For a lot of people, knowledge that one is biased about a group of people, or even an individual, is little help in grading fairly. Quantifying one’s standards is also of limited help. Most of us can just tell when a paper exhibits genuine understanding or insight, but rules for detecting it are very likely to be elusive. The human mind often operates with intuition and insight best, and, as people working with those who lack insight in an area of life find, using rules and reason can be very limited.
So what to do? Removing names from papers is not always possible or, with small groups, effective.
One thing one might try is to compensate by systematically awarding higher grades than one’s biased judgment indicates. E.g., “This feels like C work to me, but I’ll give it a B- because it is too likely that I’m being biased.” Clearly, one should worry about doing this when it will harm other students; e.g., in a competition. But otherwise, can it do harm? Unfortunately, yes, it can.
It turns out that students of groups liable to be targets of bias know when they are getting graded up because of their race or ethnicity. They experience it as a kind of invalidating practice that puts them further outside the normal practices of the majority. Or so this interesting article from the Atlantic argues.
So what to do?
Citing the original report and a researcher, the article says:
So how can this problem be solved? Harber and his colleagues found that teachers who have greater social support at school are less likely to show a positive feedback bias toward black students. The theory is that teachers with support feel less anxious about their performance and can concentrate on being fair graders.
But this comment seems odd, because it does not seem to be addressing a context in which anonymous grading can actually be helpful. The problem discussed seems to be more general that the one we started with, though I don’t think that’s a reason for dismissing the idea that inflated feedback can be damaging.