Effect of Expectation and Dis Confirmation on Post Exposure Product Evaluations an Alternative Interpretation



Comments



Description

Journal of Applied Psychology 1977, Vol. 62, No.4, 480-486 Effect of Expectation and Disconfirmation on Postexposure Product Evaluations: An Alternative Interpretation Richard L. Oliver Department of Business Administration University of Kentucky Interpretations of the effect of expectation and disconfirmation on perceived product performance are reviewed. At issue is the relative effect of the initial expectation level and the degree of positive or negative disconfirmation on affective judgments following product exposure. Although the results of prior studies suggest a dominant expectation effect, it is argued that detection of the disconfirmation phenomenon may have been clouded by a conceptual and methodological overdetermination problem. To test this notion, 243 subjects responded to expectation and disconfirmation measures in a three-stage field study of reactions to a recently introduced automobile model. These measures were later related to postexposure affect and intention variables in a hierarchical analysis of variance design. Although the results support earlier conclusions that level of expectation is related to postexposure judgments, it is also shown that the disconfirmation experience may have an independent and equally significant impact. Implications of the findings are discussed. The effect of confirmation and disconfirmation of expectations on perceived product performance has received scant attention in the literature, despite an apparent relationship between expectancy disconfirmation and product satisfaction (Anderson, 1973). This is evidenced by the fact that only five studies in the last decade exist to provide a foundation for a research tradition of the effects of expectation and disconfirmation on postexposure product reactions. This article reviews these investigations and suggests a new conceptual and empirical perspective to disconfirmation effects. Theories oj Reaction to Discrepancy A number of competing explanations have been proposed by Anderson (1973) to describe the effect of expectation and confirmation-disconfirmation on perceptions of product performance. All include the implicit assumption that consumers acquire cognitive expectations of the most probable level of Requests for reprints should be sent to Richard L, Oliver, who is now at the College of Business Administration, University of Iowa, Iowa City, Iowa 52242. 480 product performance. The extent to which these expectations are met determines the perceived disconfirmation experience. Note that one's expectations will be negatively disconfirmed if the product performs more poorly than expected, confirmed if the product performs as expected, and positively disconfirrned if performance is better than anticipated. Thus, confirmation is more properly the midpoint on a disconfirmation continuum ranging from unfavorable to favorable disconfirmation. The product performance predictions made by the competing theories differ in terms of magnitude of effect placed on expectation and disconfirmation (Anderson, 1973). For example, assimilation theory (Sherif & Hovland, 1961) and dissonance theory (Festinger, 1957) would predict that postexposure ratings are primarily a function of the expectation level because the task of recognizing disconfirmation is believed to be psychologically uncomfortable. Thus, consumers are posited to perceptually distort expectationdiscrepant performance so as to coincide with their prior expectation level. In comparison, contrast theory (Sherif & Hovland, 1961) would predict that outcomes deviating from expectations will cause the subject to favor- Thus. pp. high expectation subjects in the low performance condition rated the product higher than subjects in the low expectation-low performance group. while the low expectation-high performance subjects evaluated the recorder lower than subjects whose expectations of a high quality recording were confirmed. it is argued that the assimilation and contrast predictions derived from communication theory may not be meaningful within the context of product exposure. Rather than responding favorably to an unanticipated superior performance or unfavorably to an unexpected inferior product experience. Specifically. In summary. the findings of the first study have been questioned by Olshavsky and Miller (1972. The effect of positive disconfirmation was introduced by Olshavsky and Miller (1972) in a 2 x 2 design featuring high and low levels of subject expectations and the performance of a reel-type tape recorder. Finally. the latter study confirmed Cardozo's findings that a negative disconfirmation of performance expectations resulted in lower product ratings than were obtained under accurate expectation conditions in accord with the contrast model. Mean evaluations of the highest expectancy group were equivalent to those in the accurate treatment. In these investigations. Consider the implications of the findings to date in light of the relation between disconfirmation and perceived performance. the experimental group evaluated the coffee as more bitter than expected but less bitter than indicated by the control group. Anderson (1973) investigated the im- pact of varying levels of expectation on ratings of a ballpoint pen and found that with the exception of a high expectancy extreme." In a thorough analysis of the expectation effect. whereas a contrast model would predict that performance ratings are primarily a function of the disconfirmation experience. subjects assimilated their postexposure judgments in the direction of the expectation treatment. an assimilation model would predict that product performance perceptions are a function of the expectation level. 20-21) while Cohen and Goldberg (1970) did not provide for differing expectation levels. Olson and Dover (197S) manipulated subject expectations that an unfamiliar coffee brand would be free of bitterness and then permitted both the experimental group and a control group receiving no information to taste a very bitter coffee blend. In short. the three most recent studies provide consistent evidence in favor of the predictive superiority of the assimilation model. the weight of the evidence would appear to support the assimilation model. Unlike the results of the two prior studies. Olshavsky and Miller concluded that performance evaluations "tended to be assimilated toward manipulated expectations whether positively or negatively disconfirmed (p. The . the effects of positive and/or negative disconfirmation were compared to evaluations under conditions of accurate expectations. When compared to their own pretrial expectations and the control group ratings. Although the two earliest studies found evidence for a contrast effect. 21). Cardozo (196S) found that subjects who received a pen of expected quality rated the product significantly higher than a group expecting a higher quality writing instrument. The results showed that the contrast effect is elusive.EFFECT OF EXPECTATION AND DISCONFIRMATION 481 ably or unfavorably react to the disconfirmation experience in that a negative disconfirmation is believed to result in a poor product evaluation whereas a positive disconfirmation should cause the product to be highly appraised. the ratings of a group receiving accurate expectations were higher than those of two lower expectation groups and lower than a moderately high expectation group. In response. The Evidence Five marketing studies have investigated the effects of expectation and/or disconfirmation on product evaluations. subjects appeared to distort performance to coincide with their expectations. Accordingly. whereas Cohen and Goldberg (1970) showed that subjects who received an unaltered cup of a new coffee brand rated it higher than subjects tasting the same brand with a poor-tasting additive. In two early investigations. in that high expectations are assumed to result in an unfavorable disconfirmation and low expectations in positive disconfirmation. the quietness of an automobile). Rather. rather. 1970). in part.482 RICHARD L. a design such as the following would be required: Negative Expectation disconfirmation Confirmation Low High Positive disconfirmation As noted by Weaver and Brickman (1974). Olson & Because no such design has been tested. In no study was disconfirmation measured per se. outcome. although only two can be manipulated independently as a practical matter. OLIVER conclusions of the most recent studies indicate that a positive disconfirmation results in lower ratings and a negative disconfirmation in higher ratings than would be obtained under accurate expectations—a counterintuitive proposal. The nature of the third variable has been assumed from the following postulates: 1. the studies to date can be said to have only shown that (a) expectations influence postexposure product ratings and (b) that actual product performance influences product ratings. the inferences drawn in these studies regarding the predictive superiority of an assimilation over a contrast model or of the greater effect of expectation over disconfirmation may be due to the confounding inherent in Assumptions 1 and 2. Consequently. Conditions facilitating independent effects. 1965. An outcome given high expectations results in perceived negative disconfirmation.. Weaver and Brickman took issue with this interpretation and argued that a separate disconfirmation effect may exist independent of the outcome and expectation treatments and that studies manipulating only expectation and performance may have obscured this possibility. one's reaction to a disconfirmation experience is believed to be a function of three constructs (expectation. three varied expectation only (Anderson. Effects Postexposure Evaluations in Dover. First. 1972). 2. the assumed relation between the two variables may be attenuated for two reasons. the position taken in this article is that both expectation and disconfirmation explanations are needed to fully specify the level of postexposure evaluations in much the same way that any revised attitude can be explained in terms of initial position plus degree and direction of change. one's overall assessment of many products usually involves a number of subjective attributes including those that are objective but are judged subjectively as a practical matter (e. A defense of the notion of expectationdisconfirmation independence would require that the likelihood of either a positive. An outcome given low expectations results in perceived positive disconfirmation. The Case jor Independent. This line of thought is pursued in the following discussion. zero. Specifically. and one examined expectation and performance (Olshavsky & Miller. the magnitude and direction of the disconfirmation experience were assumed from Postulates 1 and 2. one manipulated performance (Cohen & Goldberg. the product usage experience itself may serve to interfere with the retention of expec- . and disconfirmation). To independently test the effect of disconfirmation. Cardozo. 1975). Second. In the five studies cited in the preceding discussion.g. it may be inappropriate to force an assimilation or contrast interpretation on any set of findings. Consequently.. or negative disconfirmation be the same at all expectation levels. Although performance perception ratings may. because expectation formation and disconfirmation occur at separate points in time. be based on objective criteria. Note that an axiomatic negative correlation between expectation and disconfirmation is implied. Strictly speaking. 1973. research on the disconfirmation of expectations is confounded by conceptual overdeterrnination. It is argued here that the context of many overall product performance evaluations is sufficiently subjective so that this condition is satisfied from a perceptual standpoint. postexposure evaluations are likely to be highly subjective as well. Method Sample Students at a Midwestern university voluntarily participated in a study of reactions to a recently introduced automobile model. selected from research summaries provided by the manufacturer. . and finally make a number of postexposure judgments.EFFECT OF EXPECTATION AND DISCONFIRMATION 483 tation levels and. respectively. Second. a three-stage cross-sectional design was employed whereby expectation and disconfirmation data were collected without treatment manipulations. the separate disconfirmation effect posited by Weaver and Brickman (1974) may serve to mitigate the strong expectation effect predicted by the assimilation model. subjects indicated their perceptions of the car's overall quality by responding to the item "Before you inspect (test drive) the . 1974) and because the external validity of the prior findings has not been assessed. The degree of commitment to these beliefs was believed to be similar to that obtained in prior studies. its effect may be greater than that of expectation. the final sample consisted of 243 undergraduates 57% of whom were male. As all prior studies have manipulated expectation and disconfirmation experimentally.and postexposure questionnaires where positive and negative net changes were interpreted as favorable and unfavorable disconfirmation.g. the measures were collapsed to provide . the time interval may enhance forgetting. and uncomfortable-comfortable. First. then examine the car in a test drive or stationary situation. Subjects were recruited both from campus housing units and from various locations around campus. Specifically. disconfirmation was also viewed as the average change in the attribute ratings between the pre. it may be instructive to note that the relationship between expectation and evaluation over time has not been explored in a usage context.?" on a 7-point scale ranging from very poor. Because a laboratory solution to this problem is not easily executed (Weaver & Brickman. the literature provides little precedent for measuring these constructs on a selfreport instrument. if usage takes place over a period of time. in the event that one had very high expectations. As a result. . Second. In summary. do you expect it to be . inspection only).. and disconfirmation are largely subjective. a methodology is needed that tests for the disconfirmation effect independent of expectation. because aroused disconfirmation is in closer temporal proximity to the postexposure evaluation. it is also suggested that one experiences feelings as opposed to mental calculations of overall disconfirmation based on subjective perceptions of disconfirmation on the individual attributes. disconfirmation (negative. high). positive). the position taken here assumes that when expectations. "whiter than white"). no necessary relation between expectation and disconfirmation would be expected even though one's expectation level may provide a baseline for disconfirmation in an objective performance situation. Design In an effort to adapt the traditional laboratory investigation of the effect of expectancy disconfirmation to the field. whether these feelings were. Finally. in fact. Subjects were asked to rate the car in terms of their expectations on the following scale: My expectations were: Too high: It was poorer than 1 thought 8 8 8 Accurate: It was just as I had expected 8 8 8 Too low: It was better than I thought 8 8 In accord with the second measure of expectation. To address the issues raised above.. performance. After deleting the responses of seven subjects for incompleteness. a field investigation was conducted in a manner that would allow both expectation and disconfirmation to be related to postexposure evaluations. a quasi-realistic shopping experience was used in which subjects were asked first to complete a preexposure questionnaire. expectation was also viewed as the mean rating on fifteen 7-point bipolar scales measuring subject perceptions of the car's position on attribute. The data were later collapsed into a 2 X 3 X 2 unequal « factorial design consisting of expectation (low. Examples include cramped-roomy. through average. noisy-quiet. accurate or not. Measures Expectation and disconfirmation. to very good. Thus. and nature of exposure (drive and inspection. Consequently. he could still "feel" as if the product performed better than expected (e. zero. In fact. expectation was measured on the preexposure questionnaire in two ways. The first measure of disconfirmation was obtained in a straightforward manner on the postexposure questionnaire. To create distinct categories of the independent variables. higher mean evaluations were obtained in high 1 The complete hierarchical analyses of variance tables and cell sizes are available from the author. and Intention Measures Variable 1. As measured by the partial F statistics. Moreover. * p g ..28* . 5. were selected to assess the effect of expectation and disconfirmation. the expectation measures were uncorrelated with the disconfirmation variables except when the summed attribute pretest scores were used to represent expectation and the attribute change scores were used as a proxy variable for disconfirmation. To obtain a postexposure evaluation of overall affect.01 -. to certain. .39* . two levels of expectation and three levels of disconfirmation. is more likely due to the common regression effect obtained when change scores are correlated with pretest scores (Lord.10 — .35* — .1 Table 2 shows that both the expectation and the disconfirmation effects were significant. 1962). Analysis Because of the unavoidable necessity of unequal cell sizes in the design. Moreover.01 .37* . although a margin of ± . disconfirmation had a greater impact on affect regardless of how the independent variables were measured. To show that the correlational findings generalize across disconfirmation categories and exposure conditions. those who indicated that their perceptions were accurate.01. The hierarchical technique was selected in the present case because a logical a priori ordering of main effects was considered defensible on grounds of temporal precedence and research tradition. but no effort was made to equally partition the disconfirmation measures as the natural categories of negative. 1969) was required. Affect Intention Overall expectation Pretest ratings Overall disconfirmation 1 2 3 4 Results Correlations between the raw scale scores are reported in Table 1. a nonorthogonal analysis of variance (Overall & Spiegel. which shows that postexposure affect and intention were positively related to both the overall and summed attribute expectation variables as well as both the overall and attribute change disconfirmation variables. Disconfirmation. the dependent variables were positively correlated with both the expectation and disconfirmation measures.62* . however.38* . 2. This observed negative correlation. and positive disconfirmation were theoretically denned. This same procedure was used to partition the attribute ratings change continuum.53* — . Consequently.61* . the results of the hierarchical analyses of variance on postexposure affect and intention to buy using categories derived from the overall as well as the ratings scales are shown in Table 2. Because expectations are antecedent to product trial and consequently to the disconfirmation experience. subjects were divided into those who selected the three negative disconfirmation scale points. 5 6 Correlations Between Expectation. the expectation variable was given priority in the sum of squares partitioning. Two attitudinal dimensions. Net ratings change Note. Note that even with the negatively correlated ratings scales.22* — .51* . In all cases. 4. overall affect and intention to buy. and those who chose the the rightmost positive disconfirmation categories. Affect. a condition reflected in the hierarchical positioning. however. the two exposure groups were combined for purposes of presenting the cell means in Table 2. the exposure effect was given the lowest priority. In a similar manner. through SO-SO. subjects were asked to check their perceptions of the car on a 7-point scale ranging from very unappealing (1) to very appealing ( 1 ) .25 scale points around zero net change was allowed to provide the same percentage of accurate subjects that was obtained with the overall disconfirmation scale. Because both . A median split was used to divide the sample into high and low expectation groups.484 Table 1 RICHARD L. subjects indicated their purchase intention by responding to the question "If you were 'in the market' for a car. Because no main effect attributable to the nature of the product experience was significant in any analysis. whereas the effects on intention were more equally attributable to both independent variables. N = 243. Attitudinal criteria. 3. zero. OLIVER expectation and disconfirmation were assumed to operate to the same degree in the drive and inspection groups. what are the chances that you would buy a ?" on an 11-point scale ranging from zero.54* — 6.41* . the evidence to date suggests that the expectation effect is more potent than that of disconfirmation. an assumption that was not supported by the data reported here. As noted. ratings were inversely related to negative di.22** 54. To conserve space.06 5.69 3.33 .94 1. Moreover.89** 28.05 4. Confusion in interpretation of the findings appears to be linked to a reliance on Postulates 1 and 2 to infer disconfirmation.87 3.84 2.47 6. Although significant.66** 9. 1970. As the reader may recall. The upper mean scores and F ratios were obtained using the overall expectation and disconfirmation scales to form design cells.19 5.03 .71 2. *p < . **p < .24 5. the disconfirmation experience was assumed from the treatments applied. which was attributed to the one aberrant cell mean noted above. and. Similarly. when expectation was held constant and a performance manipulation was used to create disconfirmation.93 6. Consequently.76 3.21 3. The analysis of intention did yield a marginally significant Expectation X Disconfirmation effect. In fact.11 . Olson & Dover. Olshavsky & Miller.82 1.ION Table 2 Design Cell Mean Scores and Main Effect 485 F Levels Main effect F ratios Expectation Disconfirmation Exposure 27. none were significant in the analysis of the affect criterion.02 32. 1972.22** Disconfirmation level expectation level Affect Low High Intention Low High Negative 2. when expectation was manipulated while holding performance constant. in fact.05.00 Zero 3. Conversely.83 Note. Rather.24** 28. whereas the lower figures were determined using the attribute ratings to construct cells.01. prior interpretations would suggest a negative correlation between expectation and disconfirmation.06 4.47 Positive 5. in no prior study was disconfirmation measured directly even though the various measures of performance were subjective in nature. 20). no interaction F ratios are shown. This latter finding was nearly true for the intention measure with the exception of one aberrant cell.97** 1. Mean intention evaluations in the low expectation-confirmation group were slightly lower than those obtained in the low expectation-negative disconfirmation group when the overall scales were used to categorize subjects. p. this would appear to be of minor consequence in the overall interpretation of the findings.97 3.64 4. all results show that perceived performance is a positive function of expectation and disconfirmation when other factors are held constant. 1972. these findings generally support those of the four most recent studies in that the expectation effect was also significant.>confirmation and positively related to a positive disconfirmation (Cohen & Goldberg. Thus.EFFECT OF EXPECTATION AND DISCONFIRMAT.03 3. Discussion The results have shown that the disconfirmation effect implicit in the expectation theories of consumer satisfaction can be a significant predictor of postexposure affect and intention to buy and may be viewed independently of product performance expec- tations.88 4.97* 64. it is sug- . 1973. product ratings were a positive function of expectation (Anderson. 197S). Olshavsky & Miller. higher mean scores on affect were obtained in the successively higher disconfirmation groups at both expectation levels.83 5. expectation groups at all disconfirmation levels.91 1.06** 32. as they do not refute the Olson. However. however. 1965. Psychoevery disconfirmation level. and satisfaction. ticularly when the disconfirmation experience Weaver. References Anderson. Social judgment: Assimilation and contrast effects in communication considered.Lord. 311-322. D. search suggests that two stages of marketing 10. & Dover. 19S7. P.. product performance. Journal of Personality and in previous investigations and suggests that Social Psychology. 1973). 1962. R. L. Consumer expectations. The dissonance model in post-decision product evaluation. R.486 RICHARD L.. M. & Goldberg. the Cohen. Conn. In C. 1961. N. 30. Harris (Ed. M. Journal of Marketing Research. Elementary models for measuring change. New perceptions of positive disconfirmation. K. the independent disconfirmation effect is Sherif. 7. would include claims similar to "I never Olshavsky. W. I. New Haven. D. Journal of Marketing Research... concept is implicit in many dissonance-reduc. A theory of cognitive dissonance. addition to a proper positioning of the level of advertising claims (Anderson. C. The author views these findings with mixed 9. E. Implications ity of a perceived negative disconfirmation is high. 38^4. 1974. parUniversity Press. F. feedback. effort may be required to affect postusage Cardozo. P. Effects of expectation creprevious observation that higher expectations ation and disconfirmation on belief elements of congitive structure. E. Proceedings of the Sixth beget higher postexposure judgments when Annual Conference of the Association for Conproduct performance is held constant. J. feelings. This possibility has been obscured come satisfaction. & Brickman. J. In effort. Concerning least High expectations yielded higher ratings at squares analysis of experimental data. 244-249. 1976. M. Expectancy.. & Spiegel. 1973. J. B." uct quality. Problems of meaing messages aimed at current and potential suring change. & Hovland. An sumer Research. and perceived prodexpected my Maytag washer to last so long. 72. Journal of Marketing Research. Journal marketing manager may also wish to provide of Marketing Research. inspection of Table 2 shows this quite clearly: Overall. Consumer dissatisfaction: The effect of disconfirmed expectancy on perceived product From a managerial standpoint. 1969. 1970. 2.. the benefits accruing to the user and attitude change. and disconfirmation as independent factors in outis negative. 315-321. 19-21. & Miller. this reperformance. 1976 H not be strategically sound when the probabil- . 168-17S. J.3. This York: Harper & Row. a high expectation promotional strategy may Received August 13. when logical Bulletin. postpurchase information affecting consumer Festinger. C.: Yale of inflated appeals may be diminished. E. A.). An experimental study of consumer evaluations and satisfaction maximally. W. Madison: University of Wisconsin users of the manufacturer's product and Press. expectation. OLIVER gested that other avenues of thought be pursued in future studies. 420-428. 1972. R.
Copyright © 2024 DOKUMEN.SITE Inc.