Our previous research showed that despite the use of interactive engagement (IE) techniques at our institution, the difference in performance between men and women on a conceptual learning survey persisted from pre to posttest. This paper reports on a three-part follow-up study that investigates what factors contribute to the gender gap. First, we analyze student grades in different components of the course and find that men and women's course grades are not significantly different (p>0.1), but men outscore women on exams and women outscore men on homework and participation. Second, we compare average posttest scores of men and women who score similarly on the pretest and find that there are no significant differences between men and women's average posttest scores. Finally, we analyze other factors in addition to the pretest score that could influence the posttest score and find that gender does not account for a meaningful portion of the variation in posttest scores when a measure of mathematics performance is included. These findings indicate that the gender gap exists in interactive physics classes, but may be due in large part to differences in preparation, background, and math skills as assessed by traditional survey instruments.