The NSF GRFP problem, 2020 edition

The National Science Foundation Graduate Research Fellowship Program (NSF GRFP) awards were announced today. And the familiar “Matthew Effect” pattern continues.

• The entire University of Texas system, with 167,028 undergraduate students, gets 55 awards (0.03% awards / student).
• The entire California State University system, with around 484,300 undergraduate students, gets 39 awards (0.008% awards / student).
• Harvard University, with 6,788 undergraduate students, gets 22 awards (0.3% awards / student).
• Yale University, with 5,964 undergrad students, also gets 22 awards (0.3% awards / student).
• MIT, with 4,530 undergraduates, got 68 awards (1.5% awards / student).

The success rate of Harvard and Yale is about ten times higher than the Texas system and about 45 times higher than the California State University system.

If you want a GRFP, go to an Ivy League university. Or some other well-resourced east coast university.

You can download the NSF GRFP data here. Scroll down and search for “Export options” to get an Excel spreadsheet.

It has been five years since Terry McGlynn did the important job of first pointing out how slanted the playing field is. It’s been four years since NSF announced changes to their rules to try to create more diversity. But the same pattern keeps happening.

Is this NSF’s fault? Not necessarily. The NSF doesn’t release data on how many applications come in from each institution (as far as I know), and that is almost certainly extremely uneven. The missions of something like the California State University system is very different from that of an Ivy League institution.

I still think we need to call the NSF on this every single year. Something like an NSF GRFP is the sort of thing that can make a career happen. An award from a national funding agency creates so many opportunities that it’s not clear that the goals of the NSF are best served by those award going to students from extremely well-resourced institutions,

They’re two years old now, but if this question interests you, please do see these two amazing posts by Natalie Telis.

• The price of a GRFP, part 1
• The price of a GRFP, part 2

This concentration of GRFP awards is puzzling given that for regular awards, NSF does think about its overall portfolio of awards and is aware of the problems with concentrating many awards in a few institutions. It considers factors like whether a grant is coming from an EPSCOR state, which is essentially a measure of how successful a state has been at getting NSF funding.

Related posts

Fewer shots, more diversity?
The NSF GRFP problem continues