Expected Utility and Measurability: I. The Expected Utility Hypothesis
We have argued above that an (operationally justifiable) cardinal analysis is potentially more powerful than an ordinal analysis in meeting the problems of the present enquiry. Yet the two cardinal models which we have discussed so far are unsatisfactory in several respects. First, an admittedly far- from-systematic empirical confrontation casts serious doubt on whether the scales really measure preference intensities. Second, insofar as we maintain as a tacit working assumption that the consumer tastes which are "to count" are those revealed through consumers' consistent 1 market-type choices,2____________________
This similar formulation, by the way, helps to indicate the important distinction between "consumer's sovereignty" and "freedom of choice." In the latter, there is a valuational element attached to the process of choosing as distinct from the content of choice. (Cf. Bergson, "Socialist Economics," in the Survey of Contemporary Economics, 423). Consequently, it is possible to have consumer's sovereignty without freedom of choice. For example, tastes may be revealed by, say, governmental action on behalf of the consumers; the social choices are made for, but not by, the individuals concerned.