When NIH uses affirmative action to fix a bias

Jul 20 2018 Published by under Anger, Fixing the NIH, NIH, NIH Careerism

We have just learned that in addition to the bias against black PIs when they try to get research funding (Ginther et al., 2011), Asian-American and African-American K99 applicants are also at a disadvantage. These issues trigger my usual remarks about how NIH has handled observed disparities in the past. In the spirit of pictures being worth more than words we can look up the latest update on success rates for RPG (a laundry list of research grant support mechanisms) broken down by two key factors.
First up is the success rate by the gender of the PI. As you can see very clearly, something changed in 2003. All of a sudden a sustained advantage for men disappeared. Actually two things happened. This disparity was "fixed" and the year after success rates went in the tank for everyone. There are a couple of important observations. The NIH didn't suddenly fix whatever was going on in study section, I guaranfrickentee it. I guarantee there were not also any magic changes in the pipeline or female PI pool or anything else. I guarantee you that the NIH decided to equalize success rates by heavy handed top-down affirmative action policies in the nature of "make it so" and "fix this". I do not recall ever seeing anything formal so, hey, I could be way off base. If so, I look forward to any citation of information showing change in the way they do business that coincided directly with the grants submitted for the FY2003 rounds.
The second thing to notice here is that women's success rates never exceeded that for men. Not for fifteen straight Fiscal Years. This further supports my hypothesis that the bias hasn't been fixed in some fundamental way. If it had been fixed, this would be random from year to year, correct? Sometimes the women's rates would sneak above the men's rates. That never happens. Because of course when we redress a bias, it can only ever just barely reach statistically indistinguishable parity and if god forbid the previously privileged class suffers even the tiniest little bit of disadvantage it is an outrage.
Finally, the fact that success rates went in the tanker in 2004 should remind you that men enjoyed the advantage all during the great NIH doubling! The salad days. Lots of money available and STILL it was being disproportionately sucked up by the advantaged group. You might think that when there is an interval of largesse that systems would be more generous. Good time to slip a little extra to women, underrepresented individuals or the youth, right? Ha.

Which brings me to the fate of first-time investigators versus established investigators. Oh look, the never-funded were instantly brought up to parity in 2007. In this case a few years after the post-doubling success rates went in the toilet but more or less the same pattern. Including the failure of the statistically indistiguishable success rates for the first timers to ever, in 11 straight years of funding, to exceed the rates for established investigators. Because of affirmative action instead of fixing the bias. As you will recall, the head of the NIH at that time made it very clear that he was using "make it so" top-down heavy handed quota based affirmative action to accomplish this goal.

Zerhouni created special awards for young scientists but concluded that wasn't enough. In 2007, he set a target of funding 1500 new-investigator R01s, based on the previous 5 years' average.

Some program directors grumbled at first, NIH officials say, but came on board when NIH noticed a change in behavior by peer reviewers. Told about the quotas, study sections began “punishing the young investigators with bad scores,” says Zerhouni.

"quotas".

I do not recall much in the way of discussing the "pipelines" and how we couldn't possible do anything to change the bias of study sections until a new, larger and/or better class of female or not-previously-funded investigators could be trained up. The NIH just fixed it. ish. permanently.

For FY2017 there were 16,954 applications with women PIs. 3,186 awards. If you take the ~3% gap from the interval prior to 2003, this means that the NIH is picking up some 508 research project grants from women PIs via their affirmative action process. Per year. If you apply the ~6% deficit enjoyed by first time investigators in the salad days you end up with 586 research project grants picked up by affirmative action. Now there will be some overlap of these populations. Women are PI of about 31% of applications in the data for the first graph and first timers are about 35% for the second. So very roughly women might be 181 of the affirmative action newbie apps and newbies might be 178 of the affirmative action women's apps. The estimates are close. So let's say something like 913 unique grants are picked up by the NIH just for these two overt affirmative action purposes. Each and every Fiscal Year.

Because of the fact that, for example, African-American PIs of research grants or K99 apps represent such tiny percentages of the total (2% in both cases), the number of pickups that would be necessary to equalize success rate disparities is tiny. In the K99 analysis, it was a mere 23 applications across a decade. Two per year. I don't have research grant numbers handy but if we use the data underlying the first graph, this means there were about 1,080 applications with African-American PIs in FY2017. If they hit the 19% success rate this would be about 205 applications. Ginther reported about a 13% success rate deficit, working out to 55% of the success rate enjoyed by white applicants at the time. This would correspond to a 10.5% success rate for black applicants now, or about 113 application. So 92 would be needed to make up the difference for African-American PIs assuming the Ginther disparity still holds. This would be less than one percent of the awards made.

Less than one percent. And keep in mind these are not gifts. These are making up for a screwjob. These are making up for the bias. If any applicants from male, established or white populations go unfunded to redress the bias, they are only losing their unearned advantage. Not being disadvantaged.

27 responses so far

  • GM says:

    So basically you are confirming that men are blatantly and clearly discriminated against by the system, something that everyone looking at the situation without a political bias has been aware of for quite some years now.

  • Postdoc says:

    LOL to those whose pristinely untroubled minds are certain that the ONLY bias in the system is the corrective bias, and that all the evidence of bias beforehand is simply a reflection of the privileged class’s greater merit. LOL to them, indeed.

  • GM says:

    Bias is to give advantage to people that they do not deserve based on their merit.

    There are two clear examples of such biases:

    1. Affirmative action in favor of women and non-Asian non-White people.

    2. Legacy and "development" admissions at the undergraduate level (in favor of rich people).

    That males as a group are having an unfair and undeserved advantage (note the words "unfair" and "undeserved") is very far from having been clearly demonstrated. But that they are being unfairly disadvantaged by affirmative action is an obvious fact.

    Also, the expected distribution of successful people in any highly competitive field is highly skewed towards males, for objective biological reasons, thus the burden of proof is on those claiming that there was bias when males have been traditionally more successful because of an undeserved advantage that they have and that has to be corrected. If there is no such undeserved advantage, then it logically follows that right now females are the ones having an undeserved advantage thanks to the blatant pro-female bias in the system.

    Something that everyone in the hard sciences who has observed how hiring and award allocation decisions are being made these days has plenty of anecdotal evidence in support of.

    P.S. The most truly disadvantaged groups in the US right now are Asian males and working class white males at the undergraduate admissions level. How often do you see anyone talk about that? That alone tells you much of what you need about the fairness of the system right now.

  • JL says:

    GM, you are a shitty troll.

  • GM says:

    If you have actual arguments against what I am saying, please share them with us.

    Otherwise accusing me of being a troll can easily mean you have no such arguments.

  • drugmonkey says:

    It could also mean that your ideas are laughably stupid, very old, have nothing interesting or honest* about them and generally bore us. I have a notoriously light hand on the comment editing but nobody is obligated to respond to unimaginative trolling.

    *e.g., the post just prior to this one dealt with a bias against Asian-American K99 applicants, 2/3 of whom we can presume are male, without additional breakdown of the data.

  • A Salty Scientist says:

    GM, In a previous post, you argued that women make up false sexual assault claims at a rate that far outstrips reporting of other crimes, and here you argue that women are biologically inferior to men. You obviously have a problem with women. Why should we engage your intellectually dishonest sexism when others in the room are willing to have productive conversations?

  • Postdoc says:

    Also, it is truly a thing to behold the awesome confidence that some people have to simultaneously believe that (1) an entire class is biologically inferior, and (2) that they themselves would never at all prejudge any individual in the slightest way on the basis of that individual’s membership in the inferior class. Such supreme faith in one’s own rationality!

  • David says:

    "they are only losing their unearned advantage. Not being disadvantaged." I like this phrasing. Hope you don't mind if I steal it.

  • drugmonkey says:

    Be my guest.

  • Grumpy says:

    Curious how you implement this. Have applicants self-report race/ethnicity? Those things usually come with a disclaimer that the info won't be used during review and I assume plenty of minorities abstain from filling out. Would you try and modify the language of the disclaimer and risk congressional/judicial backlash?

    Or should POs sniff this out by checking applicant's websites/Google image search? I wonder what the error rate would be.

    But I agree that doing something imperfect would be better than nothing.

  • drugmonkey says:

    Curious how you implement this. Have applicants self-report race/ethnicity?

    Where do you think the data for Ginther or that K99 analysis come from in the first place?

    I assume plenty of minorities abstain from filling out

    on what basis?

    Or should POs sniff this out by checking applicant's websites/Google image search?

    Of course not.

  • Grumpy says:

    Well I'm in an underrepresented group and I sometimes leave it blank, was just assuming others did too.

    Do POs even have access to that data? they usually come with some sort of disclaimer that they are only for reporting/record keeping and can't be used for review, no?

  • GM says:

    GM, In a previous post, you argued that women make up false sexual assault claims at a rate that far outstrips reporting of other crimes, and here you argue that women are biologically inferior to men.

    Where have I ever said that women are biologically intellectually inferior to men (which is what you are implying)? I could not have said that because I do not think it.

    Women are clearly biologically inferior to men in everything having to do with physical strength.

    They are also clearly inferior to men in terms of overall productivity over long period of time because of things like childbirths.

    None of these two statements should be in any way controversial for they are obvious facts.

    But those do not concern the intellectual side of things. There, an argument can be made about males having higher variance thus being enriched at the very very top, and that is probably true, but that concerns a very small group of people.

    So to accuse me of saying women are intellectually biologically inferior is to blatantly lie. Which would not be for the first time in these comment threads, we've had plenty of conversations where the intellectual dishonesty of the social justice koolaid drinkers has been exposed for all with a sufficiently discerning eye to see.

    Case in point -- false rape allegations. I not only claimed there are a ton of false rape allegations, I also provided extensive evidence in support of that claim. Which nobody could refute, not that many even tried.

  • drugmonkey says:

    Your "evidence" fell far short of convincing so we ignored you. because your trolling is tiresome.

  • A Salty Scientist says:

    the expected distribution of successful people in any highly competitive field is highly skewed towards males, for objective biological reasons

    I'm not really interested in debating the exact nature of your bigotry.

  • GM says:

    How is that bigotry? It's an objective fact.

    I want to be able to fly like a bird, but I am too heavy to do that, so I guess physics is bigoted against me or something of the sort, right?

    All else equal, the people who put in the most effort will come out on top.

    For objective biological reasons having to do with differential parental investment, in Homo sapiens, males can over time put in a lot more effort into any highly competitive long-term race to the top than females, so on average they are expected to predominate among the winners of that race.

    To claim that should not be the case is equivalent to claiming that the group that is disadvantaged due to the differential parental investment is sufficiently inherently superior to make up for that disadvantage. In which case you are exactly the kind of bigot you are accusing others of being, because it is you and not those others who is in fact claiming biological superiority that has not actually been demonstrated to exist.

  • jmz4 says:

    "All else equal, the people who put in the most effort will come out on top. "
    -The trite answer is that yes, this is true, but things are extremely far from equal between genders for historical and sociological reasons, in ways that far outstrip the somewhat trivial gender differences.
    The cynical answer is that this is not true. More effort does not always correspond to increased success in academia, and, in addition, the proposition "all else being equal" is an absurd hypothetical not relevant to the discussion at hand.

    Either way, it isn't an argument for correcting inequality of outcomes, even if you think equality of opportunity is present (which again, is debatable).

    There's also a strong utilitarian argument to be made for equity, or equality of outcomes, in society, even though it necessitates positive discrimination, and even in settings of equal opportunity.
    Egalitarian societies are more stable and productive, overall, than hypercompetitive meritocracies, because the latter allow for the concentration of wealth, power and prestige. That breeds resentment, which leads to iconoclasm, which leads to social collapse.

    If we're looking at scientific academia as a microcosm of this phenomena, you could easily see how a competitive meritocracy is not optimal.

    There's also the following utilitarian argument:
    Scientific talent is somewhat rare. Which means to capture the most of it, you need to cast the widest net. Ideally we'd have some advanced sorting hat algorithm that allowed everyone to be sorted to the area in which they'd be most productive. However, without that, our best bet is to let as many people as possible have an opportunity to try out for Team Science (and the various positions available therein). This relies on people *wanting* to be scientists, which requires that they be able to *see* themselves as a scientist. If their only image of that is a person they are not (e.g. a middle-aged white man), they will be less inclined to go into the field. So more prominent women and minorities as scientists means more women and minorities will train as scientists, which will increase the overall pool of talent for the meritocratic elements to work upon, improving the efficiency of the entire operation. Because we certainly do not rely on meritocratic methods to put people into the pipeline for science.

  • drugmonkey says:

    jmz4, your utilitarian bit here is spot on and is how you can tell if one of these “objective merit is best for science!” brodogs is full of crap or not.

  • GM says:

    "If their only image of that is a person they are not (e.g. a middle-aged white man), they will be less inclined to go into the field."

    Oh, please...

    Was I thinking how inspiring it is to be a "middle-aged white man" when I was 5 and decided all that science stuff is super cool? No, I wasn't (and how do you know I am a white man, anyway?). It was the science itself that fascinated me, not the people who made the discoveries. And really, do you want people to be in it for the science or because of personal reasons?

    The discrepancies in outcomes are set up at two critical times -- first, childhood, when more boys go seriously into science than girls (do not pay much attention to the numbers of undergrad and grad students, what matters is who is truly strongly committed to doing science rather than who plans to go into consulting once they get their PhD, and in the hardcore fields it is still mostly males anyway), and, second, when the time to have children comes, when the women who are still serious about their science have to make some hard choices.

    It is outright ridiculous to claim that the lack of female role models plays a role in childhood. The immediate role models at that time are teachers, who are mostly female. If anyone does not have role models, it is the boys, not girls. But that is irrelevant anyway because kids don't have any idea what the demographics of tenured faculty at research universities look like at that point in their lives. If anything is a serious influence on them at that age, it is popular culture, and in popular culture the white male scientist is, the vast majority of the time, either a villain or a figure of ridicule.

    You do not become a top scientist by deciding to go into science in your mid-20s after not paying much attention to it before that. There are a few exceptions throughout history, but they are just that, exceptions.

    When I was a little kid, I was the only boy in my class who was seriously dedicated to academics, and while I was miles ahead of everyone else, there was still some competition from a few girls. I remember very well what happened to those girls -- somewhere around the age of 13 they lost all interest in studying and instead developed a very strong interest in frequent and random physical intimacy with members of the other sex. Did that happen because of the absence of strong female role models in science? Seriously? Also note that I was from a dirt poor lower single-parent-family working class background, they were from the kind of families that you would expect successful people to come from...

  • A Salty Scientist says:

    My god GM, your anecdata are rife with confirmation bias. Did you also not notice that 13 year old boys are also hormonally challenged? This is a moving of the goal posts anyway. This post is talking about women who somehow overcame their teenage hormones and/or the draw of the consulting world (WTF is wrong with you) to actually become PIs and apply for R-grants. Which they were receiving with lower success rates--on a per grant basis--than men until ~2003. You seem to think that NIH grant review is a perfect meritocracy immune from implicit (or explicit) biases. As a white dude who has witnessed blatant sexism in the academy (so likely the tip of the iceberg), I consider this position to be one of willful ignorance.

  • JL says:

    Ok GM, I take it back. You are not a shitty troll. You are a good troll and a shitty human being.

  • GM says:

    Did you also not notice that 13 year old boys are also hormonally challenged? This is a moving of the goal posts anyway. This post is talking about women who somehow overcame their teenage hormones and/or the draw of the consulting world (WTF is wrong with you) to actually become PIs and apply for R-grants.

    As I clearly stated, the discrepancies are set up at two stages -- childhood and people's early to mid-30s.

    In childhood the main driver is variation in natural aptitude towards taking up science as one's mission in life.

    In adulthood, it is having kids.

    If you have taken time away from work to give birth and nurse and raise kids, naturally you are not going to be as competitive as someone who has not had that burden. You will have fewer papers and other accomplishments to back up your requests for funding, and accordingly you are not going to be as successful getting grants. I don't see how that is in any way a controversial statement. There is a reason why so many of the successful women in academia remain childless or if they ever have kids, it is very late in life after tenure (I have also seen cases of very successful female scientists who have had 2-3 kids before they became independent researchers, and what they have managed to accomplish given that context is simply astonishing).

    You seem to think that NIH grant review is a perfect meritocracy immune from implicit (or explicit) biases

    Not at all, I am well aware how the system works. Also, it is strange to accuse me of thinking that NIH grant review is a perfect meritocracy when my first post in the thread was decrying the fact that it is not that due to the pro-female bias that NIH has mandated to equalize outcomes.

  • Skeptic says:

    GM, you must be aware by now that you are arguing with religious fundamentalists. They are not traditional fundamentalists, but rather fundamentalists of the new state religion of egalitarianism. However, they are equally immune to facts and reason. To them, it is unquestionable dogma that all people, all over the world, and of both sexes (or is it 52 now?) are blank slates with precisely equal abilities (on average). Thus, any disparities in group outcomes MUST be due to sexism or racism (by white men, of course, but that goes without saying). You would probably have better luck trying to convince a young Earth creationist of the reality of evolution.

  • GM says:

    GM, you must be aware by now that you are arguing with religious fundamentalists.

    Oh, I've known that for many years. It has all the hallmarks of a secular religion, in fact it is one of the best examples of what a secular religion looks like that one can find throughout history.

    The truly tragic thing is that this is not a bunch of mentally deranged gender studies graduates on Tumblr, this is the drugmonkey blog that only working scientists read and post on.

    And when your working scientists have completely abandoned the epistemological foundations of what they're doing...

    I consciously try not to waste too much time fighting windmills, but the subconscious is often stronger

  • pielcanelaphd says:

    Relevant to this discussion, the GAO just released the following report, "NIH RESEARCH Action Needed to Ensure Workforce Diversity Strategic Goals Are Achieved", which includes a response from HHS: https://www.gao.gov/assets/700/693872.pdf

  • drugmonkey says:

    thanks for the tip. interesting update

Leave a Reply