Effect of Satisfaction and Its Antecedents on Consumer Preference and Intention

Richard L. Oliver, Washington University
Gerald Linda, Marsteller, Inc.
ABSTRACT - Theorists and researchers in the satisfaction area now generally agree that product satisfaction intervenes between expectancy disconfirmation and various postpurchase cognitive states including attitude and behavioral intention. Research in a variety of settings has supported the effect of expectation and its disconfirmation on satisfaction, but only a small number of studies addresses the cognitive consequences of satisfaction decisions and none reports data on choice processes such as brand selection. This study examines the influence of satisfaction and its determinants on behavioral intention and product preference in a simulated two-stage consumer usage situation. Generally it was found in both overall and summed attribute analyses that satisfaction was a function of expectation and disconfirmation, that intention was a function of satisfaction, and that preference was influenced by satisfaction and disconfirmation, the latter having the greater effect.
[ to cite ]:
Richard L. Oliver and Gerald Linda (1981) ,"Effect of Satisfaction and Its Antecedents on Consumer Preference and Intention", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 88-93.

Advances in Consumer Research Volume 8, 1981      Pages 88-93


Richard L. Oliver, Washington University

Gerald Linda, Marsteller, Inc.


Theorists and researchers in the satisfaction area now generally agree that product satisfaction intervenes between expectancy disconfirmation and various postpurchase cognitive states including attitude and behavioral intention. Research in a variety of settings has supported the effect of expectation and its disconfirmation on satisfaction, but only a small number of studies addresses the cognitive consequences of satisfaction decisions and none reports data on choice processes such as brand selection. This study examines the influence of satisfaction and its determinants on behavioral intention and product preference in a simulated two-stage consumer usage situation. Generally it was found in both overall and summed attribute analyses that satisfaction was a function of expectation and disconfirmation, that intention was a function of satisfaction, and that preference was influenced by satisfaction and disconfirmation, the latter having the greater effect.


The field of consumer satisfaction, long neglected by cognitive consumer behaviorists, is now beginning to develop a research tradition. Based on theoretical works by Engel, Kollat, and Blackwell (1968), Howard and Sheth (1969), Andreasen (1977), and Day {1977), research in both the laboratory and the field is proliferating. This has prompted reviewers (Day 1977, +lander 1977, Oliver 1977, LaTour and Peat 1979) to call for theoretical extensions and methodological refinements. Specifically, researchers have been asked to position satisfaction and its related constructs in cognitive models of consumer choice and to experiment with more coherent measures of the expectancy disconfirmation process.

Theoretical Perspective

A number of theories have been proposed to explain satisfaction decisions. These have been reviewed in detail elsewhere (Anderson 1973, Oliver 1977, LaTour and Pest 1979) and are not repeated here. In summing up the evidence to date, Oliver (1980b) has suggested that the most epistemologically efficient explanation derives from Helson's (1948, 1964) adaptation level theory which suggests that judgments of newly perceived stimuli are affected by prior experience with the general class of objects (the adaptation level) and the discrepancy perceived between the new stimulus and previously determined stimulus levels. Discrepancy, Helson noted, was a relative individually-specified phenomenon and has no necessary correspondence to objective differences.

Helson's (1964) theory may be viewed as isomorphic with the process of expectancy disconfirmation thought to occur in satisfaction judgments, differing only in terminology. One's expectation level, however determined, performs the function of an adaptation level. States of satisfaction/dissatisfaction are hypothesized to result, in part, from a comparison between one's perception of product performance and the expectation level. This discrepancy is known as (positive or negative) disconfirmation. Because the expectation level provides a baseline about which disconfirmation takes place, satisfaction is easily shown to be an additive function of expectation and disconfirmation. This interpretation was first suggested and empirically demonstrated in the industrial and social psychological literature (Ilgen 1971, Weaver and Brickman 1974) and has received limited support in consumer environments (Oliver 1977, Swan 1977, Madden, Little, and Dolich 1979). None of these studies extended the analysis to actual post-exposure choice behavior, however.

Measurement Issues

Researchers remain in disagreement over the most appropriate way to measure the antecedents of satisfaction and satisfaction itself. The concept of expectations, in particular, has been somewhat evasive from a psychological standpoint. Perhaps the most general workable definition is that of Katona (1964) who referred to expectations as a "subclass of attitudes that point to the future (p. 34)." As such, they may be seen as predictions of affect.

Early satisfaction researchers pursued a cognitive approach to the measurement of expectation and used product and attribute ratings as proxy variables for one's expected performance. Later, Olson and Dover (1979) and Oliver (1980a) adapted Fishbein (1967) scaling to this same task. Olson and Dover used only the belief elements of Fishbein's ab scales while Oliver combined the cognitive and affective dimensions to provide an evaluative belief score. Both approaches demonstrated the intended effect, but neither was a pure measure of predictive affect.

Disconfirmation has followed a similar developmental process. The discrepancy concept was initially viewed in the objective sense. Experimenters using manipulations in the laboratory assumed that disconfirmation existed whenever subjects' product experience deviated from the materials used in the expectation-creation task. Unfortunately, manipulation checks on perceived discrepancy were not taken. As a result, neither disconfirmation nor its magnitude and direction were determinable. Later researchers used the difference scores between preexposure and post-exposure attribute ratings (Oliver 1977, Swan 1977, Madden, Little, and Dolich 1979). While this approach to discrepancy was "real" in the sense that it involved a mathematical comparison between perceived performance and some prior cognition, it proved to be confounded with expectation and, on a conceptual plane, ignored the "surprise" value of disconfirmation. To address these problems, "better than expected -- worse than expected" scales have been used by Oliver (1977, 1980a) and have shown promise as predictors of satisfaction and other postexposure cognitions. This approach will be expanded here.

Finally, satisfaction has been measured using a variety of postexposure variables having affect as a common theme. Most researchers would agree, however, that product evaluations and attitudes are not operational definitions for satisfaction. Other approaches include Likert scales and "raw" satisfied-dissatisfied items. The latter have the greatest face validity but tend to produce positively skewed distributions reflecting a halo effect (Andrews and Withey 1976, Day and Bodur 1977). In comparison, multi-item Likert or semantic differential scales present the problem of finding a sufficient number of synonyms for satisfaction without also sampling the effective like-dislike domain. Interested readers are referred to some preliminary findings reported by Westbrook and Oliver (1981).


In an effort to extend further the basic satisfaction model to consumer intention and choice, a two-stage field study of consumer reactions to the preprocessing of new apparel fabrics was performed. In accord with the theoretical perspective of expectations as predictions of affect, expectations were viewed as a priori perceptions of the amount of satisfaction to be received from the product as a whole and in terms of its salient attributes. This required that measures of expectation be obtained before the exposure situation was encountered.

To maintain a consistent perspective to the expectancy disconfirmation process, disconfirmation was viewed as the perceived satisfaction deficit (surplus) after the product experience. Measured in terms of less (more) satisfied than expected, disconfirmation as viewed here focused on the comparative process and not on actual states of performance per se. The "mathematics" of discrepancy judgments were made by the subjects who responded with net feelings of fulfillment or lack thereof. As before, both overall and attribute-specific disconfirmations were elicited.

The last part of the second study phase, in keeping with the temporal nature of the model, required that satisfaction and subsequent cognitions be obtained. Care was taken to insure that the proper sequence of variable measurement was maintained to overcome partly problems associated with the necessity for concurrent measurement in this last study phase.

The approach taken here differs in a number of respects from prior studies and may provide a different perspective to satisfaction processes. First, the measurement of satisfaction and its antecedents involved consistent referents across study stages. Specifically, operationalizations of expectations of satisfaction, disconfirmation of satisfaction, and satisfaction itself were constructed with the same semantic phrases and attribute lists. Previous studies have confounded a host of different referents and algebraic difference scores in testing the basic model. While a multiple indicator approach will eventually be required for purposes of construct validation, research on satisfaction could benefit greatly from attempts to establish a coherent framework of satisfaction-related semantic concepts.

Secondly, attribute-specific satisfaction processes were compared to an overall approach in an effort to determine if the scheme suggested here operates in both modes. Attribute-specific measurement may not fully represent all salient attributes for all subjects, but does provide diagnostic information on an individual attribute basis. In contrast, overall scales serve to capture the totality of the cognitive process at the sacrifice of detail and, in addition, can be evaluated on the basis of internal consistency reliability. To date, few studies have been conducted using attribute-specific and/or multi-item overall scales for satisfaction and expectancy disconfirmation and none has compared these two variations. Hopefully, some insight into the relationship between specific and general satisfaction processes will be evident from this analysis.



A mall intercept study of male and female shoppers was conducted in a suburban Chicago shopping center. After an initial screening to determine eligibility requirements for the product group under investigation, subjects were asked to examine two pairs of men's sleeping apparel having different texture and softness. All product samples were recently manufactured and contained the sizing and other fabric conditioners used to facilitate construction of the garment. In the first stage of the study, the subjects indicated how satisfied they expected to be with both garments and garment attributes in daily normal usage and care, their interest in buying the products, and their preference for one over the other.

In the second stage of the study, the subjects were shown the identical pair of sleeping apparel after each had been laundered five times. Laundering removes the sizing and other pretreatment chemicals and generally makes fabrics softer, although to varying degrees. Subjects now responded to overall and attribute-specific disconfirmation scales on each product, then to overall and attribute-specific satisfaction scales, and finally to purchase intention and new preference.


Based on the sponsoring company's directive, 500 mall shoppers were approached on a quota basis to achieve a final sample of 250 males and 250 females. Males were required to be fifteen years of age or older (the target group) and to wear pajamas at least twice a week. Females, known to be the primary purchasers of men's pajamas for members of their families, were required to be 21 or older, to have males 15 years or older who wear pajamas at least twice a week in the household, and to usually buy the pajamas for the man in their family. The average age of males in the final sample was 38 years and that of females was 46.


Attribute-specific variables. Evaluative criteria for the attribute-specific measures were obtained by pretesting a list of 20 factors known to be related to judgments of wearing apparel on an independent sample of 400 consumers. Based on mean importance scores, a shorter list of nine attributes was judged sufficient by company management and is not reported here for proprietary reasons. Based on the Katona (1964) approach, expectations were measured by asking the respondent to indicate how satisfied "you or the male you would buy for" would be with regard to each attribute. Scales ranged from "very satisfied" to "very dissatisfied." The nine expectation scores were then summed to form the attribute expectation measure. Pre-exposure preference, in turn, was based on the subjects' actual preferred choice of one pair of sleeping apparel over that of the other.

Postexposure attribute disconfirmation was measured on 15-point scales where the midpoint defined each attribute as expected. The negative disconfirmation pole described attributes as worse (e.g., not as soft) than expected while the positive pole described them as better (e.g., softer) than expected. Attribute satisfaction was obtained after all disconfirmation measures were taken to insure that the measurement process maintained the temporal ordering of the underlying framework. The same items used for one's expectations of satisfaction were used here as well except that the lead-in to the satisfaction items simply asked the respondent to indicate how satisfied he/she was now. Finally, postexposure intention was measured on a five point "definitely would buy" to "definitely would not buy" scale while postexposure preference was obtained by asking the subjects to reconsider the pajamas a second time and to choose the pair they new preferred. The criterion used in the study was change (no change) in preference coded in 0-1 format.

Overall variables. A consistent three-referent scheme was used to measure overall expectation, disconfirmation, and satisfaction. Based on extensive work by Andrews and Withey (1976) and Campbell, Converse, and Rodgers (1976) on the meaning of satisfaction, two constructs having close (but not identical) semantic properties were combined with satisfaction (expressed as satisfied-dissatisfied) to form three-item scales. The two related constructs, happiness (happy-unhappy) and pleasantness (pleasant-unpleasant) were thought to reflect emotions close to those involved in a satisfaction response. Thus, expectation was measured by asking respondents how (satisfied, happy, pleased) they expected to be; disconfirmation was obtained on a "more than-less than" (satisfied, happy, pleased) scale; while satisfaction was an outright rating on the three items. Alpha reliability coefficients over the total sample were .85, .79, and .94 for expectation, disconfirmation, and satisfaction respectively.


Because the males in the sample were judging their own satisfaction while females were judging satisfaction vicariously for the male they would buy for, all analyses were run separately for men and women. In addition, separate analyses were performed for the overall and summed attribute scales to determine if equivalent results would be found. Finally, only the initial preferred pair of pajamas was analyzed across the two stages of the study. Results for the second (initially least preferred) pair were virtually identical to the first and are not reported here.

The models tested derive from those suggested in Oliver (1980a). His basic hypotheses were adapted to the context of the present study and appear as follows:

Satisfaction = f(Expectation, Disconfirmation)

Preference = f(Expectation, Disconfirmation, Satisfaction)

Intention = f(Expectation, Disconfirmation, Satisfaction, Preference)

Elementary path analysis was employed to test these propositions. Of interest are the magnitudes of the coefficients of determination and the comparative magnitudes of the (standardized) regression coefficients attributable to each of the independent variables within equations for each sample taken separately and for each variable across samples. Relative stability of the weights across situations would lend further support to the basic expectancy disconfirmation model.


Zero-order correlations between all variables across the two sample and measurement groups are shown in Table 1. The data are remarkably consistent across all analyses. Satisfaction was significantly correlated with its hypothesized antecedents in all cases, although disconfirmation yielded the highest correlations in the female sample. Preference (coded as change in preference) was most highly correlated with disconfirmation in all cases although significant correlations were also obtained with satisfaction and to a lesser degree, with intention. Surprisingly, preference was not correlated with expectation, a result explained more fully in the discussion section.

Finally, intention was universally correlated with all antecedents. In fact, with the exception of the preference variable, the magnitudes of the intention intercorrelations reflect the temporal nature of the proposed scheme exactly (r53 > r52 > r51). Although no sex differences were found in this analysis, the overall scales produced higher associations, particularly for the intention-satisfaction relationship.

Table 2 shows the results obtained when the suggested models were tested in a sequential path analysis. The data show, first, that satisfaction was an additive combination of expectation and disconfirmation, reflecting the low degree of multicollinearity between these measures (Table 1). Moreover, disconfirmation appeared to be the stronger of the two effects, particularly for women respondents. Between 35% and 50% of the variance in satisfaction was explained by this two-variable scheme.



The results for preference change were dominated mostly by disconfirmation and, to a lesser degree, by satisfaction. All analyses yielded roughly equivalent findings, explaining between 12% and 21% of the criterion variance. These figures compare to a theoretical maximum value of .52 for a dichotomous criterion having the probability of occurrence (.26) found in this study (Nunnally 1978). Female respondents appeared to place greater emphasis on satisfaction which accounts for the higher coefficients of determination for the female analyses.

Finally, intention was almost exclusively a function of satisfaction. The path coefficients were very high for the satisfaction variable and virtually zero for all other predictors including preference. All coefficients of determination were above .5 and it appears that the overall scales yield a more complete explanation than do the summed attribute measures.


This study has addressed a number of issues which merit further elaboration. These include the structure of the theoretical framework, level of variable measurement, sex differences in response, criterion predictability (validity), limitations of the methodology, and further research directions.

Theoretical framework.  The basic theoretical premise, Satisfaction = f(Expectation, Disconfirmation) patterned after Helson's (1964) adaptation level theory, received consistent support. It appears that expectations do provide a standard against which product deviations are perceived. This finding adds further support to a growing number of studies which suggest this perspective (Oliver 1977, Swan 1977, Gilly 1979, Linda and Oliver 1979, Madden, Little and Dolich 1979, Oliver 1980a, Swan and Trawick 1980, Westbrook 1980). It should be noted that most all of these investigations were conducted in different contexts using two-stage measurement and a varied mix of products and services including automobiles, breakfast bars, flu inoculations, newly opened department stores, and fabric cleaners.



It is also of interest to note that all studies including the one described here have found that disconfirmation is the more potent of the two effects. In terms of the magnitude of the standardized regression coefficients, a ratio of one and one-half to one is not uncommon. While methodological issues provide a partial explanation in that disconfirmation and satisfaction are typically measured in close proximity after product exposure, this sequence of events conforms to the nature of the hypothesized cognitive response. Additionally, great care was taken in the present study to segregate the two measures through questionnaire sequencing, although it is admitted that separate instruments and longer time intervals between measurement will be required before this issue is resolved.

Summed attribute vs. overall measurement.  Inspection of Table 2 shows that the overall bipolar adjective scales yielded marginally superior results to those obtained using summed attribute scales. The differences were most dramatic when coefficients of determination for the two scaling procedures were compared, particularly for the intention criterion. Sizable differences were also evident between the attribute and overall satisfaction path coefficients in the intention regressions.

Two explanations pertain. The first relates to current theoretical perspectives of cognition formation. If one adopts the approach taken by Fishbein and Ajzen (1975), the essence of constructs having formative elements is best captured in summary statements (e.g., one's overall subjective norm). This appears to be true because any list of elements will in all probability be deficient for some subjects and extraneous for others. Secondly, summed attribute lists of diverse elements cannot be expected to have high internal consistency reliability. In contrast, the bipolar adjectives selected for the overall scales demonstrated adequate to high reliability for the three constructs tested here and are probably to be preferred unless diagnostic information is desired.

Sex differences.  The data suggest that women responded with slightly greater magnitudes of effect in two cases. First, they placed greater emphasis on disconfirmation in the prediction of satisfaction and also on satisfaction in the prediction of preference. While the study was not designed to provide explanations for this observation, it may be that women were more involved in the decision. Involvement has been suggested as a moderator in reactions to product satisfaction situations (Olson and Dover 1979). Because the majority of men's sleepwear purchases are made by women, it is likely that their potentially greater involvement made them more sensitive to feelings of disconfirmation and satisfaction. The notion that disconfirmation perceptions are affected by product involvement has potential implications for practice and should be pursued in future research efforts.

Validity.  The model as proposed quite adequately predicted intention to buy. In fact, coefficients of determination for cognitive criteria in excess of .5 are fairly common among the studies cited previously. The model can also be shown to predict (in a concurrent sense) postexposure attitude and other reactions including product ratings. Unfortunately, behavioral choice (here preference) was not predicted to the same degree, even taking into account the fact that the theoretical maximum R2 value in this study was roughly .5. This is further clouded by the fact that disconfirmation and not satisfaction played a major role in one's postexposure preference. Typically, preference, like attitude, would be a significant function of satisfaction.

Further reflection suggests that the result obtained is that which would have been anticipated. Our criterion here is change in preference, not preference per se. Helson's (1966) theory suggests that expectations define the baseline for disconfirmation but that movement from level (change) is a function of disconfirmation only. Thus, postexposure attitude should be a function of expectation and satisfaction but attitude change (the difference) would be predicted best by the performance-expectation discrepancy. In fact, reanalysis of data from prior studies available to the authors suggests that this is indeed the case. This logic provides a reasonable explanation for the strong influence of disconfirmation on preference. The coefficients for satisfaction were significant, however, in three of four analyses, as hypothesized.

Limitations and future research directions.  A number of limitations which detract from the contribution value of the study requires elaboration. Each, in turn, suggests an area of improvement for future investigations. First, the study did not involve an actual purchase situation with normal usage. Rather, these two necessities were "short-circuited" in the procedure. Although the product response was accurately duplicated, the lack of a myriad of extraneous factors normally present in real usage situations may have provided an overly sterile environment for respondents, contributing to an overstatement of the theory's validity.

Closely related to the preceding comment is the issue of the timing of the measures relative to their occurrence in a natural environment. Ideally, the study should have been replicated after each washing rather than after five consecutive washings. Thus, expectations would have been measured before the first wash and disconfirmation, satisfaction, and revised expectations after the laundering. The revised expectations then become input to the cognitive response following the second washing. No study has looked at the process of satisfaction as a function of successive usage over time and work in this area is needed urgently.

A third issue surrounds the measurement of satisfaction and its antecedents. While each new study provides more focused perspective to different measurement vehicles, little consensus exists to suggest a one best approach. The three item bipolar adjective format used here demonstrated promise on two counts. First, the internal consistency appeared acceptable especially for measurement of satisfaction per se. And second, it is easily adaptable to expectation and disconfirmation items. Readers may object to the use of pleasantness and happiness as synonyms for satisfaction. The authors share similar reservations and welcome suggestions from other researchers.

Finally, the sample frame used here remains a potential source of error. A national probability sample was not used and one wonders how the results would have been effected. While the findings do not generalize beyond the mall sampling procedure used here, we have hopefully identified a close approximation to the process of satisfaction decisions used by consumers generally.


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

Andreasen, Alan R. (1977), "A Taxonomy of Consumer Satisfaction/Dissatisfaction Measures," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed., Cambridge MA: Marketing Science Institute.

Andrews, Frank M. and Withey, Stephen B.(1976), Social Indicators of Well-Being, New York: Plenum Press.

Campbell, Angus, Converse, Philip E. and Rodgers, Willard L. (1976), The Quality of American Life, New York: Russell Sage Foundation.

Day, Ralph L. (1977), "Toward a Process Model of Consumer Satisfaction," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed., Cambridge MA: Marketing Science Institute.

Day, Ralph L. and Bodur, Muzaffer (1979), "Analysis of Average Satisfaction Scores of Individuals over Product Categories,'' in New Dimensions of Consumer Satisfaction and Complaining Behavior, Ralph L. Day and H. Keith Hunt, eds., Bloomington Ind.: School of Business, Indiana University.

Engel, James F., Kollat, David T. and Blackwell; Roger D.(1968), Consumer Behavior, New York: Holt, Rinehart and Winston.

Fishbein, Martin (1967), "Attitude and the Prediction of Behavior," in Readings in Attitude Theory and Measurement, Martin Fishbein, ed., New York: John Wiley & Sons.

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

Gilly, Mary C. (1969), "Complaining Consumers: Their Satisfaction with Organizational Response: in New Dimensions of Consumer Satisfaction and Complaining Behavior, Ralph L. Day and H. Keith Hunt, ed., Bloomington Ind.: Department of Marketing, Indiana University.

Helson, Harry (1948), "Adaptation-Level as a Basis for a Quantitative Theory of Frames of Reference," Psychological Review, 55 (November), 297-313.

Helson, Harry (1964), Adaptation-Level Theory, New York: Harper & Row.

Howard,. John A. and Sheth, Jagdish N.(1969), The Theory of Buyer Behavior, New York: John Wiley & Sons.

Ilgen, Daniel R. (1971), "Satisfaction with Performance as a Function of the Initial Level of Expected Performance and the Deviation from Expectations," Organizational Behavior and Human Performance, 6 (January), 345-361.

Katona, George (1964), The Mass Consumption Society, New York: McGraw-Hill.

LaTour, Stephen A. and Peat, Nancy C.(1979), "Conceptual and Methodological Issues in Satisfaction Research," in Advances in Consumer Research, Vol. VI, William L. Wilkie, ed., Ann Arbor: Association for Consumer Research.

Linda, Gerald and Oliver, Richard L.(1979), "Multiple Brand Analysis of Expectation and Disconfirmation Effects on Satisfaction," in Proceedings of the Division 23 Program, 87th Annual Convention of the American Psychological Association, Ivan Rosa, ed.

Madden, Charles Stanley, Little, Eldon L. and Dolich, Ira J. (1979), "A Temporal Model of Consumer S/D Concepts as Net Expectations and Performance Evaluations," in New Dimensions of Consumer Satisfaction and Complaining Behavior, Ralph L. Day and H. Keith Hunt, eds., Bloomington, Ind.: School of Business, Indiana University.

Nunnally, Jum C. (1978), Psychometric Theory, Second Edition, New York: McGraw-Hill.

+lander, Folke (1977), "Consumer Satisfaction -- A Skeptic's View," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed., Cambridge, MA: Marketing Science Institute.

Oliver, Richard L. (1977), "Effect of Expectation and Disconfirmation on Postexposure Product Evaluations: An Alternative Interpretation," Journal of Applied Psychology, 62 (August), 480-486.

Oliver, Richard L. (1980a),"A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research, 57 (November), forthcoming.

Oliver, Richard L. (1980b),"Theoretical Bases of Consumer Satisfaction Research: Review, Critique and Future Direction," in Theoretical Developments in Marketing, Charles W. Lamb, Jr., and Patrick M. Dunne, eds., Chicago: American Marketing Association.

Olson, Jerry C. and Dover, Phillip (1979), "Disconfirmation of Consumer Expectations Through Product Trial," Journal of Applied Psychology, 64 (April), 179-189.

Swan, John E. (1977), "Consumer Satisfaction with a Retail Store Related to the Fulfillment of Expectations on an

Initial Shopping Trip," in Consumer Satisfaction, Dissatisfaction, and Complaining Behavior, Ralph L. Day, ed., Bloomington, Ind.: School of Business, Indiana University.

Swan, John E. and Trawick, I. Fredrick (1979), "Satisfaction Related to Predictive vs. Desired Expectations," Paper presented at the Fourth Annual Conference on Consumer Satisfaction, Dissatisfaction, and Complaining Behavior.

Weaver, Donald and Brickman, Phillip (1974),"Expectancy, Feedback and Disconfirmation as Independent Factors in Outcome Satisfaction," Journal of Personality and Social Psychology, 30 (March), 420-428.

Westbrook, Robert A. (1980), "Intrapersonal Affective Influences on Consumer Satisfaction with Products," Journal of Consumer Research, 7 (June), 49-54.

Westbrook, Robert A. and Oliver, Richard L.(1981), "Developing Better Measures of Consumer Satisfaction: Some Preliminary Results," in Advances in Consumer Research, Vol. 8, Kent B. Monroe, ed., Ann Arbor: Association for Consumer Research. (this volume).