The Intractable Conjunction Fallacy: Statistical Sophistication, Instructional Set, and Training

Describes 3 studies conducted to replicate & extend work by A. Tversky & D. Kahneman (1983) on the conjunction fallacy (CF), which occurs when someone judges the probability of a conjunction of 2 elements to be higher than that of either element alone. In Study 1, 281 college students in NH who performed better on a test of multiplicative probabilities were less likely to commit the conjunction fallacy. In Study 2, Ss completed F. J. Yates's & B. W. Carlson's (1986) urn problem as a measure of understanding of the conjunction rule (similar in structure to the conjunction task). Ss who answered the urn problem correctly were less likely to commit the CF. In Study 3, an extension of C. S. Crandall's & B. Greenfield's (1986) training study, Ss completed the urn problem & received 1 of 2 sets of instructions. Results duplicated those of Study 2, & the instructional set manipulation had no significant effect. Across the studies, many Ss fell prey to the CF, even those who demonstrated or who were provided an understanding of the conjunction rule. 3 Tables, 12 References. Adapted from the source document.