Read our COVID-19 research and news.

A postdoctoral researcher, Carlee Ashley, conducts cancer research at the Sandia National Laboratory in Albuquerque, New Mexico, in 2011. Sandia researcher and University of New Mexico professor Jeff Brinker watches.

A postdoctoral researcher, Carlee Ashley, conducts cancer research at the Sandia National Laboratory in Albuquerque, New Mexico, in 2011. Sandia researcher and University of New Mexico professor Jeff Brinker watches.

Sandia Labs/Flickr (CC BY-NC-ND 2.0)

How grad students get paid affects where they work

It’s only one study. But a novel analysis finding a link between how U.S. graduate students in the biomedical sciences are funded and their first job after earning their Ph.D. turns one piece of conventional wisdom on its head: Students supported on a research grant are more likely to take a research job than those funded by other mechanisms.

That finding, which appears in the July issue of Research Policy but was posted online last week, adds fuel to an already heated debate over the best way to prepare the next generation of biomedical scientists. It’s also causing some serious head scratching among researchers who study the U.S. biomedical workforce. To understand why, here’s some background on support for graduate education in the sciences.

Three kinds of support

First, the National Institutes of Health (NIH) is the lead federal agency in funding the training of the next generation of biomedical researchers. NIH support for graduate students typically comes in three flavors:

  • As a research assistant (RA) on a faculty member’s grant;
  • As the recipient of a competitive fellowship; or
  • As a participant in a training grant awarded to their university.

The mechanisms provide differing amounts of autonomy for the student and leverage by NIH. RAs, for instance, exist primarily to help faculty members carry out their research, and students run the risk of being treated as simply another pair of hands. Fellowships, in contrast, give students the greatest leeway in picking a lab that meshes with their interests. But that freedom may come at a cost if their advisers are less invested in developing their skills. Training grants are designed to provide students with a high-quality, well-rounded research experience that includes career counseling, and institutions must have a clear plan to achieve those goals. But their supply is limited: They are awarded to a relative handful of universities to address shortages in a particular field of science.

NIH is not the only player in graduate education. The new study found that only 21% of students in the biomedical sciences at a particular university are being supported by NIH at any given time. It’s also not unusual for students to switch from one mechanism to another during their graduate training, say, from a university fellowship for their first year of classes to an NIH-funded traineeship as they explore different disciplines, and then to an NIH-funded RA as they settle down to complete their dissertation. But as the 800-pound gorilla, NIH exerts a huge influence on how all graduate education in the biomedical sciences is financed, from the size of stipends to eligibility requirements.

RAs most common—and most questioned

RAs make up by far the biggest slice of the pie. The number of students supported with NIH-funded RAs is about double the combined total for the two other mechanisms and has quadrupled over the past 3 decades, to more than 25,000 in 2010, while the other two mechanisms have grown only modestly. The number of RAs largely tracks the NIH budget, rising sharply in good times while holding steady or even declining in times of fiscal restraint.

With U.S. universities now routinely turning out more biomedical Ph.D.s than the academic market can absorb, many leading scientists think that NIH should be more aggressive in managing the supply of scientists. In particular, several recent blue-ribbon panels have recommended that NIH shift the balance away from RAs and toward fellowships and traineeships. Their arguments are based both on the fact that NIH has more control over these mechanisms and on the belief that RAs may provide an inferior educational experience.  

But it’s not that simple. Controlling RAs is an inexact science because faculty members, not NIH, decide whether to put a student on their grant. In addition, their close connection to NIH annual budgets creates a quandary for policymakers. A rising NIH budget, something biomedical scientists crave above all else, would spawn more RAs, who would then add to the workforce glut. At the same time, nobody wants to reduce the number of RAs if the price is a falling NIH budget.

Moreover, it’s not simply a numbers game. NIH must also take into account the impact of altering the mix on the diversity of the biomedical research workforce. Specifically, foreign students are ineligible for NIH fellowships and traineeships but comprise a disproportionately large share of RAs. So reducing the number of RAs could shrink their presence in U.S. graduate schools unless the eligibility rules were changed. On the other hand, women and groups underrepresented in the scientific workforce receive a greater share of NIH-funded fellowship and traineeship slots than research assistantships. So offering more of those mechanisms might bolster their numbers.

A surprising finding

The new study, by economist Margaret Blume-Kohout, points to a silver lining for the much-maligned RAs: Students in the biomedical sciences whose primary support was an NIH-funded RA were 11% more likely to have taken a research-related job right after graduation than were NIH-funded trainees and fellows.

Blume-Kohout looked at 41,580 students who earned Ph.D.s at 121 research universities between 2001 and 2010. She was particularly interested in their professional prospects at the time they completed their doctorate. To get the answer, she combined data from a National Science Foundation (NSF) survey of graduate students on the eve of completing their degree, a follow-up sampling of the doctoral-level workforce, and institutional records of how graduate students are funded. To eliminate differences between institutions, programs, and external factors affecting the job market, she compared students who graduated from the same university in the same field and the same year.

Nobody had tried to link funding mechanisms with first jobs for newly minted Ph.D.s, so she couldn’t predict the answer. “But we did not expect the results we found,” says Blume-Kohout, a visiting professor at Mount Holyoke College in South Hadley, Massachusetts.

The study does not weigh into the debate over what students should do after earning their degrees, she says. “We are absolutely not saying that research careers are better than the alternatives.” Instead, she says, it addresses a more fundamental question: “If NIH is funding them to become scientific researchers, then the end point should be whether they have become scientific researchers. And if some of the specific mechanisms intended to make them researchers are not doing that as well, then maybe there’s a problem with the mechanism.”

One reason for the different outcomes, she speculates, could be the mentoring that students receive under the various funding mechanisms. “Previous studies have found that students on fellowships and traineeships, although they enjoy greater freedom, complain that their advisers seem less interested in what they are doing if it’s not a part of their grant,” she notes. Conversely, faculty members might well pay more attention to RAs because their productivity is essential to the project’s success—and affects their chances of getting the next grant.

A student working directly on a grant, as an RA, may also be more inclined to “buy in” to the research enterprise, Blume-Kohout speculates. It could also make them more visible to the research community, an important factor in attaining an academic research position. Her study did not measure the quality of mentoring, however. If fellows and trainees end up receiving guidance from less experienced and less productive faculty members, she says, they might have less interest in taking a research job after graduation.

Among all graduates taking research jobs, according to the study, 84% were headed to a postdoc. In addition to being more likely to take such a job, students who had been supported primarily as RAs were also more likely to take nonpostdoctoral research positions, often in industry. “That’s what told us that maybe something interesting is going on here,” she says.

Interest, but skepticism

The findings are drawing extensive interest, but also skepticism, from others who study the biomedical workforce.

The study was funded by NIH’s National Institute of General Medical Sciences (NIGMS) in  Bethesda, Maryland, which has the largest portfolio of training programs within NIH’s stable of institutes. The institute’s director, Jon Lorsch, heard Blume-Kohout describe the study in a talk last week at a workshop co-sponsored with NSF on the science of science policy held on the NIH campus. He says the study “has generated a lot of interest at NIH.” But Lorsch is far from being persuaded by her conclusion.

“We’re looking to see the results of additional studies in hopes that they will converge,” he says. “But I think it’s interesting that Meg has turned the negative implications of being on a grant on its head and said that in some circumstances there are implications that might be beneficial.”

Paula Stephan, a labor economist at Georgia State University in Atlanta who has written extensively on the academic research enterprise, thinks that the study may reflect recent efforts to give graduate students a better sense of their job prospects. “I spend a lot of time talking to unhappy postdocs, people who haven’t taken the time to think about whether it’s something they really want to do,” she says. “So maybe what this is suggesting is that fellows and trainees are exploring other career options, rather than seeing a postdoc as the automatic next step.”

Several researchers questioned whether the study accurately measures student interest in pursuing a research career. They note that graduating students may not spend much time answering the survey that provided data for the study, be unsure of their future, or be unfamiliar with how their graduate education was financed.

Blume-Kohout acknowledges those potential problems with the data but says that her econometric models support the actual findings. She also hopes that incorporating 2013 data not yet available to researchers will help her address another criticism, namely, that the study’s window is too short to provide a solid picture of the career paths of young scientists. “It would be more informative to look 7 to 10 years post-Ph.D.,” notes Howard Garrison, who tracks workforce issues at the Federation of American Societies for Experimental Biology in Bethesda, Maryland.

Even so, the experts praise Blume-Kohout for making headway on what they say is a very complex subject. “I tried to tackle some similar issues and could not find a satisfactory path,” says Jeremy Berg, Lorsch’s predecessor as head of NIGMS who is now at the the University of Pittsburgh School of Medicine in Pennsylvania. “It is a very hard problem given the available data sources.”