Compatibility Principle

DISCUSSION

Although the notion of compatibility has long been suggested as a possible cause of elicitation effects (e.g., Lichtenstein & Slovic, 1971; Slovic & MacPhillamy, 1974), the authors describe their investigations as direct tests of this hypothesis. They tested several implications in studies of prediction and preference. In each of these studies, enhancing the compatibility between a stimulus attribute and the response mode led to increased weighting of that attribute. These findings indicate that compatibility plays an important role in judgment and choice.

The testing and application of the compatibility principle require auxiliary hypotheses about the characteristics of a stimulus attribute that make it more or less compatible with a given response mode. Many features of stimulus attributes and response scales could enhance their compatibility. These include the use of the same units (e.g., grades, ranks), the direction of relations (e.g., whether the correlations between input and output variables are positive or negative), and the numeric correspondence (e.g., similarity) between the values of input and output variables. Although we do not have a general procedure for assessing compatibility, there are many situations in which the compatibility ordering could be assumed with a fair degree of confidence. For example, it seems evident that the prediction of market value in dollars is more compatible with a predictor expressed in dollars than with a predictor expressed in ranks. The same situation exists in the domain of perceptual motor performance. There is no general theory for assessing the compatibility between an information display and a control panel, yet it is evident that some input–output configurations are much more compatible than others and therefore yield better performance. . . .

The scale-compatibility notion concerns the correspondence between the scales in which the inputs and the outputs are expressed. In a previous paper (Tversky, Sattath, & Slovic, 1988), we explored a more abstract notion of compatibility that was later called strategy compatibility by Fischer and Hawkins (1988). To introduce this concept, we distinguished between qualitative and quantitative choice strategies. Qualitative strategies (e.g., dominance and min–max) are based on purely ordinal criteria, whereas quantitative strategies (e.g., multiattribute utility theory) are based on tradeoffs or weighting of the dimensions. We proposed that the qualitative strategy of selecting the option that is superior on the more important dimension is more likely to be used in the qualitative method of choice, whereas a quantitative strategy based on the tradeoffs between the dimensions is more likely to be used in the quantitative method of matching. In this sense, the prominence effect may be attributed to the compatibility between the nature of the task and the nature of the strategy it invokes. (For further discussion of strategy compatibility and its relation to scale compatibility, see Fischer & Hawkins, 1988.)

Although compatibility, like anchoring, has a powerful effect on prediction and preference, people are generally unaware of this phenomenon. Such a bias seems to operate at a very elementary level of information processing, and it is doubtful whether it can be eliminated by careful instructions or by monetary payoffs. Indeed, the use of incentives to promote careful responses has had little influence on the prevalence of preference reversals (Slovic & Lichtenstein, 1983).

The reported compatibility effects represent a major source of violations of procedure invariance – namely, the requirement that normatively equivalent elicitation procedures should yield the same ordering of options or events. The failure of procedure invariance complicates the task of the practitioner and the theorist alike. From a practical perspective, the present findings underscore the lability of judgments and choices and make the elicitation task quite problematic. If the decision-maker’s response depends critically on the method of elicitation, which method should be used, and how can it be justified? At the very least, we need to use multiple procedures (e.g., choice, pricing, rating) and compare their results. If they are consistent, we may have some basis for trusting the judgment; if they are not, further analysis is required. . . .

* This is an edited and condensed version of a chapter that first appeared in R. M. Hogarth (Ed.). Insights in decision making: Theory and applications. Copyright © 1990 University of Chicago Press. Reprinted by permission.