ABSTRACT
Coombs’ theory is similar to Thurstone’s in that it also was proposed in order to support a scaling procedure (in this case the “unfolding” procedure, from which the theory derived its name). Hence, it presumes the existence of quantitative variables of a psychological kind. It differs from Thurstone’s theory in two important respects. In the first place it concerns a special category of pair comparison judgments, namely, preferences. Second, it lacks the probabilistic mechanism inherent to Thurstone’s theory. Consider the situation where a person is making pair comparison preference judgments on a set of attitude statements. The subject is not judging which statement within each pair is the more favorable; he is indicating his preference (i.e., which statement within each pair he agrees with most). Assume that the statements lie along a single relevant quantitative dimension (e.g., the pro-anti dimension). Let each person have a point of maximum preference on this dimension. Then according to Coombs’ unfolding theory a person will prefer statement x to statement y, if and only if he perceives his point of maximum preference to be nearer to x than to y along this dimension. Thus each pair comparison preference judgment provides information about the relative distances between a subject’s point of maximum preference and the stimuli used. Because the theory postulates that all subjects agree about the location of the stimuli on this dimension, even though they might have different points of maximum preference, pair comparison preference judgments from different subjects may be pooled to provide enough distance information to infer something of the quantitative structure of this dimension. Ideally, both the location of the stimuli and the subjects’ points of maximum preference may be inferred. Something of the flavor of this procedure is given in this chapter, although the primary concern is with the logic of the theory: its assumptions, testability, and use as a basis for psychological measurement.
