More than the final grade: Analyzing multiple channels and levels of student data

Final course grades can predict retention to some extent, but a single grade cannot capture the important complexity of student achievement or the student experience. Indeed, richer data is needed to better inform us on how to improve our instruction, student learning, and the student experience. Over the course of several studies, we have collected five kinds of student data: 1) More detailed course performance data, including performance on individual assignments and individual exam items, 2) student survey data on psychological factors such as psychological stress and self-efficacy, 3) submission times of assignments and quizzes, and 4) data on scores and (when available) speed on standardized tests, and 5) data on gender, race and first-generation status. Using data from thousands of physics students, we report on complex associations and interactions among these dimensions of data and highlight how these results can challenge our assumptions of what grades measure, what we want them to measure, and to what extent traditional grading practices may be part of the systemic issues of inequity in demographic participation and performance in physics courses.