With the average time to complete a humanities Ph.D. in the range of nine years, many educators have been trying to figure out what makes some graduate students finish more quickly than others -- and why some never finish at all.

A new study by a team of researchers at the Cornell Higher Education Research Institute suggests that a key factor may well be expectations management. The study was a follow-up to an earlier study by the institute of the impact of one of the more ambitious recent efforts to reform doctoral education: the Graduate Education Initiative of the Andrew W. Mellon Foundation. That project awarded large grants throughout the 1990s to help selected graduate departments improve their Ph.D. programs.

The earlier study had documented that the efforts did result in increases in retention and slight increases in the speed at which students finished their Ph.D.'s. The new study -- based on survey interviews with students about the factors that helped them finish up -- was designed to go "inside the black box of doctoral education" and to find the factors that are most important to getting students through.

The analysis found that the key factors are policies or practices that express to the students a departmental goal of completion and of completion on a certain time frame. These policies vary from program to program, and may involve expectations about what is done in the summer, about when course work will be completed, about dissertation completion time, etc. But "clarity of expectations" was the factor that stood out, the authors write.

Daniel Denecke, director of best practices for the Council of Graduate Schools, said he thought that finding was significant. "Probably the single most important thing that programs can do to promote completion is truth in advertising," he said, about the realities of program requirements, the job market, and expectations. Departments that regularly check in with students -- especially those who are ABD -- have a big impact on completion rates, he said. "And there's not necessarily a uniform culture for advisers checking in."

Denecke cautioned, however, that a study based on students' views might miss some key factors that graduate students themselves wouldn't realize. For instance, he said that many graduate schools are trying to do a better job in the admissions process -- making sure that potential students meet potential advisers or mentors, for example -- and that these changes in the admissions process may also go a long way toward admitting graduate students who will succeed.

"Not all practices are visible to students," he said.