Compatibility Theory

How large is the amount of $100? How good is an SAT score of 1200? How large is the number of 800 people left homeless by a natural disaster? Evaluating attribute values is a cornerstone of virtually every judgment or decision that humans make. In this paper we advance and empirically test a theory of how people evaluate attribute values. We posit that individuals may rely on four processes (or inputs) when evaluating attribute values. We conjecture and provide empirical evidence that the four inputs are ordered, in the sense that an input that is perceived to be more diagnostic and/or requires less cognitive effort is more likely to drive an attribute’s evaluation. An input is diagnostic for a judgment—in this case an evaluation—to the degree that it is relevant for evaluating the target attribute value. The use of an input is effortful to the degree that it consumes cognitive resources. We propose that normally less diagnostic-more effortful inputs exert a stronger influence on evaluations when otherwise more diagnostic-less effortful inputs are unavailable. We describe the four evaluation inputs below. The first input entails an evaluation of a target value without reference to other information. The second input entails a comparison of the target stimulus’s target attribute value with another value on the same attribute. The second input has a stronger impact on evaluations when the target attribute value is not highly evaluable in isolation ((Hsee 1996; Hsee et al. 1999). The third input entails a comparison of the target value with a compatible value on another (reference) attribute. The third input has a stronger impact on evaluations when the target attribute is not highly evaluable in isolation and there is lack of information about other values on the target attribute. The fourth input entails an inference about the target value based on an incompatible value on a reference attribute. The fourth input has a stronger impact on evaluations when the target attribute value is not highly evaluable in isolation, when no comparison can be made with another value on the same attribute, and when no comparison can be made with a compatible value on a reference attribute.