A practice analysis, in which an appropriate sample of professionals i
s asked to describe what they do in their practices, is considered an
essential part of the content-related validity evidence for licensure
and certification tests. This article examines the precision of estima
tes of the mean frequencies for categories of activities in an activit
y inventory administered on two occasions about 6 weeks apart. In part
icular, it explores the variability in estimates of the mean frequenci
es for categories due to the sampling of respondents, activity stateme
nts, and occasions. The results suggest that samples of 200 respondent
s, about 10 to 20 activity statements per category, and one occasion a
re adequate for estimating the mean frequency for each category.