Advances in Consumer Research Volume 7, 1980 Pages 314-317
PREDICTING SALES PROMOTION EFFECTS: ASSIMILATION, ATTRIBUTION, OR RISK REDUCTION?
Richard L. Oliver, Washington University
Three behavioral paradigms proposed to explain partially consumer response to sales promotion activities are examined in light of a general disconfirmation framework. Reviews of the relevant literature suggest that assimilation, attribution, and risk reduction phenomena may be useful in hypothesizing promotion-induced purchase behavior. Specifically, it is proposed that special promotions create attribute-specific expectations, that these expectations involve probabilities of benefit and risk, that the expectations are subject to positive or negative disconfirmation, and that attributions are made regarding the disconfirmations. The most immediate effect of the suggested psychological process is thought to be satisfaction/dissatisfaction, while indirect effects on repurchase probabilities are also hypothesized.
Three interesting perspectives on the cognitive structure generated by sales promotion strategies have been proposed. All approach the topic from different theoretical bases using qualitatively different types of promotion. Whereas each makes a unique and much needed contribution to the literature, this diversity of viewpoints effectively precludes detailed analysis on an individual basis. In an effort to provide an integrated discussion, each major thesis will be briefly reviewed and later positioned in a "megatheory" of sales promotion effects. Hopefully, this effort will demonstrate the strengths and potential weaknesses of each of the explanatory frameworks taken alone.
Assimilation - Contrast
Raju and Hastak (1980) have proposed a thoughtful interpretation of deal-oriented sales promotions based on a review of dissonance, self perception, and assimilation-contrast theory. Their discussion of the traditional dissonance framework would suggest that a price reduction lowers the consumer's input to the purchase. The lowered input creates a situation which will motivate the consumer to lower his/her evaluation of the product after purchase so that a consonant balance between inputs and evaluated outputs is maintained. This phenomenon would predict that deals are prone to failure in the long run because of repeated low price and low evaluation associations. Raju and Hastak correctly note, however, that price concessions are not necessarily analogous to the traditional effort concept in dissonance predictions and that attitude change is only one of the many forms of dissonance reduction.
More promise was evident from the authors' review of self-perception theory. A variant of attribution theory, self-perception theory suggests that consumers introspect as to the reasons for buying a brand on deal. If a price discount is successful in attracting repeat customers, proponents of self-perception theory argue that the consumer attributed his/her purchase to a sincere liking for the brand; the deal was only a secondary consideration. If, on the other hand, the discount fails to stimulate repeat purchases, these same proponents would suggest that the initial purchase was attributed to the price reduction. When the product is no longer discounted, no attributions can be generated and no purchase would be expected. Raju and Hastak (1980) illustrate quite clearly that this very convenient "fail-safe" property of self-perception theory makes individual level predictions very difficult. For, if both attributions are possible, the success of a deal will be dependent on the proportion of self as opposed to incentive attributors in the consumer population. Moreover, they show that if variety-seeking is added as a third attribution (e.g., I bought the brand to try something new.), more possible predictions and outcomes are likely to occur.
The authors propose an alternative to dissonance and self-perception theories based on the social judgment phenomena of assimilation and contrast (Sherif and Hovland, 1961). Assimilation-contrast theory, originally based on discrepancies between a communicator's and recipient's position on an ego-involved issue, assumes that individuals have an anchor or comparison level for judgments of a cognitive or physiological nature. In addition, there exists a range of positive and negative discrepancies about this anchor in which individuals will perceive no meaningful difference (a perceptual jnd). This range, or latitude of acceptance, creates an assimilation effect in that deviations from the anchor are always believed to be of no importance. Thus, judgments of stimuli in the acceptance zone are assimilated toward the anchor. The latitude of rejection, however, involves stimuli perceived as materially different from the anchor position. It is believed that individuals will exaggerate the distance between a true position and the anchor when a stimulus falls in either the positive or negative rejection zones. This phenomenon is termed the contrast effect and is believed to be especially likely when one is ego-involved with the judgmental task.
Raju and Hastak (1980) postulate that deals succeed when they move the price attribute from the "too high" rejection region into the latitude of acceptance for the brand on deal. This is thought to combine with a negative attribution of low quality to the low price so that an adversary situation is created between the higher priced, high quality preferred brand and the reduced price brand. Post-trial disconfirmation of consumer perceptions of the brand on deal is thought to follow.
Disconfirmation is based on a comparison between the expectations generated by low price and low quality connotations and one's perception of the "objective" quality of the trial brand. If the brand on deal performs about as expected, assimilation effects will cause the consumer's final evaluation of the product to conform to expectations. This evaluation is compared to that of the preferred brand and the higher of the two determinations should be reflected in the next purchase. If the new brand is either much better or much worse than expected, the consumer's overreaction will drive his/her evaluation either very much higher or lower than originally anticipated. The net effect will be an extremely low postpurchase evaluation in the case of negative disconfirmation or an extremely high evaluation when disconfirmation is positive. The former situation will almost certainly preclude repeat purchase while the latter will probably result in a postpurchase evaluation higher than that of the preferred brand.
It should be noted that the description provided by Raju and Hastak (1980) is much more extensive and precise than that provided here. This overview will serve only to highlight the main propositions of their exposition.
Cheron and Zins (1980) have provided an attribution theory perspective to the strategy of offering free promotional goods and services with the basic product. This differs from the previous discussion in that the price reduction was a guaranteed prepurchase incentive. We focus our attention here on the situation where an expected promotion may be spontaneous (expectations confirmed), obtained upon request (expectations confirmed only if the consumer expected to have to request the promotion), and not requested and thus not obtained (expectations disconfirmed). The authors interpret these events as causal stimuli for an attribution judgment.
Various predictions regarding felt satisfaction over the offer are possible depending on whether the consumer makes internal or external attributions. If the promotion must be requested, internal attribution dominates and neutral or low satisfaction is predicted. If the promotion is expected and received spontaneously, a higher frequency of external attributions is expected; positive external attributions of this nature are posited to result in higher levels of satisfaction. If the promotion is neither requested nor received, two attribution explanations appear equally likely. First, the consumer may blame the vendor for failing to offer the promotion (an external attribution) in which case dissatisfaction would be predicted. A second possibility is that the consumer may accept responsibility for failing to request the offer (an internal attribution) in which case satisfaction should not be affected. Finally, Cheron and Zins (1980) suggest that the price of the promotional offer may moderate the relationships between either the manner in which the promotion was obtained and the ratio of internal to external attributions, or the manner of obtainment and satisfaction. It is not entirely clear from the discussion, however, why and how price affects the criteria as hypothesized.
Shimp and Bearden (1980) have proposed a model which attempts to explain the manner in which warranty concepts affect consumer confidence. While a warranty is not generally thought of as a sales promotion, short term offers of warranties extended beyond the generally accepted time limits or broadened to cover more product components than is usual definitely would be viewed as promotional offers. In fact, the authors' use of the phrase "warranty quality'' coincides with the warranty extension variable discussed here. Warranty extensions differ from other sales promotion schemes, however, in that they do not act to lower the purchase price or add to the tangible value of the product. Rather, their true value can be ascertained only after the product is consumed or disposed of. At the time of purchase, a warranty has only financial risk reduction potential.
Shimp and Bearden (1980) suggest that consumer confidence (i.e., risk reduction) is a function of warranty quality (i.e., the coverage) and the reputation of the warrantor. The relation between confidence and quality is thought to be positive and linear for highly reputable warrantors. Warrantors with poorer reputations generate less confidence. This latter statement is thought to be particularly true if the low reputation warrantor offers a high quality warranty, an incongruence which should cause consumers to be wary, at the least. Product price was thought to play some role in the model but the study results rendered its effect inconclusive.
Although not specifically stated in the paper, it should be pointed out that warranties operate to enhance value by increasing the anticipated state of second level satisfaction. A warranty does not affect the likelihood of product or attribute failure, it serves only to insure some degree of satisfaction once the failure has occurred. Consumers may assume, however, that the higher the warranty quality, the lower the likelihood of failure. Neither these consumer perceptions nor the relationship between warranty quality and failure have been put to empirical test.
The three papers reviewed in the preceding discussion can be positioned in a matrix of sales promotion effects shown in Table 1. Entries in the Table are limited to those propositions suggested in this review. Each theory is briefly summarized in terms of the predictions it would make at various stages of a four step satisfaction model tested in Oliver (1979). The four stages include the creation of expectations, product performance, disconfirmation of expectations, and decision satisfaction. Subsequent stages in the consumer behavior process including attitude change and purchase intention are assumed to derive from the satisfaction judgment.
The table shows that two theories, assimilation-contrast and risk reduction, imply that sales promotions affect prepurchase expectancies. Stated simply, a promotional offering appears to influence expectations either by detracting from the product's value (a price concession) or by adding to it. The consumer is assumed to draw inferences from this value addition/subtraction effect. Attribution theory appears to first contribute to the model at the purchase stage. Specifically, the manner in which the promotion is received (and when, in the case of rebates) is an added factor not generally considered in consumer purchase models. At this point, it appears that sales promotions not only change expectations, but that they also add a new "performance" dimension to the purchase. The vendor is now judged on the basis of delivery of the promotion as well as the product.
At the disconfirmation stage, assimilation-contrast theory would appear to provide the greatest insight into postpurchase effects of brand-sales promotion combinations, although a more parsimonious interpretation will be presented later. As suggested by Raju and Hastak (1980), the product's performance is compared to the consumer's expectations. These expectations, however originally formed, are amended, in part, by the nature of the promotion. Postpurchase evaluations within the latitude of acceptance do little to change the product's image, while high positive disconfirmation results in high affect levels. In contrast, negative disconfirmation certainly will guarantee that the product will be viewed in less favorable light than it was previously.
The Cheron and Zins (1980) paper adds a new element to the disconfirmation process. Apparently, only external attributions will trigger a reevaluation of the brand subsequent to purchase. Good and bad judgments resulting from internal attributions may change the consumer's self-evaluation, but not that of the product.
Risk reduction theory suggests that negative disconfirmation will bring to the forefront those attributes (including warranty) which have the greatest potential for, or were specifically purchased for, the prevention of the perceived loss. Operation of these risk reducing factors or an evaluation of their true postpurchase value sets the stage for a second disconfirmation, that due to the adequacy of protection afforded by a warranty, service visit, store refund, etc.
Finally, satisfaction is thought to be an immediate result of the first and/or second level disconfirmation responses. This will be most pronounced for attributions of an external nature including those to the salesperson, store, or manufacturer. Generally, only evaluations in the latitude of rejection will materially affect satisfaction and only those in the positive disconfirmation region will enhance the likelihood of repeat purchase.
HYPOTHESIZED SALES PROMOTION EFFECTS
AN ATTEMPT AT AN INTEGRATED MODEL
The model proposed here draws heavily from the author's prior work on the disconfirmation of expectations (Oliver, 1977, 1979) and on the three papers reviewed here. No exception will be taken with the general approaches suggested in the three papers although different viewpoints will be noted. The goal of this last section is to provide a sales-promotion-specific interpretation of postpurchase effects very similar to that presented by Raju and Hastak (1980).
Central to the argument presented here is the notion that consumers enter product purchase situations with attribute-specific expectations of the product's performance and that these expectations can be viewed as belief-probabilities of attribute occurrence (Olson and Dover 1979; Oliver 1979). Because the expectations are probabilities, they can be compared to the occurrence or non-occurrence of the attribute state of nature during consumption. Thus, a consumer can be asked to attach a likelihood to a gas mileage rating of 15 m.p.g.; the actual mileage obtained is then easily measured and compared to the prediction.
As noted in the preceding discussion, sales promotions act on these expectations in two ways. First, they may alter the consumer's assessment of the performance level probabilities (e.g., the possibility that a price concession will lower expectations of quality). Second, if the promotion consists of an additional promise or offer (e.g., a free gift, rebate), expectations regarding the offer itself are generated, and these new expectations are also subject to disconfirmation. Some of the changed and added expectations will concern product benefits and some will concern product risks. If an extended warranty promotion reduces consumer predictions of high maintenance costs, then the warranty offer will most likely be successful. It should be noted, however, that the consumer will also develop new expectations about the manufacturer's willingness to honor the warranty as offered.
Once the new or changed expectations are acquired, they become subject to disconfirmation. Raju and Hastak (1980) have suggested that an assimilation-contrast model may be useful in evaluating this phenomenon, an idea first proposed by Olshavsky and Miller (1972) and Anderson (1973). Unfortunately, neither study found unequivocal support for the theory. In fact, the results of all studies to date which have tested this paradigm, including that of Olson and Dover (1979), find that subjects rate products somewhere between their expectations and some more objectively obtained measure of performance. Olshavsky and Miller and Olson and Dover preferred to call this an assimilation (only) effect while Anderson speculated that contrast may operate for extreme discrepancies.
Oliver (1977) took issue with this interpretation, arguing that a more parsimonious explanation exists. It was shown, for example, that assimilation-contrast interpretations of product reactions make counter-intuitive predictions and that analogies from the communications literature, where social judgment theory was first developed, may be tenuous. Rather, if one views assimilation as an expectation effect and contrast as a direct reaction to the disconfirmation, a simple additive model incorporating only expectation and disconfirmation perceptions will be sufficient to predict postexposure product reactions. Specifically, expectations create an anchor or adaptation level (Helson, 1948) about which disconfirmation judgments are made. Products having attribute performance levels above expectations (positive disconfirmation) are rated highly while those with lower-than-expected levels (negative disconfirmation) are rated poorly. Satisfaction, in turn, can be shown to he an additive function of expectation and disconfirmation (Oliver, 1979).
The effect of a sales promotion will largely result from the expectations it creates and the subsequent disconfirmations. First and foremost is the consumer's expectation of receiving the promotion. Price concessions may occur with certainty, but "on-site" promotion expectations may not be realized (see Cheron and Zins, 1980). Second are the changed product expectations resulting from the interaction of the promotion and product attributes. Third are the expectations regarding the performance of the promotion. All play a role in subsequent repeat purchases in a manner that can be analyzed using expectation and disconfirmation, but which must be determined empirically. For example, if the promotion itself is not received or is of poorer-than-expected quality, attitudes toward the product may be negatively affected. If the image of the product is changed as a result of the promotion, then the revised expectations are subject to disconfirmation. As Raju and Hastak (1980) note, the consumer's subsequent evaluation is then contrasted with the evaluation, already formed, of the preferred brand. This comparison determines which of the two brands will be repurchased.
The role of attribution theory in our expectation disconfirmation model has not yet been elaborated. The author is of the opinion that attributions concerning deals will follow the analysis presented by Raju and Hastak (1980) regarding the reasons for purchasing the product and the suggestions by Cheron and Zins (1980) for reactions to the manner in which it was received. Regardless of the reasoning process involved, the product Consumption experience will always involve the disconfirmation of expectations. While the blame/praise for bad/good products may be vested with the manufacturer or the consumer, evaluations of the product will be made nonetheless. It may be that externally attributed results exacerbate the impact of disconfirmation while internal attributions have no product implications. Future work on attribution theory in the context of expectancy disconfirmation clearly is needed.
The author views risk reduction totally within an expectancy framework. As Peter and Ryan (1976) note, perceived risk may be best conceptualized as expected negative utility. Negative utility, in turn, is seen as a function of one's expectation of attribute-specific loss and the evaluation of that loss. As such, it dovetails nicely with the expectancy model presented here. Expectations of loss are subject to disconfirmation in both a positive sense (when the loss does not occur or is less severe than anticipated) or a negative sense (when it is more severe than anticipated). Clearly, any promotion designed to reduce the anticipated probability or severity of a risk-laden attribute can be analyzed using postpurchase disconfirmation and satisfaction as the criteria of import.
Anderson, Ralph E. (1973), "Consumer Satisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research, 10, 38-44.
Cheron, Emmanuel J. and Michel A. Zins (1980), "Extra Promotional Offerings and Consumer Attributions," in Advances in Consumer Research, Vol. 7, Jerry C. Olson, ed., Association for Consumer Research.
Helson, Harry (1948), Adaptation Level Theory, New York: Harper and Row.
Oliver, Richard L. (1977), "Effect of Expectation and Disconfirmation on Postexposure Product Evaluations: An Alternative Interpretation," Journal of Applied Psychology, 62, 480-486.
Oliver, Richard L. (1979), "Product Satisfaction as a Function of Prior Expectation and Subsequent Disconfirmation: New Evidence," in New Dimensions of Consumer Satisfaction and Complaining Behavior, Ralph L. Day and H. Keith Hunt, eds., Bloomington, Ind.: School of Business, Indiana University.
Olshavsky, Richard W. and John A. Miller (1972), "Consumer Expectations, Product Performance, and Perceived Product Quality," Journal of Marketing Research, 9, 19-21.
Olson, Jerry C. and Philip Dover (1979), "Disconfirmation of Consumer Expectations Through Product Trial," Journal of Applied Psychology, 64, 179-189.
Peter, J. Paul and Michael J. Ryan (1976), "An Investigation of Perceived Risk at the Brand Level," Journal of Marketing Research, 13, 184-188.
Raju, P. S. and Manoj Hastak (1980), "Consumer Response to Deals: A Discussion of Theoretical Perspectives," in Advances in Consumer Research, Vol. 7, Jerry C. Olson, ed., Association for Consumer Research.
Sherif, Muzafer and Carl I. Hovland (1961), Social Judgment: Assimilation and Contrast Effects in Communications and Attitude Change, New Haven: Yale University Press.
Shimp, Terence A. and William O. Bearden (1980), "Warranty and Other Extrinsic Cue Interaction Effects on Consumers' Confidence," in Advances in Consumer Research, Vol. 7, Jerry C. Olson, ed. Association for Consumer Research.