Racial bias in NIH Grant review?
August 18, 2011
oh boy.
This is going to be explosive. Jocelyn Kaiser reports:
But an in-depth analysis of grant data from the U.S. National Institutes of Health (NIH) on page 1015 in this issue of Science finds that the problem goes much deeper than impressions. Black Ph.D. scientists—and not other minorities—were far less likely to receive NIH funding for a research idea than a white scientist from a similar institution with the same research record. The gap was large: A black scientist’s chance of winning NIH funding was 10 percentage points lower than that of a white scientist. [emphasis added-DM]
The report by Ginther et al is here, the key figure is below:
Let the race-splainin’ begin….
[h/t: Academic Lurker]
__
Additional commentary:
Sally Rockey, Office of Extramural Research
Tom Insel, NIMH
Updated:
Bashir
Chronicle of Higher Ed
National Public Radio
DrugMonkey 
August 18, 2011 at 9:35 am
What’s there to ‘splain? There’s obviously some bias here on the reviewers’ part.
LikeLike
August 18, 2011 at 9:36 am
Deadlinky? It says “not found”.
LikeLike
August 18, 2011 at 9:37 am
Ohhhh namnezia, you are so naive….
LikeLike
August 18, 2011 at 9:43 am
Well, I’m sure that if we don’t think too hard, we can come up with some explanation off the top of our heads and assume that the original study didn’t take those variables into account, because there was no in-depth discussion of those variables in the soundbite version we read.
This is the internet, after all.
LikeLike
August 18, 2011 at 10:26 am
The apparent effect on Asian PIs goes away (statistically) if the analysis is limited to those who were US citizens upon award of their doctorate.
LikeLike
August 18, 2011 at 10:51 am
White folks submit twice as many grants?
LikeLike
August 18, 2011 at 10:55 am
The counter arguments will essentially be the following:
-How do you know that group isn’t really just inferior? Or has gotten a lot of prior special help, that cause them to perform poorly when a truly objective measure is used?
-How could their be racial bias if I’ve never noticed it? I don’t recall anyone dropping racial slurs during a grant review. For there to be racial bias someone would have to stand up in a meeting and say something along the lines of “I am racially biased. I am scoring this grant slightly lower because of it.”
LikeLike
August 18, 2011 at 11:18 am
I don’t recall anyone dropping racial slurs during a grant review. For there to be racial bias someone would have to stand up in a meeting and say something along the lines of “I am racially biased. I am scoring this grant slightly lower because of it.”
Not to mention that some panels have good to over-representation of African-American scientists thanks to the diversity rules of panel appointment, Bashir. And there is no way in hell that every reviewer sees a PI name and knows or notices that the person is African-American. It is a thorny problem that the NIH is now facing, if they want to get to the bottom of this.
LikeLike
August 18, 2011 at 11:19 am
White folks submit twice as many grants?
I’m still trying to wrap my head around the “applicants” versus “application” numbers myself, brooksie. this is a distinction I’ve always wanted to see in the more general success-rate type numbers……
LikeLike
August 18, 2011 at 12:04 pm
It’s possible that what this reflects has nothing to do with how NIH reviews grants. It may be all in how scientists are trained and receive professional development. The idea being that black scientist do not on the whole receive the same level of advanced professional development, particularly grantsmithing.
That would not surprise me one bit.
LikeLike
August 18, 2011 at 12:35 pm
But marc, take the difference in resubmission rate. Even if down to mentoring it still interacts with the need to resubmit in the first place. That’s on the NIH.
LikeLike
August 18, 2011 at 12:41 pm
Or maybe white scientists just have more white buddies who sit on study section.
LikeLike
August 18, 2011 at 12:50 pm
Study section service decreases African-American PIs’ chances of being triaged, no such effect for White, Asian or Hispanic PIs, Isis. Chance to become buddies…or more likely to benefit from the “training” this provides?
LikeLike
August 18, 2011 at 1:08 pm
I have no idea for the reason, but Isis’ comment seems plausible- something relating to the size of the professional network or club. This is hard to control for- I need to check the study because I think they controlled for institution rank but maybe only at the post-doc level.
I wonder if to get at the bottom of this there could be a reviewer questionnaire. “Are you familiar with this applicant based on [check boxes] published work, attendance at meetings, same grad institution, same post-doc institution? Are you familiar with this applicants post-doc mentor…etc.” Something to get a handle on difficult to control variables.
Obviously it isn’t a good situation.
LikeLike
August 18, 2011 at 3:21 pm
I also suspect buddy effects. Buddy effects could reflect circles of trainers and trainees, but it could also reflect marginalization within research groups (i.e. minority scientists being less likely to be part of the buddy network that exists within a research group).
I also think, though, that straight out bias plays a role. It actually doesn’t matter if the reviewer is also a minority — the data suggests that most of us have an implicit bias against the certain minority groups in certain situations (I’m citing to the implicit association test data, which I find convincing). And, of course, the same kind of bias exists throughout their training periods.
I’m actually surprised to see less bias effects on hispanics and have a sneaking suspicion that this is because a sufficient number of hispanics in science are white, and, potentially not detectably different from non-hispanics.
LikeLike
August 18, 2011 at 3:34 pm
Funnily enough neurolover, the accompanying commentary from Tabak and Collins appears to be proposing IAT testing of reviewers!
LikeLike
August 18, 2011 at 4:59 pm
Question about the IAT: I took an IAT online, but I knew what the test was looking for, and I was nervous about the test finding racism. So, it was in no way a blind test. Has any research been done to determine if IAT results are valid when the test-takers aren’t blinded?
I’d like to think that the results were invalid because of that, but I have no idea if I’m right.
LikeLike
August 18, 2011 at 5:18 pm
Despite being at a historically-Black institution, I have a relatively limited sample size to draw from to offer any hypotheses at this point.
However, I put a good amount of credence in the opinions of my senior African-American colleagues such as Otis Brawley. Otis has always told it like it is and offers a variation on the buddy system – from The New York Times:
Dr. Otis W. Brawley, who is the chief medical officer at the American Cancer Society and is black, said the cause was not overt racism. “It’s not that they’re out to deny blacks funding,” said Dr. Brawley, who worked as an administrator at the National Cancer Institute, part of the N.I.H., in the 1990s.
Rather, it is more likely an unconscious bias, he said, with the reviewers more likely to give the benefit of the doubt to someone they are familiar with, and with black researchers tending to keep a low profile in the scientific world.
LikeLike
August 18, 2011 at 5:45 pm
To rephrase Otis Brawley, then: the problem isn’t racism, but the fact that the review system is totally broken. Reviewers are biased towards whom they know; they know fewer blacks, so they award fewer grants to blacks.
Of course, the hand-wringing, wailing and gnashing of teeth will all be about racism. (Actually, it already is: see Rocky and Insel links in DM’s post.) Which means that the true problem – that the review process doesn’t work as advertised, no matter how many Sally Rockies they pay to jump up and down and scream that it does – will remain unsolved.
LikeLike
August 18, 2011 at 7:55 pm
Grumble, that is an insufficient explanation given that other minority groups examined in the study don’t appear to suffer the same deficit in NIH awards.
LikeLike
August 19, 2011 at 2:21 am
This is pretty alarming. I worry about an unrecognized confounding factor, but the initial list seemed pretty relevant. Digging in and understanding this phenomenon would be a good idea.
LikeLike
August 19, 2011 at 3:12 am
I wonder what those numbers are for my social science…
1) People would recognize a black person’s name… there are so few of them!
2) Likely due to discrimination at the attending graduate school level (some of my colleagues disagree here, but I pull out Spence models), the quality of the black PhDs in my discipline is much higher than that of the average PhD. They blow us average folks away in productivity and quality and of work some of them are household names.
LikeLike
August 19, 2011 at 4:43 am
“Grumble, that is an insufficient explanation given that other minority groups examined in the study don’t appear to suffer the same deficit in NIH awards.”
The explanation is (potentially) sufficient *if* there are fewer blacks in science than other minorities. My very informal, anecdotal impression is that there are plenty of Asian scientists, a fair number of hispanics, and very few blacks.
One solution, then, would be to put more effort into increasing the numbers of black graduate students. That will eventually translate into more black faculty, more black grant applicants, more blacks in the good-old-boy network, and hence a more favorable view, among reviewers, of applications from blacks.
That will diminish the problem of a lower rate of grant success by black applicants. But will it diminish the real problem, which is that grant review is so strongly influenced by whom the reviewers know rather than the quality of the proposed science?
LikeLike
August 19, 2011 at 6:20 am
Isis says:
This makes a lot of sense.
LikeLike
August 19, 2011 at 10:39 am
Could another piece of the puzzle be tokenism, i.e. black scientists being pressured into more service work (committees, advising, etc) in order to increase diversity or the appearance thereof, thus leaving less time for grant-related activities? I realize that plays into an assumption that there is an actual difference in the quality of applications, rather than (or in addition to) a difference in the review process.
LikeLike
August 19, 2011 at 11:28 am
Due to long running historical inequities and the persistent and aggressive natural selection imposed by institutional racism, black scientists that succeed in becoming independent investigators occupy a state of intellectual mind that is more disposed to asking the hard questions and attacking the hard problems.
This, of course, places them at a distinct disadvantage when applying for NIH grants.
LikeLike
August 20, 2011 at 12:28 am
Give the disparity in size between the participants in the small groups and the large groups; this sounds like Simpson’s Paradox might play a roll here.
/Wiki
http://en.wikipedia.org/wiki/Simpson's_paradox
LikeLike
August 20, 2011 at 4:25 am
Unfamiliar with this but reading Wikipedia it appears necessary that the effect disappear or reverse when you examine subgroupings. That didn’t happen here, they present different models that account for various factors and the results are consistent with the entire sample.
The Asian disparity that disappears with nationality-upon-PhD-award might be related though…?
LikeLike
August 20, 2011 at 7:16 am
The Asian disparity may come down to language and writing because I guess I would presume those in the US for Ph.D. would have greater experience with English.
I think the study should have included attempts to normalize to post-doc and Ph.D. institution- this could get at community/buddy effects.
LikeLike
August 28, 2011 at 5:23 am
Additional comment from cackle of radness.
http://t.co/MLeiZer
LikeLike
December 5, 2011 at 7:32 am
[…] Bashir DrugMonkey New York Times Science NIH blog NIMH […]
LikeLike
January 11, 2012 at 5:13 am
[…] Remember the Ginther et al. (2011) report on NIH Grant awards to Principal Investigators sorted by race and ethnicity? The one that showed African-American PIs suffered worse success rates even when controlling for a number of obvious potential contributing factors? […]
LikeLike
April 6, 2012 at 9:48 am
[…] the wake of the Ginther report, this is a very nice step forward. It was not something I had considered before as a response to […]
LikeLike
January 13, 2014 at 11:13 am
[…] you know I am distinctly unimpressed with the NIH's response to the Ginther report which identified a disparity in the success rate of African-American PIs when submitting grant […]
LikeLike