These limitations leave many important questions unaddressed. Is the origin of GPDs most closely related
to subject matter, instructional style, student background, or mode of assessment? To obtain initial insight
into these many possibilities, we have conducted uniform measurements of gendered performance difference across a multi-disciplinary array of 116 large enrollment courses at the University of Michigan. These courses vary substantially in subject matter, instructional style, and mode of assessment. As at other major universities, this array of courses forms a kind of grand quasiexperiment, the results of which hint at possible explanations for GPDs and suggest an array of future observations and experiments. In what follows, we provide a detailed explanation of our methods for quantifying student performance and
measuring GPDs. We then describe patterns observed in the distribution of GPDs across the courses we study, discuss possible explanations for these patterns, and conclude with suggestions for future study and action.