Decision Uncertainty, Expected-Loss Minimization and the Compromise Effect

Shibin Sheng, Virginia Polytechnic Institute & State University
Andrew M. Parker, Virginia Polytechnic Institute & State University
Kent Nakamoto, Virginia Polytechnic Institute & State University
EXTENDED ABSTRACT - Simonson (1989) first introduced the concept of a Acompromise effect@ into the consumer-behavioral literature as a specific type of context effect. It suggests that a brand in a two-alternative set can gain more market share following the addition of an adjacent competitor that makes the brand a compromise choice within the choice set. Only a few explanations of compromise effects have appeared in recent years (Simonson 1989; Simonson and Tversky 1992; Wernerfelt 1995).
[ to cite ]:
Shibin Sheng, Andrew M. Parker, and Kent Nakamoto (2003) ,"Decision Uncertainty, Expected-Loss Minimization and the Compromise Effect", in NA - Advances in Consumer Research Volume 30, eds. Punam Anand Keller and Dennis W. Rook, Valdosta, GA : Association for Consumer Research, Pages: 47.

Advances in Consumer Research Volume 30, 2003     Page 47

DECISION UNCERTAINTY, EXPECTED-LOSS MINIMIZATION AND THE COMPROMISE EFFECT

Shibin Sheng, Virginia Polytechnic Institute & State University

Andrew M. Parker, Virginia Polytechnic Institute & State University

Kent Nakamoto, Virginia Polytechnic Institute & State University

EXTENDED ABSTRACT -

Simonson (1989) first introduced the concept of a "compromise effect" into the consumer-behavioral literature as a specific type of context effect. It suggests that a brand in a two-alternative set can gain more market share following the addition of an adjacent competitor that makes the brand a compromise choice within the choice set. Only a few explanations of compromise effects have appeared in recent years (Simonson 1989; Simonson and Tversky 1992; Wernerfelt 1995).

We suspect one of the reasons for the lack of research attention to compromise effects is that marketing scholars may categorize compromise effects as a special case of attraction effects, without any theoretical distinction. Few have discussed the conceptual difference between these two types of context effects. However, a clarification of theoretic distinction between attraction and compromise effects may help us understand the mechanism and determinants of compromise effect. The attraction effect refers to the phenomenon that the addition of a dominated alternative to a choice set can increase the choice likelihood of existing alternatives, if the existing alternative is superior to the new entrant on both attribute dimensions. Both attraction and compromise effects deal with the result of the addition of a third alternative on an existing two-alternative choice set, and both have the same outcome in terms of an increase of market share of the target brand. However, the context of attraction and compromise effects is different. The essential distinction is that the third brand added into the choice set is completely dominated by the target brand in attraction-effect situations, whereas perceived equivalent to the target brand in compromise-effect situations.

The compromise effect has seen limited explanation of its underlying mechanism. We propose Expected-Loss Minimization under decision uncertainty as the underlying mechanism. A consumer’s shopping decision almost always involves uncertainty. To some extent, uncertainty may reflect a consumer’s perceived risk. A number of definitions of risk have appeared in the psychological literature. In this study, we use expected loss (the logical equivalent of expected value) as definition of risk of shopping decision in this situation. Traditionally, in defining risk indicators, loss has meant essentially what it means within prospect theory. That is, a prospect yields a loss if it results in an outcome that falls short of some specified reference point. In the compromise effect context, a consumer may use a target outcome or aspiration level as the reference point. We provide a proof to show that a consumer may minimize the expected-loss by choosing the compromise option given the decision uncertainty. Therefore, we hypothesize that the higher a consumer’s decision uncertainty, the more likely he or she will select the compromise option as a vehicle to minimize expected loss of decision.

We ran a measurement development study prior to the main study to obtain reliable and valid measure of decision uncertainty. The development of a psychological decision-uncertainty instrument began with a literature review that generated an item pool, designed to measure hypothesized components of decision uncertainty. Nineteen items were generated to reflect five facets of consumers’ uncertainty: decision process uncertainty, inadequacy of information, product uncertainty, prediction of decision, and justifiability. A self-administered questionnaire for scale development was given to an initial sample of fifty-four undergraduate students enrolled in a marketing course at a state university. Based on the result of an exploratory factor analysis, eight items remained in the instrument. Using just these eight items, decision uncertainty exhibited high reliability for all six products (Cronbach a>.85).149 undergraduate students in a state university participated in the main study for extra credit for a marketing class. They were presented with six product classes. In each product class, there were three alternatives, differing on two attributes. All other attributes in a product class were defined as identical. Subjects examined the alternatives in each set, and then made their shopping decision by choosing one brand, followed by completing the decision uncertainty scale.

The compromise effect suggests that the market share of the compromise option will increase as a result of addition of the third brand to the choice set. In our study, the compromise effect is significant in four out of six product classes. In order to test the hypothesized positive relationship between decision uncertainty and compromise effect, we use logistic regression as the analytical tool because the dependent variable in this study is dichotomous (it is coded as "1" if a compromise option is chosen by the subjects, and as "0" otherwise). As predicted by our hypothesis, the coefficients of decision uncertainty for all the product classes are positive, with all being significant except one product. This finding suggests that decision uncertainty significantly increases the likelihood of the compromise option being chosen in the shopping decision process.

Of course, there are several limitations in this study. We employed undergraduate students as subjects in a paper-and-pencil task. Moreover, brands are described on only two or three attributes, without physical presentation of products. These experimental conditions limit the external validity of findings in the study. The compromise effect is not a universal phenomenon occurring in all circumstances. It might be influenced by particular characteristics of the decision maker, the choice task, and the product. Studies about how these characteristics influence the compromise effect may provide insightful avenues to understand this phenomenon.

----------------------------------------