Can student evaluations be made more representative? Testing alternative strategies
通过随机实验，测试改变评价提示或时间能否提高学生评价的代表性，发现低GPA、少数族裔等学生参与率低，延迟评价可改善种族代表性但降低总体回复率。
Student evaluations are regularly used to assess the quality of faculty teaching. However, low response rates and unrepresentative student samples can bias the conclusions drawn from them. We conduct a randomized experiment at a selective liberal arts college in the United States to test whether student evaluations can be made more representative of the broader student population through changes to the prompt or timing at which feedback is solicited. We find that Pell Grant recipients, students with low GPAs, and those later in their college careers are less likely to complete teaching evaluations and to write more than three words in response to qualitative questions evaluating the instructor across all study conditions. Black, Hispanic, and multi-racial students are less likely than white students to complete traditional evaluations solicited at the end-of-semester. Providing an alternate prompt, which asks students to articulate their own criteria for effective teaching, increases the likelihood that students write more than three words and increases the total words written, conditional on responding. Delaying the solicitation of feedback until the following semester decreases overall response rates, particularly for seniors, but does so to a lesser degree for Black, Hispanic, and multi-racial students than white students, resulting in a more racially representative sample overall. This suggests a potential tradeoff between overall response rates and the representation of minority student voices in providing student feedback on teaching.