Thoughts on Comprehensive Exams, Part 3

In my first two posts on departmental exams, I talked about the different types of exams and the ways in which they are administered. Today I'm going to ask the question every grad student that has ever studied for a departmental exam has asked: "Why are we doing this?"

While departmental qualifying or comprehensive exams are generally the norm in physics and astronomy programs, there are a growing number of departments that have de-emphasized or even completely removed the departmental exam from their PhD requirements. Departmental exams became were popularized in the formative days of American universities by some well-known exams in European universities, most notably the Tripos exams at Cambridge. Prior to these written exams becoming popular, exams were generally given orally - a tradition that lives on in the thesis defense. These written exams allowed professors to directly compare students in a qualitative manner that simply wasn't possible with oral examinations.

If you refer to my first post on this subject, I discuss the purpose of the two models of departmental exams. While both models serve unique purposes which I won't repeat here, they both serve the purpose of what I will call academic quality control. By this I mean that every PhD program in existence admits some fraction of students who for some reason do not have the ability/desire/preparation/gumption/"special sauce" to be a successful physicist. This stems from the fact that admissions and acceptance decisions are made based on extremely limited (if any) personal contact. Departmental exams provide a way to remove those people from the program for their own good as well as the good of the department. We have all met 8th and 9th year graduate students (not those that went on leave for a year, had a baby, etc.) who could have done themselves and the department a favor if they had left with a Master's degree after a couple years. Sometimes failing the exam once can cause a student to re-evaluate their career choices. And in cases where it is obvious to the faculty that a student will not be successful but the student refuses to face the music, the exam provides a qualitative way to dismiss the student without a single professor being the bad guy by flunking them in a class. Of course the exams are not perfectly effective. As in any quality control program, sometimes the departmental exams let people that aren't qualified through or reject those who are qualified. At CU, the exams are reviewed and approved by a committee of graduate students that have already passed the test in an effort to make the exam appropriately challenging and fair.

Even with the best designed departmental exams, there is a growing feeling in the physics and astronomy community that departmental exams are more about tradition than effective education. Many departments, such as the physics departments at Cornell and Harvard, have entirely removed the written exam component of their PhD requirements. The belief is that by making graduate classes more rigorous and improving student-adviser relationships, the quality control factor can be accomplished in other ways. In some cases professors may make classes less rigorous when there is a departmental exam, thinking that the exam will make-up for what their class lacks in homework or tests.

Another argument against the exams is that they provide a unrealistic setting to test a student's abilities. In all likelihood, I will never be asked to spend 7 hours answering questions about a whole variety of astrophysical topics anywhere except on my comprehensive exam. Oral exams - especially when coupled with a research project/presentation - more accurately resemble how knowledge is used in the real-world of academic research. Enrico Fermi was famous for being able to apply general knowledge from a variety of fields to a problem and come up with some very impressive back-of-the-envelope calculations. This is a powerful tool that any successful physicist or astronomer should have in their grasp, but perhaps that would be better tested in an oral setting or by contact with an adviser than with a written exam.

Perhaps the most interesting thing about the hole debate is that it is all based on anecdotal evidence and supposition rather than data. Physics and astronomy are data-driven fields and yet there is no data on a crucial question in educating future physicists and astronomers.

Generally, my opinion is that written departmental exams are less effective than rigorous classes, oral exams coupled with research projects, and good student-adviser relationships. There may be some departments that need some sort of exam, but for the most part I think they are exist because of tradition, not because of effectiveness.