ABSTRACT
Rational Choice and the Framing of Decisions S263
3. The extension of the present analysis to prospects with many (nonzero) outcomes involves two additional steps. First, we assume that continuous (or multivalued) distribu tions are approximated, in the framing phase, by discrete distributions with a relatively small number of outcomes. For example, a uniform distribution on the interval (0, 90) may be represented by the discrete prospect (0, .1; 10, .1; . . . ; 90, .1). Second, in the multiple-outcome case the weighting function, irp(p,), must depend on the probability vector p, not only on the component p„ / = 1 , . . . , n. For example, Quiggin (1982) uses the function irp(p,) = ‘n(Pi)fMpi) + . . . + ir(p„)]. As in the two-outcome case, the weighting function is assumed to satisfy subcertainty, *irp(p,) + . . . + ttp(p„) == 1, and subproportionality.
