Causal Attributions and Persuasion: the Case of Disconfirmed Expectancies

ABSTRACT - The persuasiveness of an apocryphal advertiser was tested by comparing the post-message attributions of subjects with their pre-message expectations in an experimental setting. Mixed results suggest that disconfirmed expectations lead to enhanced message acceptance and "entity" attributions E when pre-message expectancies anticipate reporting bias. Modality manipulations further suggest the superior efficacy of a hidden camera over a "typical purchaser" in enhancing the credibility of both the corporate sponsor of the advertising and its-spokesperson.

Citation:

James M. Hunt, Teresa J. Domzal, and Jerome B. Kernan (1982) ,"Causal Attributions and Persuasion: the Case of Disconfirmed Expectancies", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 287-292.

Advances in Consumer Research Volume 9, 1982      Pages 287-292

CAUSAL ATTRIBUTIONS AND PERSUASION: THE CASE OF DISCONFIRMED EXPECTANCIES

James M. Hunt, University of Florida

Teresa J. Domzal, George Mason University

Jerome B. Kernan, University of Cincinnati

ABSTRACT -

The persuasiveness of an apocryphal advertiser was tested by comparing the post-message attributions of subjects with their pre-message expectations in an experimental setting. Mixed results suggest that disconfirmed expectations lead to enhanced message acceptance and "entity" attributions E when pre-message expectancies anticipate reporting bias. Modality manipulations further suggest the superior efficacy of a hidden camera over a "typical purchaser" in enhancing the credibility of both the corporate sponsor of the advertising and its-spokesperson.

INTRODUCTION

Communicators in general and advertisers in particular have long been interested in how individuals assess the veridicality of persuasive messages. Attribution theory (Bem 1972; Heider 1958; Jones and Davis 1965; Kelley 1967, 1972a 1972b, 1973; Kelley and Michela 1980), which suggests that the inferences people make regarding the causal nature of a message influence their judgments about the accuracy of that message, is a useful perspective for undertaking the analysis of persuasion.

This study is an attempt to bring attribution theory to bear on promotion -- "the efforts of sellers to persuade prospective buyers to accept the seller's information [as a veridical interpretation of the environment] and store it in retrievable form" (Kernan and Sommers 1967). It focuses on the attribution/persuasion process, paying particular attention to the related concept of source credibility.

BACKGROUND

Attribution theory has been characterized as the single most pervasive influence on social psychology during the 1970s (Cialdini, Petty and Cacioppo 1981) and its presence in consumer research is no longer a novelty (Burnkrant 197 Mizerski, Golden and Kernan 1970; Scott 1978). It is not surprising that the majority of consumer-related attributional studies deal with promotion-like phenomena (e.g. Golden 1977; Hansen and Scott 1976; Settle 1972; Settle and Golden 1974; Smith and Hunt 1978; Sparkman and Locander 1980) since, at base, attribution theory is information processing theory.

A direct link between attributional analysis-and the persuasion process has been proposed by Eagley, Wood and Chaiken (1978), who argue that message recipients should be viewed as problem solvers who are attempting to maximize the validity of their judgments. According to the model. < key factor in the persuasion process is the message recipient's presumption about the likely position a communicator will espouse. This expectancy is based on pre-message cues regarding the communicator's traits and the extant situational pressures. If the expectancy is subsequently confirmed by the message (if the source "says what the receiver expects"), the validity of the message arguments is discounted, the credibility of the source is questioned, and little persuasion results. When pre-message expectancies are disconfirmed, however, the source is seen somehow to be credible to be telling the "real truth" about the object in question.

The Eagley, Wood and Chaiken model does not predict results radically different from the more traditional paradigms (e. g. Festinger and Maccoby 1964; Hovland and Mandel 1952; Jones et al. 1971; Koeske and Crano 1968; McPeek and Edwards 1975; Mills and Jellison 1967; Steiner and Field 1960; Thibaut and Rieken 1955; Walster, Aronson and Abrahams 1966; Walster and Festinger 1962) and it focuses on what Cialdini and Petty (1981) call an "elastic" opinion shift--one that responds almost wholly to situational demands. Its value to consumer analysis rests largely in how it hypothesizes persuasive effects. Two features are noteworthy. First, explicit recognition is given to the effect on a recipient simply expecting to have to deal with a persuasive communication. Second, this anticipatory process is explicated through attributional analysis.

To assert that there are anticipatory message effects is to recognize the existence of pre-message cognitive processing. Attributional analysis simply explicates this, commonly by opining the invocation of causal schemata (Kelley 1973) in this case idiosyncratic summaries of the likely causes for a source to take one stance or another, relative to some issue. In the case of a persuasive (promotional) message, a recipient likely anticipates a presenter (a spokesperson, a corporation) to reflect bias, since the seller obviously has something to gain by message acceptance. For the most part, this is "reporting bias" (the difference between what the source knows about the entity being described and the way he actually describes it) as opposed to "knowledge bias" (the difference between the truth about an entity and the source's understanding of that entity). Or, as-Eagley, Wood and Chaiken (1978) suggest, one is far more likely to question the trustworthiness of a commercial message than the expertise of its source.

Succinctly stated, the Eagley, Wood and Chaiken model argues that communicators can increase their credibility--and thus persuasive efficacy--by transmitting messages that disconfirm recipients' pre-message expectancies, when those expectancies manifest reporting bias. The present study attempts an operational test of that model by explicitly measuring the pre-message schemata and expectancies that attend realistic, consumer-related advertisements. The question-is whether advertisements that disconfirm premessage expectancies that don't say what one would expect an advertisement to say about a product--engender "entity" attributions (those which infer the cause of the message to be the products per se, rather than some situational, reporting bias).

METHOD

Subjects

A total of 150 students (female and male) was selected from the student body at the University of Cincinnati to participate in an "advertising evaluation" task. These subjects were assigned randomly to one of five groups: an "expectancy" group or one of four different "persuasion" (treatment) groups.

Procedure

Subjects were informed that they were participating in a study designed to test advertisements for several new products that were soon to be test marketed in various cities throughout the country. They were then told that their task would entail: (1) reading background information about several manufacturer/advertisers; (2) reading background material about several new advertisements; (3) viewing these advertisements; and (4) responding to questions to assess their reactions.

After receiving these instructions, each subject received a booklet containing experimental materials. For the first phase of the experiment, the booklets presented background material concerning the "Telco Electronics Corporation," a firm that purportedly had recently developed a new clock radio. It was stated that this corporation was now in the process of testing various advertising formats for television audiences, and that subjects would be viewing the content of one of these advertisements in storyboard format.

Next, subjects read "background facts" about one of "several" possible advertisements. The format was described as one in which a customer (spokesperson) endorses the Telco product. Following this, the background material described how this spokesperson was to be introduced to the television audience. Finally, the background section ended with the spokesperson about to "report his experience with the product.

At this point, subjects viewed a sample mock advertisement having several picture frames accompanied by a script. After that, they completed a questionnaire designed to assess the dependent measures of the study.

Expectancy subjects were treated identically, except they did not actually view the mock advertisement. Instead, after reading the background materials, they responded to a set of items pertaining to their expectancies about what would be said in the advertisement and their pre-message schemata.

Promotional Message

The sample advertisement viewed by each subject consisted of six frames. Each of these frames depicted the spokesperson--a man in his mid-twenties - making various statements (directed toward the viewers of the message) about the Telco clock radio as he was standing at a store counter.

The statements consisted of information related to four product ascertained through two pilot studies and two statements dealing with the spokesperson's overall evaluation of the product. These latter statements (appearing in the first and last frames) described the product as an overall superior clock radio.

Independent Variables

Manipulation of the expectancy confirmed/disconfirmed factor was carried out by varying the information subjects were exposed to in the commercials. Subjects either received an advertisement that made superior claims (expectancy confirmed) on all product features--(1) sound of the radio, (2) accuracy of the clock-radio, (3) clock and alarm reliability, and (4) styling--or they received a message that made such claims on only three of these attributes (expectancy disconfirmed). On the fourth attribute, sound, the information given to subjects in the disconfirmed group was that the Telco product "did not have the best sound.

A second factor was used to elaborate the expectancy confirmed/disconfirmed factor. This factor involved the type of spokesperson used to convey the commercial message. Half the subjects were exposed to a message from a spokesperson described as a typical customer who was in the process of buying a Telco clock radio. The other half viewed a commercial conveyed by a customer who had been filmed by a "hidden camera." This manipulation was carried out through the information presented in the background section dealing with the advertising format.

Variation of this second factor was thought to be a further manipulation of reporting bias. Essentially, the hidden camera condition involves a speaker in a modality different from that associated with "typical" advertisements, where the spokesperson usually appears as though s/he is following a script. As such, -it was anticipated that the hidden camera condition would lead to more entity attributions and thus have a greater persuasive effect.

Given these manipulations, a 2 (expectancy confirmed vs. disconfirmed) x 2 ("typical purchaser" spokesperson vs. "hidden camera" spokesperson) between-subjects design resulted. Expectancy subjects responded to background material representing the "typical purchaser" cell.

Dependent Measures

Expectancies. Expectancy subjects were asked to indicate their "agreement" or "disagreement" to a series of items dealing with their judgments as to what the "typical purchaser" would say in the advertisement they were about to read. Each of these items -- rated on a 15-point bipolar scale ranging from "strongly agree" to "strongly disagree" -- corresponded to specific product features.

Causal Schemata. Subjects were asked to judge the importance of three factors in influencing the seller's claims as conveyed in the ad: (1) "the real facts about the clock radio" (entity-related factor); (2) "the firm's true feelings about the clock radio" (internal/seller-related factor); and (3) "the firm's desire to sell products" situational/bias-related factor). Two different scales were used to assess this item. Subjects were asked first to respond to the importance of each factor on a 15-point bipolar scale ranging from "extremely important" to "extremely unimportant."- In addition, of the three factors mentioned, subjects selected the factor that was "the most important in influencing the Telco Corporation to produce and televise such a commercial."

In addition to the items above, subjects were asked to respond to a similar set of items involving the actual speaker. Four causal factors (again measured on a 15-point bipolar scale) were used: (1) the pay the speaker received for appearing in the ad; (2) the speaker's true feelings; (3) the real facts about the clock radio; and (4) the speaker's opportunity to promote himself on TV. The items used for both this assessment of causal schemata and the one above were ascertained from two previous pilot studies.

Pre-message Schemata. Expectancy subjects responded to the same set of items as did the subjects in the persuasion group. However, the items that expectancy subjects responded to were prefaced by the following statements. "Although you do not know for certain what this (the typical purchaser) person will say in the ad, assume that he actually says what YOU personally expect him to say. In that case, how important do you think each of the following factors would be in influencing him to make such a statement?"

Message Acceptance. Prior to receiving the questions pertaining to causal schemata, subjects indicated their agreement to a series of 15-point scales, anchored by "strongly agree" vs. "strongly disagree." As was the case with expectancy subjects, each belief statement corresponded to one of the various product characteristics discussed above. Additionally, subjects rated the "overall superiority" of the Telco clock radio. This, also, was done on a 15-point scale.

Source Credibility. Source credibility was measured at two levels: (1) the level of the individual spokesperson; and (2) the level of the corporation. Regarding the spokesperson, subjects were asked their judgment as to how "honest" and how "sincere" he was. Both items were scored on a 15-point (strongly agree vs. strongly disagree) type scale. Credibility of the company was assessed on a 7-point bipolar scale. This measure entailed responses to an item that had to do with how "honest" the corporation was as an advertiser.

Other Measures. In attempting to assess reporting bias, subjects judged two items. The first asked subjects to indicate their agreement (on a 15-point scale) with the statement that the spokesperson "tried to report his true private opinion about the clock radio." The second item involved whether subjects thought the spokesperson would state that the clock-radio was a "superior" clock-radio during "the course of a Private conversation."

RESULTS

[Space limitations prohibit a complete display of findings. For additional elaboration, contact James Hunt, Department of Marketing, University of Florida, Gainesville 32611.]

Data analysis was conducted by using a 2 (expectancy confirmed vs. disconfirmed) x 2 (typical purchaser spokesperson vs. hidden camera) analysis of variance. In addition, chi-square analysis was used to assess the dichotomous measures of schemata.

Design Requirements

One requirement of the design was that expectancy subjects exhibit bias-related expectancies - i.e., they expected claims about the product to be relatively superior ones. To establish this, a comparison was made between (1) subjects' expectations that in the ad the spokesperson would describe the product as being a "superior clock radio" and (2) their expectations as to what he would say about the product in a private conversation. A paired t-test showed these to be different --t(28)=6.56, p<.001 in the hypothesized direction. In addition, the average value of attribute-related expectancies fell into higher ranges of the 15-point scale: 11.73 for sound quality; 12.79 for product accuracy; 12.73 for reliability; and 11.04 for styling.

Message Acceptance

In three of the four performance attributes, persuasion subjects accepted the promotional message more when they received a message that disconfirmed expectations. This is evident from Table 1. Belief scores on the first attribute (sound quality), which was used to manipulate expectancy confirmation/ disconfirmation, were significantly lower -- as hypothesized -- in the disconfirmed group:F (1,110) = 117.39, p <.0001. In the case of the "accuracy" dimension, belief scores for the subjects in the disconfirmed group were significantly greater than those of the confirmed group:F (1,110) = 4.63, p<.03. Similar results were obtained for the reliability feature F(1,110) = 5.17, p<.02. No difference was found between the confirmed and disconfirmed group in terms of "styling," the fourth product attribute. Finally, none of the expectancy confirmation x spokesperson type interaction terms approached significance, nor did the main effects of spokesperson type.

Causal Schemata

Subjects' judgments regarding why the corporation would sponsor the advertisement differed as a function of expectancy confirmation. The group receiving a message that disconfirmed their expectations made more attributions to the entity-related factor ("real facts") than did the group that received a message confirming their expectations. This is exhibited in Table 2. Further, the number of subjects who attributed the cause of the ad to the selling motive (situational bias) was less in the disconfirmed group than in the confirmed group. These differences were assessed through a test of independence using a (Yates corrected) chi-square statistic (X2 = 2.824, df=2, p <.09). Similar results were found over the spokesperson type factor. Subjects exposed to the "hidden camera" spokesperson made more entity-related and fewer situational attributions than did the group that received the message from a "typical purchaser" spokesperson (corrected x2 = 2.633, df=1, p<.10).

TABLE 1

PERSUASION SUBJECTS' MEAN BELIEF SCORES BY EXPECTANCY CONFIRMATION AND TYPE OF SPOKESPERSON

TABLE 2

FREQUENCY OF SUBJECTS' RATINGS OF FACTORS INFLUENCING THE CORPORATION

TABLE 3

FREQUENCY OF SUBJECTS' RATINGS OF FACTORS INFLUENCING THE SPOKESPERSON

No expectancy confirmation differences were found to exist with respect to the reasons for the spokesperson's statement. Spokesperson type, however, did have an impact on subjects' inferences about why the spokesperson said what he did in the ad. This can be seen in Table 3. The group that was exposed to the "hidden camera" speaker made fewer situational attributions (pay and self promotion) and more attributions to the spokesperson's "true feelings" about the product (X2 = 11.955, df = 3, p <.01).

Analysis of schemata in terms of importance scores for each causal factor was done through a 2 (expectancy confirmed vs expectancy disconfirmed) x 2 ("typical purchaser" spokesperson vs. "bidden camera" spokesperson) analysis of variance. With respect to the expectancy confirmation factor, no significant main effects were found. There were, however, several hypothesized main effects of the spokesperson type factor. Subjects who were exposed to the hidden camera spokesperson judged the speaker's statement to be caused: less by "pay," F(1,110) = 9.55, p <.01; less by "the opportunity to promote himself" (p <.001); and more by his "true feelings about the clock radio" (p <.01). In addition, these same subjects rated the "desire to sell products" (situational factor) to be less important in influencing the corporation than did those exposed to the "typical purchaser" (p <.06).

Pre-message Schemata

Comparing the expectancy group to both the expectancy confirmed and disconfirmed groups, the findings are at odds with the hypothesized results. As can be seen from Table 2, the expectancy group differs considerably from the expectancy confirmed group in their explanations of why the corporation would/did produce the ad (X2 = 14.625, d.f. = 2, p <.001). Further, the expectancy group differed only marginally from expectancy disconfirmed group (p <.14). That difference, however, was opposite to the hypothesized difference - i.e., the expectancy disconfirmed group made proportionately more attributions to the situational factor (selling motivation).

Regarding the spokesperson's statement, as would be hypothesized, the expectancy group did not differ from the expectancy confirmed group in terms of causal explanations. However, contrary to our anticipation, the expectancy group did not differ from the expectancy disconfirmed group, either.

Reporting Bias

Mean scores for the two measures of reporting bias appear in Table 4. Analysis of variance indicated that the only factor having a significant main effect on either of these dependent measures was that of spokesperson type. As discussed above, the first measure dealt with subjects' estimates of what the speaker would say about the clock radio (in terms of overall superiority) during the course of a private conversation. The results showed that subjects who received the ad from the "typical purchaser" judged this statement to be less "superior" than did those who received this ad from the "hidden camera" speaker: F(1,110)3.14, p <.10. Accordingly, the bias scores for subjects in the "typical purchaser" group were significantly higher than those in the "hidden camera" condition.

Similar findings resulted when subjects were asked to judge the spokesperson's freedom "to express his true opinion in the ad. Subjects who were exposed to the "hidden camera" spokesperson exhibited higher ratings on this item than did those who received the ad from the "typical purchaser:" F(1,110) = 6.03, p <.05. Thus, the mean bias scores of those in the "hidden camera" condition were significantly lower.

Source Credibility

A 2 x 2 analysis of variance was performed to assess the effects of both treatment factors on subjects' judgments of source credibility. In terms of subjects' ratings of the spokesperson's "honesty" and "sincerity," the expectancy confirmation factor was found to have no significant main effect (see Table 4). The means of these two measures were, however, in the hypothesized order. The second factor, spokesperson type, did have a significant main effect on ratings of both honesty and sincerity. Subjects in the hidden-camera group judged the speaker to be more honest: F(1,110) = 5.39, p <.05; and more sincere:F (1,110) = 3.50, p <.10; than did those who received the same message from the typical purchaser. Neither interaction was significant in these analyses.

Similar results were found in terms of the corporation's credibility. The expectancy confirmation factor had no significant main effect on subjects' ratings of how "honest" Telco was in its advertising. There was, however, a marginally significant main effect of spokesperson type. Subjects exposed to an ad from a spokesperson on hidden camera rated Telco to be more "honest" than did those in the "typical purchaser" group, F(1,110) - 2.51, p <.12. Again, no significant interaction was found.

TABLE 4

MEAN RESPONSE TO CREDIBILITY AND BIAS ITEMS

DISCUSSION

Whether, and to what extend, promotion leads to message acceptance depends in part on the degree to which the actual message contrasts with bias-related expectancies. Subjects who received an advertising message that made claims of superiority on every product attribute exhibited less message acceptance than did those who received a message that made a "less than superior" claim on one attribute. Subjects in the latter group rated the clock radio as being more reliable and accurate and less likely to have good quality sound. From this, it can be reasoned that message acceptance is dependent, in part, on whether the message confirms or disconfirms bias-related expectations which, in the case of promotion, appear to be ones that are relatively superior.

That subjects have such expectations concerning promotional messages is suggested by the marked difference between expectancy subjects' ratings of what the speaker was about to say in the ad and their expectations about what he would say if he were stating his "true" opinion in a private conversation. Implicit in this difference is variation over modality, or situation. Message recipients expect the "effect" the speaker's statement interpreting some entity, or issue) to differ over the two situational contexts. And this difference is thought to represent the situational, or reporting, bias that becomes part of message recipients' causal schemata

Attributional Processing

Utilization of bias-related schemata was reflected in subjects' post-message attributional processing. To a large extent, the difference in persuasion subjects' message acceptance appears to be due to the postulated differences in causal inferences. As expected, subjects who received a message that consisted of all positive information about the product (confirmation of expectancies) were more like to explain that message in terms of situational bias--i.e the corporation's desire to sell products. Subjects who were exposed to a message that contained negative attribute information- i.e., a message that presumably disconfirmed expectancies--were more likely to attribute the message t "real facts" about the product and less likely to assign cause to the situational factor.

It should be noted that this result also could be explain solely in terms of Kelley's (1973) discounting and augmentation principles (cf. Hansen and Scott 1976; Settle and Golden 1974). These rules, however are post-dictive and do not require the use of pre-message schemata or expectancies The theoretical structure presented here rests on a premessage message recipients are thought to utilize pre-message schemata, which result in pre-message expectancies. Messages that confirm those expectancies are taken to be evidence confirming pre-message schemata. When a promotional message confirms bias-related expectancies message recipients should explain that message in terms of their bias-related pre-message schemata. On the other hand, when a seller's message disconfirms such expectancies, message recipients have evidence indicating their pre-message reasoning should be rejected. Accordingly, their explanation of the seller's message should emphasize factors other than the biasing element(s).

Alas, the measures of pre-message and post-message schemata did not fully support the above reasoning, however. Two comparisons are relevant here: the expectancy group versus the expectancy-confirmed group; and the expectancy group versus the expectancy-disconfirmed group. According to the reasoning presented above, the pre-message schemata of the expectancy group should be similar to the post-message schemata of the expectancy-confirmed group and significantly different from the schemata of the expectancy-disconfirmed group. In terms of attributions about the corporation, however, the results were opposite to what was anticipated Expectancy-confirmed subjects seemed to have assimilated the actual message more toward the dominant pre-message cause- the situational bias. These subjects appear to have elaborated (up-dated) their schemata in terms of the cause that appears "most plausible" in explaining messages of superior claims. If such an "up-dating" process did occur, it was not as prevalent, of course, in the case of the expectancy-disconfirmed group. It is difficult, however, to draw any strong conclusions along these lines.

Related results concerning source credibility and situational bias clearly offer no support for the reasoning presented above. Although the majority of these items were found to vary in the expected direction, their lack of significance prevents any kind of statement other than nonsupport.

Spokesperson Type

The results associated with the second factor, spokesperson type, are again somewhat confusing in terms of attribution theory. As pointed out above, this factor was seen as a manipulation over the modality, or situation, presumed to be part of message recipients' causal schemata. Accordingly, it was anticipated that less reporting bias would be associated with the "hidden camera" speaker than with the "typical purchaser." The majority of the results surrounding this hypothesis were congenial to Kelley's model. The "hidden camera" spokesperson was rated as less biased in reporting his true opinion. In addition, he was rated as more credible. Similar results were found regarding the corporation's credibility. Clearer evidence of the postulated attributional process was exhibited by persuasion subjects' post-message schemata. The group exposed to the "hidden-camera" spokesperson was more likely to attribute the cause of the message to "real facts" about the product and less likely to make attributions to the seller's desire to sell products. This result. however. did not seem to explain differences in message acceptance, which were not significant across the spokesperson type factor.

On Methodology

Two comments, each relating to this study's methodology, seem to warrant specific attention. The first of these has to do with comparisons between the expectancy group and the persuasion group(s). As noted above, it would be more tidy if these groups' causal schemata behaved according to hypothesized i.e., if the pre-message schemata of the expectancy group were similar to the post-message schemata of the expectancy-confirmed (persuasion) group and different from those of the expectancy-disconfirmed (persuasion) group. That these comparisons did not emerge as hypothesized, however, might well be a function of the study's execution rather than a fault with the conception of pre-message schemata as mediators of marketing communication. Inasmuch as the post-dictive (confirmed/disconfirmed) comparison did behave as hypothesized, it might well be that we simply failed to operationalize the pre/ post comparison properly.

Finally, there seems to have been a problem with the "styling" attribute in the study. While obviously this is an important attribute, it cannot be treated in the same way that accuracy, reliability, and other "endorsable" qualities can if. Hirschman 1981). The (now obvious) fact is that subjects' belief scores for styling had nothing to do with expectancy confirmation; rather, they simply looked at the clock radio (there were three depictions in each ad treatment) and made up their own minds. "Who said what' about the product's styling paled in comparison to each subject's own sensory experience.

REFERENCES

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

Burnkrant, R.E. (1975), "Attribution Theory in Marketing Research: Problems and Prospects," in M.J. Schlinger (ed.), Advances in Consumer Research, 2, 465-9.

Cialdini, R.B. and Petty, R.E. (1981), "Anticipatory Opinion Effects," in R.E. Petty, T.X. Ostrom and T.C. Brock (eds.), Cognitive Responses in Persuasion. Hillsdale, N.] : Lawrence Erlbaum Associates. 217-35.

Cialdini, R.B., Petty, R.E. and Cacioppo, J.J. (1981), "Attitude and Attitude Change," in Annual Review of Psychology, 329 357-404.

Eagley, A.R., Wood, W. and Chaiken, S. (1978), "Causal Inferences about Communicators and their Effect on Opinion Change," Journal of Personality and Social Psychology, 369 424-35.

Festinger, L. and MacCoby, N. (1964), "On Resistance to Persuasive Communications," Journal of Abnormal and Social Psychology. 79. 350-66.

Golden, L.L. (1977), "Attribution Theory Implications for Advertisement Claim Credibility," Journal of Marketing Research, 14, 115-7.

Hansen, R.A. and Scott, C.A. (1976), "Comments on Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 13, 193-7.

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

Hirschman. E.C. (1981), "Cognitive Complexity and the Perception of Intangible Attributes," in K.C. Bernhardt, et al. (eds.), Educators' Conference Proceedings. Chicago: American Marketing Association , 158-61.

Hovland, C., and Mandell, W. (1952), "An Experimental Comparison of Conclusion Drawing by the Communicator and by the Audience," Journal of Abnormal and Social Psychology, 47, 581-8.

Jones, E.E. and Davis, K.E. (1965), "From Acts of Dispositions," in L. Berkowitz (ed.), Advances in Experimental Social Psychology, 2. New York: Academic Press, 219-66.

Jones, E.E., Worchel, S., Goethals, G.R. and Grumet, J. (1971), "Prior Expectancy and Behavioral Extremity as Determinants of Attitude Attribution," Journal of Experimental Social Psychology, 5, 59-80.

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

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

Kelley, R.H. (1972b), "Causal Schemata and the Attribution Process," in E.E. Jones et al. (eds.), Attribution: Perceiving the Causes of Behavior. Morristown, N.J.: General Learning Press, 151-74.

Kelley, R.H. (1973), "The Process of Causal Attribution," American Psychologist, 28, 107-28.

Kelley, H.H. and Michela, J.L. (1980), "Attribution Theory and Research," in Annual Review of Psychology, 31, 457-501.

Kernan, J.B., and Sommers, H.S. (1967), "Meaning, Value, and the Theory of Promotion," Journal of Communication, 17, 109-35.

Koeske, G.F., and Crano, W.D. (1968), "The Effect of Congruous and Incongruous Source Statement Combinations Upon the Judged Credibility of a Communication," Journal of Experimental Social Psychology, 4, 384-99.

McPeek, R.W., and Edwards, J.D. (1975), "Expectancy Disconfirmation and Attitude Change," Journal of Social Psychology, 96, 193-208.

Mills, J. and Jellison, J.M. (1967), "Effect on Opinion Change of How Desirable the Communication is to the Audience the Communicator Addressed," Journal of Personality and Social Psychology, 6, 98-101.

Mizerski, R.W., Golden, L.L. and Kernan, J.B. (1979), "The Attribution Process in Consumer Decision Making," Journal of Consumer Research, 6, 123-40.

Scott, C.A. (1978), "Attribution Theory in Consumer Research: Scope, Issues, and Contributions," in S.C. Jain (ed.), Research Frontiers in Marketing: Dialogues and Directions. Chicago: American Marketing Association, 169-73.

Settle, R.B., (1972), "Attribution Theory and Acceptance of Information," Journal of Marketing Research, 9, 85-8.

Settle, R.B.-and Golden, L.L. (1974), "Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 11, 181-85.

Smith, Robert E. and Hunt, S.D. (1978), "Attributional Processes and Effects in Promotional Situations," Journal of Consumer Research, 5, 149-58.

Sparkman, R.H. Jr. and Locander, W.B. (1980), "Attribution Theory and Advertising Effectiveness," Journal of Consumer Research, 7, 219-24.

Steiner, I.D. and Field, W.L. (1960), "Role Assignment and Interpersonal Influence," Journal of Abnormal and Social Psychology, 61, 239-46.

Thibaut, J.W. and Rieken, H.W. (1955), "Some Determinants and Consequences of the Perception of Social Causality, Journal of Personality, 24, 113-33.

Walster, E.E., Aronson, E. and Abrahams, D. (1966), "On Increasing the Persuasiveness of a Low Prestige Communicator," Journal of Experimental Social Psychology, 2, 325-42.

Walster, E.e. and Festinger, L. (1962), "The Effectiveness of 'Overhead' Persuasive Communications," Journal of Abnormal and Social Psychology, 65, 395-402.

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