I had a math teacher in high school who could be fairly described as abusive. He would cut weak students out of the herd and verbally eviscerate them, as if public excoriation could somehow impart knowledge of the subject. I hated that teacher and still do, but now I wonder about him, too. If he actually cared about teaching math (and I believe that he did), what purpose could he have seen in his own pedagogy? He had to believe that fear was the greatest motivator.
With that dark thought in mind, let me turn to Ph.D. comprehensive exams. Everyone who has ever taken comps recalls something about the experience—the exams are a fertile source of graduate-school war stories, not least because they inspire so much fear. But what are they for? And how do they meet their goals?
Comps seem as old as graduate school itself, so that tampering with them might appear the equivalent of rewriting the sacred scrolls. But comprehensive exams in the United States date back only to the late 1930s. They proliferated when graduate populations grew too large after World War II to manage by dissertation alone. Between the GI Bill and Sputnik-related research spending, the number of Ph.D.'s tripled in the 1960s. Against that backdrop, the dissertation defense became more of a formality, creating a need for so-called barrier exams beforehand.
The Ph.D. comprehensive exams mark the barrier that separates graduate course work from the dissertation phase. The purpose of the exams is to determine whether a student should be permitted to "advance to candidacy"—that is, to go on and write a thesis. That basic fact is worth keeping in mind.
A recent study of time-to-degree in U.S. graduate programs found that most attrition takes place in the first few years. That is, advanced students who are hopelessly tangled in their dissertations are less likely to withdraw from graduate school than are their unsnarled colleagues who are near the beginning of their studies. It's not hard to infer the reason. As the cliché goes, lots of people throw good money after bad. They feel too committed to walk away. We should try to pre-empt that unpleasant possibility.
Bureaucratically speaking, the comprehensive-exam stage represents the last procedural chance to dismiss a graduate student whose competence is in doubt.
I've seen lots of professors (including, regrettably, myself) give a weak candidate a break on the comps, only to see that candidate struggle horribly afterward. Mercy does such students no favors. To pass a weak student is essentially to deceive him or her—because, as any parent can tell you, the action (awarding the passing grade) matters more than the words ("We were going to fail you, but we gave you a break so you could shape up").
Easing weak students through a graduate program that makes its most onerous demands during its end stages (dissertation and job search) is a kind of torture. If weak candidates do finish somehow—and they are often borne to the finish line leaning heavily on the shoulders of committee members—they invariably arrive underprepared to seek the jobs they want. Such endgames are a mortifying spectacle, and they ill serve the students.
It follows that professors should use the comps for what they were designed for: to identify students who are best suited for dissertation work along with those who are not. Those who can thrive should go on. Those who won't, should not.
But how can we best identify good dissertators? Subjecting them to an oral inquisition, under immensely stressful conditions, that covers everything they've ever learned in the field has little to do with the act of dissertation writing. The comprehensive exam should therefore bear a concrete relation to the dissertation and the work—that is, the research methods and practices—that will be required to complete it.
Comprehensive exams in some fields already do that pretty well. Many biology programs, for example, require students to defend a research proposal of their own authorship before a committee of faculty members. The purpose of such an exercise, declares the Web site of the University of Tennessee's biology department, is to "simulate, as much as possible, the professional demands that a scientist will face while pursuing a research career."
Daniel Simons, a psychology professor at the University of Illinois at Urbana-Champaign, makes a similar point about exam practice in his own discipline when he says, "We don't want students to stop and just read. They should read as they research."
Approaches to the comps in the social sciences and humanities vary widely. Old-fashioned written and oral exams still abound. Those exams look almost entirely backward. Their underlying commandment is, "Tell me what you know." But having a student look backward at a disciplinary canon doesn't offer much of a clue to the student's ability to do what comes next. Dissertations look forward. They require research and synthesis in the service of something new. How does the ritual torture of an oral exam (or a grueling sit-down written test) serve that goal?
I don't mean to suggest that the traditional comprehensive exam can have no purpose. I read plenty of interesting books in the months preceding my orals. But I was already doing that. As the English professors Cary Nelson and Stephen Watt put it, "What do such exams accomplish that the 12 to 16 courses doctoral students take don't?"
My own orals didn't require me to develop or demonstrate any sort of competency that might have helped me with the dissertation to follow. Nor did the exam permit me to show much more than evidence of my wide reading, coupled with a certain verbal agility under pressure. Moreover, two hours of conversation was a real letdown after six months of study. I shook hands with the members of my committee and thought, "Is that all?" Then I decided to go traveling for a year.
Comps need to look forward as well as backward. Some humanities departments have devised exams that do just that. The University of Maryland's department of American studies has a comprehensive exam that includes an "interdisciplinary synthesis paper" based on the student's own research, and it's explicitly designed as a segue to the formal dissertation proposal. The English department at the University of West Virginia administers an exam drawn from a reading list based on the student's dissertation topic, with works selected by the student in consultation with faculty members. Written and oral exams built from that list direct students toward the questions that their dissertations will raise.
Those are good examples of what can be done, but it's possible to go even further. Comprehensive exams take a long time to prepare for, after all, and may add significantly to a student's time-to-degree. In the humanities, it now takes an average of nine years to earn the Ph.D.; one way that we might save students time (and money) is to streamline the admission to candidacy.
Accordingly, some departments have adopted portfolio-based systems that unfold in a series of stages rather than one watershed moment. The American-studies department at Saint Louis University, for example, administers a written exam early on, but assessment of a student's candidacy rests more fundamentally on a research paper ("worked up into publishable form," wrote Matthew Mancini, chair of the department, in an e-mail), followed by a literature review of three fields that the student chooses with a dissertation topic in mind. The system is "forward-looking" as well as developmental, says Mancini, and "rigorous without being inhumane."
The history department at Duke University also moved a few years ago from what a former graduate director, John Thompson, calls "the big, scary exam sometime in the distant future" to a similarly phased model. It moves students along faster, he noted in an e-mail, and avoids "post-prelim depression"—another contributor to lengthened time-to-degree.
Who needs big, scary exams, anyway?
When we examine prospective dissertators, we are looking for a certain kind of fitness for the task ahead. The comprehensive examination is a rite of passage, but that doesn't mean it should scare the hell out of everyone who takes it, nor should it add unnecessary time to an already long passage to the Ph.D.
Comps need to make sense. Let's examine our graduate students, by all means, but let's also examine what we do ourselves.