Understanding Gender Differences to Improve Equity in Computer Programming Education.

The increasing wealth of data available from large online programming courses offers exciting opportunities to improve equity in computer science education. In particular, it can provide insight into potential causes of differences between groups, informing future investigations, course development and educational policies. In this paper, we analyse differences between male and female high school and primary school students across five programming courses of different difficult levels (n=14,570), including block-based and traditional text-based courses. We examine differences in course enrolment, completion rates and student behaviour while solving the exercises, and consider the impact of grade level, enrolment and course type. We find that, while females are underrepresented in the courses, performance differences only occur in specific circumstances, and in both directions. This suggests that the differences between male and female students are not fundamental, and that there is potential to address them by better understanding these circumstances.