Attributing equity gaps to course structure in introductory physics

We add to a growing literature suggesting that demographic grade gaps should be attributed to biases embedded in the courses themselves. Changes in the structure of two different introductory physics classes were made while leaving the topics covered and the level of coverage unchanged. First, a class where conceptual issues were studied before doing any complicated calculations had zero final exam grade gap between students from underrepresented racial or ethnic groups and their peers. Next, four classes that offered students a retake exam each week between the regular bi-weekly exams during the term had zero gender gap in course grades. Our analysis indicates that demographic grade gaps can be attributed to the course structure (a course deficit model) rather than to student preparation (a student deficit model).

Figure 1

The URM grade gap on the final exam is the URM average final exam score minus the average final exam score of their peers in the same class. The final exam distribution is normalized to standard deviation $=1$ and the results for the four different classes taking the same final exam are shown. Classes II, III, and IV were taught chapter by chapter in the usual way and class I was taught concepts first. Class I and II were taught by the same instructor using exactly the same materials (lecture slides, student activities, and homework) but just arranged differently in time in the two classes. The error bars are standard errors.

Figure 2

Comparing the concepts-first class with the three chapter-by-chapter classes grouped together. Again, the URM grade gap on the final exam is positive if URM students outperformed their peers. For each class organization the bare uncontrolled URM gap is shown as well as the URM gap after controlling for incoming math and physics understandings of the students. The error bars are standard errors.

Figure 3

The gender gap for the course grade is the average course grade for female students minus the average course grade for male students in the same class. The course grade distribution already has a standard deviation of about one so it was not normalized. These classes all had course grades largely determined by exam grades and the classes that allowed retake exams had little or no gender grade gap. The error bars are standard errors.