ROLPH E. ANDERSON*

Four psychological theories ore considered in determining the effects of dis-confirmed expectations on perceived product performance ond consumer satis-faction. Results reveal that too great a gap between high consumer expectationsand actual product performance may cause a less favorable evaluation of a

product than a somewhat lower level of disparity.

Consumer Dissatisfaction: The Effect ofDisconfirmed Expectancy onPerceived Product Performance

CONSUMER DISSATISEACTION

It seems incongruous that consumerism could havebecome such a powerful prevailing force when the mar-keting concept, i.e., "customer satisfaction at a profit,"has been so highly publicized as a successful businesscredo since the early 195O's. To date, most of what hasbeen written about the consumer movement, its un-derlying eauses, and the appropriate response for busi-ness has been based on little more than speculation. Toreally understand the underlying reasons for consum-erism, one must follow Alderson's edict to set upfalsifiable propositions or testable hypotheses concern-ing the problems, then obtain empirical evidence to sup-port or refute these hypotheses [1]. Today's consum-erism is such a complex force—so interrelated withother ecological, social, political, ethical, economic, andtechnological problems—that, at this stage, it ean bestudied empirically only one step at a time. Probablythe most fundamental question being raised about con-sumerism is: What are the sources of consumer dissat-isfaction?

No satisfactory literal definition has yet been devel-oped for consumer satisfaction or dissatisfaction in theliterature of marketing. The Random House Dictionarystates: "dissatisfaction results from contemplating whatfalls short of one's wishes or expectations. . . ." Con-sumer dissatisfaction, then, might be measured by thedegree of disparity between expeetations and perceivedproduet performance.

* Rolph E. Anderson is Associate Professor of B"usiness Man-agement at Old Dominion University.

There is much conflicting evidence in the journals ofpsychology regarding the effects on individuals of dis-confirmed expectancies, and this critical question fordesigning promotional mixes has been virtually ignoredin the marketing literature. Only two experiments (withconflicting results) have been published to date [4, 19].There may be significant policy implications for qualitycontrol, price, promotion, and other elements of themarketing mix depending upon whether consumer ex-pectations are too higli, produet performances too low,or promotional messages misplaced, i.e., aimed at gen-erating an inappropriate level of expectations. Cor-porate promotional mixes may be ereating unrealisticexpeetations for produets which result in consumer dis-satisfaction upon purchase and use of the products.Buskirk and Rothe go so far as to say: "It is this senseof frustration and bitterness on the part of consumerswho have been promised much and have realized less,that may properly be called the driving force behindconsumerism" [3, p. 62].

THEORETICAL MODELSIn predicting the effects on product evaluation and

customer satisfaction of disparity between expeetationsand actual or objective product performance, at leastfour psychological theories may be considered, namely:(1) cognitive dissonance (assimilation), (2) contrast,(3) generalized negativity, and (4) assimilation-contrast.

Dissonance or assimilation theory posits that any dis-erepaney between expectations and produet perform-anee will be minimized or assimilated by the consum-er's adjusting his perception of the produet to be moreconsistent (less dissonant) with his expectations.

38

Journal of Marketitig Research,Vol. X (February 1973}, 38-44

CONSUMER DISSATISFACTION: EFFECT OF DISCONFIRMED EXPECTANCY 39

Contrast theory assumes that the customer will mag-nify the difference between the product reeeived and theproduct expected; i.e., if the objective performance ofthe product fails to meet his expectations, the customerwill evaluate the product less favorably than if he hadno prior expectations for it. Contrast is thus the con-verse of assimilation.

The generalized negativity thesis is that any discrep-ancy between expectations and reality results in a gen-eralized negative hedonic state, causing the product toreceive a more unfavorable rating than if it had coin-cided with expectations. Even if the product's perform-ance exceeds the customer's expectations, it will beperceived as less satisfying than its objective perform-ance would justify.

Finally, the assimilation-contrast approach maintainsthat there arc zones or latitudes of acceptance and re-jection in consumer perceptions. If the disparity be-tween expectations and product performance is suf-ficiently small to fall into the consumer's latitude ofacceptance, he will tend to assimilate the difference byrating the product more in line with expectations thanits objective performance justifies. However, if the dis-crepancy between expectations and actual product per-formance is so large that it falls into the zone of rejec-tion, then a contrast effect comes into play and theeonsumer magnifies the perceived disparity between theproduct and his expectations for it.

Which theory is correct? What is the effect on con-sumer product perception and satisfaction when ex-pectancies are disconfirmed? To answer this question,it is necessary to review these theories in more depth.

Cognitive Dissonance (Assimilation)

An unconfirmed expectancy, according to Festinger'stheory of cognitive dissonance, creates a state of dis-sonance or "psychological discomfort" because the out-come contradicts the cotLsumer's original hypothesis[13]. The theory suggests that an individual has cognitiveelements (or "knowledges") about his past behavior, hisbeliefs and attitudes, and his environments [20], Con-sumers continually receive various kinds of product in-formation from their own experience, associates, adver-tisements, and salesmen. These bits of information arecognitions which consumers like to have consistent withone another [15]. When an individual receives two ideaswhich are psychologically dissonant, he attempts to re-duce this mental discomfort by changing or distortingone or both of the cognitions to make them more con-sonant. The stronger the cognitive dissonance, the moremotivated he is to reduce dissonance by changing thecognitive element [2].

As applied to marketing, if there is a disparity be-tween expectations for a product and the objective per-formance of that product, the consumer is stimulatedto reduce the psychological tension generated by chang-ing his perception of the product to bring it more into

Figure 1THEORIES OF DISCONFIRMATION OF EXPECTATIONS

line with his expectations. Therefore, if this propositionis true, the promotional mix for a product should sub-stantially lead expectations above product performanceto obtain a higher consumer evaluation or perceptionof the company's product. This concept, illustrated inFigure 1 by the dotted line, shows that perceived productperformance is always between objective performanceand expectations, except when all three coincide. Con-siderable controversy and some disaffection with thetheory of cognitive dissonance have developed in re-cent years due to the accumulation of an increasingamount of contradictory evidence [8, 12, 17, 21]. Onemajor criticism is that the theory assumes that the indi-vidual, instead of learning from his purchasing mistakes,actually increases the probability of making them againthrough his efforts to reduce post-purchase dissonanceby justification and rationalization of his decisions [9].

Contrast \

Even in the studies supporting assimilation theory,some individuals tend to shift their attitudes and evalua-tions away from expectations aroused by communica-tions if inconsistent with reality [16]. When expectationsare not matched by actual product performance, con-trast theory presumes that the surprise effect or con-trast between expectations and outcome will cause theconsumer to exaggerate or magnify the disparity.

Contrast theory would predict consumer product per-ceptions as shown by the dashed line in Figure 1. Thistheory suggests that slight understatement of the prod-uct's qualities in advertising might lead to higher cus-tomer satisfaction with the company's product. Ofcourse, the advertisement could not so understate theproduct's qualities that customers bypass it for an-other brand. Several studies tend support to the possiblesuccess of this promotional strategy flO. 14. 16, 22, 23,241. One of the two marketing studies reported to dateprovided some support for contrast theory albeit find-ings were divided into support of assimilation theory as

40 JOURNAL OF MARKETING RESEARCH, FEBRUARY 1973

well [5]. In attempting to reconcile the difference be-tween assimilation theory and contrast theory, Cardozointroduced another variable, shopping effort, into thedecision process. Specifically, he found that customerproduct evaluation or satisfaction is lower when theproduct does not measure up to expectations, but satis-faction rises as effort expended to obtain the productincreases. Unfortunately, Cardozo's results may havebeen affected by a methodological error since his ex-perimental subjects rated the product (ballpoint pens)based on expectations generated from leafing througheither high (average price was $1.95) or low (averageprice was $.39) price product catalogs. Subsequentevaluations on a scale of 0 to 100 as compared to theproducts in the catalog were not comparable since in-dividual subject rating criteria shifted depending uponthe catalog assigned [4, p. 246]. For example, a ratingof 50 in the high-expectation condition indicated aver-age quality for a $1.95 pen, but this same rating wouldalso indicate average quality for a $.39 pen in the low-exjx^ctation condition.

Generalized Negativity

A classic study of the consequences of unconfirmedexpectations was conducted by Carlsmith and Aronson[7]. In a test of their hypothesis that any disconfirma-tion of an expeeted result will be perceived as lesspleasant or less satisfying than if the expectancy hadbeen confirmed, they asked individuals to taste bitterand sweet solutions, manipulated their expectations re-garding the tastes, and measured the ratings under thevarious conditions. It was assumed that bitter was anunpleasant taste and sweet was a pleasant taste for themajority of subjects. The Carlsmith and Aronson find-ings seem to suggest opposing theories. When the sweet

solution was expected and the bitter solution tasted, adisconfirmed expectancy resulted in a rating of more bit-ter which would support contrast theory. On the otherhand, when the bitter solution was expected but thesweet solution came up, a disconfirmed expectancy re-sulted in a rating of less sweet or assimilation towardthe expected taste in support of assimilation theory.Carlsmith and Aronson explain this apparent confiict byarguing that any disconfirmed expectancy results in ahedonically negative state which is generalized to ob-jects in the environment. Thus, one can make the follow-ing prediction: If a customer expects a particular per-formance from a product but a different performanceoccurs, he will judge the product to be less pleasant thanif he had no previous expectancy.

The Carlsmith and Aronson explanation implies thatpromotional claims aimed at target customers shouldseek to create expectations which are consistent withactual product performance. In Figure 1, the theory ofgeneralized negativity is depicted by the line of alternat-ing dots and dashes. Note that only when expectationsand product performance coincide is the consumer'sevaluation of the product as favorable as its objectiveperformance.

A ssimilation-Contrast

A final theory for consideration in attempting to pre-dict the effects on consumer satisfaction of disparitiesbetween expectations and objective product perform-ance is assimilation-contrast. As its name implies, itcombines the theories of assimilation and contrast. Workby Hovland, Harvey, and Sherif provides support for thecontention that product performance differing onlyslightly from one's expectations tends to result in dis-placement of product perceptions toward expectations

Table 1PREDICTION MATRIX FOR HYPOTHESES

Treatments^

Ti (with take-measure)Null hypothesisAssimilationContrastGeneralized negativityAssimilation-contrast

Ti (without take-measure)Null hypothesisAssimilationContrastGeneralized negativityAssimilation-contrast

NoneCa

StandardStandardStandardStandardStandard

StandardStandardStandardStandardStandard

Very lowCl

StandardVery lowVery highVery lowVery high

StandardVery lowVery highVery lowVery high

Conditions^

LowCa

StandardLowHighLowLow

StandardLowHighLowLow

Accurate

c,

StandardOnOnOnOn

StandardOnOnOnOn

High

c.

StandardHighLowLowHigh

StandardHighLowLowHigh

Very high

c»

StandardVery highVery lowVery lowVery low

StandardVery highVery lowVery lowVery low

' Conditions consist of various levels of expectancies generated by product information.^ Treatments consist of the presence or absence of a "take-measure" in the form of a questionnaire regarding expectancies gen-

erated following receipt ol" product information.

CONSUMER DISSATISFACTION: EFFECT OF DISCONFIRMED EXPECTANCY 41

(assimilation effect), while large variances between one'sexpectations and actual product performance tend tobe exaggerated (contrast effect) [16]. The theory as-sumes that individuals have ranges or latitudes of ac-ceptance, rejection, and neutrality. Whether assimila-tion or contrast effects develop is a function of therelative disparity between expectations and actual prod-uct performance.

Assimilation-contrast theory suggests that promo-tional messages should create expectations for the prod-uct as high as possible without creating a level of dis-parity between expectations and objective performancewhich falls outside the consumer's range of acceptance.In accord with assimilation-contrast theory, consumerperceptions of product performance would take theform of the S-shaped curve in Figure 1.

HYPOTHESES

In order to discover which, if any, of the above out-lined theories best describes the true relationships be-tween these important consumer variables, five hypoth-eses were tested, as follows:

1. Null Hypothesis—product perceptions are notsignificantly different for various levels of expecta-tions.

2. Assimilation—product perceptions will vary directlywith the level of expectations.

3. Contrast—product perceptions will vary inverselywith the level of expectations.

4. Generalized Negativity—product perceptions willalways be negative when there is disparity betweenexpectations and actual product performance, andthe degree of negativity will vary directly with theamount of disparity.

5. Assitnilation-Contrast—product perceptions willvary directly with expectations over a range aroundactual performance, but above and below thisthreshold, product perceptions will vary inverselywith the level of expectations.

METHODOLOGY

A 2 X 6 factorial research design was used to test thefive hypotheses. Manipulation of the independent vari-able (expectations) was accomplished by randomly as-signing subjects to one of five different levels of per-suasive product information, or no product information.As verified in pretesting, condition one (Ci) substantiallyunderstated the product's features, C2 slightly under-stated the features, C;̂ described the product accurately,C4 slightly overstated the quality of the product's fea-tures, and Cfl substantially overstated its features. Sub-jects (Ss) in Co were given no product information, butinstead they received a communication unrelated to theexperiment. In addition, a "take-measure" in the formof a self-administered questionnaire was given, afterpresentation of product information but before seeingthe product, to half the Ss in each condition to ensure

that expectations were being created in the desired di-rection and intensity. The basic experimental designand prediction matrix for the experiment hypothesesare provided in Table 1.

Evaluations of identical, unmarked ballpoint pensselling at retail for about $1.00 each were obtained ingroup administration from 144 students enrolled in anundergraduate marketing course. To increase involve-ment, Ss were told that they could keep the particularpen randomly assigned them for evaluation, whether itturned out to be one of the expensive pens or not. Sswere permitted to inspect and test the product for thesame length of time, then record their reactions on amodified logarithmic product rating scale anchored indollars and cents distributed in small ranges from $.04to $64.00, as illustrated in Figure 2. (To faciUtate anal-ysis, these ratings were later converted to integers bysequentially numbering the rows of the product ratingscale.) Ss rated the ballpoint pen on 15 visual featuresand performance characteristics. The unweighted meanof these ratings by each S constituted the first dependentvariable. The second dependent variable consisted ofan overall rating by Ss on the pen's combined charac-teristics. This was a weighted mean since each S wasable to assign certain product features more weight

Figure 2

PRODUCT RATING SCALE

ON THIS PEN,EACH OF THESECHARACTERISTICSIS ABOUT LIKETHE ONES OF PENSCOSTING BETWEEN

Si.04/*

.06

.!»

.12

.19

.25

.37

.50

.75

$ 1.00

1.50

2.00

3.00

4.00

£.00

e.oo

12.00

16.00

24.00

32.00

46.00

64.00

EXACTLY HOW MUCH DO YOU THINK THIS flALlPOINI PEN IS WCWTH7 J

42

than others in determining his overall evaluation. Lastly,Ss estimated the ballpoint pen's price to obtain the thirddependent variable.

RESULTS

The mean responses by condition and treatment forthe experiment are presented in Table 2. Manipulationof the independent variable (product information) cre-ated expectations in the desired direction with signifi-cant differences of intensity. Mean scores for the de-pendent variables (product ratings) show productperceptions are in the direction of expectations untilmanipulation at the "very high" level, which caused areversal and downturn in product ratings for all threemeasurement variables.

On each dependent variable, Ss gave the product amore favorable evaluation when it was accurately de-scribed in C;i than when no product information wasprovided in Cn. Duncan's New Multiple Range Test[11] revealed that there was a significant difference atthe .01 level between mean scores for C3 and Co- Re-sults of the one-way analysis of variance (treatmentvariable collapsed since Ss were not sensitized by the"take-measure") indicated that Ss responded with sig-nificant differences in their evaluations or perceptionsof the product depending upon their level of expecta-tions, as can be seen in Table 3.

Table 2MEAN RESPONSE BY CONDITION AND TREATMENT

Manipulation

ExpectationsDependent variable"

A; B

C

Product ratingsTreatment l̂ -

ABC

Treatment 2''ABC

Combined"ABC

CiVerylow

2.973.253.67

6.436.587.00

6.226.586.92

6.336.586.96

GLow

4.314.255.42

7.227.837.92

6.847.177.92

7.037.507.92

CoNone

7.248.427.50

7.398.008.17

6.967.338.50

7.177.678.33

CsAccu-rate

8.548.929.58

8.569.009.58

8.949.259.58

8.759.139.58

c,High

11.7012.4212.08

9.219.75

10.00

10.0510.5010.75

9,6310.1310.38

CsVeryhigh

15.9216.6717.08

9.009.259.83

8.388.67

10.08

8.698.969.96

Dependent variable A = Product features, B = Combinedcharacteristics, C = Price.

^ Treatment 1 = Take-measure; Treatment 2 = No take-measure.

"Treatment variable collapsed.

JOURNAL OF MARKETING RESEARCH, FEBRUARY 1973

Relating Experimental Results to Theoretical Models

Data for each of the three dependent variables wereplotted by each of the six conditions. Inspection of theplots indicated conformity with assimilation theory un-til reaching C5, which showed a decline in product rat-ings for all three dependent variables in accord with theassimilation-contrast theory of disconfirmation of ex-pectations.

The tests for linearity in Table 4 showed deviationsfrom linearity to be highly significant for the relation-ship between expectations and perceived product per-formance for each of the dependent variables. Signifi-cant deviations from linearity discounted both the nullhypothesis and the possibility of assimilation theory de-scribing the data. The positive slope of the plotted dataruled out contrast theory which demands a negativelysloped relationship between expectations and productperceptions. Finally, the deviation from linearity orkink in the curve occurred at the wrong point to supportthe Carlsmith-Aronson hypothesis of generalized nega-tivity. Therefore, the data best fit assimilation-contrasttheory, since product ratings were assimilated towardexpectations until the "very high" condition when con-trast effect began, causing a downturn away from ex-pectations in evaluations of the product.

CONCLUSIONS

The experiment revealed that there is a point beyondwhich consumer will not accept increasing disparity be-tween product claims and actual performance, at leastfor certain relatively simple or easily understood prod-ucts. When this threshold of rejection is reached, con-sumers will perceive the product less favorably than at aslightly lower level of expectations. More complexproducts, where there is considerable ambiguity and un-certainty in making judgments, may yield different re-sults as consumers may tend to be more dependent onthe information provided them. Olshavsky and Millerinvestigated the effects on product evaluations of bothoverstatement and understatement of product quality fora reei-ty[>e tape recorder and found assimilation theorysupported in each case [19].

Product Commitment

Contrast effects did not appear in C,, "very low" ex-pectations. This result may be partially explained by"floor effects" which prevented manipulation of expec-tations far enough below the relatively low cost item (adollar ballpoint pen) to cause sufficient surprise or ex-hilaration upon seeing and trying the product. Due toinsufficient commitment to low-priced products, it sim-ply may not be possible to obtain contrast effects atvery low expectation levels. Freedman has shown thatpersonal involvement or commitment to the product isa determinant of the level of disparity necessary to pro-duce maximum perceptual change [14]. With strong

CONSUMER DISSATISFACTION; EFFECT OF DISCONFIRMED EXPECTANCY 43

commitment, even a small disparity between expecta-tions and product performance may fall outside theconsumer's range of acceptance.

The failure of an upturn to occur in Ci at the lowend of the range of expectations is not sufficiently rele-vant to marketers to merit much discussion since noadvertiser would drastically disparage his product if hewished to stimulate a high volume of sales. Though re-peat purchasing rates might be high, once consumerstried the product, no such negative promotional strategywould be likely to achieve the desired initial trial rate.

A tnount of Information

Ratings in C3 were consistently higher than ratingsin Co, thus another important conclusion from the ex-periment is that a more favorable evaluation is obtainedwhen a product is accurately described than when littleor no product information is provided. This finding sup-ports Cardozo's suggestion that the mere processing ofinformation may lead to a more favorable evaluation ofthe product, not only because customers have greaterknowledge on which to base evaluation, but also becausethe processing of information about products constitutesa form of commitment to the products [6]. It wouldappear that marketers who provide more relevant in-formation about their products will generate higherevaluations and customer satisfaction for their productsthan those marketers who depend on persuasive mes-sage content that provides only minimal information.Of course, the amount of information customers maywillingly process without boredom or confusion nodoubt varies widely among individuals and products.But, since consumers perceive selectively in some or-ganized fashion, it is probably better to provide toomuch product information than too little. One studyfound that when marketers provide insufficient productinformation for consumers, they merely encourage orforce consumers to seek information about the productfrom other sources, such as friends, associates, inde-pendent rating organizations, consumer groups, govern-ment agencies, and competitors [18].

IMPLICATIONS FOR MARKETING

Results of the experiment have important implica-tions for positioning the level of promotional claims.Assuming that consumer dissatisfaction is a function ofthe disparity between expectations and perceived prod-uct performance, unrealistic consumer expectations gen-erated by excessive promotional exaggeration can resultin consumer dissatisfaction. Since consumer expectationsapparently affect satisfaction with a product, the mar-keter who wishes to understand and favorably influencecustomer satisfaction with his ofTcring may be able to doso by understanding and influencing customer expecta-tions. That is, if consumer expectations are a functionof promotion, and customer satisfaction is a function of

Table 3ANALYSIS OF VARIANCE: PRODUCT PERCEPTIONS

(TREATMENT VARIABLE COLLAPSED)

Sources

Product featuresBetween conditionsWithin conditions

Combined characteristicsBetween conditionsWithin conditions

PriceBetween conditionsWithin conditions

d.f.

5138

5138

5138

MeanSquares

38.773.85

40.464.34

42.283.82

F-ratio

10.06^

9.31»

n.08»

» p < .01.

expectations, then satisfaction may be considered a func-tion of promotion.

Some consumers may have unrealisticaDy high ex-pectations for product performance even without theadded boost of promotional claims. Distribution of un-realistic expectancies among consumers may stem partlyfrom widespread faith in the achievements and possi-bilities of research and technology, resulting in the feel-ing that products can and should be made to performflawlessly. As history demonstrates, standards and per-formance expectations for all categories of products(from automobiles to clothing) tend to steadily rise andpeople become increasingly less tolerant of product de-ficiencies.

Enlightened company executives might profitably re-assess the often implicit sales-oriented view that thehigher the level of consumer product expectations stim-ulated by promotion the better because the primary ob-jective of the promotional mix is to sell the product.This attitude can be contagious among competing com-panies. However, the accumulation of consumer dissat-isfaction may eventually erupt in a demand for moreconsumer protection legislation, resulting in a more re-strictive marketing environment.

EUTURE RESEARCH

As indicated by the Olshavsky and Miller findings, theapplicability of the different theories of expectationsmay vary across product classes [19]. More researchneeds to be conducted with a variety of products andservices. Disconfirmation of expectations for productsfor which consumers make deep personal and financialcommitments may have substantially different effects onconsumer perceptions of performance than less personal,lower cost, and less ego-related goods. A logical exten-sion of the present study and previously reported ex-periments is consideration of time. Does the consumerbecome more objective in evaluating products he haspurchased as he gathers information and feedback fromvarious sources, or does he become more committed to

JOURNAL OF MARKETING RESEARCH, FEBRUARY 1973

Table 4DEVIATION FROM LINEARITY: RELATIONSHIP BETWEEN

EXPECTATIONS AND PRODUCT PERCEPTIONS

6. -. "Customer Satisfaction: Laboratory Study and Mar-

' Sources1

Product featuresBetween groupsLinear regressionDeviation from lin-

ear regression(DJ

Error term (E)D/E

Combined characteristicsBetween groupsLinear regressionDeviation from lin-

ear regression(D)

Error term (E)D/E

PriceBetween groupsLinear regressionDeviation from lin-

ear regression(D)

Error term (E)D/E

d.f.

413

115

413

115

413

115

Sums ofsqtiares

177,19117.6559,54

189,75117.1772.58

203.58154.7048.88

Meansiitiares

19.85

3.88

24,19

4.56

16.29

3.91

5.12"

5,30-

4.17'

• p < .01.

products over time and thus more satisfied with extendedusage? It might also be important to determine if signifi-cant differences between consumer reactions to expec-tations-performance disparity can be attributed to psy-chographic variables.

REFERENCES

1. Alderson, Wroe. Dynamic Marketing Behavior. Homewood,111.: Riehard D. Irwin, 1965.

2. Brehm, Jack W. and Arthur R. Cohen. Explorations in Cog-nitive Dissonance. New York: John Wiley & Sons, 1962.

3. Buskirk, Richard H. and James T. Rothe. "Consumerism—An Interpretation," Jottrnal of Marketing, 34 (October1970), 61-5.

4. Cardozo. Richard N. "An Experimental Study of CustomerEffort, Expectation, and Satisfaction," Journal of MarketingResearch, 2 (August 1965), 244-9.

5. . "An Experimental Study of Customer Effort, Ex-pectation, and Satisfaction," unpublished doctoral disserta-tiewi. University of Minnesota, 1964.

keting Action," Proceedings. Fall Conference, AmericanMarketing Association, 1964, 283-9.

7. Carlsmith, J. Merrill and Elliot Aronson. "Some HedonicConsequences of the Confirmation and Disconfirmation ofExpectancies," Journal of Abnormal and Social Psychology,66(Eebruary 1963), 151-6.

8. Chapanis, Natalia P. and Alphonse Chapanis. "CognitiveDissonance: Eive Years Later," Psychological Bulletin, 61(January 1964), 1-22.

9. Cohen, Joel B. and Marvin E. Goldberg. "The DissonanceModel in Post-Decision Product Evaluation," Journal ofMarketing Research, 7 (August 1970), 315-21.

10. Diab, Lutfy N. "Some Limitations of Existing Scales in theMeasurement of Social Attitudes," Psychological Reports,17 (October 1965), 427-30.

11. Duncan, David B. "Multiple Range and Multiple F Tests,"Biometrics, 11 (March 1955), 1-42.

12. Feldman, Shel, ed. Cognitive Consistency: MotivationalAntecedents and Behavioral Consequences. New York: Aca-demie Press, 1966.

13. Festinger, Leon. A Theory of Cognitive Dissonance. NewYork: Harper & Row, 1957.

14. Freedman, Jonathan L. "Involvement, Discrepancy andChange," Jottrnal of Abnormal and Social Psychology, 69(September 1964), 290-5.

15. HoUoway, Robert J. "An Experiment on Consumer Dis-sonance,"/ourna/o/Mor^f/i/ig, 31 (January 1967), 39^3 .

16. Hovland, Carl L, O. J. Harvey, and Muzafer Sherif. "As-similation and Contrast Effects in Reactions to Communi-cation and Attitude Change," Journal of Abnormal andSocial Psychology. 55 (July 1957), 244-52.

17. Insko, Chester A. Theories of Attitude Change. New York:Appleton-Century-Crofts, 1967.

18. Katona, George and Eva Mueller. "A Study of PurchaseDecisions," in Lincoln Clark, ed.. Consumer Behavior: TheDynamics of Consumer Reaction, I. New York: New YorkUniversity Press, 1955, 74-5.

19. Olshavsky, Richard W. and John A. Miller. "ConsumerExpectations, Product Performance, and Perceived ProductQuality," Journal of Marketing Research, 9 (February1972). 19-21.

20. Oshikawa, Sadaomi. 'The Theory of Cognitive Dissonanceand Experimental Research," Journal of Marketing Re-search. 5 (November 1968), 429-30.

21. Rosenberg, Milton I. "When Dissonanee Fails: On Elimi-nating Evaluation Apprehension from Attitude Measure-ment," Journal of Personality and Social Psychology, 1(January 1965), 28-42.

22. Sherif, Muzafer and Carl I. Hovland. Social Jtidginent:Assimilation and Contrast Effects in Communication andAttitude Change. New Haven: Yale University Press, 1961.

23. Spector, Aaron J. "Expectations, Fulfillment, and Morale,"Journal of Abnormal and Social Psychology, 52 (January1956). 51-6.

24. VVhittaker. James O. "Attitude Change and Communication-Attitude Discrepancy," Jottrnal of Social Psychology, 65(February 1965), 141-7.