Some researchers interpret the "Doctor Fox" studies as the most serious threat to the validity of student ratings of teaching effectiveness. The Doctor Fox effect has two premises. First, lecturer expressiveness is the determining factor in student ratings. Second, ratings cannot differentiate between lectures with different amounts of content when lectures are delivered in a highly expressive manner. The implication of the Doctor Fox effect is that student ratings are not accurate. This study investigated the effects of expressiveness, frequency of lecturer behavior, and prior knowledge of items on student ratings of lectures. The following research questions were asked: (1) Does lecturer expressiveness affect ratings? (2) Does lecturer expressiveness differentially affect ratings on low-inference and high-inference items? (3) Are students' reports of the occurrence of lecturer behaviors affected by expressiveness? (4) Does prior knowledge of rating items influence ratings? The four videotapes developed for the study differed along two dimensions: expressiveness and frequency of several lecturer behaviors. The research design was a three-factor experiment with two levels of each factor. The factors were expressiveness (high and low), frequency of behavior (high and low), and prior knowledge of rating items (none and prior). Two hundred students, 25 per group, rated the videotaped lectures on eight low-inference and eight high-inference items. A series of t tests indicated that students gave better ratings to high-expressive lectures on all high-inference items and five of the low-inference items. An analysis of variance indicated that the effects of expressiveness were greater for high-inference items, although both item types were affected by expressiveness. A series of t tests on the low-inference items indicated that ratings reflected the frequency with which behaviors occurred. Prior knowledge of rating items did not affect ratings. The results were consistent with prior studies in that expressiveness had strong and consistent effects on ratings. However, ratings accurately reflected the frequency of lecturer behaviors. Students were not overwhelmed by expressiveness, nor did they report that the lecturer engaged in behaviors solely on the basis of an expressive lecturing style.