While it is unclear why, the attrition rate for students in Ph.D. programs may be dropping -- at least in fields outside of the humanities, where it appears stalled.

That finding is among the most striking in a major study of Ph.D. completion and attrition rates being released today by the Council of Graduate Schools. The finding on attrition is significant because one of the major reasons for the study and one of the top concerns in graduate schools is that so many students never finish -- leaving some fields facing shortages of doctorates and leaving many students who drop out feeling like they wasted years of their lives.

The data on Ph.D. attrition rates are part of a larger analysis of 10 years of statistics on who starts and finishes Ph.D. programs. Much of the summary data on completion rates was released in July and confirmed what is no secret to English Ph.D. students: It takes much longer for a humanities Ph.D. to finish up, on average, than for those in any other broad area.

Even after 10 years, a majority of humanities Ph.D. students have not finished their degrees, while nearly two-thirds of those in engineering have done so. After seven years, a majority of engineering and life sciences students have wrapped up, while that's not the case for even 30 percent in the humanities. (The social sciences are in the middle, and there are also gaps within the sciences, social sciences and humanities, with psychology outperforming sociology, for example.)

While the numbers on overall rates have been refined, the general trends are unchanged. But the more detailed report being issued today does provide some hope for those who have been concerned about attrition rates among doctoral students. The study analyzed three cohorts of doctoral students -- those who started in 1992, 1995 and 1998 -- and tracked attrition rates through 1995, 1998 and 2001, respectively. (As with other data in the study, information was collected from 30 universities.)

While not enough time has passed to see what proportion of the students in the latter cohorts eventually finished, there are equal time periods to study attrition in the first few years of a program. The data show large changes in the attrition rates in mathematics and physical sciences, and the social sciences, and smaller changes everywhere except the humanities, where the decrease in rates is 0.3 percent.

For the earliest cohorts, the humanities doctoral students had lower attrition rates, but there is some debate about whether that is the pull of programs or the more lucrative job opportunities that exist for people in engineering or the sciences who have a few years of graduate education, but no doctoral degree.

Four-Year Cumulative Attrition Rates for 3 Cohorts of Ph.D. Students

Robert Sowell, vice president for programs and operations at the Council of Graduate Schools, said that he was encouraged by those shifts (except for the humanities), but that the study thus far did not offer explanations for why the changes would vary so much by discipline. During the periods of the three cohorts, however, he said many graduate schools started to devote more attention to preventing attrition and that he hoped the data show the success of those efforts. He also said that he could not identify from the available data why humanities attrition rates would show so little change during periods that the other groups improved.

He said that the current data did not have demographic breakdowns -- and that would be the focus of the next data analysis.

The idea behind the entire project, he stressed, was to get data that could be used at the campus level to improve completion rates. "To improve completion rates, we've got to know where we are," he said.

Cumulative 10-Year Completion Rates for Students Who Entered Ph.D. Programs 1992-3 through 1994-5