NSF Graduate Fellowships and the distant mirage of an equitable pipeline
It’s that time of year again. Congrats to the 2000 students who are recipients of the GRFP! From talking to so many panelists about their experiences, it’s clear that they could fund so more people, and every single one of them would be quite worthy of the support.
If there was such a thing as a Blog Citation Classic list for this site, then discussions about equitable distributions of NSF graduate fellowships would definitely be on there.
I can concisely encapsulate these concerns: Your odds of personally knowing someone who got a GRFP from your undergrad years might be best predicted by the size of the endowment of that institution. NSF is working hard to be inclusive with respect to gender, ethnicity, and various axes of diversity, but the bottom line is that students attending wealthier and more prestigious undergraduate institutions are more likely to end up with fellowships.
Since I started talking about this four years ago, I’m used to a bunch of folks coming out of the woodwork to tell me that my concerns are misplaced. They’ll say, but WAIT! You can’t give fellowships to people who don’t apply! It’s not NSF’s fault! The program officers are doing their best! The reviewers are wonderful! For example, this guy is as regular as clockwork:

This isn’t wrong. But it’s desperately eager to miss the point. This dude thinks I’m just trying to point fingers and whine, but I’m here working to discuss the problem and find a solution to a very difficult problem. So hey, step aside, man. We’re rolling up our sleeves. At least I’m gratified that while these guys are dismissing me, there are folks in NSF and in the scientific community that actually are paying attention to our discussion of this equity problem and how to deal with it.
NSF doesn’t release information about the composition of the applicant pool. We, the public, don’t know which institutions applicants are coming from, we only know who gets awarded. (And they have fully legitimate reasons for limiting access to this information. NSF’s really good about transparency, in general.) But even without this data, I’m quite darn sure that the universities that aren’t producing many undergrad awardees aren’t heavily represented in the application pool either. This year, the phalanx of awardees from MIT who landed awards this year vastly outnumber the dribble of awardees from the California State University (CSU) system, which enrolls about a half million students at the moment.
Yeah, it’s a pool problem. No shit, Sherlock McScientist.
So what are we gonna do about that pool problem, eh? We’ve known about it for a long time, and it isn’t fixing itself. Given the structure of the current program, it’s a capacity building problem. The undergraduate institutions that we all want to be generating more awardees aren’t generating enough applications.
Okay, so let’s look at generating institutional capacity to create a large number of applications.
Let’s be clear, when it comes to social mobility and increasing access to opportunity, regional state universities are where it’s at. Have a look at this Social Mobility Ranking list of universities and the CSU rocks. In my book, as metrics go, this is a helluva lot more useful than the US News and those other rankings. (My campus ranks 24th in the nation, according to these folks, by the way.)
So while regional state universities are doing the best at social mobility in US, we’re still struggling to get folks funded with prestigious national fellowships. That’s just a leap too far, apparently.
Let’s take a look to see what kind of institutions are punching above their weight. If you’re at all vaguely familiar with these data and the sociology of science and higher ed, then you’ll know that it’s the high-endowment small liberal art colleges (SLACs) that disproportionately produce PhD students and GRFP recipients. Here’s the 2019 data, if you’re not inclined to take my word for it.
I went to undergrad at a place like that. Each cohort of students at this entire college is about 600 students, and only a fraction of them are STEM majors. This year, there are three people who did their undergrads there who got the GRFP. (And congrats to them! I personally know one of them, and they’re a spectacular scientist. All of them are quite deserving, of course.) Meanwhile, my department has more bio majors than an entire cohort at that institution. And my university didn’t generate any GRFP recipients this go-round. To be fair and clear, I’m not even sure if anybody applied. I think a few applications might have gone out, but not to my knowledge at this writing.
I’m sure that particular SLAC generated well more than three applicants, if they landed three awards. And my campus generated far fewer applications. So, hey, as Dr. Sherlock McScientist keeps saying, nowhere to blame, right? Right. Let’s get past the blame, and talk about what’s going on, and then figure out how to fix it. How about everybody join in on the heavy lifting, shall we?
Why do some places generate far more applicants than other places? It’s because some places provide more support for students. Better access to research opportunities, more individualized mentorship from faculty, more support with academic writing, smaller classes and more professors who know students well enough to write more impressive letters. They might even have an office dedicated to helping students apply for awards like these. These students are also more likely to be working fewer hours in non-academic jobs, are less likely to have academics in their family, and more likely to ask for and expect assistance.
The universities that can’t offer all of these resources to all students often have some externally funded programs that try to serve the same function. Programs like LSAMP, McNair, MBRS, S-STEM, and so on. (Meanwhile, NIH decided to pull a bunch of this funding from regional state universities and let R1s control the purse strings. What a mistake.) These grants can’t reproduce the full ecosystem of support that happens in a high-endowment SLAC, but it’s better than not having it. But grants come, grants go, and in the midst of a large institution, these site grants only support a small fraction of individuals.
One thing we can do to fill this gap is to see if we can leverage existing resources to get more mentorship, advising, and writing support from other institutions. This has helped some students so far, and I’m working more to secure funding and institutionalize this. We’ll see. It’s another stopgap measure, but better than the status quo at least. (If you’re with a private foundation and want to support a program that can have huge impact for relatively low cost, please do give me a ring. Seriously. We’ll name it after you, if you wish.)
Having this support to generate students who are prepared to submit GRFP applications is all about the benjamins, because that support costs money. But wait, well, it’s not all about the benjamins. It’s also about institutional priorities, and leadership, and administrative stability. So it could be about allocating those benjamins to things that matter, and pooling resources to maximize impact.
Let’s own the reality that regional state universities are never going to be like high-endowment SLACs. Let’s take a look back at the GRFP recipient data, and see which regional state universities are punching above their weight. Hoooo, look, CSU Monterey Bay has two awardees this year! And they consistently are sending their undergraduates off to graduate programs at a solid clip. What are they doing better than most others?
How can we have what they have up at CSUMB? I took a field trip to find out. They gracefully spent hours getting me up to speed on how they run their shop, how they built it, and helping me strategize how to do something like this at CSUDH. They have what my Vice Provost called the “gold standard” of undergraduate research offices. Their Undergraduate Research Opportunities Center has a staff of nine or so. And they seriously got it going on. They’ve got the same site grants that we have at our university. The difference is that they’re not operating these grants out of separate silos, but it’s all run through a central office, which can use the efficiency of scale to support more people on campus. And when a grant runs out, they’ve gotten their administration to keep staff on with institutional base funds. This office provides so much more support undergraduates seeking research careers than comparison institutions. Because they’ve made it a fiscal and administrative priority. Like nearly all good things, this spectacular resource was built by small group of people with a clear vision who worked collaboratively to make things better.
It’s a bummer to realize that, despite an extraordinary level of excellence, CSUMB still produces fewer GRFP recipients per capita than wealthier institutions up the road in the Bay Area that you have definitely heard of. They’re doing it so well at CSUMB, and they’re still not able to replicate the privilege of high-endowment institutions. This is no surprise. Because whenever we innovate effective methods for supporting students on a budget, then these other places can just take our approaches and do it easier and maybe better with more money. Privilege will always win out.
This is why I’m talking about the equity problem, in addition to working to build capacity at the institutional level. Because no matter how great low-endowment institutions are at supporting undergraduate researchers, they’re still going to be experiencing a relative disadvantage. If we just pull the Dr. McScientist route, and say “Well, just do a better job preparing applicants to apply!” that doesn’t fix the actual equity problem.
I’m only concerned about blame if the party to blame can be shamed into fixing things. This situation is more complex. We’ve got to work together on this one. (The only ones here to blame are the ones who are too quick to absolve involved parties from the responsibility to step up their game because the status quo isn’t working well enough. And it’s easier just to just sidestep the blowhards and get to work.)
This is a pipeline problem. It’s not that the pipeline doesn’t work well enough. It’s that it works too well. Because the pipeline is built to serve the folks who built it. We can’t fix the pipeline problem by just building more pipeline at underfunded institutions (though by gosh that’s what my day job is), because the pipeline is always going to flow more smoothly at better funded institutions. The real fix to the pipeline problem is to dismantle the pipeline.
What does a dismantled pipeline to broaden participation look like? Can we just restructure the program to reduce the barriers to application, so that more applications come in from the underrepresented institutions? That’s a difficult problem because NSF is already overwhelmed with the amount of work to review the applications that it already receives. I think there are a variety of more creative solutions. (For example, just as NSF limits how the number of institutional applications for some award categories, they can limit the number of applications per number of enrolled students per institution. Which clearly would piss off places like MIT and Harvard and Harvey Mudd and whatnot, who get so many awards, but on the other hand, it would definitely result in a more equitable distribution of awards to highly worthy applicants. Or, the application could be restructured in a way so that the barriers to applying aren’t as great. Or NSF could create a program to fund institutional infrastructure to support research mentorship targeting institutions with low rates of application, and so on. But big organizations, even ones as nimble and data-driven as NSF, are slow to change, and a lot of smart people working from within are doing what they can. I know they listen, and I thank y’all at NSF for being so receptive).
I don’t know entirely what that looks like, though it will take all parties involved working together to make opportunities more equitably accessible. And, yes, that includes fellowship-granting organizations, to reconsider what their application and selection process looks like. I know there’s active effort going into tweaking processes for this effect, but as long as we keep saying it’s on underfunded institutions to level the playing surface, that then the fellowships will keep rolling towards students who had the social capital to access high-endowment institutions. I’m not prescribing the steps it takes to fix this, beyond what I’ve already said. I’m saying we need to have a broader conversation about it. We need to keep talking about it, and this growing conversation is the foundation for positive change.