The National Science Foundation’s Graduate Research Fellowship (GRF) is an extremely prestigous, three-year fellowship awarded to U.S. citizens pursuing postgraduate studies in science, technology, engineering, math and STEM eduction.
It is well documented that GRF award rates are not equitable across institution categories. Students who complete their undergraduate degrees at prestigous and/or R1 institutions receive a disproportionate number of awards; likewise, students currently enrolled in postgraduate studies at prestigous or R1 school receive an outsized number of awards.
This page presents some visualizations of these trends, based on 2021 GRF award data from the NSF’s website.
The first way we can look at these data is to simply examine the total number of GRFs awarded by undergraduate institution. In the bar graph below, I have color-coded each bar based on “institution type”: Ivy League, Ivy League-equivalent, historically Black college or university (HBCU), tribal college or university (TCU), and other. You can mouse over each bar to see which institution the bar corresponds to, and how many GRF awardees that institution had in 2021.
From this first visualization, you can see that Ivy League universities dominate in terms of total number of awards.
Of course, we know that colleges and universities vary wildly in the number of students that they have. Here, we can see the number of GRFs awarded on a per-student basis.
The Ivy Leagues still dominate (though to a reduced extent, and Columbia and Penn have really fallen off), but we now see that small, selective liberal arts colleges are dominant as well.
One factor that might be causing these patterns is endowment size. Here we can see the relationship between endowment dollars per student and GRF awards per student.
The relationship between the two variables is highly significant (p < 2.2e-16), and moderately positive (R = 0.55). Some institutions really overperform, like Harvey Mudd College, and others, like Princeton, underperform.
NSF also reports awardees’ current institutions. This data is somewhat less clear to interpret, because it reflects a mixture of awardees’ postgraduate institutions, and undergraduate institutions for those awardees who are still undergrads. Nevertheless, we can examine the trends in this data as above.
When looking purely at raw numbers of awards, we see that Ivy League universities again dominate. Compared to the effect of undergraduate institution, current students at HBCUs are awarded even fewer GRFs, demonstrating inequity by current institution type.
When visualized as number of GRFs awarded per student, the Ivy League institutions are even more dominant.
Finally, we can consider the relationship between endowment per student and number of GRFs awarded per student, by the student’s current location. Here, we see an even stronger relationship between endowment and award rate than we did for undergraduate institutions, suggesting that the resources available, or possibly the prestige associated with, a student’s current institution is even more important than those of their undergraduate institution.