If you’re doing basic scientific research in the US, here’s a new must read. This preprint by Chen et al. dropped on Friday, entitled “Decades of systemic racial disparities in funding rates at the National Science Foundation.”
Using over 20 years of data on funding rates, they demonstrate that white PIs have been getting funded at higher rates than non-white PIs. It feels like the scope of this preprint is similar to what Ginther et al. documented for NIH in 2011. Since that time, the Ginther Gap has been central to discussions involving disparities at NIH.
I think the figures speak for themselves, so I’m going to just share some of them:
Figure 1 shows that white folks have consistently been getting more funding across the years.
Figure 5 shows how ad-hoc review scores are particularly lower for black scientists:
Figure 3 shows that funding rates in general are lower for research awards compared to non-research awards, and that the disparities in these research awards are greater:
This should come as zero surprise to us. It should be expected. Not because of anything notably different about NSF, but simply because it’s an institution here and racism is baked into (perhaps) everything. I think Dr. Dzirasa says this well:
One thing raised in the discussion is that disaggregated data are needed to understand the nature of this challenge and to design appropriate interventions. For example, how does this intersect with gender? (Also, those of us in Small Ponds are presumably also interested with how this breaks down by institution type. I can hypothesize that the disparities would be greater in non-R1s, but I can also hypothesize just as credibly that they would be smaller. Or perhaps there’s no difference. If you’re targeting interventions, this is stuff is needed to ensure that resources are well targeted.)
There are a lot of straightforward interventions that are suggested in the discussion of the paper. One of the bigger suggested lifts is to update our perspective on the myth of meritocracy. As Chen et al. explain, “The use of merit review criteria to find and fund ‘the best ideas and the best people’ is motivated by a shared understanding that the integrity of research knowledge relies on individual and collective adherence to principles of objectivity, honesty, and fairness. However, a vast body of research shows that systems designed to facilitate impartiality and merit-based rewarding can instead perpetuate the very biases they seek to prevent.”
One thing to keep in mind that this is a preprint, and it’s not clear how additional peer review might result in revisions. However this manuscript is clearly getting lots of peer review. However, so far it has a lot more “likes” on twitter than downloads from the website. So it’s getting a lot of conversation, but not we still need a lot of us to dig in.
Drugmonkey has posted a more detailed take on Chen et al. Considering his substantial experience with the impact (or the lack thereof) of the Ginther report over at NIH, then he’s well qualified to interpret what’s going on over at NSF. I’ll leave you with his bottom line: “Go read it. Study. Think about it.“
One problem I see w/ interpreting such data is that racial/ethic information isn’t included in NSF proposals sent out for external review. That said, clearly names can sometimes reveal this information, and at a very high rate for some groups.
The evidence of bias is the evidence of bias. While the paper presents some evidence that this is (partially) based on ad hoc reviewer ratings, it goes much much deeper than reviewers knowing or not knowing the identity of the authors of the proposal. For example, one things we’ve learned from the Ginther situation at NIH is that scientists of color are getting funded at lower rates because the things that they identify as scientific priorities are not as high priority for NIH. They’re putting in proposals for areas that are being relatively underfunded. This paper isn’t a “reviewer gotcha” – it’s a heads up that there are structurally wrong problems that require structural solutions.
This is interesting and simultaneously discouraging. I think it indicates that reviewers have an subconscious bias to some degree. I’m wondering what it is about minority proposals is leading to a lack of funding. The verbage or diction? Or is it simply trickle down effects as white scientists have usually had more opportunities in life to train more for a goal like this. As the other comment mentions, it’s not like the reviewers can consciously tell someone’s race from their proposal. I have not read the preprint at all yet, but I am guessing that this may have something to do with the fact that a higher number of PI’s at R1 institutions happen to be white, which may be contributing to this disparity.
I have a lot of thoughts here, I don’t want to go crazy writing it all but I definitely feel that those of us trying to do research at “Non-R1” institutions, which are often MSI’s and/or HSI’s are subtly pushed to apply to the “non-research” grant categories. The ones that train teachers, or give students scholarship money, or pay for mentors from nearby businesses to work with students, etc. All worthy things to fund, but NOT the kinds of things that are going to make a PUI/small college lab or department into a place for students to learn to be scientists from mentor-professors.
I get that MSI’s and HSI’s are the same as PI’s of color, but the two systemic problems seem related, and definitely to the “research” vs. “non-research” issue that Chen et al identified.
Does anybody else feel this at their institution or in their attempts at grant-acquisition? Thanks- JS
should say “MSI’s/HSI’s are NOT the same as PI’s of color” oops. -JS
Just for myself, the admins on my campus aren’t necessarily exerting a greater pressure to get training/education/site grants more than research grants, but a lot of us do generally feel a greater pressure because there is so much pressing need for student support that the university can’t afford to provide.