Effects of Expectation Creation and Disconfirmation on Belief Elements of Cognitive Structure

Jerry C. Olson, The Pennsylvania State University (student), The Pennsylvania State University
Philip Dover,
ABSTRACT - Previous research on consumer expectations is briefly reviewed. A major problem in this work is the lack of a precise and theoretically rich conceptual definition of expectations, since most other problems stem from the resultant ambiguity regarding the expectancy construct. It is suggested that expectations are best considered as individual belief elements in a cognitive structure. In a longitudinal experiment, expectations about an attribute of a new brand of coffee were created by exposure to three ad-like messages and then were disconfirmed through actual product trial. The effects of these manipulations on belief-expectancies and other elements of cognitive structure are examined, as are various theoretical perspectives for future research.
[ to cite ]:
Jerry C. Olson and Philip Dover (1976) ,"Effects of Expectation Creation and Disconfirmation on Belief Elements of Cognitive Structure", in NA - Advances in Consumer Research Volume 03, eds. Beverlee B. Anderson, Cincinnati, OH : Association for Consumer Research, Pages: 168-175.

Advances in Consumer Research Volume 3, 1976      Pages 168-175

EFFECTS OF EXPECTATION CREATION AND DISCONFIRMATION ON BELIEF ELEMENTS OF COGNITIVE STRUCTURE

Jerry C. Olson, The Pennsylvania State University

Philip Dover (student), The Pennsylvania State University

[This research was funded by a grant from the Pennsylvania State University to the senior author and Andrew Mitchell.]

ABSTRACT -

Previous research on consumer expectations is briefly reviewed. A major problem in this work is the lack of a precise and theoretically rich conceptual definition of expectations, since most other problems stem from the resultant ambiguity regarding the expectancy construct. It is suggested that expectations are best considered as individual belief elements in a cognitive structure. In a longitudinal experiment, expectations about an attribute of a new brand of coffee were created by exposure to three ad-like messages and then were disconfirmed through actual product trial. The effects of these manipulations on belief-expectancies and other elements of cognitive structure are examined, as are various theoretical perspectives for future research.

Much has been written about the importance of consumer expectations, particularly their impact on purchasing behavior and the national economy (cf. Katona, 1974; Kelley and Scheewe, 1975) and their effects on consumer dissatisfaction (cf. Anderson and Hair, 1972). However, despite the seaming relevance of expectancies to several major issues, only a handful of empirical studies in the marketing literature deal explicitly with consumer expectations (Anderson, 1973; Cardozo, 1965; Cohen and Goldberg, 1970; Olshavsky and Miller, 1972; Swan, 1972; Woodside, 1972). Most of this research has concentrated on the effects of a product usage experience that disconfirms pre-trial expectations.

In a critical review of this research, Olson, Toy and Dover (1975) identified several theoretical and methodological problems which may be inhibiting consumer behaviorists from examining important expectancy phenomena. The present paper summarizes these problem areas and describes an experimental study which focuses on the cognitive processes underlying expectancy formation and disconfirmation. This research was designed to eliminate many of the problems found in earlier studies and to explore theoretical bases for future work.

PROBLEMS IN PREVIOUS RESEARCH

Manipulation of Expectations

In each of the marketing expectancy studies, product expectations were supposedly created by providing consumers with product attribute and/or performance information prior to actual trial. The usual manipulation involved a written communication about specific characteristics of the product. For instance, Olshavsky and Miller (1972) presented consumers with a list of features about audio tape decks, including price, brand name, and several technical/performance attributes. Similarly broad informational manipulations of consumer expectations were used by Cardozo (1965), Woodside (1972), and Anderson (1973).

These "shotgun approaches" to expectancy creation indicate a conceptual vagueness regarding the expectation construct. In none of the studies did the author(s) specify precisely what type of expectation was to be created by the manipulation used. Moreover, logical or theoretical rationales were not offered regarding why the manipulation should be effective. Rather, it seemed to be implicitly assumed that exposure to information about product attributes would, by some unspecified process, create an expectancy regarding the product. Other evidence for the conceptual uncertainty regarding the expectancy construct is the almost total lack of measures to determine whether or not the informational manipulations actually created expectations (only Anderson, 1973, used manipulation checks). Of course, empirical manipulation checks are difficult to design without a precise idea of what is being manipulated.

Three other problems involving the expectancy-creating manipulations should be briefly noted. First, there was a general failure in the past research to identify and use alternative external sources of consumer expectations. In addition to general written information about product features, product expectations may be created by exposure to advertisements and word-of-mouth, by observation of the product, and through actual product usage experiences. Second, although expectancies are probably formed most often by a series of exposures to such sources over time, none of the expectancy research incorporated this longitudinal dimension. Third, the expectancy manipulation and its position within the overall experimental procedures typically were somewhat artificial and transparent. For instance, in all of the studies except Cohen and Goldberg (1970), consumer subjects were given the informational expectancy manipulation immediately prior to the product usage experience which in turn was immediately followed by the post-trial questionnaire. Such short time periods between experimental events are somewhat artificial and also tend to increase the transparency of the experimental purpose, perhaps creating demand characteristic artifacts. None of the studies reported use of debriefings or other procedures to determine whether or not some respondents had discovered the experimental purpose.

Theoretical Perspective

Much of the expectancy research may be criticized for a lack of attention to alternative theoretical bases for disconfirmation phenomena. Beyond the typical broad references to dissonance theory, only Anderson (1973) examined various alternative theoretical explanations for his results.

Conceptual Definition of Expectation

A major problem in the consumer expectancy literature, and a primary cause of the problems discussed above, is the lack of a conceptually precise definition of an "expectation." Partly because of the resultant ambiguity regarding the expectancy construct, investigators have used various manipulations and theoretical perspectives in their research. However, broadly speaking, in all of the previous studies expectancies seem to have been inferred from consumers' general, overall evaluative judgments of the product. Examples of this "conceptual approach" include Cardozo's overall rating of product value, Cohen and Goldberg's overall product preference raging, Olshavsky and Miller's rating of overall product performance, and Anderson's overall product rating and aggregated index based on separate ratings of 15 product attributes. Thus, although explicit definitions of expectancies were not found in the previous research literature, expectations seem to have been implicitly conceptualized as global product evaluations. Most theorists agree that such evaluative judgments are attitudes toward the product (cf. Fishbein and Ajzen, 1975). Thus, the past research has implicitly considered consumer expectations as general, overall product attitudes.

We believe this conceptual perspective to be somewhat vague, potentially misleading, and therefore dysfunctional. By considering expectancies as global product attitudes, researchers tend to ignore the cognitive processes underlying attitude formation and attitude change due to a disconfirmation experience. In order to focus research on these cognitive attitude processes, the expectation construct must be given a precise meaning that will unambiguously identify its cognitive nature and position it theoretically vis-a-vis other cognitive elements. To that end, we recommend that consumer, product expectations be conceptualized as individual belief elements in a consumer's cognitive structure regarding the product.

As a scientific construct, belief has a precise and generally accepted definition; that is, one's subjective judgment regarding the likelihood of a relationship between two concepts (cf. Fishbein, 1967). Thus, an expectation is the perceived likelihood that a product possesses a certain characteristic or attribute, or will lead to a particular event or outcome. Essentially, beliefs and expectancies are the same cognitive construct, although some may wish to use the term expectation only when referring to a belief judgment regarding a "future event or state of affairs." We should note that this conceptual approach is not new; for example, Tolman (1932) considered expectancies to be "cognitions" or "subjective probabilities that one event is associated with (or follows from) some other event (Fishbein and Ajzen, 1975, p. 28)."

By considering expectancies as beliefs, research on consumer expectations is placed squarely into the .theoretically-rich literature dealing with cognitive structure (cf. Bettman, Capon and Lutz, 1975; Bither and Miller, 1968) and attitude formation and change (cf. Lutz, 1975; Olson and Mitchell, 1974). Such a conceptual perspective is useful in several ways. First, based upon the fairly extensive body of research which has examined belief formation processes, manipulations of consumer belief-expectations may be more easily operationalized. Second, a cognitive perspective to expectancy research forces attention on the elements of cognitive structure (e.g., belief-expectations, the evaluative aspect of belief-expectancies, overall evaluation or attitudes toward the product, behavioral intention, etc.), and on their interrelationships. Third, based upon the massive literature dealing with the relationships between beliefs, attitudes, behavioral intention, and actual behavior, consumer expectations may be theoretically linked with subsequent overt behaviors such as purchase. Fourth, by considering expectations as cognitive beliefs, various cognitive theories or learning theory may be used to explain the acquisition of overall product attitude (Olson and Mitchell, 1974). Then, if desired, this attitude may be considered as a summary or global expectation about the product.

PURPOSE OF PRESENT RESEARCH

The present study represents an initial examination of consumer expectations as attribute-specific product beliefs and of the disconfirmation of expectations from the more theoretically justifiable perspective of cognitive structure modification. This conceptual approach leads to an explicit consideration of the processes by which belief-expectations are created and of the effects that such a cognitive structure has on product judgments following a disconfirming usage experience. Clearly, this conceptual perspective requires a longitudinal research design in which each subject's cognitive structure is monitored over time as a function of various expectancy-creating and expectancy-disconfirming informational inputs. The present longitudinal, individual-level approach contrasts with the earlier research, all of which has used between-groups, aggregate data analyses. In summary, the present research is an attempt to operationalize the suggestions presented above and, particularly, to illustrate the advantages of a more cognitively oriented approach to research concerning the acquisition and disconfirmation of consumer expectations. Furthermore, although competing theoretical explanations for disconfirmation effects are not explicitly tested in this study, a second purpose is to examine relationships between selected constructs from which useful theoretical bases for future research may be inductively derived.

METHODS

Research Design

The primary purpose of this research was to examine the process of expectancy creation and the subsequent effect of these beliefs on product evaluations following an expectancy-disconfirming usage experience. Therefore, the major manipulation was four "trials" in which consumer subjects were exposed to product information that could be acquired and integrated into their cognitive structures (see the research design presented in Table 1). In trials 1, 2, and 3, subjects in the experimental group were presented with three different, ad-like communications, each of which stressed that the product possessed a specific product characteristic and which should create belief-expectancies to that effect. Trial 4 consisted of an actual product usage experience which was purposefully devised so as to disconfirm the manipulated expectancy beliefs. A separate control group received none of the three expectancy-creating messages, but did try the product in the same usage experience setting. Following each informational exposure (i.e., each trial), extensive measures of cognitive structure were taken.

TABLE 1

EXPERIMENTAL DESIGN

Product

Several authors have suggested that some degree of ego-involvement with a product may be necessary in order that disconfirmation effects are clearly evidenced (Cohen and Goldberg, 1970; Anderson, 1973). With this in mind, ground coffee was chosen as a product thought to be relatively high in ego-involvement for the subjects used in this research. Additional considerations also influenced the choice of ground coffee. Coffee has a relatively limited cognitive structure, in terms of the number of salient belief attributes, thus simplifying the measurement requirements. Moreover, the authors had access to a new brand of coffee (of below-average quality) which was not available in the local area and, therefore, was unknown to the subjects. In order to increase the mundane realism of the experiment, the actual brand name of this product was used throughout the study. However, for purposes of this paper, the product is referred to as "Coffee."

Subjects

The subjects for this study were 38 adult women recruited from several small social groups which met regularly on a bi-weekly basis. Most of these women were married, with children in their family, and all were regular consumers of ground coffee. The 20 women in the experimental group were promised $5.00 and the 18 control group women were promised $2.50 for their respective participation. Consumers in both groups were paid upon completion of the study.

Dependent Variables

Although numerous methods for measuring belief elements of cognitive structure have been used in marketing research (e.g., Fishbein and Raven, 1962; Sheth, 1974; Wilkie and Pessemier, 1973), certain dissatisfactions with these procedures exist (cf. Cohen, Fishbein and Ahtola, 1972). Recently, a seemingly improved model and measurement procedure was proposed by Ahtola (1975). This "vector model" maintains the theoretical properties of the belief construct while allowing a more realistic measurement of a consumer's belief structure regarding a product. In order to implement this procedure, detailed knowledge about consumers' cognitive structures is required. The major belief-expectancy dimensions (or salient product attributes) in a cognitive structure, as well as the specific perceptual categories or levels along each dimension, must be determined (cf. Mazis, Ahtola, Klippel, 1975). Once these cognitive aspects are identified, product beliefs are measured not for a specific dimension (e.g., sweetness), but for each discriminable level or category along that dimension (e.g., slightly sweet, very sweet, etc.).

In order to identify these aspects of cognitive structure for ground coffee, extensive pilot interviews were conducted with 15 housewife subjects using a modified repertory grid procedure (for details of this method see Wilson and Dover, 1975). Five attributes of ground coffee and their respective discriminable levels were identified as the most frequently appearing elements of cognitive structure (see in Table 2).

TABLE 2

BELIEF-EXPECTANCY ELEMENTS OF COGNITIVE STRUCTURE FOR GROUND COFFEE

These cognitive dimensions were represented in a questionnaire using the same procedure as Ahtola (1975). Briefly, subjects were to assign 10 points among the 4 or 5 levels of each belief-expectancy dimension, such that the number of points granted to each attribute level represented the strength of belief about the likelihood that the product possessed that particular level of the attribute dimension. Other cognitive elements measured in the questionnaire included confidence in each belief-expectancy rating, the evaluative aspect of each attribute level along each dimension, the overall evaluation of (or the attitude toward) the product, and the behavioral (purchase) intention toward the product.

Expectancy Manipulations

The specific belief expectancy dimension chosen for manipulation was coffee bitterness. In the repertory grid pretesting, bitterness generally seemed to be the most important characteristic of ground coffee, with "not bitter," of course, the most desirable level of bitterness. Further pretesting suggested that coffee drinkers could best distinguish four levels along the bitterness dimension--not at all, somewhat, fairly, and very bitter (see Table 2).

Based upon this knowledge, three different messages were created using an "every-day" informal writing style to simulate advertising copy. Each of these messages emphasized that "Coffee" had no bitterness. Because other product attributes were not explicitly mentioned in the messages, only expectancies regarding coffee bitterness were directly manipulated. Each message was typed on official-appearing stationery made up with the company logo as a letterhead. Communication A simulated a typical consumer print ad and described the product in general non-technical terms emphasizing its lack of bitterness. Communications B and C were slightly more technical discussions of "Coffee's" lack of bitterness. The former summarized the results of a consumer survey and the latter described opinions of expert taste-testers.

Experimental Procedures

Subjects were first contacted at their respective group meetings and were asked to participate in a study of the effectiveness of marketing procedures used to introduce a new product, in this case a new brand of ground coffee. Interest was high and virtually all those contacted agreed to participate.

Subjects in the experimental group were told that manufacturers often send out information about their products in order to inform consumers of the product and its particular characteristics. Further, they were told that the researchers had obtained several of these communications for a new brand of coffee not yet available in the local area. These subjects were told that the present study was being conducted in order to obtain their reactions to these communications and to the coffee itself. The control group subjects were merely told that a new brand of coffee had been developed, but was not yet available in the local area. They were told that the present study was to obtain their tasting reactions to a sample of the coffee at one of their future meetings.

The study began about two weeks after the initial contact. At that time, the first of the communications was personally delivered to each experimental group respondent, along with the questionnaire containing the dependent variables. Subjects were told to read the communication at their leisure and then immediately to complete the questionnaire and mail it back to the researchers. Four days later, the second communication and an identical questionnaire were delivered to the subjects, followed in four days by the third communication and questionnaire. No problems occurred regarding prompt return of the questionnaires.

Then, approximately four days later, the experimental group met for one of its regular meetings. During the meeting, consumer subjects were taken singly to another room for the tasting of the "Coffee." In order to create an expectancy-disconfirming trial experience with the product, a large pot of coffee was made up using 50% more ground coffee than called for by the package directions. In earlier pilot tests, this composition had been judged to have substantial bitterness.

For the tasting, subjects were seated at a table and given one-half cup of black "Coffee" (i.e., no cream or sugar was allowed). Subjects were told that taste sensitivity was enhanced when cream and/or sugar were omitted. In this way a consistent product trial experience was maintained for all subjects. Subjects were allowed to sip and sample the taste and aroma of the coffee for approximately two minutes and then responded to a questionnaire containing the same cognitive structure measures taken in the previous three trials as well as several other measures not relevant to this paper. Identical tasting procedures and measuring instruments were used for the control group subjects when they met for one of their regular meetings.

All subjects were debriefed individually in their homes approximately two weeks later.

RESULTS AND DISCUSSION

Evaluation of Demand Characteristic Effects

Longitudinal designs involving multiple exposures and measurements tend to increase the potential for demand characteristic artifacts (Olson, 1974; Sawyer, 1975). In order to roughly determine the biasing potential of such artifacts in the present study, subjects were asked, Upon completion of the research, for their opinion of its purpose. All repeated essentially the original explanation given them. No subjects mentioned that the study might deal with the effects of a disconfirmation experience or with the development of product beliefs and their impact on later product judgments.

Manipulation Checks--Expectations

The first three measurements of belief strength for the "not at all bitter" expectancy may be considered as manipulation checks for the three expectancy-creating communications. On the 10-point scale of belief strength, the mean ratings for the belief expectancy that the "coffee" was "not at all bitter" were 7.45, 6.65, and 6.70, over the first three trials. Although the mean expectancy strength decreased slightly after the first communication, the trend was not significant (F=1.28, df-2/57, p >.10). Thus, the communications were successful in creating a rather strong belief-expectation that the "Coffee" was not bitter.

Manipulation Checks--Disconfirmation

In fact, however, the "Coffee" was rather bitter as evidenced by the control group's post trial ratings. As can be seen by looking ahead to Figure 2, the control subjects, who had no specific expectancy manipulations prior to tasting, judged the "Coffee" to be more bitter, at all four bitterness belief levels, than did the experimental group. Although only the not-at-all-bitter belief-expectancy yielded a marginally significant difference between the experimental and control groups (t=1.68, p=.10), the overall pattern of ratings provides indirect evidence that the coffee tasting experience did disconfirm the experimental group subjects' "unrealistic" expectations.

Effects on Belief Elements of Cognitive Structure

The variety of measures taken over the four experimental trials resulted in a large amount of data regarding several cognitive elements and their interrelationships. However, this paper deals only with those results related to belief-expectancies about "Coffee" bitterness and certain other closely-related belief elements in consumer's cognitive structures.

In order to clearly interpret the effects of the messages and disconfirmation experience on the bitterness belief-expectancies, one must consider the evaluative responses to each bitterness belief level. Table 3 presents the mean evaluation ratings of each level of bitterness (for the generic product, ground coffee) taken at each of the four experimental trials and for the control group (post-trial). Three results in these data deserve mention. First, for the experimental group, the evaluative ratings are quite stable over the four trials. None of the four one-way ANOVA's yielded significant differences. Second, none of the control group's evaluations of the bitterness attribute are significantly different from the grand mean of the experimental group's evaluations (all p's >.10). Third, there are wide differences in evaluative responses between the four belief levels for coffee bitterness. Based upon the means presented in Table 3, the evaluative aspects (al) of these four bitterness belief-expectancies can be qualitatively labeled as: not at all bitter--very good, slightly bitter--neutral, fairly bitter--fairly bad, very bitter--extremely bad. These distinct differences in evaluation provide further support for treating belief components of cognitive structure in terms of specific categories, levels, or amounts of each attribute dimension.

Figure 1 presents the mean ratings of belief-expectancy strength for the four bitterness levels as a function of the expectancy-creating messages and the disconfirming trial experience. For each bitterness level the changes in expectancy strength over the four trials were analyzed by a one-way, within-subjects ANOVA.

TABLE 3

MEAN EVALUATION LEVELS FOR FOUR BELIEF-EXPECTANCIES REGARDING "COFFEE" BITTERNESS

FIGURE 1

CHANGES IN BELIEF STRENGTH FOR BELIEF-EXPECTANCIES REGARDING COFFEE BITTERNESS

These analyses indicated significant changes in belief-expectations for the "not-at-all-bitter" (F = 3.29, df = 3/57, p < .03) and the "fairly bitter" (F = 2.61, df = 3/-57, p <.06) belief levels, but not for the other two bitterness beliefs. Using the Bonferroni method for pairwise comparisons, no significant differences were found between the three pre-trial expectations for the not-at-all-bitter and fairly-bitter belief levels. Therefore, these pre-trial expectancies were averaged and the single belief-expectation was compared with the post-trial belief strength. For both the not-at-all-bitter and fairly-bitter attribute levels, the disconfirming trial experience caused a significant change in belief-expectations (both p's<.03). In an evaluative sense, both beliefs changed in more negative direction (i.e., each Biai combination became more negative, or less ositive). In Figure 1 note also that the nonsignificant changes in post-trial belief-expectations for the other two bitterness levels are consistent with the above-mentioned results. Thus, the entire bitterness belief-expectancy vector (i.e., SBiai) became more negative following the disconfirmation experience. In summary, the overall pattern of belief-expectancy changes is consistent (in an evaluative sense) for all four bitterness levels, with the greatest change in each occurring as a function of the disconfirming usage experience. [Although space does not allow a thorough discussion, it is interesting to note that all the other cognitive structure elements measured (e.g., attitude--Ao, behavioral intent--BI, and the belief vectors them- o selves--SBiai) evidenced this same pattern of change.]

FIGURE 2

COGNITIVE CHANGES IN BELIEF-EXPECTANCIES AS A FUNCTION OF A DISCONFIRMING TRIAL EXPERIENCE

Thus, as Cohen and Goldberg (1970) found and as might be intuitively expected, product usage experience appears to have a strong impact on belief-expectancy elements of cognitive structure. These data suggest that the information consumers derive from product trial (or longer-term usage experience) may have a greater effect on the acquisition and/or change of cognitive structure elements than does information from sources such as advertising or word-of-mouth. Of course, variables such as product-specific and general self-confidence could be expected to mediate the cognitive impact of such experientially-derived information.

The changes in belief-expectancies become more clearly interpretable when compared with the belief ratings obtained from the control group, which received no pretrial manipulation of expectancies regarding the "Coffee.'' In Figure 2, the consistent effect of the disconfirming trial experience on post-trial belief strength "moved away from" the pretrial expectancy and "towards," but did not achieve, the "objective" post-trial belief strength of the control group. It should be noted that, although the latter result is true in terms of absolute magnitudes of belief strength, only the not-at-all-bitter belief-expectancy evidenced a significant difference in post-trial ratings (p=.10) between the experimental and control groups. The direction, if not the magnitude, of each post-trial be-lief-expectancy change constitutes a perfect example of an assimilation effect (Sherif and Hovland, 1961). In disconfirmation phenomena, assimilation occurs when a stimulus is judged closer to one's own position (pretrial belief-expectancy) than it objectively is (determined by the control group's post-trial belief strength rating). These findings of consistent assimilation effects are in agreement with the results of virtually all the earlier disconfirmation research in marketing.

However, before concluding that assimilation-contrast is the theoretical basis for explaining disconfirmation effects, it must be noted that at least one alternative theory--namely, cognitive dissonance--also yields predictions that are consistent with the pattern of results reported above and in earlier research. Interestingly, however, the two theories do differ in terms of specific cognitive phenomena of primary interest in a disconfirmation study. For instance, given data as presented in Figure 2, assimilation-contrast theory explains the post-trial belief "R" as a movement from "C"' (the "objective" post-trial belief of the control group) caused by the pretrial belief-expectancy "E", whereas dissonance theory focuses on the "E-to-R" belief changes and requires no control group "C".

Because dissonance and assimilation-contrast theories generally predict the same pattern of results, it is quite difficult to devise a clear, distinguishing test between the two. Only when expectations are very strongly disconfirmed do the two theories generate differing predictions. For disconfirmations of great magnitude, assimilation-contrast theory predicts a contrast effect--i.e., a post-trial belief rating "further away from" one's expectation than an objective belief judgment. In comparison, dissonance theory would continue to predict a post-trial belief "closer" to one's expectation than an objective judgment. It is interesting to note that no marketing-related study has found evidence for a contrast effect, although Anderson (1973) incorrectly identified a contrast effect in his extreme disconfirmation condition (see Olson, Toy and Dover, 1975).

Reasoning inductively, the present and earlier results suggest that usage experience with most consumer products may not create the extreme disconfirmations necessary to produce a contrast effect. Alternatively, it may be that post-trial cognitive changes following a disconfirming experience are actually caused by dissonance reduction rather than assimilation processes. Mediating or moderating variables which may be useful in determining whether or not contrast effects can be obtained in consumer disconfirmation situations include: ego-involvement with the product, effort expended and involvement in the choice-usage situation, and the general ambiguity of the product judgments.

A rather different conceptual perspective for viewing disconfirmation phenomena might be derived from the various information processing models of the information integration process. The theoretical alternatives include information averaging models (cf. Anderson, 1970; Bettman, Capon and Lutz, 1975), congruity theory (Osgood and Tannenbaum, 1955), attitude models (cf. Fishbein, 1963; Rosenberg, 1956), and balance theory (Heider, 1946; Newcomb, 1953), among others. Most of the extant research involving these models has examined the outcomes of integrating information about different stimulus attributes (e.g., price and style) or different cognitive elements (e.g., affective and cognitive). However, it may be possible to modify or extend these theoretical approaches in order to handle situations involving disconfirmation of expectations in which pre- and post-trial information regarding the same dimensions must he integrated.

FIGURE 3

CHANGES IN BELIEF STRENGTH FOR BELIEF-EXPECTANCIES REGARDING RELATIVE EXPENSIVENESS

Briefly, two other theories may be useful for developing further research regarding the disconfirmation of consumer expectations. Helson's (1964) adaptation level theory has been applied in a price perception context (Nwokoye, 1975) and may be relevant to understanding belief-expectation development. Comparison level theory (cf. Thibaut and Kelley, 1959) has been used by Swan (1972) in a consumer expectation study focusing-on information search processes.

Finally, a cognitive phenomena involving several interesting theoretical belief-related issues warrants brief discussion. If it is assumed that belief dimensions are somehow interrelated in a cognitive structure, the creation of a belief-expectancy regarding one product attribute may be considered to have "spill-over" effects on belief-expectancies about other attribute dimensions (called inferential belief formation by Fishbein and Ajzen, 1975). Unfortunately, few consumer behaviorists have considered cognitive spill-over effects in their research (for an exception, see Lutz, 1975).

The present study allowed for the determination of cognitive spill-over effects from the bitterness belief-expectancy manipulations including measures of four other belief dimensions about ground coffee (see Table 2). Due to space constraints, only two of these belief vectors are examined in this paper. Figures 3 and 4 present the changes in belief-expectancies for the "level of expense" and "strength of flavor" attributes over the four experimental trials. Both figures indicate that the three expectancy-creating messages regarding "Coffee's" bitterness had some spill-over effects in terms of creating other belief expectancy elements in respondents' cognitive structure. Note that the belief strength ratings for "level of expense" and "strength of flavor" even though not directly manipulated, generally seem to approach a relatively stable or asymptotic pretrial level after the second message exposure. This pattern is similar to that found for the manipulated bitterness belief vector.

FIGURE 4

CHANGES IN BELIEF STRENGTH FOR BELIEF-EXPECTANCIES REGARDING STRENGTH OF FLAVOR

The cognitive process by which related belief-expectancies are acquired, even though information specific to those dimensions has not been processed, is one of great interest. Cognitive spill-over effects have obvious methodological and theoretical implications for those interested in attitude change, advertising effectiveness, or tests of various models of cognitive structure such as expectancy-value models. Although we will not speculate regarding the form of the internal information integration (or generation?) processes underlying these observed spill-over phenomena, several variables seem likely to be positively related to the amount of spill-over obtained. Among these candidates for incorporation into future research are: the amount of experience with a product, the complexity of the cognitive structure (e.g., number of belief dimensions) for a product, the level of confidence in one's belief-expectancy ratings (perhaps indicating the ambiguity or instability of certain belief dimensions), and the centrality of the manipulated belief in one's cognitive structure.

CONCLUSIONS

The conceptualization of consumer expectancies as specific belief elements within a cognitive structure appears to have been successful. Moreover, these belief-expectancies seem measurable with sufficient precision using the "vector model" methodology that relatively small and subtle belief changes can be identified.

Considering consumer product expectancies as a type of belief has obvious advantages, foremost among which is the ability to rely on established and proposed theoretical linkages between beliefs and other cognitive constructs. This facilitates theorizing about expectancy creation and disconfirmation, as well as developing of testable models of expectancy-related phenomena. Although several potentially useful theoretical perspectives have been discussed, most with a cognitive structure orientation, the task of developing clear-cut tests of competing theories will clearly be difficult. Presently, the major conceptual perspectives seem to be dissonance and assimilation-contrast theories. It may prove useful to base future investigations upon notions derived from information processing research.

REFERENCES

Olli T. Ahtola, "The Vector Model of Preferences: An Alternative to the Fishbein Model," Journal of Marketing Research, 12 (February 1975), 52-9.

Ralph E. Anderson, "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance,'' Journal of Marketing Research, 10 (February 1973), 38-44.

Ralph E. Anderson and Joseph F. Hair, "Consumerism, Consumer Expectations, and Perceived Product Performance," Proceedings, Third Annual Conference of the Association for Consumer Research (1972), 812-815.

James R. Bettman, Noel Capon, and Richard J. Lutz, "Cognitive Algebra in Multi-Attribute Models," Journal of Marketing Research, 12 (May 1975), 151-164.

Stewart W. Bither, and Stephen J. Miller, "A Cognitive Theory View of Brand Preference," Proceedings, Fall Conference of the American Marketing Association (1969), 280-286.

Donald T. Campbell: and Julian C. Stanley, Experimental and Quasi-Experimental Designs for Research (Chicago: Rand McNally, 1963).

Richard N. Cardozo, "An Experimental Study of Customer Effort, Expectation and Satisfaction," Journal of Marketing Research, 2 (August 1965), 244-249.

Joel B. Cohen, and Marvin E. Goldberg, "The Dissonance Model in Post-Decision Product Evaluation," Journal of Marketing Research, 7 (August 1970), 315-21.

Joel B. Cohen, Martin Fishbein, and Olli T. Ahtola, "The Nature of Uses of Expectancy-Value Models in Consumer Attitude Research," Journal of Marketing Research, 9 (November 1972), 456-60.

Martin Fishbein, "An Investigation of the Relationships Between Beliefs About an Object and the Attitude Toward that Object," Human Relations, 16 (August 1963), 233-240.

Martin Fishbein, "A Consideration of Beliefs and Their Role in Attitude Measurement," in Martin Fishbein (ed.), Readings in Attitude Theory and Measurement. (New York: Wiley, 1967), 257-66.

Martin Fishbein and Bertram H. Raven, "The AB Scales: An Operational Definition of Beliefs and Attitudes," Human Relations, 15 (1962), 35-44.

Martin Fishbein and Icek Ajzen, Belief, Attitude, Intention and Behavior (Reading, Massachusetts: Addison-Wesley Publishing Company, 1975).

Fritz Heider, "Attitudes and Cognitive Organization," Journal of Psychology. 21 (January 1946), 107-12.

Harry Helson, "Current Trends and Issues in Adaption-Level Theory," American Psychologist, 19 (1964), 26-38.

George Katona, "Psychology and Consumer Economics," Journal of Consumer Research, 1 (June 1974), 1-8.

Eugene J. Kelley, and Rusty Scheewe, "Buyer Behavior in a Stagflation/Shortages Economy," Journal of Marketing, 39 (April 1975), 44-50.

Richard J. Lutz, "Changing Brand Attitudes Through Modification of Cognitive Structure," Journal of Consumer Research, 1 (.March 1975), 49-59.

Michael B. Mazis, Olli T. Ahtola and R. Eugene Klippel, "A Comparison of Four Multi-Attribute Models in the Prediction of Consumer Attitudes," Journal of Consumer Research, 2 (June 1975), 38-52.

Theodore M Newcomb, "An Approach to the Study of Communication Acts," Psychological Review, 60 (1953), 393-404.

Nonyelu Nwokoye, "An Experimental Inquiry Into the Effects of Parameters of Price Structure on Buyers' Price Judgments: A Condensation of a Doctoral Dissertation," School of Business Administration, University of Massachusetts, Amherst: Working Paper #75-12 (May 1975).

Richard N. Olshavsky: and John A. Miller, "Consumer Expectations, Product Performance, and Perceived Product Quality," Journal of Marketing Research, 9 (February 1972), 19-21.

Jerry C. Olson, "Cue Properties of Price: Literature Review and Theoretical Consideration," Working Series in Marketing Research, College of Business Administration, Pennsylvania State University (1974), Paper #20.

Jerry C. Olson, and Andrew A. Mitchell, "The Process of Attitude Acquisition: The Value of Developmental Approach to Consumer Attitude Research," in M. J. Schlinger (ed.), Advances in Consumer Research: Volume II. (Chicago: Association for Consumer Research, 1974), 249-264.

Jerry C. Olson, Daniel R. Toy, and Philip Dover, "Consumer Expectations: Literature Review and Theoretical Considerations," Working Series in Marketing Research, College of Business Administration, Pennsylvania State University (1975), Paper #36.

Charles E. Osgood, and Percy H. Tannenbaum, "The Principle of Congruity in the Prediction of Attitude Change," Psychological Review, 62 (1955), 42-55.

Milton J. Rosenberg, "Cognitive Structure and Attitude Affect," Journal of Abnormal and Social Psychology, 53 (November 1956), 367-72.

Alan G. Sawyer, "Demand Artifacts in Laboratory Research on Consumer Behavior," Journal of Consumer Research, 1 (March 1975), 20-29.

Muzafer Sherif and Carl I. Hovland, Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change (New Haven, Connecticut: Yale University Press, 1961).

Jagdish N. Sheth, "A Field Study of Attitude Structure and Attitude Behavior Relationship," in J. N. Sheth (ed.), Models of Buyer Behavior (New York: Harper & Row, 1974).

John E. Swan, "Search Behavior Related to Expectations Concerning Brand Performance," Journal of Applied Psychology, 56 (August 1972), 332-335.

J. W. Thibaut, and H. H. Kelley, The Social Psychology of Groups (New York: Wiley, 1959).

Edward Chace Tolman, Purposive Behavior in Animals and Men (New York: Appleton-Century-Crofts, 1932).

William L. Wilkie, and Edgar A. Pessemier, "Issues in Marketing's Use of Multi-Attribute Attitude Models," Journal of Marketing Research, 10 (November 1973), 428-41.

David T. Wilson, and Philip Dover, "A Test of the Repertory Grid Technique as a Means of Developing Attributes for Use in Choice Models," Working Series in Marketing Research, College of Business Administration, Pennsylvania State University (1975), Paper #33.

Arch G. Woodside, "Positive Disconfirmation of Expectation and the Effect of Effort on Evaluation," Proceedings, 80th Annual Convention, APA (1972), 743-744.

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