Information on the biomedical labor market is necessary both for the formulation of policies that ensure its sustainable future as well as for informing individual career decisions. After announcement in Science, a coalition of universities pledged to release information on all of their biomedical graduate students AND postdocs.

 

The first set of data was released on February 1st 2018, focusing on admissions and demographics data about Ph.D. students. The next set of data released July 1st 2018 includes:

  • Number of postdoctoral researchers
  • Demographics of postdoctoral scholars by gender, underrepresented minority status, and citizenship
  • For some institutions, length of postdoc and career outcomes.

The data can be accessed from this page by institution and we have updated our career outcomes tracking resource with this information. Data is reported by institution and again FoR congratulates UCSFJohns HopkinsUniversity of WisconsinFred Hutchinson Cancer Research CenterUniversity of PennsylvaniaUniversity of Maryland Baltimore CountyCornell University and Weill CornellDuke UniversityMIT, and University of Michigan for leading this movement and releasing this data.

 

Below, we discuss some highlights of these datasets that illustrate the importance of having this data available. We are urging other universities to join the NGLS coalition to demonstrate their commitment to transparency and stewardship of the biomedical research enterprise – interested institutions can get in touch with the Coalition at CNGLS@JHU.EDU

 

The coalition has laid out a roadmap with important milestones for releasing trainee information in a progressive fashion, and the next data release scheduled is October 1st for time in postdoctoral training at each institution.

 

Highlights of the latest data release

Demographic representations

It is important to remember that, in contrast to graduate admissions which are carried out at a program- or department-wide level, the decision to hire a postdoc usually rests entirely with a Principal Investigator, with Human Resources complying with the hiring decision of the PI. Therefore the demographics of postdocs more accurately reflect the hiring behavior of PIs than of institutions per se. For example, we might therefore expect that as more of the professoriate (especially the senior professoriate) is male, phenomena such as those reported in “Elite male faculty in the life sciences employ fewer women” may be seen in the data coming out from an institution, or department as a whole.

 

Hopefully therefore such data can help an institution appreciate what interventions may need to be made in the diversity of their postdoctoral population, over which they exert less control than they would in graduate admissions.

 

Overall, the total postdoctoral population tends to be male, largely foreign, and have a lower population of underrepresented minorities than would be expected in the U.S. population. These demographics can all be seen to vary by department/school on a deeper dive into the data, as shown below for some examples:

 

UCSF – all postdocs

 

Duke – all postdocs

 

Fred Hutchinson Cancer Research Center – all postdocs

 

UMBC (summed over all years) – all postdocs

 

University of Wisconsin-Madison – all postdocs

 

University of Michigan – life sciences vs other disciplines

 

MIT – all postdocs vs Biology department for gender and URM/international status

 

 

Cornell – all postdocs

 

Cornell Life Sciences

Cornell Social Sciences

Length of postdoc

While this is not expected until the next release of data on October 1st, some institutions have included length of postdoc or time spent in postdoc so far in their data releases, and some of these are for the institution overall, whereas some are broken down into schools and departments. It is important to remember that the length of a postdoc is expected to vary by field (generally expected to be around one year in the humanities and on average 4-5 years in the life sciences) and also some institutional policies “cap” the length of the postdoc at 5-6 years, associated with a change in title, if not necessarily a change in duties. These factors could therefore affect interpretations of data about the length of the postdoc.

 

Nevertheless, some of the most striking data has been shared by UCSF, perhaps warranting further study in the area of retention. UCSF has captured the duration of someone’s postdoc each year that they leave (separation year) and has plotted this out for all postdocs across all fields of study:

 

 

This varies year-to-year, which may have some effect on the trends we now describe. Nevertheless, over time, the median length of a postdoc for someone from an underrepresented minority has been less than that for the well-represented population. While the sample size is small, and in 2015 both groups had a median postdoc length of 2.8 years, in some years, such as 2017, there is a near-twofold difference in median length:

 

2015:

 

2017:

 

 

 

 

An interesting trend can also be seen in the international postdoc population, who in 2011 had a longer median postdoc length than the domestic population, but over time a trend clearly emerges that reverses these positions:

 

2011:

 

 

2017:

 

That UCSF has gathered and shared these data are so crucial to enabling us to ask further questions about what is happening during the postdoc. Are career decisions being made earlier? Are successful transitions being made into satisfying careers? Or are there issues with retaining people in positions and creating inclusive environments? Are these trends even meaningful? This data is critical to assessing the state of the postdoc, and then being able to use that data to assess whether to, and then to justify, making interventions at the institution.

 

Likewise, the median duration of the postdoc seems to be short at Cornell in the life sciences – tracking where people go after leaving at various years of postdoctoral experience (e.g. faculty/industry/another postdoc/exiting the workforce) could shed light on the behaviors of different cohorts, and the value of postdoctoral experience:

 

Cornell Life Sciences

 

Trends in postdoctoral length are also important to track using such data collection efforts – is the apparent doubling of the length of the social sciences postdoc at Cornell in a matter of years a cause for concern, or explained by anomalous data/a particular cohort? With such data, institutions are again better able to make interventions if necessary:

 

Cornell Social Sciences

 

Postdoctoral Outcomes

The Fred Hutchinson Cancer Research Center has also presented outcomes of all their postdocs:

 

Have you found anything interesting or have questions about what this data means? Please comment below or tweet at @FoRsymp!