pleasant/unpleasant, likable/not likable, and good/bad. And, attitude toward the endorser was measured with four items: interesting/uninteresting, pleasant/unpleasant, likable/not likable, and good/bad.

RESULTS

The means of the measures for the correspondent inferences were summed and averaged to form an index of correspondent inference, with a coefficient alpha=.89. A t-test was then performed, comparing participants’ inferences against the scale midpoint (5.00). This base value represents a neutral impression of the celebrity’s brand attitude (i.e., neither a favorable nor an unfavorable brand attitude). These results are depicted in Figure 1.

As expected, there was a significant difference in correspondent inferences in the nonpaid, free choice conditions for both the extreme endorsement ad and moderate endorsement ad (M=7.55 vs. 5.00), t=6.87, p<.001 and (M=6.61 vs. 5.00), t=4.35, p <.001, showing that individuals made strong correspondent inferences when they believed that the celebrity was appearing in the ad free of charge or by free choice. In the paid conditions, correspondent inferences were much weaker, as expected, but still significant from the neutral value for the extreme endorsement ad (M=5.97 vs. 5.00), t=2.62, p<.02 and marginally significant from the neutral value for the moderate ad (M=5.68 vs. 5.00), t=1.83, p<.10. Individuals still made correspondent inferences about the celebrity’s attitudes and preferences for the product, even when they were told that the celebrity was being paid $ million dollars to promote the product.

FIGURE 1

INFERRED ATTITUDES OF THE ENDORSER AS A FUNCTION OF PAYMENT OR NONPAYMENT

In the nonpaid condition, there was a marginally significant difference between the extreme endorsement ad and the moderate endorsement ad, t=-1.78, p<.07, although in the paid condition it was nonsignificant (t<1). There was no significant difference found overall between the extreme advertisement and the moderate advertisement, showing that individuals still made correspondent inferences regardless of the "intensity" of the celebrity endorsement (i.e., the number of product benefits specified by the celebrity) in the advertisement.

The means of the measures for attitude toward the ad, attitude toward the brand, and attitude toward the endorser were each summed and average to form an index for each attitude, with coefficient alphas of .91, .92, and .96, respectively. Given this, correlation analysis was done comparing these attitude indices and the correspondent inference index. Table 1 shows the correlations for these items.

The correlations between correspondent inference and attitudes toward the ad, the brand, and the endorser are all positive and significant, showing that there is a strong relation between correspondent inferences and evaluations about product. As inferences about the celebrity’s true brand attitudes and preferences increase in favorability, the participant’s attitude toward the brand increases in favorability. This is also true for attitude toward the ad. The high and positive correlation shows that as participants make increasingly stronger favorable inferences about the celebrity’s attitudes and preferences, their own favorable attitude toward the advertisement increases. Even the attitude toward the endorser is positively correlated with correspondent inferences of the participants, increasing in favorability as more favorable celebrity attitudes are inferred.

DISCUSSION

The results of the present experiment indicate that consumers are quite willing to make strong dispositional inferences on the basis of weak data. Even when a celebrity’s behavior is highly constrained (e.g., $6 million dollar contract), and is therefore uninformative, consumers assume that the celebrity, nevertheless, has a very favorable attitude toward the endorsed brand. Moreover, as inferences about the celebrity’s brand attitude became more favorable, consumers’ attitudes toward the ad, the brand, and the endorser increased in favorability.

The present experiment indicates that correspondence bias occurs when consumers evaluate celebrity endorsed advertisements and provides one possible answer to why celebrity advertising is an effective advertising tool. Results show how people ascribe positive attitudes and preferences for a product to the celebrity endorser, and because of the correspondence bias, they do this even when they know that the celebrity has been paid millions of dollars to promote the product. This is especially interesting in light of the fact that contemporary consumers are aware that celebrities are often given large endorsement fees, and it appears this knowledge maynot diminish the effectiveness of the ads.

TABLE 1

CORRELATION BETWEEN CORRESPONDENT INFERENCES AND ATTITUDE MEASURES

A possible alternative explanation for the strong inferences that were found is that consumers assume the celebrity endorser likes the brand and the product, regardless of whether she endorses the product or not. People may assume this without even seeing an advertisement featuring the celebrity. In other words, a neutral attitude may not actually be "neutral." Future research should include a control condition with no endorsement ad to establish a between-condition measure of a neutral attitude. But, we believe this study is a positive, primary step toward understanding how consumers form judgments about celebrity endorsers in advertising.

Also important from the study is the strong correlation between people’s correspondence inferences and attitudes toward the ad and attitudes toward the brand. These findings show that as individuals make increasingly stronger correspondent inferences about the celebrity’s attitudes and preferences for the product, their own evaluations of advertisement and the brand increase. This has important implications for justifying the use of celebrities in ads. A strong relation between attitudes toward the ad and the brand and correspondent inferences suggests that advertisers can use celebrity endorsers to illicit strong, positive attitudes toward their products. Of course, the advertiser hopes that the celebrity endorser elicits positive correspondent inferences. It follows intuitively that selecting the "right" celebrity for an advertisement is important, as McCracken also suggests (1989). As consumers form correspondent inferences about celebrities in ads and this subsequently influences attitudes and evaluations about the product, there is an implied necessity that the celebrity be appealing to the viewer and appropriate for the product. This generally supports similar findings in the extant literature on celebrity advertising.

Finally, this study contributes additional support for the theory of correspondent inferences and the correspondence bias and suggests that further study of how this process affects consumer judgment may be warranted. By increasing our understanding about how consumers make attributions and what types of attribution errors they make when evaluating and selecting brands, marketers may be able to better plan and execute marketing programs.

REFERENCES

Allison, Scott T., Diane M. Mackie, Michelle M. Muller and Leila T. Worth (1993), "Sequential Correspondence Biases and Perceptions of Change: The Castro Studies Revisited," Personality and Social Psychology Bulletin, 19 (2) (April), 151-157.

Atkin, Charles (1983), "Effectiveness of Celebrity Endorsers," Journal of Advertising Research, 23 (1) (February), 57-61.

Bettman, J.R. (1979), In An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley.

Fein, S., J.L. Hilton and D.T. Miller (1990) "Suspicion of Ulterior Motivation and the Correspondence Bias," Journal of Personality & Social Psychology, 58, 753-764.

Fiske, Susan and Shelly Taylor (1991), "Attribution Theory," In Social Cognition, Reading, MA: Addison-Wesley, 22-41.

Fleming, John H. and John M. Darley (1993), "Actors and Observers Revisited: Correspondence Bias, Counterfactual Surprise, and Discounting in Successive Judgment of Constrained Behavior," Social Cognition, 11 (4), 367-397.

Frieden, Jon B. (1984), "Advertising Spokesperson Effects: An Examination of Endorser Type and Gender on Two Audiences," Journal of Advertising Research, 24 (5), 33-41.

Friedman, Hershey H. (1976), "The Effectiveness of Advertisement Utilizing Four Types of Endorsers," Journal of Advertising, 5 (3) (Summer), 22-24.

Gilbert, Daniel T. and Edward E. Jones (1986), "Perceiver Induced Constraint: Interpretations of Self-generated Reality," Journal of Personality and Social Psychology, 50, 269-280.

Gilbert Daniel T. and Patrick S. Malone (1995), "The Correspondence Bias," Psychological Bulletin, 117 (1), 21-38.

Jones, Edward E. and K.E. Davis (1965), "From Acts to Dispositions: The Attribution Process in Person Perception," In Advances in Experimental Social Psychology, L. Berkowitz (Ed.), San Diego, CA: Academic Press, 2, 219-266.

Jones, Edward E. and V.A. Harris (1967), "The Attribution of Attitudes," Journal of Experimental Social Psychology, 3, 1-24.

Jones, Edward E. (1979), "The Rocky Road from Acts to Dispositions," American Psychologist, 34, 107-117.

Jones, Edward E. (1986), "Interpreting Interpersonal Behavior: The Effects of Expectancies," Science, 234 (October), 41-46.

Kamins, Michael A. (1990), "An Investigation Into the Match-Up Hypothesis in Celebrity Advertising: When Beauty May Be Only Skin Deep," Journal of Advertising, 19 (1), 4-13.

Kelley, H.H. (1967), "Attribution Theory in Social Psychology," In Nebraska Symposium on Motivation, Levine, D. (Ed.), Lincoln: Univ. of Nebraska Press, 15, 192-238.

Lawson, Timothy J. and Arthur G. Miller (1996), "Correspondence Bias and Attitude Toward Abortion," Current Psychology: Developmental, Learning Personality Social," 14 (4), 339-349.

Kahneman, D. (1973), Attention and Effort, Englewood Cliffs, NJ: Prentice-Hall.

Kardes, Frank R. (1993), "Consumer Inference: Determinants, Consequences, and Implications for Advertising," In Advertising Exposure, Memory and Choice, Mitchell, Andrew A. (Ed.), Hillsdale, NJ: Erlbaum, 163-191.

Madden, Thomas J., Chris T. Allen, and Jacquelyn L. Twible (1988), "Attitude Toward the Ad: An Assessment of Diverse Measurement Indices Under Different Processing Sets," Journal of Marketing Research, 25 (August), 242-252.

McCracken, Grant (1986), "Culture and Consumption: A Theoretical Account of the Structure and Movement of the Cultural Meaning of Consumer Goods," Journal of Consumer Research, 13 (June), 71-84.

McCracken, Grant (1988), Culture and Consumption, Bloomington, IN: Indiana University Press.

McCracken, Grant (1989), "Who Is the Celebrity Endorser? Cultural Foundations of the Endorsement Process," Journal of Consumer Research, 16 (December), 310-321.

Myers, David (1994), "The Fundamental Attribution Error," In Exploring Social Psychology, 53-61.

Pechmann, Cornelia (1992), "Predicting When Two-Sided Ads Will Be More Effective Than One-Sided Ads: The Role of Correlational and Correspondent Inferences," Journal of Marketing Research, 29, (November), 441-453.

Ross, Lee (1977), "The Intuitive Psychologist and His Shortcomings," In Advances in Experimental Social Psychology, Berkowitz, L. (Ed.), San Diego, CA: Academic Press, 10, 173-220.

----------------------------------------