Brad DeLong's Semi-Daily Journal: A Weblog

I once took a summer session course in Dynamics and I think a take home exam led to my getting an A. The students in the course had begged the professor to give a take home exam. The professor told them he would, but they might regret it, since if he gave one it would be really hard. It was the last exam before the final, and it was really hard. I spent a lot of time on it, and I think that it was in doing that exam that I really learned the material. I had a B going into the final. The teacher had said that anyone getting an A on the final would get an A and I did. I don't think I would have aced the final if it weren't for the take home exam before it.

You're (naturally) left-handed, Brad? No wonder! I'm left-handed too. One day the secret left-handedocracy will consolidate its control over the levers of government, and we will declare our dominance openly!

Dear Brad,

From the "World's Silliest Dog" on, your posts are a treasure. But, so too were the well intended comments. Can there be a way of getting back to a reasonably free flow of comments?

Anne

I always avoided take home exams in law school because they were usually 24 hour take homes, which means a viscious all nighter or a poor grade. To offset this, some professors made a rule that no more than six or eight hours should be taken on the test. But everybody cheats and puts in 16 to 24 hours, so you face the ethical quandry of being honest and having a poor test compared to others, or to cheat on the timing because everyone else will. It was a lot easier just to avoid these situations.

And sure life is a take home test to some extent. But if I send an associate out to do a research assignment, when she's done, I expect her to come back to me and describe the issue in a knowledgable way, without fumbling through books, etc. How different is that from taking a course for three months, and then being expected to be readily familiar with the material.

the closed-book exam complaint applies in engineering school too. when I was in school losers in the required sophomore classes lottery got an electrical engineering prof who gave exams with 10 questions, each of which required a single number for the answer. the exams were closed-book and there was no partial credit for work and method, so if you happened to forget one of numerous memorized formulas, e.g. something given in any table of integrals, or if you made a decimal point or addition error, or just a simple typo, you got zip for that problem. this kind of testing has nothing to do with engineering in the real world.

I think an in-class closed book exam has its place and its merits. However what I object to is the guessing-game aspect of the situation where the student has to try to figure out what the teacher is going to pick out to ask about.

Before I give a test, I give the questions the week before, so the students can prepare and research their answers, but they have to write the test fresh.

"pick it up at 1:00, and return it at 5:00."

At this point, why not have an in-class exam where students can type?

A test that I could have typed would have been great. I've often said that my only qualification to be a doctor is my handwriting.

It's important in thinking about this post, though, to remember that at a LOT of law schools, such as UChicago, pretty much all in-class exams are open book. UM seems to be the exception - I was surprised by this post, in fact - as I know Columbia, Harvard, Yale, Georgetown all generally make in-class exams open book. With that in mind, the reasons for closed-book exams cited in this post seem odd given that so many professors at so many schools don't seem to have these concerns.

When I was in business school at the University of Chicago, the dominant approach to exams was to hold in-class closed book exams, but to allow each student a single 8.5 x 11 page of handwritten notes.

That way you didn't need to memorize formulae or lists of factors/facts, but you still had to be able to operate largely free of outside resources. And the one cheat sheet that you were allowed had to be produced by you yourself. Given the massive volume of material covered in most quarter-length courses, this forced the student to be very thoughtful about what the most important ideas and analytical tools in the course were.

I typically found that after spending several hours preparing my cheat sheet - thinking long and hard about what the most important topics were, weighing the most important topics against a list of topics I felt were the easiest to screw up, and finally distilling the essence of those topics into as little space as possible, that I generally didn't have to even look at my cheat sheet when I got to them exam. The process of preparation caused me to internalize the content, and I ended up doing quite well on most of my exams.

And as a further benefit, I've found that by keeping all of my one page cheat sheets I can quickly refresh my memory on the subjects I learned in school. A half hour or so reviewing those simple, concise documents can put me right back to the level of understanding I had on the day of my final exam several years ago. This has been invaluable in my professional life as a management consultant, where I often find that, for example, I need to "get smart" on operations optimization quickly having not thought about it in quite some time.

"However what I object to is the guessing-game aspect of the situation where the student has to try to figure out what the teacher is going to pick out to ask about." =====

The presumes that the student has answer all the questions in order to get a good grade. In engineering school it is pretty common to give exams most student cannot finish in time. So they (should) end up picking the questions they're most qualified to answer.

Did you happen to see that 1898 exam for British 11-year olds published in The Spectator the other week? If you can get more than 40% you are doing well:

http://www.kimdutoit.com/dr/weblog.php?id=P5535?id=5535_0_1_0

Learning is active not passive, thus you need weekly assignments preferably a significant part of the grade so students don't screw themselves by skipping them. I don't know anything about law school but in math and the sciences those weekly assigments are problem sets. Then you should have a final at the end of the course to make sure students review everything and get an overview of what the course is about i.e. it is not just a set of weekly hurdles solved with the help of the textbook. Ideally the exam would test to see if students have a basic grasp of the course and in addition problems whose solution involves many concepts of the course not just the narrow focus of the weekly problem set. That still leaves the problem of how do you get students to review their solutions to the weekly problems. I guess that's one advantage of the Oxford/Cambridge tutor system

As someone who spent between about 7am Thursday till about 3 pm Friday working on an Applied Math take home exam, I can say that take home exams are generally a bad idea. What they do the best is testing how much time you can dedicate to the exam, not how much you know. Open-book, in-class exams work the best because if you know the material in the book you'll fin it and if you don't you'll probably spend the whole test in a futile race to find one fact. I can atest to this, as I've been on the side of knowing the material and searching futilely for an answer.

It's not all that difficult to allow a typewriter option for in-class exams. A friend of mine (who also has bad handwriting) did it for all his law school exams, and wished he had started doing it earlier in his undergraduate career. He just got an electric typewriter with no memory, and asked the instructor ahead of time, so the instructor could arrange for a room with electrical outlets and another proctor. (And big enough tables to support the typewriters. The little elbow desks aren't good enough.)

So far I have only had 2 closed book exams in all of law school. One was for Decedent's Estates which was INSANE. Who can remember who gets a reversionary interest in a dog and how many lives must a lived person's life have lived to apply the rule against perpetuities? Honestly now. I read every single damn word of that book and listened to every ridiculous story about a man, a woman, a divorce, and a pretermitted child or a revocable trust and still only got a B. A damn B. Half the people in that class were 3rd years that didn't read but clearly had better memorizing skills than I do. I find it patently unfair to be tested on something akin to my childhood ability to remember whether there is a triangle or a dog on the other side of that little square piece of cardboard to match up with my triangle.

Thank god my pending Tax exam is completely open book or I would never be able to remember how to calculate a casualty loss deduction for some idiot with a really low AGI that happens to drive a Porsche. The ridiculousness of the hypotheticals always distracts me.

Recent graduate of Harvard Law School. I believe all of my exams were open book. In my experience, this is the way it should be. There is no way you have time during an exam to learn material or even find things in your books that you didn't already know were there. Open book creates better answers and a higher level of studying.