A synthesis based on persuasive arguments theory but including a revis
ed social comparison component is proposed to account for group polari
zation. According to the proposal, group members choose between risky
and cautious alternatives based on the proportion of known arguments s
upportive of each but argue exclusively for their chosen alternative d
uring group discussion. This implies that the proportion of risky vers
us cautious arguments in discussion will be more extreme than the prop
ortion in participants' lists of arguments on both sides of the issue,
but no more extreme than the proportion in participants' lists of the
''reasons for'' their chosen alternative. Research results generally
supported this implication.