The quandary is complex, but can be simply stated. “The problem is many students who begin Ph.D. programs do not complete them,” said William Russel, dean of the graduate school at Princeton University. And so far, he continued, “Our traction on this issue is still limited.”
A group of people involved in managing and funding graduate education gathered to grapple with that issue and others Monday at a “A Fresh Look at Ph.D. Education,” an all-day workshop sponsored by the Council of Graduate Schools  and the National Science Foundation  at NSF’s headquarters in Arlington, Va. The workshop began with a presentation on the council’s data, released in December, on Ph.D. completion and attrition rates , showing that over all, only about 57 percent of students who start their Ph.D.'s complete them within 10 years, though there are significant variations by discipline.
Robert Sowell, vice president of programs and operations for the council, also provided a preview of preliminary data on completion rates by race and gender -- which find, for instance, that male students are completing their degrees at higher rates than females in all broad fields except the social sciences and humanities. And, interestingly, while African-Americans generally have lower Ph.D. completion rates than white students in engineering and the sciences, they complete life sciences degrees at about the same rate -- potentially a significant finding in terms of identifying interventions that work, as one participant emphasized.
Speakers subsequently switched to just that topic, offering perspectives on the relative successes of interventions on degree completion rates on particular campuses. And the day concluded with presentations on the content of the Ph.D. degree itself, and what basic skills could or should be taught across disciplines.
Talking about increasing completion rates, graduate administrators from Duke University and the University of California at Los Angeles told of interventions that their universities put in place in the 1990s -- and changes in the completion and attrition data on their campuses since that time.
Lewis Siegel, vice provost at Duke and the CGS-NSF dean in residence, began by identifying some of the issues at Duke: that many students didn’t know what they were getting into relative to the differences between undergraduate and graduate education and also the nature of the academic job market; that faculty relied too heavily on Graduate Record Examination scores and college grade point averages in considering applicants; and that student funding was tied to departmental needs, "designed to achieve maximum service at lowest cost.”
"Some students taught three courses a semester for a stipend that did not meet the cost of living in Durham, North Carolina," Siegel said. "It was unbelievable."
In response, Siegel said interventions introduced after 1995 included reducing the emphasis on GREs and GPAs in selecting students, and publicly posting data on placement rates, time to degree, and completion, all in the name of transparency. For its arts and science disciplines, Duke increased per-student funding, Siegel said, while cutting some programs in size by half. The university also changed the budget formula, he explained. Rather than fund graduate students based on departmental needs for teaching assistantships, they devised a new formula, Siegel said, that didn’t take a department’s teaching needs into account at all.
Instead, the formula considers factors like the number of faculty who have supervised Ph.D. dissertations in recent years and relative completion rates, and it rewards departments that garner external recognition and funding with extra internal monies. Departments that found themselves without sufficient numbers of graduate students to fill their teaching needs could apply for funding to hire temporary faculty or postdoctoral fellows -- not graduate students.
The university also began subsidizing child care for graduate students, based on financial need, Siegel said. Comparing groups of students that began their degrees before and after the interventions were put in place (those that began in 1992-94 compared to 1998-2000), Duke's Ph.D. completion rates within seven years rose from 35 to 46 percent in the humanities, and in the social sciences, from 51 to 63 percent. Duke had more mixed results in the sciences, with rates increasing in some disciplines, and decreasing in others.
Claudia Mitchell-Kernan, vice chancellor for graduate education at UCLA, also reported significant increases after interventions were introduced in the 1990s there, including a doubling of fellowship funding. The university also built a database to track time to degree in various departments, and incorporated items on time to degree and completion rates among the performance indicators that departments judge themselves by and are judged upon. “Most faculty members did not a have a clue what time to degree was in their respective departments," she said.
“Shining light on these programs really seemed to make a difference.”
Mitchell-Kernan stressed that it’s difficult to determine what particular interventions, if any, caused the changes in completion rates. "You can't isolate the effects of the interventions from all the other things in our environment," she said. Lewis Pyenson, dean of the graduate college at Western Michigan University, was skeptical of the significance of the changes reported Monday, saying that the level of analysis relative to other parallel events happening in (recent) history was “pretty thin.” But, beyond that, he pointed out that many of the interventions described Monday require significant investments of money -- and so are not applicable to the majority of institutions in the United States, like his own, that are producing Ph.D.s under intense budgetary pressures.
Questions of Content
After lunch Monday, the conversation shifted for awhile away from completion rates and toward content. Yehuda Elkana, president and rector of Central European University, in Budapest, stressed the need to train students to embrace contradictions in science, and to ask students epistemological questions. Why, for instance, in creating a particular statistical model, did students choose the parameters they chose? Can they answer that question? (From his experience, it seems that the best and brightest by and large cannot.)
And Mary Ritter, pro-rector of postgraduate and international affairs at Imperial College, London, described its relatively newfound focus on teaching doctoral students and postdocs “transferable skills," with the support of a government initiative and associated funding. Imperial College now requires that all of its graduate students complete a certain number of workshops offered within seven broad skills areas: research skills and techniques, the research environment (covering topics like peer review, pressure for results, and obligation to the public), research management, personal effectiveness, communications skills, networking and team working, and career management. The college offers more than 40 different workshops in topics like science and the media, the commercialization of research, negotiation skills, writing skills, thesis writing and stress management. Imperial College also sponsors intensive three-day residential workshops on transferable skills that each of its approximately 500 first-year Ph.D. students complete in groups of 30 to 35 at a time.
“This is a useful addition, and it helps to enhance the research training,” Ritter said of transferable skills and the doctoral degree.
In a final presentation intended to portray the student and postdoc’s point of view, Crispin Taylor, executive director of the American Society of Plant Biologists, returned to many of the themes of the morning relative to attrition and completion. “Particularly when we’re talking in a fairly blasé way about 7, 10, 11-year completion rates, these are not good for bringing people into graduate programs,” Taylor said. “It takes a long time before you’re going to have a ‘real job,’ particularly if you’re going to stay in the academy.”
“It just seems to be absolutely crystal clear that the Ph.D. takes too long,” added UCLA’s Mitchell-Kernan. And while graduate schools have “fiddled around the edges," the core doctoral curriculum has largely remained static, without critical questions being asked about what needs to be included and what could be eliminated -- saving student time. What about courses, she asked, that students rank as unhelpful year after year in their evaluations? Why do they still stay in the curriculum?
“That kind of focus, really on the content of the program, is really needed. But it’s really hard to do from the graduate dean’s office.”