A Test of the Relationship Between Trait and Causal Attribution

Citation:

Richard W. Mizerski (1975) ,"A Test of the Relationship Between Trait and Causal Attribution", in NA - Advances in Consumer Research Volume 02, eds. Mary Jane Schlinger, Ann Abor, MI : Association for Consumer Research, Pages: 471-480.

Advances in Consumer Research Volume 2, 1975      Pages 471-480

A TEST OF THE RELATIONSHIP BETWEEN TRAIT AND CAUSAL ATTRIBUTION

Richard W. Mizerski, Arizona State University

[Richard W. Mizerski, Assistant Professor of Advertising and Marketing, Arizona State University.]

Attribution principles, developed to predict trait attribution in person perception, were applied to the causal attribution of information about products in order to test an attribution theory explanation for unfavorable information's disproportionate influence in consumer decision-making. Subjects received either favorable or unfavorable ratings about one of two products. It was hypothesized that subjects who received unfavorable information would make more extreme internal attributions (perceive the information as "caused" by the product only), as well as have more confidence in their attributions. The results show that the predicted attributions appeared for one of the two products, with no significant differences in attributional confidence. Probable causes and implications of the findings are discussed.

In this era of consumer protection, marketers are being faced with an unparalleled barrage of unfavorable information about their products or services. Consumer-oriented publications (e.g., Consumer Reports), counter advertising by organizations such as Stern Concern, comparative advertising by competitors, corrective advertising, and widely publicized tests of products by government agencies (e.g., E. P. A. gas consumption tests of new automobiles), are just a few of the sources that have augmented word-of-mouth advertising as potent media for unfavorable product information. While the chances of a business encountering potentially ruinous and possibly untrue information are increasing, the marketing literature has just begun to seriously analyze its impact.

The limited marketing literature that deals with unfavorable information suggests that it wields an inordinate influence in an individual's decision-making process. Arndt (1968) reported that only 18% of the subjects in his study purchased a test product after they received unfavorable word-of-mouth communication; yet 54% of those who received favorable word-of-mouth communication purchased the same item. Reynolds and Darden (1972) elicited responses from women who rejected a fashion innovation (the midi clothing style), and found that 93% perceived their discussions about the fashion as unfavorable. Although neither study controlled for the amount of each type of information, they did provide empirical data on unfavorable information's differential impact.

Social psychologists have provided more rigorous analyses of the perceptions of favorable and unfavorable information in their studies of impression formation. This research (e.g., Anderson, 1965; Feldman, 1966) typically has subjects evaluate a stimulus person based upon manipulated sets of equally polarized favorable and unfavorable adjectives. Their findings substantiate those of Arndt in marketing, and show the disproportionate influence of unfavorable information (adjectives) in forming overall impressions.

An understanding of the differential strength of favorable (usually the firm's communications) and unfavorable information could help marketers gauge the total effect of communications about their products, and ultimately improve the promotion strategy of their companies. Unfortunately, attempts to explain this perceived differential in strength have not been successful. A number of reasons such as unfavorable information's greater surprisingness (Feldman, 1966), or its less frequent use (Zajonc, 1968), have been proposed. However, these reasons have been almost totally eliminated by Kanouse and Hanson (1971).

Another possible explanation is based upon the concept that individuals may differ in their processing of favorable and unfavorable information, and that the difference centers on the source or "cause" to which the information is attributed. This area of information processing is referred to as attribution theory, and involves "an examination of the process whereby people attribute characteristics, intentions, feelings, and traits to objects or events in the real world" (Kanouse and Hanson, 1971). This is accomplished by an analysis of phenomenal causality; i.e., noting the determinants and consequences of causes for particular actions or events (Jones and Davis, 1965). In terms of processing information, attribution theory suggests that an individual accepts and is influenced by information from another person only if he attributes the report to the entity being described ("internal" cause). Therefore, an accurate report about an entity is one that is perceived to be "caused" by that entity. In many cases, however, an individual may believe that the information was "caused" by another factor or factors such as the possible social pressure to flatter an individual (one possible "external" cause).

One principle of attribution theory is that "the role of a given cause in producing a given effect is discounted if other plausible causes are also present" (Kelly, 1973, p. 113). This concept, often re,erred to as the "discounting principle," has been demonstrated in experiments of compliant behavior (Thibaut and Riecken, 1955), and in the perceived sincerity of complimentary ratings (Jones, Gergen, and Jones, 1963). Both studies suggested that their subjects perceived more external causes for favorable behavior or information (causes other than the entity on which information was being provided), and thus the possibility of the information having an internal (originating in the entity) cause should be discounted. This discounting supposedly leads to a stronger inference of external causation. These results led Kelly to propose that "complimentary remarks made about a person under the guise of accurately reporting one's feelings about him . . . may be attributed to some other causes such as the ambient social situation" (1971, p. 18). By applying the discounting principle, favorable information should have a greater tendency (than unfavorable information) to be attributed to external causes.

The principle of "social desirability of effects" (Jones and Davis, 1965) provides a possible explanation for the attribution of unfavorable information. Viewed as a modification of the discounting principle (Kelly, 1973), social desirability of effects has two assumptions:

1. In general, people intend desirable effects from their actions.

2. Socially desirable effects provide relatively little information about causes. When a person adheres to social norms (e.g., intending desirable effects), one is unsure whether the person's behavior was caused by internal (true intentions, beliefs) or external (society, peer pressure) causes. Therefore, socially undesirable behavior tends to rule out external causes, and leaves internal causes as the reason(s) for the effect (Hastorf. et al. 1970).

Kanouse and Hanson (1971) have provided an interpretation and extension of this rationale. They suggest that because of social norms of discussions about people or objects, most information is favorable in an individual's social environment. Therefore, unfavorable information about traits or behavior would stand out and have a tendency to be attributed to the person or object being evaluated (internal cause). An example of this attribution behavior would be the following:

If an individual believes that most people are sincere, he is likely to be relatively unimpressed with the information that person x is sincere. Given that sincerity is the norm, credit attaches to social pressure or to the simple fact of being human rather than to person x. Insincerity, however, is another matter. By standing in contrast to the norm, the insincere individual invites attributions of responsibility for this trait.(p.57)

Both of the attribution principles, social desirability of effects and the discounting principle, have evolved from studies dealing with the attribution of traits in person perception. Their application depends upon perceived social norms of discussion about individuals. It would seem that the perception of products (based upon information about their attributes) involves an analogous process, one in which the principles of attribution should apply if there are similar perceived norms of discussions about products. A possible problem in generalizing these concepts from a person to a product domain is that the attribution of a trait has always been inferred rather than measured directly. This raises a question as to whether there would be significant differences in attribution if the subjects were asked to actually specify the cause for the information. In order to test the appropriateness of generalizing the concepts, the following hypothesis was developed:

Subjects will tend to attribute favorable information about a product to "external" causes, and unfavorable information about a product to "internal" causes.

If the differences in attribution do exist for processing favorable and unfavorable product information, this differential should affect information influence. An individual would seemingly give more weight to information he felt reflected a product's characteristics (the suggested attribution of unfavorable information). than to information that tends to be attributed to external influences such as social norms (the suggested attribution of favorable information).

Another facet of attribution theory may augment a theoretical explanation of unfavorable information's differential influence. Jones and Davis (1965) proposed that more confident and extreme attributions follow from observed effects that deviate from the norm (the suggested perception of unfavorable information). They argue that confidence and extremity of trait attribution were "equivalent and inseparable" measures. In other words, one can make highly confident trait attributions only when the stimulus person does something distinctive, thus also justifying an extreme trait attribution. If this is the case, one would logically expect that individuals would tend to impute a greater weight to information in which they had more confidence. The following hypothesis will test the role of confidence in the causal attribution of information about products:

Subjects will have more confidence in causal attributions about unfavorable information than with causal attributions about favorable information.

EXPERIMENT 1

METHODOLOGY AND RESULTS

In order to determine if there seem to be social norms about product discussions, and thus some consensus about the type of information (favorable or unfavorable) expected, 97 subjects [The subjects were similar to those used in the main (#2) experiment.] were asked whether they would expect "favorable" or "unfavorable" information from: (1) a student who had just tested a new automobile, and (2) a student who had just previewed a new movie. [These are the same products used in the main experiment.] The responses to these questions (Table 1) show that a majority of subjects did expect favorable information about each product (78% for the automobile, 59% for the movie). Since similar tests in trait attribution have not been conducted, there are no previous measures from which to determine whether either product would have the minimum expectancy necessary in order to apply the proposed attribution principles. However, a simple majority does suggest a feasible application.

TABLE 1

EXPECTED INFORMATION TYPE

EXPERIMENT 2

METHODOLOGY

The main experiment was performed on 9 classes (300 subjects) [For the analysis, 30 subjects were deleted for missing data and for improper use of a scale used in research external to this paper. The latter deletion was necessary because improper scale use affected the data collected for this study.] of upper division undergraduate and graduate students at the University of Florida College of Business Administration. Each student was told that the study involved testing how people evaluate, and are affected by, the opinions of others. They then received evaluative information on three salient attributes (chosen by the Fishbein and Ravens 1962, method) of a product to which they were randomly assigned. The automobile attributes were maintenance costs, comfort and gas mileage. Acting, plot and the photography were the three movie attributes. The information was presented in the form of personal ratings, supposedly reprinted from another student who was randomly chosen to test and evaluate the product. Two of the three attributes were given a neutral rating, with only one attribute rated either favorably or unfavorably. The modifiers used for rating were scaled for equal polarity and affective opposite meaning by replicating a study by Myers and Warner (1968). Following the information treatment, subjects were asked to respond to an attribution question similar to the following:

To what extent do you feel that other reasons--reasons having nothing to do with the automobile tested--influenced the student's opinion about . . .

         Other reasons had no effect on the opinion      Other reasons were the only cause for the opinion

                     Maintenance costs:       1   2    3     4    5    6    7

The smaller the scale value chosen, the more internal ("caused" by the product only) the attribution. After the attribution question, the respondents were asked to rate their confidence in each attribution. A seven point scale ranging from "no confidence" (#1) to "complete confidence" (#7), recorded their confidence scores.

RESULTS

The first hypothesis was supported for the movie only. Mean attribution scores for the two products are presented in Table 2. While more extreme internal attributions (smaller mean scores) were hypothesized to occur in response to unfavorable information for both products, only the movie scores are in the predicted direction. In addition, unfavorable information produced more extreme internal attributions across products. An unweighted means analysis of variance (ANOVA) was used in order to detect significant differences between treatments (Table 3), and revealed a main effect of attribute (F = 7.46, p < .001), and an interaction of product and information (F = 4.33, p < .038). The main effect of attribute shows that the subjects perceived a difference between the attributes in terms of each being the cause for their individual bogus ratings. However, these differences were independent of the information type received and do not constitute a test of the hypothesis. In order to detect the source of the information x product interaction, a Newman-Keuls Multiple Range Test was performed (Table 4). The results of this test show that only the movie's attribution scores were significantly different in the two information conditions, with unfavorable information producing the predicted stronger internal attributions.

TABLE 2

MEAN ATTRIBUTION SCORES

TABLE 3

ANALYSIS OF VARIANCE: MEAN ATTRIBUTION SCORES

TABLE 4

RESULTS OF A NEWMAN-KEULS MULTIPLE RANGE TEST ON MEAN ATTRIBUTION SCORES

The second hypothesis was not supported for either product. In fact, the mean confidence scores (Table 5) show favorable information producing greater confidence, a result opposite that predicted. However, an unequal means ANOVA failed to find any significant differences between the group means, which suggests that the differences are due to chance.

TABLE 5

MEAN CONFIDENCE SCORES

While the predicted significant differences in confidence failed to appear, it is still of some interest to determine whether the concept of "correspondence" is generalizable to the causal attribution of information about products. A series of polynomial regression analyses were applied to the data in order to measure the degree of association between the extremity of causal attribution, and the confidence in that attribution. A high degree of association, particularly for the movie, would lend support to the Jones and Davis position of "inseparability" (i.e., if the same principles are the basis for each phenomenon, their reactions to stimulus should be highly related). Results of the regressions (Table 6) reveals that three of the four information x product conditions show significant relationships. However, the number and mix of nonlinear components, as well as the inconsistency of their appearance across conditions, reveal that the relationship is very complex. On balance, it does appear that causal attribution and confidence in the attribution are related, but the idea that they are "equivalent and inseparable" (as suggested in trait attribution), is not supported by the data.

DISCUSSION

It is interesting to note that the movie, not the automobile, conformed to the predictions of attribution theory. Since more subjects expected favorable information about the automobile than about the movie (Table 1), the automobile would seem to be the more obvious product for application of the attribution principles. Perhaps the different degree of objectivity of judgment for the two product's attributes accounts for the hypothesis being supported for the movie only. The movie's attributes called for more subjective evaluation than two of the three automobile attributes. It may be that the more objective the criteria, the less the perceived difference in causal attribution. Subjects who were interviewed after the experiment seemed to think that attributes such as maintenance costs and gas mileage involved rather straight-forward ratings based upon "hard" empirical data, which would leave little room for "other" (external) causes to effect the evaluation. On the other hand, evaluating attributes such as acting skill, the plot, and the quality of the photography depend on more subjective criteria (external causes) such as past experience and learned tastes. This suggests that the principles developed in trait attribution may be particularly applicable in situations where subjective evaluations are important [Perhaps the success in applying these principles in trait attribution stems from those studies' test subjects' perception of the generally subjective nature of evaluating others.] such as with many products using social, psychological, or any emotional appeal (e.g., see Settle and Mizerski, 1973). Future research should include some measure of the subjects' perception Os the ability to objectively evaluate a product's attributes so that this possibility could be tested.

TABLE 6

POLYNOMIAL REGRESSION: ATTRIBUTION ON CONFIDENCE IN THE ATTRIBUTION

The failure of negative information to evoke more confident attributions may have been a function of the very limited stimulus information provided the subjects. Many respondents noted that their confidence in the attribution would have been substantially altered if they had been given more information about the bogus rater. Of course, the object of the study was to eliminate those factors and test only for the effect of attribute and information. However, future studies could manipulate information about the source (e.g., product experience, or demographic profile), or even the source of the message (government, independent testing organization, consumer group, etc.) In order to give subjects a more realistic information matrix. These manipulations would also be interesting in terms of the resulting causal attributions.

Finally, the tested application of two principles of trait attribution was partially successful, and does suggest at least part of an explanation for the disproportionate influence of unfavorable information. Future research could tie these and other principles of trait attribution to established models of decision-making (e.g., Mizerski, 1974) in order to provide a more rigorous analysis of the concept of causal attribution in consumer purchase decisions.

REFERENCES

Anderson, Norman H. Averaging versus adding as a stimulus--combination rule in impression formation. Journal of Personality and Social Psychology, 1965. 2. 1-9.

Arndt, Johan. Perceived risk, sociometric integration, and word-of-mouth in the adaption of a new food product. In D. F. Cox (Ed), Risk taking and information handling in consumer behavior. Boston: Division of Research, Graduate School of Business Administration. Harvard University. 1968.

Feldman, Sheldon. Motivational aspects of attitudinal elements and their place in cognitive interaction. In Sheldon Feldman (Ed), Cognitive consistency. New York: Academic Press 1966.

Fishbein, M. and Raven, B. The AB scales: An operational definition of belief and attitude. In M. Fishbein (Ed), Readings in attitude theorY and measurement. New York: John Wiley and Sons, Inc., 1962.

Hastorf, A.H., Schneider, D.J. and Polefka, J. Person perception. Reading, Mass.: Addison Wesley, 1970.

Jones, E.E., Gergen, K.J. and Jones, R.G. Tactics of integration among leaders and subordinates in a status hierarchy. Psychological Monographs, 1963, 77.

Jones, E.E., and Davis, K.E, From acts to dispositions: the attribution process in person perception. In L. Berkowitz (ed.),Advances in experimental social psychology. Vol. 2. New York: Academic Press, 1965.

Kanouse, D.E., and Hanson, L.R. Jr. Negativity in evaluations. In Edward E. Jones,et. al. (Eds.), Attribution: Perceiving the causes of behavior. Morristown, N.J.: General Learning Press, 1971.

Kelly, H.H. Attribution in social interaction. In Edward E. Jones,et. al. (Eds.),Attribution: Perceiving the causes of behavior. Morristown, N.J.: General Learning Press, 1971.

Kelly, H.H. The processes of causal attribution. American Psychologist, 1973, 28, 107-128.

Mizerski, R.W. Attribution theory and consumer processing of unfavorable information about products. Unpublished doctoral dissertation, University of Florida, 1974.

Myers, J.H., and Warner, G. Semantic properties of selected evaluation adjectives. Journal of Marketing Research, 1968, 5, 409-412.

Reynolds, F.D., and Darden, W.R. Why the midi failed. Journal of Advertising Research, 1972, 12, 39-46.

Settle, R.B., and Mizerski, R.W. Differential response to objective and social information in advertisements. In ConcePtual and methodological foundations of marketing. (Proceedings of the 1973 Fall A.M.A. conference.) Chicago: American Marketing Association, 1974.

Thibaut, J.W., and Riecken, H.W. Some determinants and consequences of the social causality. Journal of Personality, 1955, 24, 113-133.

Zajonc, R.B. Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 1968, 7, 1-29.

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