It’s unusual for a science journal to publish a paper indicting the National Institutes of Health (NIH) for racism, but these are unusual times. The journal happens to be the very prestigious organ Cell, in which nineteen women scientists (I believe they’re all women of color) have indicted the NIH for funding black scientists at only 55% the rate of white ones. Further, they explicitly attribute this disparity to racism against blacks. (Asians aren’t mentioned, so I’m not sure whether they were excluded from the data or lumped in with whites.)
The paper, however, is misguided in several important ways. First, as I mentioned, it blames the funding disparity—which has persisted for over a decade—on racism, despite the fact that there are several other explanations, two of which have been supported by data. Second, while the authors cite data that, they say, shows racial bias against scientists, they omit the most relevant data for their contention, which shows no bias against NIH funding of black or female investigators in an extensive trial. They also fail to cite data showing that a lot of the funding disparity is based on racial differences in the kinds of problems investigated, and problems which have different funding rates for investigators of all races. Third, the group (henceforth, “Stevens et al.”) recommends a number of fixes of the system that seem to me manifestly unfair, even as forms of affirmative action. In fact, some of them may be illegal given that the NIH is a governmental organization.
The automatic ascribing of inequity (differential representation) to racism when there are other hypothesis is what I discussed yesterday as the “Diversity/Bias” fallacy, though I’ve since learned that people have given it another name. I claim no priority: it’s an obvious fallacy, especially to a scientist used to sussing out all possible explanations.
As I’ve said, I believe in the value of affirmative action in soliciting graduate students and hiring professors, as well as in other areas of social advancement. But I’ve also stated that this kind of preferential treatment in academics, which I see as valuable for several reasons, should stop at the time someone is hired as a faculty member. At that point, since the hired person has been deemed qualified for the faculty/research position, the affirmative action should cease and the candidate should compete for grants, promotions, and honors on an equal basis with everyone else. On this basis I would thus object to the methods suggested by this consortium to boost the scores and funding for black scientists. (See part 2 of this post for those methods.) Stevens et al. also neglect other disadvantaged groups like Hispanics—perhaps because data on funding disparities don’t exist.
You can see the paper (it’s really a scientific “op-ed”) by clicking on the screenshot below, and you can find the pdf here . The full reference is at the bottom.
Here are the authors’ statements (indented) that indict the NIH and modern science for systemic racism.
We are at a historic moment in time: a mainstream awakening to the pain that stems from racial injustice, with our scientific communities openly acknowledging that our practices promote racial inequity and disparity (Barber et al., 2020; Cell Editorial Team, 2020).
The evidence adduced above consists of the preponderance of white patients in clinical investigations and medical databases (which could reflect bias and racism), and some claims of grant disparity due to “systemic racism,” which, upon investigating those references, don’t show systemic racism or even any bias against blacks. More assertions:
The first study documenting racial disparity in NIH funding hit the field like a shockwave in 2011 (Ginther et al., 2011). This study showed that award probability for Black principal investigators (PIs) in 2000–2006 was ∼55% that of white PIs of similar academic achievement (Ginther et al., 2011). NIH scrambled to study potential reasons for this injustice (Barber et al., 2020; Erosheva et al., 2020).
Note that they characterize this as an “injustice”, which implicitly means racism, unless the disparity itself is seen as “unjust”. (If you believe in true “equity” with success rates of groups in the same proportion as those groups occur in the general proportion, and disparity of outcome might be considered “unjust”.)
Here’s an explicit mention of the field’s racism:
The NIH director and leadership must recognize that its previous approaches, most of which have focused on filling the “pipeline” without simultaneously addressing our profession’s systemic racism, have failed. NIH must change course.
Direct indictment of racism in the NIH funding system:
We ask: if racism is present in academia, how can it not be present in NIH grant review and research, which are performed by academics?
. . . . For example, the NIH should study the cultural competency and unconscious bias harbored by its reviewers, differential review practices and funding disparity between NIH and the National Science Foundation (NSF), and why “matching criteria” (Erosheva et al., 2020) affecting the funding disparity gap.
. . . Silence is complicity. The continued persistence of a racial funding disparity suggests that the scientific workforce, including the NIH leadership, does not understand nor is adequately equipped to recognize and respond to this racism.
There’s this, including some self-indictment:
Faculty colleagues, we respectfully suggest that it is time for us to acknowledge that we—yes each of us, including many of the authors here—have unintentionally contributed to racial inequity in our profession. As just one example of the insidious nature of systemic racism, many studies have shown that we judge CVs and resumes differently based on the name of the applicant alone with both racial and gender bias, even if these CVs are otherwise identical (Eaton et al., 2020, Henry et al., 2017).
And I find this particularly invidious: the authors demand, in their list of anti-racist changes in NIH funding procedure, to score grants higher if the proposal contains diverse (i.e., black) investigators:
We ask: why is “diversity of the investigator team” not a scorable criterion in NIH grant review and priority for funding?
Be careful with responding, because one answer is racist and the other is not.
Seriously? This is a debatable issue, and one answer does not automatically make one “racist.”
I’ll talk about the c.v. data in a second, and will claim that it’s largely irrelevant to the entire funding procedure of the NIH.
The paper has one figure arguing that antiracist reviewers are needed to eliminate the funding and career disparity between black and white scientists—implying that racist reviewers expressing “group think” account for that disparity.
Note above that the cited evidence for racism in judging grants rests on two papers showing differences when scientists judge c.v.s of black and white investigators. I could access only one of these, the Eaton et al. paper, which does show disparities between sexes and races in hireability, likeability, and judgements of competence when made-up c.v.s of potential postdocs were assessed. (Note that these aren’t grant applicants or potential faculty members being assessed, but postdoctoral candidates.)
The result: Men were judged significantly more competent and hireable, as postdocs while women were judged more likable. For race, “White and Asian candidates were rated as more competent and hireable than Black and Latinx candidates across departments. Likeability ratings were not found to differ significantly by applicant race.”
So this does show apparent sexism and racism in assessing c.v.s of graduate students that are made up. However, another survey of faculty and student hiring preferences show that minority candidates and women are preferred above white candidates and males, especially by faculty. (This is for hiring faculty members, not postdocs.) That study was done not using fake c.v.s but assessing attitudes, and if you were wedded to the hypothesis of racism, one could say that the respondents were simply distorting their own views.
However, all of these disparate results become irrelevant when you look at an important study that, surprisingly, was not even cited by Stevens et al. themselves. It shows, through the use of reviewers assessing (as an exercise) actual NIH proposals whose authors were randomly identified as black, white, male, or female (names were switched around), that there was neither gender nor racial bias in scoring proposals. Since c.v.s are part of NIH proposals, these assessments would include weighing the candidates’ c.v.s and research productivity. Most important, this is an actual study, done in large numbers, of how grant proposals are adjudicated by the NIH (the authors used reviewers who had previously reviewed for the NIH). The study was published in 2019, so the authors of the Cell piece above certainly knew of it. They just omitted it.
Click on the screenshot to read.
I don’t want to go into detail here, but the researchers used genuine NIH proposals that were either funded or unfunded (24 of each), and sent three proposals to each of 412 scientists who had reviewed NIH grants before. The names were changed to reflect names associated with gender or race, and precautions were taken to prevent the reviewers from checking whether the applicants actually existed. You can read this for yourself, but here’s the conclusion from the abstract:
We find little to no race or gender bias in initial R01 evaluations, and additionally find that any bias that might have been present must be negligible in size. This conclusion was robust to a wide array of statistical model specifications. Pragmatically, important bias may be present in other aspects of the granting process, but our evidence suggests that it is not present in the initial round of R01 reviews.
It’s puzzling that this result isn’t mentioned by Stevens et al., since it’s the most important study bearing on the possibility of bias in NIH reviewing.
So if there’s no bias in reviewing, why do black investigators get NIH funding only 55% as often as do white investigators? One possibility, which is supported by another study, is that black investigators choose areas of application that generally have lower funding rates. And, indeed, the paper below supports that (click on screenshot; see a shorter summary here):
Here’s a summary from the abstract:
Notably, AA/B [African-American/Black] applicants tend to propose research on topics with lower award rates. These topics include research at the community and population level, as opposed to more fundamental and mechanistic investigations; the latter tend to have higher award rates. Topic choice alone accounts for over 20% of the funding gap after controlling for multiple variables, including the applicant’s prior achievements.
In other words, black investigators tend to propose “applied” studies, including those involving community intervention, health disparities, fertility, adolescent health, and so on, that are directed more towards solving social problems than other proposals involving “pure” research, and these studies are funded at lower rates—regardless of the investigator’s ethnicity or sex—than are “pure research” studies.
Of course, one could argue that this involves bias of a sort as well—bias against applied research. One could also argue that this kind of work is seen as less important because it’s considered an area of interest to minorities.
Now this factor accounts for only 20% of the disparity in funding between black and white scientists. An earlier study that I haven’t yet seen shows that “rack record” (i.e., accomplishments as recorded on the NIH c.v., which includes papers published that resulted from previous grants or other funding) to have an even larger effect on rate of funding. That shows that the track record of black scientists is rated lower than white scientists in funding, but, as we saw above, racism itself, as opposed to this index of previous accomplishment, wasn’t found to contribute to funding scores. Track record is not a funding problem, but a “pipeline” problem whose solution is complex. But, as I’ve suggested, widening the pipeline—assuring equality of opportunity for all at the outset of life—is the best and most permanent way of treating people fairly.
The study above suggests at least one route for reducing the funding disparity: upgrade the importance of “applied” research involving community intervention, public health, and so on. Because those fields disproportionately attract black investigators, you could fund more minority scientists simply by diverting more money to these areas. In this way you increase equity without any form of affirmative action. (Of course, upgrading areas that tend to attract minority investigators could be seen as a form of affirmative action.)
I see this post is getting too long, so I’ll break it into two sections. In this one we’ve seen that the claim of systemic racism is not a data-supported explanation for the lower funding rate of black investigators. What the data show for funding is that race or sex don’t seem to be important, but that track record and choice of research area do. To create more equity in funding then, one has to consider interventions that would improve the track record of black scientists (this is a complex problem that begins well before grants are submitted) or give more money to areas of applied research.
However, the Cell paper of Stevens et al. suggest affirmative-action interventions instead: when evaluating grants, we should give black investigators higher scores, special interventions, and different treatment. In the next and final bit of this analysis, I’ll discuss these recommendations.
Stevens, K. R., K. S. Masters, P. I. Imoukhuede, K. A. Haynes, L. A. Setton, E. Cosgriff-Hernandez, M. A. Lediju Bell, P. Rangamani, S. E. Sakiyama-Elbert, S. D. Finley, R. K. Willits, A. N. Koppes, N. C. Chesler, K. L. Christman, J. B. Allen, J. Y. Wong, H. El-Samad, T. A. Desai, and O. Eniola-Adefeso. 2021. Fund Black scientists. Cell. DOI: https://doi.org/10.1016/j.cell.2021.01.011