Consider two judgment procedures for selecting an answer from a set of
multiple alternatives. One could answer a question either by includin
g likely alternatives from the initial set of alternatives or by elimi
nating the least likely alternatives from that same initial set. An in
teresting question is whether the two judgment processes are equivalen
t and yield the same final selection. The results from two studies ind
icate that individuals generate significantly larger sets of candidate
s in an elimination process than in an inclusion process, with concurr
ent increase in accuracy. We show that this finding is a logical conse
quence of the non-complementarity of elimination and inclusion, and su
ggest a screening model with two criteria to explain the results. (C)
1997 by John Wiley & Sons, Ltd.