Last month the National Science Foundation (NSF) reported that U.S. universities awarded a record number of Ph.D. degrees in 2014: 54,070, with 75% conferred in science and engineering. Those new graduates face gloomy job prospects, media reports said—“The Ever-Tightening Job Market for Ph.D.s” was a typical headline—citing data that only 61% had lined up a “definite employment commitment.” And things have gotten worse in recent years, many stories added, noting the 2014 employment number was lower than the 69% reported for the class of 2009.
Don’t believe everything you read. Dig deeper into the available data and you’ll discover that almost all of those new Ph.D.s are gainfully employed, thank you very much. But it helps to know what to look for, and where.
The 61% figure comes from the Survey of Earned Doctorates (SED), an annual census of soon-to-be graduates. But NSF also conducts a biennial sampling of all Ph.D.-level scientists and engineers in the United States. It’s called the Survey of Doctorate Recipients (SDR), and the 2013 report pegged the unemployment rate for the entire sector at a minuscule 2.1%. (By comparison, the national jobless rate that year stood at about 7.5%.)
The jobless rate for young scientists is no higher. A special analysis of the 2010 SDR data found that only 2.1% of Ph.D. scientists and engineers were unemployed 2 years after earning their degrees. And that number drops to 1.9% for those 3 to 5 years beyond their degree.
The nonexistent jobs crisis is a reminder of the dangers of taking government data at face value and using them for unintended purposes. The SED—the survey that prompted the press coverage—was never designed to measure the employment status of new graduates, says Mark Fiegener, a project officer at NSF’s headquarters in Arlington, Virginia. (NSF actually contracts out both surveys to the National Opinion Research Center [NORC], an independent research organization at the University of Chicago in Illinois. The SED reaches every graduating Ph.D. student, and has a response rate of 93%. The SDR is a sampling of 120,000 people with science and engineering Ph.D.s.)
One problem with trying to use the SED to gauge the job status of graduates is that students may fill out the survey months before they actually receive a degree and, thus, may not yet be focused on the next stage of their career. Another problem is that the students’ answers may not be an accurate depiction of their job prospects.
Take someone who defended her dissertation and filled out the SED in December, Fiegener says. That would be before the winter-spring academic hiring season. So she would be more likely to report “no commitment”—even if she lands a job by the time she graduates in May. In contrast, someone who completes the survey in March may have just returned from a job fair with a firm offer. So his answer would be yes.
Fiegener says a true employment survey would avoid that problem by using a reference date, such as, “Were you working for pay on 1 April?” But the SED does not, because postgraduation plans are only part of a broader suite of questions about the cost of education, field of study, and other aspects of the graduate experience.
Another complicating factor is how the SED defines unemployment. The government standard requires meeting two criteria: someone who is looking for a job and doesn’t have one. But for the SED, it’s enough to not have a job. That leaves out people who aren’t looking for work—for whatever reason—a category that applies to 4.5% of the supposed “jobless” graduates.
An even bigger factor is those who are actively negotiating with an employer but haven’t closed the deal. Some 8.8% of the respondents find themselves in that situation. Removing both categories from the SED tally would lower the 2014 Ph.D. jobless rate to 25.3%, far below the 38.6% that spawned the dire media warnings about a lost generation of Ph.D.s.
A closer look at the SED also casts doubt on the media’s assertion that conditions are worsening. The 61% that the 2014 SED report counted as employed is actually a composite of both those with a firm offer and those returning to a job they held before entering graduate school. And although the total dropped from 69% in 2009, the percentage heading to a new job after their Ph.D. has stayed the same for the past decade, at about 50%. It’s also worth noting that the overall total held steady for 15 years, a period spanning both the 2001 dot-com bust and the 2008 financial collapse, before dipping in 2014. That suggests the recent drop may be an anomaly.
The SDR—NSF’s broader survey, which paints a sunnier picture of Ph.D. employment—follows a sample of Ph.D.s for their entire careers. Although it reveals some volatility over the last 2 decades in the jobless rate for early-career scientists, the rate remains very low. For those within 2 years of earning their Ph.D., it stood as low as 1.3% in 1999 and 2010, and reached a recent high of 3.4% in 2012, found an analysis by NORC for the Federation of American Societies for Experimental Biology in Bethesda, Maryland. Still, even that number means more than 96 out of 100 new Ph.D.s were employed.
Demographers may be aware of all these caveats. But they can easily trip up a reporter trying to dig into the latest data on employment trends. The ambiguity is worth keeping in mind the next time you read an article warning about the dismal job market for scientists.
This story originally appeared in the 20 May issue of Science.