Making proclamations about the scientific enterprise based on sparse employment and career data about junior scientists has become a common endeavor. But this approach is fundamentally flawed. The unemployment data for Ph.D. scientists is so beset by caveats, exceptions, and holes that it is essentially useless for informing the ongoing discussion about whether the current research enterprise serves trainees as it should.
A case in point is Jeffrey Mervis’s recent piece “‘Employment crisis’ for new Ph.D.s is an illusion.” Mervis used data from two National Science Foundation (NSF) surveys to show that, despite recent headlines about a Ph.D. unemployment crisis, the vast majority of recent Ph.D. recipients are “gainfully employed” 2 to 5 years after earning their degrees. It’s true that the survey data suggest high employment levels for Ph.D. holders, but focusing simply on whether or not they are employed misses the larger question: Is this employment in fact gainful? Do these Ph.D.s hold jobs that call on their training and make it a worthwhile investment—both for the individual graduate and for the larger research enterprise, which uses public funding and invests significant effort in training them?
The short answer is that we just don’t know. Mervis drew on the best available data we have at the moment: the Survey of Earned Doctorates (SED) and the Survey of Doctorate Recipients (SDR). The SED acts as a census, taking a snapshot of those earning doctoral degrees in the United States at the time their degree is conferred, and the SDR aims to follow a subset of Ph.D.-level scientists in the United States through their careers. But neither the SED nor the SDR nor any other large-scale data sets tell us if Ph.D.s are underemployed, either in positions that do not justify their lengthy training or where they do not enjoy reasonable job satisfaction. Simply put, the available data render a discussion of whether a Ph.D. is a good investment futile.
Because after many years of graduate work, with annual stipends often below $30,000 and frequent expectations of working significantly more than 40 hours per week, almost any kind of work would be appealing to a freshly minted Ph.D. In other words, many Ph.D.s take jobs that are likely far from their ideal. The increasing number of adjunct faculty members, for example, suggests that Ph.D.s are willing to accept part-time, low-paid positions. As an educated, experienced population with a diverse skillset, we should not be surprised that most Ph.D.s can find some employment, but this fact alone does not justify their training.
There is another problem with focusing on employment within the first 5 years after graduation: This period encompasses the time that many trainees spend in postdoctoral positions. Postdocs are intended to be temporary posts that are still part of the training process in academia, acting as crucial career stepping stones to the careers these researchers ultimately desire. Unfortunately, for many, completing a postdoc does not helpfully advance their careers. (It’s also worth noting that, according to the 2010 SDR, 11% of postdocs in the life sciences and 17% in the physical sciences reported lack of other employment as their primary reason for doing a postdoc.) Although a postdoc is technically employment, it is somewhat misleading to lump together this distinct pool of trainees holding explicitly temporary positions with other Ph.D. holders who have moved on to permanent positions that are likely to offer more stable, predictable career progression options.
Tracking postdocs’ career paths is quite challenging; we cannot even pin down the number of biomedical postdocs to within a factor of two. The best solutions to date for figuring out their career outcomes include trawling the Internet to track postdocs from labs with funding from National Institutes of Health T32 institutional training grants, which require that all lab personnel be listed, and manually curating a postdoc outcome database using piecemeal data from a variety of sources, both of which are very labor intensive and potentially error prone. Moreover, these methods leave many postdocs effectively untraceable. With so many Ph.D.s on a career trajectory that is so poorly studied, and about which recent data paints a worrying picture, talking about unemployment becomes convoluted and, to some extent, meaningless.
All that is certain from the currently available unemployment data is that almost all of the Ph.D.-level scientists who are counted by the NSF surveys—which exclude those with Ph.D.s from non-U.S. institutions and other graduate-level degrees such as M.D.s—are employed in some way, shape, or form. But that doesn’t mean that there aren’t very real problems with the academic enterprise that must be faced. As scientists, analysts, and policymakers, we must treat the data—and the lack thereof—with care as we work to reform the system so that it serves everyone.