What can we learn from quantitative teaching assistant evaluations?

The teaching assistant plays an important role in teaching computer science at large research-intensive universities. We conducted an exploratory study examining what quantitative teaching evaluations can tell us about our TAs. We found that evaluations were highly coarse-grained: students did not differentiate between the different criteria in the evaluations, and that TAs working in pairs were evaluated as one unit. We found that poor TA evaluations were negatively correlated to student retention. We also found that TAs teaching more than three lab sections a week had lower evaluations. Finally, we found no correlation between TA evaluations and TA experience, students' final grades, and TA gender. We note that quantitative evaluations paint an incomplete picture of TAs' performance, and that more work is needed to provide TAs with formative assessment.