(My previous post on the topic is here.)
The responses to last week’s PNAS study on gender bias in science have been satisfying, for the most part. I’ve gotten used to avalanches of knee-jerk reactions every time a study on science and gender comes out. This time, there is a good deal (relatively speaking) of subdued and contemplative silence, at least among the actual scientists; the denials seem diminished in quantity. The effect might not be obvious to a bystander, but is quite noticeable to someone who has been following the debates for a while. I hope that this is a good silence, that some of us are taking the time to sit down and actually think about it.
This of course doesn’t mean that the subject has suddenly become totally uncontroversial. As Sean Carroll says in comments:
At least the trolls have moved on from “there is no discrimination” to “discrimination is rationally justified.” Progress!
I’ll be more specific. The wonderful, wonderful thing about the Yale study is that it allows us to have this discussion without being called “paranoid,” “hypersensitive,” or “emotionally unbalanced.” It feels refreshing and different to read long, argumentative comment threads on the subject and never see those words.
The discrimination apologists argue that, given the same “official” credentials, the rational employer will give preference to a man over a woman, because babies, pregnancies, dolls, biological differences, innate abilities, bell curves, life priorities, and other similar perennials.
Then there are press responses. The New York Times ran an article on the Yale study, then followed up with a discussion page. Here’s what one of the participants contributed:
There is little to suggest that colleges and universities are systematically discriminating against women or discouraging them from pursuing STEM disciplines. […]Why should we focus on achieving balance in STEM fields, while ignoring the overall imbalance in higher education as men fall farther behind? Factors other than sexism are likely the cause as to why fewer women pursue STEM fields. When students choose majors, they take into account myriad factors, such as their interests, aptitudes and career aspirations. Some research suggests, for example, that women with high-levels of quantitative skills are also likely to have high aptitudes in other areas, while men with high STEM-aptitudes tend to be less talented in other areas.
That, right there, is why I usually stay away from this type of debates. Let’s recap what the study actually said: that given identical paperwork from two hypothetical job candidates, one male and one female, the woman was judged as less competent and offered a lower salary. This is not about whether girls, statistically speaking, are less interested in science. It’s about a specific candidate who had already met the prerequisites, got a degree, demonstrated interest and skill in research, stated his or her career priorities clearly and explicitly, and was received much better when his name was John instead of Jennifer.
This is uncomfortable to write about. It’s much easier to shift focus to the “larger picture” of why there are so few women in science, so that we can go right back to reiterating and arguing the same well worn tropes yet again. The editors might have the best intentions for all I know, believing that they provide a valuable service by broadening the scope of the discussion and allowing a wider perspective. They end up regressing the conversation instead. In this particular debate, we need more specifics and less generalities.
It’s important to keep in mind that the bias uncovered by the study was unconscious and involuntary. We have all reasons to believe that the participating scientists acted in good faith and based their evaluations on their best judgement of the competence of the candidate. We’re busy people. If I had been selected to participate in the study, I would have likely spent 5-10 minutes tops reading the CV and writing my response, assuming that I had the time to respond at all. I’m guessing that most of the respondents made similar snap judgements, without taking the time to make up an argument about the female candidate possibly getting pregnant, not having the right priorities in life, or placing not far enough out on some bell curve.
Instead, the respondents felt that the female candidate was less competent, even though the resume was otherwise identical. We can’t know for certain what they were thinking at the time, but the study found that the “hireability” ratings and recommended salary offers were closely correlated with the competence ratings, suggesting that this was indeed the main factor. If they’re anything like most scientists I’ve met, they would profess strong belief that we all should be judged purely on merit. Certainly, they would be absolutely convinced that this is how they judge everyone else. They might also say that they would gladly hire more women if better candidates presented themselves.
Let’s be clear about the terminology here. Just because we are not aware of the biases we carry, because nobody is declaring an all-out war on women in science, does not mean that discrimination is not taking place and that nobody is getting hurt. If, in the absence of case-specific evidence pointing either way, our gut feeling or whatever tells us to go with the male candidate, that’s bias. If we act on it and hire the male candidate every time, that’s discrimination. The absence of conscious intent to discriminate does not matter. Women are still being excluded and set aside, for reasons that are none of their fault except for having been born female.
Sexism is not perpetuated only by old men who outlived the times that had shaped them. It’s not the exclusive domain of evil people who also eat babies and torture kittens. All of us partake in it to some extent. The challenge, and the way to make progress, is to minimize it by becoming more aware and taking steps to counteract the bias.
The talk of babies, pregnancies and bell curves is not harmless. Here’s why. Bias does not come from nowhere. Among other things, we have an involuntary, reflexive, and often unfortunate tendency to associate X with Y – and act accordingly – just because they get mentioned together often enough. We see this exploited endlessly in marketing and political campaigns. And we see it here as well, in the constant talk of the women in science problem, innate abilities and childbearing.
What I’m saying is that I would have preferred NYT to respond differently. They could have followed up on the subject by exploring women’s historic contributions to science, especially where such contributions were downplayed and undervalued by their contemporaries. (Lise Meitner and Rosalind Franklin come to mind, to give just two obvious examples.) They could have profiled contemporary women scientists, with attention paid to a couple of things that often go wrong on such occasions: focus on professional career and scientific achievements, not on looks, wardrobe, marital status and baby count. A science-oriented journal (not necessarily NYT) might go further and start a longer, more substantial debate, in a series of articles that would take the time to go past the usual superficial arguments. It’s been established beyond doubt that we won’t get anywhere with this in 1000 words or less.
The other way to make progress, of course, is for women to be “twice as good,” or at least significantly enough better that we can’t be dismissed so easily. That’s what many women in science have been doing all along. It takes a toll on us. It’s not a good solution. Unfortunately, sometimes it’s the only one we’ve got.
More to come.
The Accidental Mathematician 
Very good points, Izabella. You point out quite correctly that the biased environment is not created solely by the neanderthals, but by also by the rest of us, to one extent or another. What I believe is not discussed nearly often enough is the inadequacy of a typical response to problems of the type you describe. It would be nice if people in our profession took this opportunity to engage in self-examination and work on correcting the problem starting with themselves. Unfortunately, I do not believe that this is very likely to happen. Instead, we may well get a heavy handed administrative response in the form of poorly thought out programs and statistical targets that are not based on anything in particular. To put it simply, the study under discussion puts us face to face with some very uncomfortable facts. It remains to be seen whether any real effort will be put into dealing with the underlying bias in an effective way.
I wish more administrators had a genuine problem-solving attitude to this: here’s what’s wrong, what can we do about it? Not just lip service and CYA procedures, but an actual effort to improve the situation that makes use of our creative problem-solving skills. In our own research, we understand complexity and difficulty, the importance of getting to the bottom of the problem and exploring different options. We know that if a naive or stupid application of a method doesn’t work, then maybe we have to try again and do something more refined and sophisticated – it doesn’t necessarily mean that the whole thing is useless and we have to abandon it. And so on. But I don’t see even a minimum of that creativity and flexibility in dealing with gender bias.
I’ve seen affirmative action attempts that, translated to a mathematical setting, would look like this. We’re trying to prove the Riemann hypothesis. So, let’s graph the zeta function on a calculator, with the viewing field set to [-10,10] x [-10,10]. I don’t see any extra zeroes, so the hypothesis is true. What? You mean there could be zeroes elsewhere? OK, I give up. Nothing can be done here.
These are all very good points. It is probably worthwhile to add that – I am citing the NYT article – the study showed female professors to be just as biased as their male colleagues. (This supports several of your points.)
There was a type of response to the article that does strike me as reasonable: the study is a cause for serious concern, and deserves to be redone with a much larger n (the sample size was n=127) and more controls (a variety of male and female names, to control for other conscious or subconscious associations, e.g., class).