Assigning p values to all transitions in a matrix based on their z sco
res is problematic on 2 counts: The z scores may not be normally distr
ibuted,and transitions are interconnected. Permutation tests, which re
quire fewer assumptions, are an attractive alternative to the standard
asymptotic methods for assigning significance, Moreover, when asympto
tic z scores are only somewhat above their critical value and sequence
s are short, often the exact probabilities of permutation tests are no
t less than .05. Log-linear and permutation methods may be used to win
now the set of transitions initially identified as significant even fu
rther. A computer program that program the performs these tests is ava
ilable from the authors.