Merit versus diversity in NHMRC grants

So there are two distinct issues. Males dominate the industry at higher levels of experience, especially L3. There are more male applicants, so there are more grants awarded to males. Second, there is a modest disparity in success rates in favour of males, which may or may not be explained by quality.

On the first issue, lack of females at senior levels, it has been rather ludicrously claimed that "the largest contributing factor to lack of retention of women in STEMM at higher levels is that women receive fewer grants … due to gender bias in the peer-review system." So a lack of grants causes females to drop out? Surely, the first order effect is that fewer women in STEMM leads to the fewer grants. Talk about chicken and egg!

This "problem of too many males" will likely dissipate over the next twenty years as those junior females become more senior researchers. If there are forces that discourage women, such as family care duties, then female quotas are not the way to fix it. Why? Because many of thefemales who receivetheaffirmative action may not even have children or may be rich enough to afford a full-time nanny. Indeed, if these women are still in the industry at mid or senior levels then they are, almost by definition, less likely to have been subject to the systemic biases that lead to female attrition. Is the NHMRC going to ask female applicants if they have kids and how much time they spend taking care of them? It would probably be illegal to do so. And what about the (likely much smaller number of) males who have put their career second because of family responsibilities. Why do they not get a free ride?

This is a pervasive issue in affirmative action. If you use a crude proxy for disadvantage like sex or race then you deliberately ignore the cause of disadvantage, for instance lack of child care or a remote geographical location. Unless the systemic bias is against all females equally, a sex bonus will advantage some females who are already advantaged, and disadvantage some already disadvantaged males. Sex-based intervention should be an absolute last resort. It is as ludicrous as giving a height-based salary bonus to shorter people because males are taller and also are paid more. You kind of miss the point if you legislate to correct the gender skew with a height bonus. Short males will be the winners and tall females will be the losers.

The second problem of gender differentials in success rates deserves some serious study, but there was nothing in the NHMRC discussion paper that got to the heart of the matter. Some throw-away lines argue that it is "difficult to see and measure the biases of those who are evaluating applicants for appointments, promotions and grants." In fact, it is not as difficult as claimed. Gender-blind versions of an application could easily be used to get a sense of this.

Projects are assessed on the quality of the researcher, based on their track record, and on the project itself. Are there any gender (or more generally identity) blind measures of researcher quality that could be used? I would be strongly in favour of this and have lobbied hard to include identity-free assessment of job applicants here at University of Melbourne, with some success.

Is it possible the quality of the female-led projects was lower? The data in the NHMRC discussion paper suggest a modest difference. If so, is it an artifact of bias? And if so, how can we remove it?

If you mention quality and merit in some places, you are liable to be shouted down, but if the NHMRC are not in the quality and merit business, then they need to be closed down. The task may not be straightforward but it is not impossible either. Journals publish papers, universities appoint and promote. And there is a market to challenge and replace journals and universities that systematically get it wrong. Granting agencies need to get it right, too, especially in medical research, where huge patents are in the offing.

Identity is a fuzzy concept and irrelevant to merit-based grants

Does it really matter how many grant-holders are female, or male, so long as the allocation is merit based? To some it does matter, but not to me – or anyone with a clear moral compass.

If 100% of successful applicants were originally from Iceland I would not care at all in principle. Certainly such an extreme skew in recipient demographics could be a possible indicator of a systemic problem. There might be some subtle Icelandic bias that I would look for and correct if I found it. But 'too many' Icelandic researchers is not a problem in itself. Alas, too many male researchers seems to be a problem.

But there is another more important point to make in relation to the entire framing of the issue by the NHMRC. Why are grants being classified according to the sex of the lead researcher? The leader is one of many who benefit from a grant, but so do the others working on that project. L3 grants, for instance, tend to be larger and employ many junior post-docs. Most of these will likely be women.