Although P(A&B|X) can never exceed P(A|X) (the conjunction rule), it is possible for P(X|A&B) to exceed P(X|A). Hence, people who rank A&B as more probably than A are not necessarily violating any normative rule if the ranking is done in terms of the probability of these events to yield an event X. Wolford, Taylor, and Beck (1990) have argue that this indeed is what happens in some problems (e.g. Tversky & Kahneman's (1983) Linda Problem). The claim made here is that the Linda problem is hard to reconcile with this interpretation; that there is little if any evidence that subjects utilize this interpretation; and that in any case, representativeness can account for all Linda problem results.
- Prof Dev