Disconfirmation Effects on Consumer Satisfaction and Decision Making Processes

Ellen M. Moore, University of Georgia
F. Kelly Shuptrine, University of South Carolina
ABSTRACT - An assessment was made of the relationships among belief-expectations, attitudes, intentions, and satisfaction/dissatisfaction concerning behavior toward nonprofit organizations to replicate and further examine previous work by Oliver (1980b). Using three measurement techniques each for disconfirmation and for satisfaction, the findings suggest that Oliver's (1980b) proposed scheme does exist, but probably in a more complex model.
[ to cite ]:
Ellen M. Moore and F. Kelly Shuptrine (1984) ,"Disconfirmation Effects on Consumer Satisfaction and Decision Making Processes", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 299-304.

Advances in Consumer Research Volume 11, 1984      Pages 299-304


Ellen M. Moore, University of Georgia

F. Kelly Shuptrine, University of South Carolina


An assessment was made of the relationships among belief-expectations, attitudes, intentions, and satisfaction/dissatisfaction concerning behavior toward nonprofit organizations to replicate and further examine previous work by Oliver (1980b). Using three measurement techniques each for disconfirmation and for satisfaction, the findings suggest that Oliver's (1980b) proposed scheme does exist, but probably in a more complex model.


Recently, substantial research has been generated regarding the antecedent processes of satisfaction/dissatisfaction decisions and the consequences of these decisions. Some of the areas of consumer satisfaction/dissatisfaction that have been examined relate to produce expectations (e.g., Olson and Dover 1979). Research in this area has also looked beyond physical products and applied models of consumer satisfaction to other relevant human expectations that may be disconfirmed (e.g., Oliver 1980b). The purpose of the present study is to reexamine Oliver's (1980b) model to assess the underlying determinants of consumer satisfaction and decision making processes. Additionally, an important contribution to research in this area can be made by comparing three measurement techniques of disconfirmation and for three measurement techniques of satisfaction and by further examination of the antecedents and consequences of satisfaction decisions.


The disconfirmation of beliefs paradigm, as suggested in the satisfaction/dissatisfaction literature by Oliver (1980a 1980b; 1981) will be used as the conceptual framework for examining the determinants of consumer satisfaction and decisions to contribute to nonprofit organizations. The research development in the field on consumer satisfaction is based on works by Engel, Kollat, and Blackwell (1968), Howard and Sheth (1969), Andreasen (1977), and Day (1977) and a series of experiments (e.g., Cardozo 1965; Olshavsky and Miller 1972; Woodside 1972; Anderson 1973; Oliver 1977; 1980b).

Theories proposed to explain satisfaction decisions have bee reviewed by Oliver (1980a) and suggest that Helson's (1964) adaption level theory is an efficient explanation of satisfaction decisions. The level of expectations of product performance may be seen as an adaption level (i.e., one perceives stimuli only in relation to an adapted standard). Oliver (1980b) cited literature suggesting that expectations create a frame of reference for comparative judgments. Most researchers agree that expectations are a factor in postpurchase evaluation, but disagreement exists regarding the measures of the antecedents of satisfaction and satisfaction itself (Oliver and Linda 1981).

Satisfaction has beer. shown to be an additive function of expectations and disconfirmation because expectation level provides the reference about which disconfirmation takes place. However, limited research (Oliver 1980b; Oliver and Linda 1981) has extended the satisfaction model to actual post-exposure consumer intentions and choice behavior. Most of the literature on postpurchase satisfaction has examined consumer complaints and repurchase. Oliver (1980b) summarized the research findings of postpurchase satisfaction interacting with belief-expectations, attitudes and disconfirmation and operationalized the model to include intentions whereby:

Attitude (t1) = f (Expectations)    (1)

Satisfaction = f (Expectations, Disconfirmation)    (2)

Attitude (t2) = f (Attitude (t1), Satisfaction)     (3)

Intention (t1) = f [Attitude (t1)]     (4)

Intentions (t2) = f [Intentions( t1), Satisfaction, Attitude (t2)]    (5)

Expectations are conceptualized as predictions of product performance when consumed (based on Helson's adaptation or comparison level). Disconfirmation is based on the process of comparing perceived product performance with expectations and if performance meets, goes beyond, or falls short of expectations.

This model is used to explore consumer contributions to nonprofit organizations and to further assess the cognitive processes of consumer decision making. The present study replicates Oliver's (1980b) work to reexamine the model while using three measurement methods of disconfirmation and three measurement methods of satisfaction to further examine these relationships.



A three-stage field study was conducted using two national nonprofit voluntary health organizations involving both structured questionnaires and an experimental task. Only data from the questionnaires are utilized in this study. At the first stage of each study, subjects were mailed a questionnaire (Part I) measuring prior belief-expectations, attitudes, and intentions toward one of the two nonprofit agencies. The second questionnaire (Part II) includes measures of disconfirmation, satisfaction belief-expectations. attitudes, and future intentions.

For each organization, Part I of the questionnaires preceded and Part II of the questionnaires followed the national/ annual fund raising drive for the organization. Responses were not felt to be affected because of the number of questionnaires received by panel members, the variety of topics covered, and the length of time that evolved from the mailing of Part I and Part II.


The two nonprofit organizations chosen from the study are voluntary health organizations: the American Heart Association (AHA) and the American Cancer Society (ACS). The ACS is the top-ranked individual receiver of philanthropic support to national health agencies in 1980--ACS contributions exceeded $150 million (American Association Fund-Raising Counsel 1981). Even though heart disease is the number one cause of all deaths and two and one-half times more prevalent than cancer (American Heart Association 1982), the AHA is ranked number two (contributions exceeding $90 million).

Subjects and Response Rates

A repeat mail survey of approximately 1,545 members of a consumer panel was conducted. Panel members are located throughout two southeastern states. The panel households were selected to be representative of urban and rural household in a two state area with annual incomes in excess of $5,000. The panel is slightly upscale for the states as a whole in terms of education, job type and income when compared with available Bureau of Census averages. However, when only urban households with annual incomes in excess of $5,000 are considered, the panel profile resembles that of the population rather closely.

The questionnaires were matched for the three time periods to obtain a sample size of 246 (44% response rate) for the AHA study and 313 respondents (31.4% response rate) for the ACS study. The percentage response rate was less for the ACS study because there had been significant expansion in the number of panel members between the AHA and the ACS surveys. Of these respondents, 183 (74.4%) contributed to the AHA and 207 (66.1%) contributed to the ACS. Only the data for those subjects that contributed are relevant to this study.


Numerous attributes of the organizations and of gift-giving behavior were originally generated from a three-step elicitation process to prevent the imposition of researcher preconceptions on the dimensions of the concepts under study. First, attributes were generated from informal interviews and from the literature. Open-ended questions utilizing the free elicitation method were used to guide the interviews for general voluntary health organization attributes and specific voluntary health organization attributes. Secondly, a pretest elicitation was administered to a convenience sample of approximately 75 adults to supplement the salient attributes suggested by the literature review and the informal interviews. Last, approximately 100 people in two local neighborhoods were asked their feelings on seven-point, important-unimportant scales about twenty of the most often stated characteristics of voluntary health organizations and twenty of the most often stated characteristics about giving to voluntary health organizations. This process resulted in the use of fourteen salient dimensions to underlie consumer beliefs. A pretest on a convenience sample (e.g., faculty, graduate students, non-academic friends and neighbors) was used for each instrument which led to a number of minor changes.

The pregiving variables were expectations, attitude toward the act of giving at time 1, and behavioral intentions at time 1. Expectations were measured as the perceived belief probabilities attributed to consequences of giving to a nonprofit organization on fourteen, seven-point bipolar scales. Evaluation dimensions were assessed but not included assuming unit positive evaluations for each attribute and over time stability of attribute evaluations (Bearden and Teel 1983). Overall expectations were the sum of the fourteen expectation measures (Fishbein and Ajzen 1975; Olson and Dover 1976a; Oliver 1980b). Attitudes toward giving to the organization (Aact ) were measured using ten, seven-point semantic differential scales such as "foolish-wise", "helpful-not helpful". Behavioral intentions were measured using an overall assessment of contributing to the organization in the next twelve months using four, seven-point bipolar scales (e.g., uncertain-certain, likely-unlikely).

Postgiving belief-expectations, attitudes and intentions were measured again on scales identical to those in the initial survey. Three methods were used to operationalize the disconfirmation structure. [If a subject did not give, s/he was instructed to skip this section and answer similar questions measuring disconfirmation of nongivers. This same procedure was followed when measuring satisfaction.] These methods are noted as follows:

Disconfirmation 1: ATTRIBUTE SPECIFIC SCALE-the sum score of fourteen, seven-point, individual attribute disconfirmation scales.

Disconfirmation 2: BETTER THAN-WORSE THAN SCALE-the summary judgment of overall disconfirmation on three, seven-point, better than expected-worse than expected scales, and

Disconfirmation 3: DIFFERENCE SCALE--the difference between before and after rating judgments of summed belief expectations.

One of the earliest approaches to measuring disconfirmation was the Difference Scale (Oliver 1977; Swan 1977) and has been shown consistently to be highly correlated with satisfaction. The Better Than-Worse Than Scale has been used in a large number of different research studies (Oliver 1977, 1980b, 1981; Oliver and Linda, 1981; Westbrook and Oliver 1981; Swan and Trawick 1980; Westbrook 1980) in which the results parallel and, in some cases were more discriminating than the Difference Scale. However, it is posited that the Attribute Specific Scale may prove to be a superior measure of disconfirmation (as suggested by Oliver 1980b) since a post evaluation is made of each attribute.

Attribute satisfaction was obtained after all disconfirmation measures to insure the temporal ordering. The three satisfaction measures are as follows

Satisfaction A: MULTIPLE-ITEM SCALE-the sum score of a three-item, seven-point Likert-type satisfaction scale,

Satisfaction B: PERCENT SCALE-an eleven-point, overall assessment percent scale of satisfaction, and

Satisfaction C: DELIGHTED-TERRIBLE SCALE-a seven-point delighted-terrible scale of satisfaction.

The Multiple-Item Scale has been recently used in satisfaction research (e.g., Oliver 1980b, Bearden and Teel 1983) and is thought to be preferable because it better captures the construct of concern than a single item. However, the Percent Scale is more predominate because it is a single-item measure and simple to use. Westbrook (1980) provides evidence from three separate studies with durable goods and banking service which suggest the Delighted-Terrible Scale may be a superior measure. The Multiple-Item Scale is posited to be a better measure because it is presumed to be a more complete representation of satisfaction.

Oliver (1980b) used Disconfirmation 2 and Satisfaction A in the study being examined. The three measures of disconfirmation and of satisfaction were used to provide confirmation of previous research while determining which scale may better reflect the construct.

Data Analysis

A series of regression equations with variables arrayed in order of their suggested temporal sequence was used to examine the cognitive model of satisfaction for those individuals who have given to the AHA or the ACS. This method is designed to test the interrelationships among expectations, disconfirmation, satisfaction, attitudes and intentions and is parallel to the analyses used by Oliver (1980b).

The system of tested equations arrays the variables in order of their suggested temporal precedence. A recursive system (i.e., "no two variables are reciprocally related in such a way that each affects and depends on the other, and no variable 'feedback' upon itself through an indirect concatenation of causal linkages however circuitous," Duncan 1975, p. 25) was chosen for three reasons: (1) the path goal coefficients are unique in that only one solution to the estimates is possible, (2) the test is considered to be a fairly stringent analysis of a temporally ordered system in that "troublesome" paths cannot be eliminated a prior, and (3) from a heuristic standpoint, some evidence attesting to the nature of the adaption level may emerge (Oliver 1980b, p. 463). A caveat is in order regarding the possible effects of multicollinearity. As in other attitudinal studies, significant intercorrelations were found between a number of the Predictor variables. [Variable intercorrelational matrices for both the AHA and the ACS groups were too extensive for reproduction. The most problematic interrelations were between the Difference Scale of disconfirmation and expectations.]


Reliability estimates (coefficient alpha), means and standard deviations for the multiple-item measures for both studies are provided in Table 1.



The mean scores and standard deviations represent positive evaluations and variability for each scale. In general, the results of the computed reliabilities show high internal consistency for the scales.

The results obtained when the variables were entered into the regression analysis are shown in Table 2. [Data were also collected on the expectations measures in the second time period. The average correlations across the two studies between this measure of performance beliefs and follow-up overall measure of disconfirmation, attitudes, and intentions were .35, .77, and .51, respectively.] The coefficients for the first equation in each pair represent data for the American Heart Association (AHA) while the data in the second equation of each pair represent the American Cancer Society (ACS).



In examining the results of the pregiving variables, attitude is a function of expectations, and intention is a function of expectations and attitudes. These results were consistent for both organizations. Further analysis, however, reveals contrary findings to what Oliver (1980b) found in his study for the disconfirmation equation. Evident for both agencies, disconfirmation is not independent of the pregiving measures. Specifically, expectations are significantly related to disconfirmation with the exception of the Better Than-Worse Than Scale for the ACS. [The negative coefficient for expectation in the Difference Scale equation (Disconfirmation 3) is a result of a "squeezing effect" or the relative change in difference scores over time. Consider computation is


If a bi(t1) is lower (unfavorable), for example, scaled at -3, than at time 2 if it increases in favorability, say to +1, then the absolute difference is 4. If another b (tl) score is higher (more favorable), for example, scaled at +1, than at time 2 if it increases in favorability to +3, then the absolute difference is 2. Therefore, when expectations are at a low value at time 1, there is a probability of a larger difference score at time 2. And when expectations are at a higher value at time 1, then there is probability of a smaller difference score at time 2. So, the negative coefficients in the Difference Scale equation follow--as expectations at time 1 increase, disconfirmation at time 2 tends to decrease.] Also, the preexposure attitude for the Attribute Specific Scale for the AHA and the Difference Scale for the ACS is significantly related to disconfirmation. An explanation for these findings will be presented in the discussion section.

Satisfaction, in turn, was a function of disconfirmation and a linear combination of pregiving variables for each method with both agencies. Overall, this is true for each measure of satisfaction. Attitude appears to be a main determinant of satisfaction with the Multiple-Item Scale for both agencies, but for the ACS only with the Percent Scale and no relation with the Delighted-Terrible Scale. The expectation measure with the Difference Scale of disconfirmation is highly significant for both agencies with each measure of satisfaction. Overall, the disconfirmation measure as hypothesized seems to provide the most impact on satisfaction for both agencies with each method.

Examining postgiving attitude indicates, as hypothesized, that satisfaction (each method) is a major determinant. However, disconfirmation is also significant for both organizations with each method. Pregiving attitude, while not the main determinant, is significantly related, also as hypothesized, to postgiving attitude. Additionally, expectations with each method of satisfaction are very significant determinants when using the Difference Scale of disconfirmation.

Postgiving intentions are affected by postgiving attitudes, satisfaction (Multiple-Item and Delighted-Terrible Scales) and preexposure intentions. These results were consistent for both agencies with each disconfirmation method. Unlike the Multiple-Item and Delighted-Terrible Scales, the Percent Scale of satisfaction was not significant for either agency. Similar to Oliver's (1980b) results, postgiving intention yielded a negative pregiving attitude coefficient which may be explained by suppressor effects (Darlington 1968; Cohen and Cohen 1975).


The before and after analysis using a series of regression equations provide additional support of the findings of earlier studies, e.g., Swan (1977) a-d Oliver (1977; 1980b). Additional variables, however, are consistently significant --not as hypothesized -- and suggest that the proposed theoretical scheme of Oliver's (1980b) may not be a complete representation of the cognitive processes. Contrary to Oliver's posited model, disconfirmation (each method) was significantly related to postgiving attitude with each method of satisfaction. The results also suggest that expectations are significantly related to disconfirmation indicating that it is not independent of the pregiving variables. Oliver measured disconfirmation using overall better than-worse than expected scales (perceived actual performance is not recognized as a measurable construct).Disconfirmation is measured by these scales as a linear combination of prior expectations and perceived performance. Consequently, prior expectations are imbedded in the disconfirmation measure. For example, x (expectations) are additively related to the sum of x and y (perceived performance) to obtain x + v (disconfirmation), and it follows that x and (x + y) will be correlated because x is contained in both variables. Therefore, it is surprising that Oliver found no preexposure measure correlated with disconfirmation.

Using the Difference Scale of disconfirmation resulted in the most significant influence of expectations on any of the three satisfaction measures. This result was also evident for expectation's influence on postgiving attitude. Overall, the coefficients for the Difference Scale are greater than the coefficients obtained with the Attribute-Specific Scale and the Better Than-Worse Than Scale. Consequently, the Difference Scale of disconfirmation is consistent with past research for the expectation measure in the satisfaction equation, while the Attribute-Specific measure, operated contrary as posited. Thus, the Difference Scale appears to better reflect the construct than the Attitude-Specific Scale and the Better Than-Worse Than Scale.

Similarly, the three measures of satisfaction provided fairly consistent results with one main exception. The Percent Scale was not significantly related to postgiving intentions as the Multiple-Item and Delighted-Terrible Scales. The data indicated the coefficients using the Multiple-Item Scale of satisfaction are greater than the coefficients for the Percent Scale or the Delighted-Terrible Scale. Therefore, the Multiple-Item Scale appears to represent the construct better than the Percent Scale or the Delighted-Terrible Scale and operates as Posited.

In summary, the cognitive structure as depicted by Oliver (1980b) may not be complete. Belief-expectations are shown to be related to postgiving attitudes and pregiving attitudes are related to satisfaction. In addition, belief-expectations are related to pregiving intentions and disconfirmation is related to postgiving attitudes. Oliver (1980b) found support for this latter relationship in his research but has suggested that expectations, attitudes, and intentions may be highly correlated and are part of the dimensions of an overall effective attitude measure. common response bias could also be a contributing factor.

From the results of this study, careful consideration should be given to the impact of belief-expectations in describing the individual's decision making process in the nonprofit setting (as well as similar relationships in the profit-making sector). Additional studies in the nonprofit and profit sector should be conducted to test this cognitive motel. Further, the estimation procedure used in this and Oliver's (1980b) study assume no measurement error. In future research, it would be worthwhile to examine measurement error on the structural model using a latent variable structural modeling method. Also, it appears each of the three sets of measures purport to tap the same underlying construct, although Disconfirmation 3 and Satisfaction A may be better. Thus, the differences tapped should be accounted for in a measurement model and separated from structural model error. Structural equation analysis (e.g. LISREL) would enable a better test of the alternative measured

Performing a second study as a systematic replication of the first did provide some strength for the study. Similar results suggest the findings may be generalizable. The study was limited by its focus on only two nonprofit health agencies. The study was limited by its focus on only two nonprofit health agencies. Additionally, a caveat is in order acknowledging the possibility of testing effects in any before and after study.

Existing profit sector marketing techniques have been successfully applied again to the nonprofit sector. This should increase nonprofit management's acceptance and knowledge of the potential value of marketing techniques for nonprofit agencies. Also of interest to management is the implication that an individual's expectations influence his/her attitudes about giving to the organization. This, in turn, affects intentions to give. Consequently, it is important that favorable belief expectations and attitudes exist to generate positive behavior.


American Heart Association (1982), Heart Facts, National Center, Dallas, TX.

American Association of Fund-Raising Counsel, Inc. (1981), Giving USA: 1981 Annual Report, American Association of Fund-Raising Counsel, Inc., New York, NY.

Anderson, Ralph E. (1973), "Consumer Dissatisfaction: The Effect of Disconfirmation 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, 11-35.

Bearden, William O. and Jesse E. Teel (1983), "Selected Determinants of Consumer Satisfaction and Complaint Reports," Journal of Marketing Research, 20 (February), 21-28.

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

Cohen, Jacob and Patricia Cohen (1975), Applied MultiPle Regression/Correlation Analysis for the Behavioral Sciences. New York John Wiley and Sons. 73-122.

Darlington, Richard B. (1978), "Multiple Regression in Psychological Research and Practice," Psychological Bulletin, 69 (March), 161-182.

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, 153-183.

Duncan, Otis Dudley (1975), Introduction to Structural Equation Models, New York: Academic Press.

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

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

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

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

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), "Theoretical Bases of Consumer Satisfaction Research: Review, Critique, and Future Direction," in Theoretical Development in .Marketing, Charles W. Lamb, Jr. and Patrick H. Dunne, eds. Chicago: American Marketing Association, 206-210.

Oliver, Richard L. (1980b), "A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research, 17 (November), 460-469.

Oliver, Richard L. (1981), "Measurement and Evaluation of Satisfaction Processes in Retail Settings," Journal of Retailing, 57 (Fall), 25-48.

Oliver, Richard L. and Gerald Linda (1981), "Effect of Satisfaction and Its Antecedents on Consumer Preferences and Intention," in Advances in Consumer Research, Vol. 8, Kent B. Monroe, ed. Ann Arbor, MI: Association for Consumer Research. 88-93.

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

Olson, Jerry C. and Philip A. Dover (1979), "Disconfirmation of Consumer Expectations Through Product Trail," 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, IN: Indiana University School of Business, 10-17.

Swan, John E. and I. Frederick Trawick (1980), "Satisfaction Related to Predictive vs. Desired Expectations," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior, H. Keith Hunt and Ralph L. Day, eds. Bloomington, IN: Indiana University School of Business, 7-12.

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

Westbrook, Robert A. and Richard Oliver(1981), "Developing Better Measures of Consumer Satisfaction: Some Preliminary Results," in Advances in Consumer Research, Vol. 8, Kent B. Monroe, ed. Ann Arbor, MI: Association for Consumer Research, 94-99.

Woodside, Arch G. (1972), "Positive Disconfirmation of Expectations and the Effect of Effort on Evaluation," Proceedings: 80th Annual Convention of the American Psychological Association, 7 (Part 2), 743-744.