Models of commitment make two assumptions: there is a first mover, and
his action is perfectly observed by the subsequent mover. The purpose
of this apper is to disentangle these two assumptions, in order to se
e if a strategic benefit from commitment remains when the first mover'
s choice is imperfectly observed. The basic finding is that the first-
mover advantage is eliminated when there is even a slight amount of no
ise associated with the observation of the first mover's selection. (C
) 1995 Academic Press, Inc.