On Using Attribution Theory to Understand Advertising Effects

ABSTRACT - Three experimental studies were reported in this session on attribution theory and advertising effects. The design of these studies and the results are discussed in terms of the contribution to attribution theory and to our understanding of how advertising works.

Citation:

Carol A. Scott (1982) ,"On Using Attribution Theory to Understand Advertising Effects", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 293-295.

Advances in Consumer Research Volume 9, 1982      Pages 293-295

ON USING ATTRIBUTION THEORY TO UNDERSTAND ADVERTISING EFFECTS

Carol A. Scott, University of California, Los Angeles

ABSTRACT -

Three experimental studies were reported in this session on attribution theory and advertising effects. The design of these studies and the results are discussed in terms of the contribution to attribution theory and to our understanding of how advertising works.

INTRODUCTION

It has been almost a decade since consumer researchers first used attribution theory to explicate some effects of advertising messages (e.g., Settle, 1972; Settle and Golden, 1974). During this period authors sought to establish the applicability of attribution concepts to the advertising communications context and to validate some predictions derived from the various attribution theory frameworks. Despite the fact that attribution theory became an extremely popular subject of research in social psychology and despite the early intense interest of consumer researchers, attribution theory has never become a dominant force in the consumer behavior literature. Although some might claim that the field was simply overrun with more cognitively-based information processing concerns, others might also note the seemingly difficult task of finding the right place for attribution theory concepts within the sort of substantive problem areas generally investigated by consumer researchers.

The three papers presented in this session all bring attribution ideas to bear on the substantive area of advertising effectiveness. Two of the papers, those by Sparkman and by Hunt, Domzal, and Kernan, test some ways of influencing attributions about the source of communications messages and illustrate perhaps why attribution research seems to be fading away. These studies apply attribution theory with considerable skill. but in doing so they show how inadequate the theory is for dealing with some of the important questions concerning the effects of advertising messages and their ultimate persuasive power. The third paper, by Dillon, Allen, Weinberger, and Madden, is methodologically based, but ironically may show one way out of the theoretical thicket. These authors explore some fundamental premises of attribution theory, and in doing so they indirectly show us how we may extend it and integrate it with other views of how communications work.

My comments on the three papers will deal with selected conceptual issues, and not primarily with methodological o philosophical ones. I begin with an examination of two assumptions underlying the papers by Sparkman and by Hunt et al. that affect the application of attribution theory to the area of communications. The second section expands the discussion to explore the limitations of current applications for explaining advertising effectiveness.

INFLUENCING ATTRIBUTIONS

Sparkman argues for the superiority of the discounting principle, derived from Kelley (1972) and to some extent from Bem (1972), over other attributional frameworks (e.g., Jones and Davis, 1965; Kelley, 1967, 1972) for explaining and predicting the effects of communications. He notes that these other frameworks focus attention on the distinction between internal or dispositional and external or situational causes for a person's behavior, while advertising messages may be perceived as being caused by a variety of external causes (characteristics of the product, being paid for presenting the message). Sparkman argues that attribution of the message to some of these external factors (e.g., to the characteristics of the product) will result in greater message acceptance. Rather than working to eliminate attributions to all external causes, the advertising strategist should seek to minimize only some of them.

Hunt, Domzal, and Kernan also assume the superiority of object or entity attributions. Basing their study on a series of experiments conducted by Eagley and her colleagues (cf., Eagley and Chaiken, 1975; Eagley, Wood, and Chaiken, 1978), these authors reasoned that people expect advertising messages to be characterized by reporting bias (i.e., the source reports facts or opinions that he doesn't believe or that he knows are not accurate). Following Eagley et al. (1978), they predicted that disconfirming this expectation would result in receivers perceiving the source as more truthful and the source's behavior as caused more by the entity or object than by the demands of the advertising situation. These perceptions, in turn, should lead to a greater acceptance of the message.

Although these two studies have different theoretical foundations, they both examine factors that should affect attributions of a source's behavior and, in turn, the persuasiveness of the source's message. Their common features also include at least two assumptions that may or may not be warranted. First, the authors seem to assume at least implicitly an automatic, hydraulic process of attribution. That is, they assume that attributing a spokesperson's statements--less to external constraints such as monetary gain will automatically result in attributing the statement more to the characteristics of the object being discussed. In many advertising contexts, however, the choice may not be so simple or so clear cut. A receiver may perceive other causes as plausible alternatives to monetary payment or product quality. Or, the receiver may not feel that reducing the probability of one cause has any effect at all on the probability of other causes. Sources may have many reasons for presenting a message, and these reasons may be somewhat independent of one another. Only if one is operating under a single cause model of the world in which an event has one and only one cause would the probabilities be perfectly related to each other in an inverse way.

Sparkman's data provides some interesting evidence relevant to this point. In his study, the discounting group did attribute Mr. Sinatra's behavior less to the money he might be paid than did the non-discounting group [It is interesting to note that even the discounting group neither agreed nor disagreed that money was a cause (3 the scale midpoint). This is surprising in view of the fact that these subjects were told that Sinatra was being paid $1.00 and in view of Sinatra's already large personal wealth which should obviate any monetary motivation. If one cannot erase the perception of monetary motivation, then perhaps increasing advertising effectiveness will depend more on enhancing the perception of other causes than on the negation of the monetary incentive. Ant, if one cannot erase the money attribution even in this pointed fashion, then it may be impossible to do so in more typical advertising settings.] (3.11 vs. 2.28). But, they did not turn overwhelmingly to an attribution of his behavior to the quality of Chrysler automobiles (2.51 vs. 2.91). Even though this difference is statistically significant, the effect size (as opposed to the level of significance, which is affected by the large cell sizes and the reduction in variance tue to the decision to replace missing values with a midpoint score of 3) is not dramatic. Subjects in this study may have attributed Mr. Sinatra's behavior more to a desire to save American jobs, a desire to help Mr. Iacocca in his valiant battle. or a multitude of other factors that were not measured.

An assumption of a hydraulic mechanism might be appropriate where one postulates and measures a conceptually exhaustive set of potential attributions. In some of the earlier attribution studies, for example, vaguer but more inclusive categories were used (e.g., "something about the person," etc., as used by McArthur, 1972). In fact, subjects were allowed to use combinations of categories in order to be sure that all possible attributions were recorded. In advertising and communications research, however, we have generally tried to be more specific, as it is undoubtedly true that the advertising strategist will find information on specific exemplars of such broad categories more useful in developing advertisements. But, moving from the general, exhaustive categories to more specific and less exhaustive factors causes some difficulties. At the very least, the researcher must assume a greater responsibility for asking about and measuring attributions to all possible exemplars. Hunt et al , for example, did ask about several possible causes. Unfortunately, they present no evidence to suggest that this list exhausted the perceived possibilities of receivers.

Second, both Sparkman and Hunt et al. seem to assume that receivers accept_ statements of fact those statements_ made by a source that are not attributable to a reporting bias. That is, both authors discuss attributions to the entity or object rather than attributions to the source's true beliefs about the entity or object. I do not wish to argue with Sparkman as to whether such attributions are external or internal, primarily because, as he points out, this is not the most important question from the advertiser's point of view. [There are those who consider a desire for money an "internal" motivation, and the distinction between internal and external causes can often be difficult to defend (cf., Kruglanski, 1975, and Calder, 1977).] Rather, I to wish to argue, as I have in the past (Hansen and Scott, 1976) that an attribution of a source's statement to an object is incomplete and unlikely to be useful in guiding the development of advertising messages or predicting what attributions will lead to greater message acceptance.

A communicator may report his or her true perceptions of a product. But, the receiver may or may not accept these true perceptions as accurate representations of reality. Mr. Sinatra may tell you what he really thinks of a Chrysler automobile, and a consumer filmed by a hidden camera may tell you his true opinion of a clock radio. But, what do they know about cars or clock radios? These spokespeople may or may not be perceived as possessing any degree of expertise. Or, these spokespeople may be viewed as possessing certain values or having certain tastes not shared by the receiving audience. As in Kelley's (1967) old example about movies, this consumer may like this clock radio, but since he likes all clock ratios or likes all bright pink radios his opinion is not very informative for others.

For me, McArthur's (1972) interpretation of Kelley's cube (1967) offers the most complete and most useful classification system. Here, an event such as an advertising message can be attributed to something about the object, something about the particular person (dispositional qualities, idiosyncratic traits), or something about the particular circumstances. Ruling out "something about the circumstances" does not mean that an attributor must leap from an attribution of the message to the source's true reaction to the object to inferences about reality This classification scheme allows for the case of a trustworthy but unexpert source, a trustworthy but not similar or attractive source (e.g., Mary Tyler Moore may like this stuff, but who wants to be like Mary Tyler Moore?"), and other combinations. At the very least, it reminds us that an attribution of the source's behavior may not be the same thing as an attribution of a quality to an object, and that trustworthiness alone may not be sufficient for persuasion.

Hunt et al. provide some evidence to support this more limited role for trustworthiness. First, the disclaiming manipulation did result in greater message acceptance on two of three relevant attributes, but it did not affect ratings of subject trustworthiness (honesty, sincerity) or attributions of the spokesperson's behavior. [In this discussion, I use only conventional levels of significance (i.e., p < .05 or better).] The hidden camera manipulation, on the other hand, did affect attributions of the spokesperson's message to such factors as payment, promotion of himself, and his true feelings, and did influence ratings of the spokesperson's honesty and freedom in the ad. The ratings show that although fewer people rated payment as the most important cause of the spokesperson's message and more people rated "true feelings" as the most important in the hidden camera as opposed to the typical consumer treatment, there is not much difference between the number of people citing "real facts" as most important in the two cells (8.47% vs. 8.93%). Despite the indications of greater perceived trustworthiness, the "hidden camera" spokesperson did not result in greater message acceptance.

These findings may be due to the particular nature of the spokesperson in the Hunt et al. study. In many source credibility studies, expertise is held constant but generally moderate to high when trustworthiness is varied. In the Hunt et al. study, however, the "typical consumer" may not have been perceived as terribly knowledgeable about such things as the accuracy and reliability of clock ratios. Thus, whereas most studies show increased persuasiveness resulting from increased trustworthiness, the current work may indicate that an increase in trustworthiness in the absence of a sufficient amount of expertise does not increase persuasiveness. It is unfortunate that no measures of expertise were included in the study. Such measures would provide evidence to suggest why the disclaiming manipulation did increase message acceptance as well as provide support for speculations about the effects of the hidden camera treatment.

PROBLEMS OF THE PAST, DIRECTIONS FOR THE FUTURE

Sparkman and Hunt et al., following in the footsteps of most consumer researchers, have applied attribution theory in a competent fashion. It is important to note, however, that these studies, like almost all others in the consumer behavior literature (and many in the social psychology literature) are applications of the theory, and not tests of it. Neither of these studies was designed to be able to falsify attribution theory or its underlying assumptions. Rather, these studies sought to determine whether some actions that an advertiser might take (e.g. 9 use a hidden camera) operate in the same way that other means of removing discounting cues do in other settings. Is the hidden camera manipulation strong enough to remove the perception of monetary or other motivations for presenting the message, for example? Thus, if the hidden camera manipulation had not worked as predicted, we would have questioned the manipulation rather than the presumed attributional processes.

It is equally important to note that the central focus of the experiments was not the overall effectiveness of the advertising message. Since attribution theory was not developed as a theory of communication, it should come as" surprise that it can be applied only to a portion of the advertising process, and that there are questions that it cannot handle. It simply is unfortunate that the questions it cannot handle may be precisely those that come to the advertising strategist's mind when she considers the ultimate persuasive impact of a message or to the researcher' mind when she begins with the idea of investigating the effectiveness of advertising.

The strategist is likely to ask about what factors will make the message persuasive, and not necessarily about how to create certain attributions, unless the connection between the attributions and persuasion is made fairly clear. To date, researchers have chosen to focus on factors which result in certain attributions and to ignore, for the most part, the question of which kinds of attributions or patterns of attributions about the source's message lead to the greatest degree of message acceptance and behavior consistent with it. If we focused on this latter question, perhaps would be examining a much broader range of attributions in an attempt to find those most closely associated with persuasion. Rather than measuring only attributions to the circumstances or the product, or only to the circumstance, the person, or the object, we might search for of categories of attributions that are important for the communications context. Rather than looking for the most important cause, we might look instead for patterns of causes. Rather than assuming that one category of attribution is always best, we might develop theories and hypotheses about when, and under what conditions, certain attributional patterns will be associated with greater persuasion. For example, we might investigate when an attribution to the idiosyncratic tastes and values of the spokesperson results i a stronger intention to purchase the product that an attribution to the "objective" properties of the product.

Dillon et al. do not take on these questions directly either. But, these authors develop and present a methodology that might find useful in subsequent investigations of them. this report, the authors examine an underlying premise of applications of attribution theory to advertising message Rather than asking for particular attributions, they collected what might be considered to be the raw materials from which attribution-type inferences are derived, i.e., the thoughts elicited by a message. These thoughts were then analyzed to determine whether the patterns of though were consistent with the kinds of causal schema proposed others and assumed to underlie attributional thinking. I patterns of thoughts were then related to attitudes and purchase intentions.

Certainly there are methodological questions one might ra about the procedures followed by Dillon et al. Most of these are noted by the authors themselves in their discussion section, and so they need not be elaborated upon further here. The contribution of interest to me is the ability to represent thought patterns which can be described and analyzed for their association with persuasion. The elicitation technique which must be used as a first step this procedure may force us to consider possible causal features that we might not have thought of on our own, ar thus may guard against an incomplete list of potential attributions. The emphasis on thought patterns may indicate when message recipients operate under multiple sufficient or multiple necessary, or other causal assumptions. Fin. with the new richness of data, we may become more interested in which attributions are related to persuasion and thus warrant further investigation.

SUMMARY

Attribution theory has much to contribute to our understanding of the communications and persuasion process. But, our research to date has focused on a very narrow aspect of it, i.e., testing various operationalizations of factors postulated to affect certain attributions of a message source's behavior. I have suggested in these comments that we must be careful in thinking through the number and type of potential attributions and that we begin to focus more of our attention on the consequences of different attributions. This latter effort may require us to move beyond the current confines of attributional theory, but this seems a necessary step if our goal is to gain insight into advertising effectiveness.

REFERENCES

Bem, Daryl J. (1972), "Self-Perception Theory," in L. Berkowitz ted.), Advances in Experimental Social Psychology, Vol. 6, New York: Academic Press, pp. 1-62.

Calder, Bobby J. (1977), "Endogenous-Exogenous Versus Internal-External Attribution Explanations," Personality and Social Psychology Bulletin, 3, pp. 400-406.

Eagley, Alice R., and Chaiken, Shelly (1975), "An Attribution An lysis of the Effect of Communicator Characteristics on Opinion Change: The Case of Communicator Attractiveness," Journal of Personality and Social Psychology, 32, pp. 136-144.

Eagley, Alice R., Wood, Wendy, and Chaiken, Shelly (1978), "Causal Inferences About Communicators and Their Effect on Opinion Change," Journal of Personality and Social Psychology, 36, pp. 424-435.

Hansen, Robert A., and Scott, Carol A. (1976), "Comments on Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 13, pp. 193-197.

Jones, E. E., and Davis, K. A. (1965), "From Acts to Dispositions," in L. Berkowitz (ed.), Advances in Experimental Social Psychology, Vol. 2, New York: Academic Press, pp. 219-266.

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

Kelley, H. H. (1972), "Attribution in Social Interaction," in E. E. Jones et al., (eds.), Attribution: Perceiving the Causes of Behavior, Morristown, NJ: General Learning Press, pp. 1-26.

Kruglanski, A. W. (1975), "The Endogenous-Exogenous Partition in Attribution Theory," Psychological Review, 82, pp. 387-406.

McArthur, L. A. (1972), "The How and Whet of Why: Some Determinants and Consequences of Causal Attribution," Journal of Personality and Social Psychology, 22, pp. 171-193.

Settle, R. B. (1972), "Attribution Theory and the Acceptance of Information," Journal of Marketing Research, 9, pp. 85-88.

Settle, R. B. and Golden, Linda L. (1974), "Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 11, pp. 181-185.

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