Whenever we talk about law school grading around here, it usually involves a law professor being incredibly lazy when it comes time to perform his or her most important function regarding a student’s likely job prospects. Or it involves a law school trying to arbitrarily inflate its grades in a desperate attempt to enhance its employment stats.
Sadly, these stories don’t reflect any effort on the part of legal academia to actually come up with a grading system that is fundamentally fair and useful to the students who rely on it. That law school grades are somewhat arbitrary is just a feature of the system that we all kind of accept, even as we know that employers place significant weight on law school grades when handing out scarce legal jobs.
Given all that, I wanted to take some time on a Friday afternoon to consider the proposals of one law professor who has actually thought through some modest ways to make grading exams something less of a random crapshoot…
University of Maryland School of Law Professor Aaron Zelinsky, writing on Concurring Opinions, came up with a couple of ideas to reform law school grading based in part on some lessons from cognitive psychology:
First, grade by question, not by exam….
Second, randomize the grading order across questions.
The first suggestion deals with something called the “Halo Effect.” It means that a professor disproportionately weights the first answer higher than all other answers. If the first answer is strong, the professor will give the student the benefit of the doubt later on in the exam. If the first answer is a drooling mess, then later answers are downgraded even if they are independently very good.
The second suggestion of randomizing the grading order across questions helps with professors who are so weak-minded that they become bored giving out the same grade or point value over and over again. If the first test gets 5 out of 5, and the second test gets 5 out of 5, maybe the third test gets 4 out of 5 just because the professor is sick of giving out fives. It sounds stupid, but this is how people think, so randomizing the order of the tests at least gives the third test a chance to get out of an unfavorable position.
I don’t actually like Zelinsky’s solutions, but at least he’s trying to think of real reforms.
On the first point: I think the Halo Effect is a good thing, and on my exams I always tried to be Master freaking Chief. Look, exams are stupid. They are, fundamentally, a stylized way of testing whether or not you are good at testing as much as they are testing whether or not you’ve actually learned anything. My thought was always that if an exam was going to be something that more resembled a discussion of knowledge and less resembled a useless regurgitation drill, it should have a flow to it. So the first question would be a more broad look at everything and the subsequent questions would build upon or drill down on those previous questions in some way. In effect, I always wanted my exam to be read as a whole — if for no other reason than it seemed like a gigantic waste of everybody’s time if I ended up repeating points I made in question 1 in question 3 just because they might be applicable there too.
Or, to put it even more simply, I knew how to take tests, so I put my best foot forward first.
If we’re reading exams in a way that’s truly question-independent, then really all we’re doing is taking very long fill in the blank tests. There’s more objectivity in that, but even less space for intellectual discourse. And I thought the point of school was to think, not recite.
On the second point about distributive bias: I mean, can we not expect professors to fight this? I know it’s hard. I get bored trying to grade five-minute law revue videos once a year while I’m drinking. But grading five-minute law revue videos is also not the most important thing I do for my job every semester.
Let me ask the question this way: how many times does the average law professor read a student’s exam? Is it more than once? Is it more than twice? Now how many times does that same professor read over drafts of his law review article?
Instead of having a “trick” to get around distributive bias, a professor could just take a “first run” through all the exams, grade them, then read them again, and re-grade them if need be, now that he has a full picture of where the class stands as a whole. Then apply the curve. That could also work, couldn’t it?
Look, law professors have a lot to do, but until they start calling up their buddies and getting students jobs, grading exams is their most important job. I know the incentives are all messed up so that they don’t see it that way, and I know it’s actually very different in college or in other graduate school programs. But in law school, where grades are the thing that people need to get their jobs to make law school worth it, the most important professional function of a law professor is to write and grade exams as fairly as possible.
I’m happy that Zelinsky is trying to think about how law professors might do that better. I wish there were like one law dean who was as concerned about this crucial student service.
Grading Lessons from Cognitive Psychology [Concurring Opinions]