Essay: Anonymous Grading

Anonymous grading: Using online tools to help avoid bias

As teachers, we all want to ensure that grading is fair and that no students have advantages that others do not. However, we are human, and having structures in place to help us avoid bias can be highly valuable. I encountered just such a problem teaching an upper division experimental design (statistics) course. Of my 27 students, five were friends of mine. I know from my previous years of experience how hard it can be to avoid subtle, unconscious bias towards students one has a closer relationship with, such as those who speak up or come to section, let alone those who are your friends.

My solution was to use the “Quiz & Survey” tool on bSpace, which allows students to submit work anonymously. This method required students to prepare their assignments in electronic form; some students had less experience with this, so I offered extra help on some necessary (and externally valuable) skills (e.g. formatting formulae, including screenshots). I also used the Quiz & Survey tool to ask my students for feedback on the system at the end of the semester.

Assessment of the success of this solution was complex. While it may have produced fairer grading, I wanted to ensure that it would not obstruct the learning environment to the point where the process was a net loss. I therefore set out to assess success both in subjective and objective ways. Subjectively, I assessed my students’ and my sense of how well the system worked, both for fairness and difficulty of use. Responding to the survey question “How often did the electronic submission cause you frustration?” over 2/3 of respondents selected “not much at all,” and all the others selected “sometimes.” Positive student comments included “Great system to keep things fair,” and “Since some of us became better acquainted with you than others, I think the anonymous grading was good.” Others suggested additional advantages of the system: “Saves paper and much energy,” “saves having to find a printer,” and “you can submit work at any time.” Some students indicated initial difficulty which they felt was resolved as the semester progressed. One student commented that “Anonymous grading is nice because it takes away any biases. On the other hand, … [with consistent writing styles] I’m curious how anonymous the grading was able to remain.” Responding to this legitimate concern,I can report that reading an electronically submitted word processed document or pdf helped me to distance myself from the creator of that document, even if I could eventually distinguish who it was.

Additional student concerns about the downsides of impersonal grading were mitigated by my one-on-one, engaging pedagogical style in sections. One student was worried that if a particular student needed more help, knowing their identity would have allowed me to give more directed feedback on their assignments, “But… this is not really the case in this class, as people are not shy to ask for help.” Finally, one concern was that “some people are… spending a lot of time in office hours… I think that should be taken into consideration [in their grade],” and I addressed this at the end of the semester by reporting effort in section to the instructor when discussing final grades. My own subjective assessment is that the combination of electronic and anonymous grading did help me to reduce unconscious bias. When I graded the students’ midterms, I was surprised by the identity of the person with the top score, though I was otherwise aware of my students’ relative abilities.

To complement these subjective assessments, I also used an objective measure of success regarding fairness in grading, statistically comparing the average grade on assignments for my friends versus students I had not previously known. (It seemed appropriate to use statistical analysis on the results of a class on statistical analysis.) Using a robust randomization test (due to the small and unbalanced sample sizes), I found no significant difference (p>0.05) between the two groups.

Overall, I believe this tool effectively reduced my bias in grading. Several lessons emerge: first, more impersonal grading should be balanced by open and inclusive section/office hours. Second, students may need assistance in preparing assignments electronically. Third, this method may vary in effectiveness for courses with more writing, fewer students, or those with multiple GSIs. I encourage GSIs to make use of this resource and see how it performs in other situations. While the bSpace staff were instrumental in suggesting this tool and in assisting me, I also encourage the campus to develop flexible, easier-to-use tools along these lines. Having help in teaching more ethically is invaluable.