The theory-perseverance bias was investigated in 3 related studies. In
Study 1, a test of the hypothesis that group discussion attenuates th
is bias, male and female Canadian undergraduates examined information
suggesting that either a positive or a negative relationship exists be
tween risk preference and firefighting performance; they were then inf
ormed that this information was fictitious. They then recorded their b
eliefs regarding the actual relationship between these variables (a) i
mmediately, (b) after 7 min of thought, (c) after a 7-min group discus
sion, or (d) after perusing a transcript of one of the discussions. On
ly the participants in the first condition made biased judgments. Stud
y 2, which used Vinokur and Burnstein's (1974) procedures and which ag
ain drew a sample of Canadian students, provided support for a persuas
ive-arguments explanation of this group-discussion effect. The results
of Study 3, which used procedures similar to those of Study 1, ruled
out a social-comparison explanation for this effect. Factors that migh
t enhance or diminish the effects of group discussion on this bias are
suggested.