How to further improve the NSF graduate fellowship program?
Small tweaks have targeted equity and access, but what bigger changes might work?
Here’s some really useful news that came out of the US National Science Foundation:
A new program targets funding towards applicants to the NSF Graduate Research Fellowship Program (GRFP) who received an Honorable Mention in the past three years and are currently in grad school in one of the ESPCoR states (see below). They might be in for some serious funding, according to this new solicitation. Please get the word out! Because stuff like this often flies under the radar during the summer, I’d like to make sure that eligible folks make avail of this opportunity.

I like this tweak on the program that is designed to get more funding towards people and places that have been (geographically) undersupported. If you’re looking for a way to build on the existing program without very little bureaucratic overhead to make more of a difference, I think this is a creative way to go about it, and good on them for making this happen.
The ongoing online conversation about the equity and access involving the GRFP trended towards a negative tenor. I think that this is because there are real issues involving this program that aren’t being addressed, either in policy or in actual conversation from those who are responsible for policy. To some extent, I think I kicked off this conversation by pointing out how students doing their undergrads in the largest university systems are getting a very small proportion of awards compared to the small fraction of students that are coming from well-endowed prestigious universities, which is still very much true a decade later. I think a lot of the negative tenor comes from the simple fact that gripes, complaints, and trauma get more attention than times when people and institutions are quietly effective and make positive changes. For example, complaints about gender equity in the NSF Waterman got lots of attention, but when NSF actually fixed the problem, there are no tweets about that going viral, eh?
I think this conversation needs to continue and I’d like to direct it into a positive trajectory. A lot of the complaints and concerns about equity and access for the GRFP cannot be readily addressed through immediate changes in policy or practice. I think there is an argument to be made that the program could use a major overhaul and be rebuilt from the bottom up, I also realize that this is not going to happen anytime soon because of bureaucratic inertia, and also I think there is a major risk involved in disrupting something that is working in many ways, even though there are some obvious laws that we keep harping on.
Which means that instead of merely complaining that the GRFP is simply one huge behemoth of a Matthew Effect that is perpetuating some of our biggest structural inequities, we need to “yes, and” that conversation to suggest practicable changes that would move things in the right direction.
For example, broadening the base of funding for people who are in grad school in EPSCoR jurisdictions is a good thing. Is it the exact change I’d make? I don’t think so?. Will it broaden representation in terms of identity beyond geography? I’m not sure, and I’d like to hear thoughts on that from people who have considered this. But it is a change that can be made to increase funding to places that have been getting less of it, in a way that fits institutional priorities and without creating a huge amount of bureaucratic overhead to make it happen.
Keep in mind here is that the staff of NSF are generally maxed out on the amount of time that they have available to make and run awards. There’s a small team that operates the GRFP year-round, but the review process seem to be a short term all-hands-on-deck operation once per year. While people are very quick to point out massive problems with how the review process is conducted, unless everything is radically restructured including a huge boost to the staff dedicating their time to it (which isn’t gonna happen, I suspect) then we need to think about changes are that are implementable.
Another thing I’d like to point out here is that a lot of the review problems are not because of any problem with values or priorities at NSF, but those within our own community who are establishing the community norms and standards on which we value and elevate students. The reviewers — that’s us — are the ones who are ranking proposals. If we think the rankings aren’t done appropriately, that’s on us. If you look at the instructions that NSF provides to reviewers about what to prioritize and how to holistically look at applications with respect to achievement relative to opportunity, I think the agency is doing pretty well here. For this problem, I’m far more disappointed with my peer reviewers than I am with the agency.
So, then, what kinds of changes might we suggest in the GRFP policies and process that the folks at NSF would be able to look at seriously and consider adopting, taking into account their workload, resources, bureaucracy, and constraints of federal law?
That’s not an easy question. Which is, perhaps, why the program hasn’t changed so much over the past nine years. To some extent I know that some folks have the attitude of, “If it ain’t broke, don’t fix it,” but I also there are plenty of people who do think it’s broke because of the equity problems but the fix is politically and administratively elusive.
Some lessons can be learned from the time when NSF used to fund students explicitly through a Minority Graduate Fellowship, but that ended in 1998 because of growing hostility to affirmative action and mounting legal and political pressure. NSF did manage to bring back the same level of representation in GRFPs without the use of the Minority Graduate Fellowship, but it took more than a decade of effort to make that happen. There are not legal limits to what NSF can do, but there are also policy constraints and practical considerations as well. Ultimately they’re NSF is reporting to congress and the GRFP is very small fraction of the budget of the entire organization. If they’re redirecting funding in certain ways that generate the ire of certain congresscritters, then that imperils the entire project. I understand that people on the raw end of these inequities aren’t placed in a position where they need to juggle these particular difficult realities because they have other difficult realities to deal with, but if we’re going to pose solutions, we need to come up with ones that can actually move forward in practical terms.
What might we suggest?
Here’s where I think might should start: taking into account the baccalaureate institution of applicants. Some applicants are current grad students, and some applicants are undergrads. While all applications require students list the institution where the applicant did their bachelor’s degree, this only factors into the application insofar as it may be subjectively considered by reviewers. However, if we are working to broaden participation and extend support to more students who come from backgrounds that are underrepresented in STEM, then it would make sense to find a way to make sure that a high proportion of awards to go students who are attending undergraduate institutions that serve this population. Considering the absurdly huge proportion of GRFPs to go people who, when they were in high school, managed to find their way into wealthy and exclusionary institutions for college, it might make sense to limit the number of awards that can go to students who completed (or about to complete) their bachelor’s degree at any particular institution. Most universities only produce a few GRFP fellows, which a small number of universities produce a ton. To fix that, what one could do is limit the number of applications based on the institution. Let’s say that the institution’s grant office had to invite nominated students to apply, and the number of nominations was determined by the size of undergraduate enrollment. This would increase the representation of students attending regional public universities but scale back the deluge of applications from R1s and small liberal arts colleges. Another way to do this, to have a less extreme effect, would be to have a fixed number of applications were baccalaureate institution, such as five. At many regional public universities, which probably don’t even generate five applicants per year, the creation of such nomination slots might even spur campuses to increase outgoing applications. Each institution’s research office could be responsible for deciding how these slots get allocated by their own internal criteria, and would include slots for recent graduates currently in graduate programs. What’s cool about this is that it would redistribute the applications more equitably but without increasing review or much more administrative burden. It’s eminently do-able and isn’t coded to target funding towards ethnic minorities or other things that might annoy congress.
Here’s another idea: change the application process so that undergraduates are not compelled to pretend that they’re going to attend a particular university even though they need to turn in their GRFP application before they’ve applied to the graduate program! This is a huge disincentive to undergraduates, especially those without social capital, because it’s entirely unclear to undergrad applicants what they’re supposed to be doing. This might seem like a minor change to some people but I’ve seen a lot of undergrads just stop working on their applications when they realize they are tacitly expected to write research statements that have enough detail that they are often tied to a particular location, laboratory, or university.
Another idea: stop holding panels over the winter break! There are a variety of way to do this, but let me tell you that people with high teaching loads are highly disincentivized to participate in panels when they’re going to be expected to do panels in the middle of their winter break, or right at the end or beginning of a semester. And you want panels with more faculty at universities with higher teaching loads, because you’ll get in more diverse perspectives.
Also: How about NSF support a centralized organization that provides support to applicants who are packaging together their applications? When I’ve talked to some folks about this at NSF, they’ve indicated that supporting applicants is the responsibility of the institution. But let’s own the fact that if we’re trying to genuinely broaden representation, that means we need to increase the quantity and quality of applications coming from students who are enrolled in institutions that have less infrastructural support for developing competitive applications (and are not in laboratories with doctoral students who support them). While programs like EEB Mentor Match, GEMS, and Cientifico Latino do their best to connect students with people who have the bandwidth and expertise to help them develop competitive applications, wouldn’t it be cool if NSF competitively funded hubs designed to do this (and perhaps more broadly, foster career advancement for STEM students at ERIs). Right now the GRANTED program can help this kind of work but only at the institutional level.
What are some other kinds of tweaks that you think would help improve equity and access for GRFPs that wouldn’t involve massive restructuring the whole thing? Thoughts?