In Stage 1, 4 groups of 4 undergraduates were assigned to a 2 * 2 block having 2 conditions of observing rate (paced and unpaced) and 2 fault rates (fast and slow). The probabilities of faults on the 3 sources were .6, .1, and .3. At the slow fault rate, unpaced S's observing of the sources was independent of source probability, but became structured more in favor of source probabilities when fault rate was increased or when observing rate was slowed by pacing. With an expanded design giving a 4 * 3 block, the trend towards selection on the basis of probability reached an asymptote where sources were observed with probabilities about .46, .22, and .32. It appeared that the variable governing the change in observing was the probability of a fault on any observation, P(F/O), which is affected by both fault rate and pacing changes. A function relating P(F/O) to the change in observing for pacing changes was derived, and the corresponding data for fault rate changes were found to fit that function. (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1969 American Psychological Association.
