Research shows that teachers interpret small differences in student evaluations of teaching as meaningful even when available statistical information indicates that the differences are not reliable. The current research explored the effect of statistical training on college teachers' tendency to over-interpret student evaluation differences. A sample of 225 English, mathematics and psychology teachers evaluated statistical reports of student evaluation data corresponding to a fictional teacher's courses. Teachers from the three disciplines varied significantly in their statistics training, but there were no differences in their tendency to rate small, non-significant differences in student evaluations as indicative of the need to make improvements in a course. These results illustrate that statistical knowledge alone does not prevent the over-interpretation of student evaluation means.