For people who care about the future of the professoriate, graduate education is vital. But the best national data about graduate education -- the highly respected "Survey of Earned Doctorates" -- focus on just what the study's name implies: those who have crossed the finish line.

On Wednesday, a new book was published that provides what experts say is an unprecedented look at how students race, walk or crawl to the finish line -- or fail to. Three Magic Letters: Getting to Ph.D. (Johns Hopkins University Press) is the result of a decade-long project in which more than 9,000 graduate students, enrolled at 21 top research universities, provided detailed information about their experiences.

Among the findings:

• More than 30 percent of all graduate students never feel that they have a faculty mentor.
• Two-thirds of graduate students enter Ph.D. programs without any debt, suggesting that those concerned about expanding the pipeline to graduate education should pay attention to the affordability of undergraduate education.
• Students rate their social interaction with faculty members as high in the engineering, sciences, mathematics and education -- and relatively low in the social sciences and humanities.
• In rating the quality of academic interactions, students in the humanities think highly of their professors while those in the social sciences and math and science are more critical.
• Significant gaps exist in the experiences of minority and female graduate students -- from admissions to getting teaching or research assistant jobs to publishing research while still in graduate school. Generally, these gaps do not favor minority students.

Debra W. Stewart, president of the Council of Graduate Schools, called the book "the first effort to describe and interpret the empirical realities of the doctoral education process from the perspective of different socioeconomic groups, in different broad fields, and across a variety of universities." Stewart said she hoped the data could lead to real reflection in graduate programs. "This is the kind of work that encourages me to believe we are actually learning something upon which we can make policy decisions in graduate education."

The study was conducted by Michael T. Nettles, vice president for policy evaluation and research at the Educational Testing Service, and Catherine M. Millett, a research scientist at ETS. The work started when both authors were at the University of Michigan and the project was backed along the way by numerous groups, including the National Science Foundation. The universities in the study included leading private institutions (Harvard and Princeton Universities), publics (Universities of Michigan, North Carolina at Chapel Hill, and Texas at Austin), and historically black institutions (Howard and Clark Atlanta Universities).

In an interview, Wednesday, Nettles and Millett said that they started the project focused more on the financing of graduate education, but that their scope gradually grew, in part because the project resonated so strongly with the students they were interviewing. The survey they gave graduate students did not contain any open-ended questions, but participants were so engaged by the study that many ended up writing long letters to Nettles and Millett, some of which are quoted in the book.

One of the things that Nettles and Millett said that they were most proud of was the extent to which graduate students are given a voice in their book to question widely held assumptions. For example, most professors consider that having or being a mentor is a natural part of graduate school, if not the essence of graduate school. But the study found that just over 30 percent of students never perceive themselves as having one. (For purposes of their survey, Nettles and Millett defined mentor as "someone on the faculty to whom students turned for advice, to review a paper, or for general support and encouragement" and while they said advisers could be mentors, they did not assume that the two were one and the same, as they clearly do not for many students.)

Nettles said that he found it "worrisome" that so many graduate students never feel that they have established a mentor relationship. Such a tie, he said, is not just about moral support, but can have an impact on whether students finish a program, get good advice, and feel happy about their education. In addition, the study found that most students with mentors find them early in their graduate programs -- such that those who don't find someone early on may be destined never to find one.

While the percentages on mentors are not unique to any one demographic group, there are some notable gaps in certain fields. In math and science, a smaller share of black students (57 percent) than white students (76 percent) reported having mentors.

The study also found a strong preference among female and minority Ph.D. candidates for mentors and advisers who are from their same groups. The demand for such mentors is particularly hard to fill for the many institutions that lack a critical mass of black faculty members, the authors write, creating "a vicious cycle" in which black students can't find black mentors, and -- if they don't finish -- leave fewer potential mentors for the next cohort.

Issues of race and gender continue to be of intense interest and controversy in higher education, and the new book provides key data in this area.

In looking at admissions-related data, the most notable gaps are by race and ethnicity on Graduate Record Examination scores. Across disciplines, black applicants lagged significantly behind members of other groups. White, Asian and international students did the best, while Latino scores were between those groups and those of black applicants. Generally, there was a correlation between having higher GRE scores and obtaining fellowships upon admission, but the authors found that black and Latino students were likely to receive fellowship offers with lower GRE scores.

While minority students do well in getting fellowship offers, they are significantly less likely, the study found,  to obtain research and teaching assistantships, especially in math and science. For instance, in engineering, 36 percent of black students were offered research assistantships, compared to 69 percent for Asian students. In teaching assistantship offers, only 53 percent of African Americans received offers in math and science fields, compared to 74 percent over all. (Latino students, while doing as well as black students in fellowships, do not appear to suffer the same gap in obtaining teaching and research assistant positions.)

In the book, the authors caution against assuming the black students make out well in this trend -- receiving fellowships and so not needing to worry about getting the teaching and research jobs. In fact, the authors note, those jobs provide opportunities for building the kinds of mentor relationships with faculty members that students need and provide a leg up in both teaching and research.

Other data in the study suggest that this gap is significant in some fields. For example, among science and math graduate students, 17 percent of black students reported publishing a journal article, while the figures for other groups were much higher: Asians (49 percent), Latinos (42 percent), whites (47 percent).

In terms of female graduate students, Millett said that what stood out was the depth of the talent pool -- while there are some variations among fields, women are entering graduate education with extremely strong credentials, including in the sciences and engineering. To the extent gaps are visible, they come in later stages, and in concerns about inadequate mentoring, especially the lack of other women to be mentors.

"It's clear that it will be other female faculty members who will help these women reach their full potential," Millett said.

Nettles said that he hoped the book would serve practical uses and not be treated simply as a work of scholarship. Graduate deans and program directors, he said, should be able to review the national results and then go to their campuses and ask some important questions: "How are we funding students and what is the result? Do people feel they have mentors and why not if they don't?" Or ultimately the big question: "How do you get people through?"