A new study came out about women in academic STEM. The authors write that women are favored 2:1 across disciplines when hiring STEM tenure track positions. Briefly, the authors sent out surveys asking for faculty from four disciplines (biology, economics, engineering, and psychology) to assess three job candidates – an inferior foil candidate (male) and two equally qualified superstars (one male, one female). In some cases, they included information about children and marital status. They also varied whether the candidates were described using masculine or feminine adjectives (assertive vs easy to work with, for instance). Over 800 faculty replied, pretty much equally split between men and women.
Now, this is a subject that is near and dear to my heart. I am lucky, I’ve only very rarely been subjected to overt sexism myself, and it was always pretty mild (and usually in Russia). But I have witnessed it more often, heard stories from many sources, and been called a rabble-rouser for bringing the issues up. I’m kind of proud of that last bit. Blame my parents, ex-60s era radicals that they are. So, I keep an eye out for papers like this.
I should state up front that I was very skeptical of this paper from the get-go. Yes, the premise seems off from my own experiences, and those of colleagues. But even more than that, the authors of this paper published another article last year, with an accompanying op-ed stating “Academic Science Isn’t Sexist.” The article and op-ed were, at best, flawed, featuring this gem:
As children, girls tend to show more interest in living things (such as people and animals), while boys tend to prefer playing with machines and building things.
I won’t go into details – this post is about their new article – but Rebecca Shulman, Rachel Bernstein, Emily Willingham, Jonathan Eisen, and Kelly J. Baker all have excellent takedowns of the methods and approach. Suffice to say, I was primed to disagree.
And, boy-howdy, do I ever.
The first sentence, “Women considering careers in academic science confront stark portrayals of the treacherous journey to becoming professors.” And you know what? Men do too. This is not a good time to consider becoming a professor. NSF has a 4-8% funding rate, depending on program (anecdotal, from what people have told me). NIH is worse. Tenure depends on getting big, nationally-competitive grants, and usually you need multiple. Hours may be flexible, but they are long. Pay is crappy until you get a tenure-track position, but even then it is not great. You will move, and move often, before you land a tenure-track job. Every tenure-track job has dozens, if not hundreds, of applicants. Academic science right now is tough.
Data is not the plural of anecdote, but I will say I was not warned about the “treacherous journey” while being explicitly female. Yes, I knew funding was terrible. People were encouraging. They wanted more women to participate. My cohort only had one guy in it, out of nine. I’m more aware of the structural obstacles that women and POC and other under-represented minorities face now, five years in. I think that’s true for most of my friends, too. But, if you are interested in STEM, I do not think many people will tell you not to do it because of sexism. Most conversations are positive – “How can we fix this?” – rather than negative – “God, this sucks.”
Williams and Ceci repeatedly state that the message “hiring is sexist” discourages women from applying for tenure track positions. Their previous work had a similar point, about citation, publishing, promotion and retention. Perhaps that single message is driving women away from STEM. But there are far more factors driving the divergence between rates of graduating women doctorates in STEM, and the hiring of women assistant professors. I’d like to think we, by and large, won’t be deterred because we hear hiring practices are sexist. Personally, I’m more deterred by the low funding rates and moving every two years until I get that magical tenure-track position.
I don’t particularly like this message-driven justification for these experiments, but you know what? The experiments themselves are important.
First, though, a little umbrage about how they designate the disciplines. Biology and psychology are called “non-math-intensive.” I cannot say anything about psychology, but biology? There’s a lot of math. Biostatistics and bioinformatics are growing disciplines because biologists are now dealing with huge datasets and complicated modelling to describe everything from fish populations to environmental metagenomics. Ecologists are way better at math than I am.
What’s more, the presence of math is probably not a good predictor for whether a field tends to more preferentially exclude women. Otherwise, as Meg Urry states in this great lecture, there would not be a disparity in representation between astronomy and physics. They do the same work, need the same skillset. Yet astronomy has better representation of women. Separating biology and psychology from engineering and economics makes sense – women representation differs strongly between the two groups. But setting them up in contrast based on how “math-intensive” the disciplines are is silly, and leads the reader to equate the non-math-intensive disciplines as feminine and math a deterrent to women. This is mostly semantics, but I think if you’re publishing a paper on gender biases you need to be careful about such details.
The authors also include in their design the marital status of the applicants, whether the spouses had a job, and whether the applicants had children. Certainly, these are the types of things that can influence hiring decisions. But they aren’t supposed to be. A job applicant is protected from answering these questions. Faculty, hiring committees, even graduate students are not allowed, at all, to ask about marital status or children. That doesn’t mean the hiring committee doesn’t know, or that someone won’t break the rules. It is still against most university guidelines for faculty to ask. And such information would not be provided in a job application. Even if that information was volunteered in conversation, I absolutely do not think it would be included in any official documents presented to faculty for assessment. Seriously. I cannot emphasize enough how shady that seems to me.
The faculty surveyed knew that these were not actual job candidates. They knew it was part of a study. From the supplemental material, an excerpt from what was sent to faculty in the survey,
“Imagine you are on your department’s personnel/search committee. Your department plans to hire one person at the entry assistant-professor level. Your committee has struggled to narrow the applicant pool to three short-listed candidates (below), each of whom works in a hot area with an eminent advisor. The search committee evaluated each candidate’s research record, and the entire faculty rated each candidate’s job talk and interview on a 1-to-10 scale; average ratings are reported below. Now you must rank the candidates in order of hiring preference. Please read the search committee chair’s notes below and rate each candidate. The notes include comments made by some candidates regarding partner-hire and family issues, including the need for guaranteed slots at university daycare. If the candidate did not mention family issues, the chair did not discuss them.”
Reaching way back to my undergraduate years, when I took “Qualitative Methods in Geography”, we learned about surveys. Bias blindspot. People think that they are less biased than they actually are, less swayed by those biases than the average American. They are more objective than their colleagues. Everyone wants to think well of themselves, so they think they aren’t racist, sexism, bigoted. Everyone wants to think themselves resistant to the implicit biases that have been instilled by the cultural landscape.
Imagine a colleague receives a survey about hiring practices in academia. The exact questions motivating this survey aren’t known, but you know that surveyors are assessing something about hiring decisions. Your colleague, she’s pretty fair. But you know she’s said something about single parents not having time to really devote themselves to their job at an R1 university. How could they? There aren’t enough hours in the day! She has a bias. Do you think, though, that she’s going to admit to that in a survey? Do you think she is going to do anything but try to be as absolutely, unimpeachably “fair” as she can be?
Do you think she would come to the same conclusion if this were a hiring decision in your department? Maybe. Maybe not. I’m skeptical.
Note: I’ve used “she” here because half of the respondents were women and I try not to default to “he”, but I’m not trying to imply that this effect is any more or less pronounced in men or women. Implicit bias influences women and men roughly equally.
I suspect that the preference for women is at least partially owing to over compensation in order to appear unbiased. I also think including anything about “lifestyle” – marital status, children, etc – probably leads people towards the idea that the study is about gender, or something related. I also just saw someone on Twitter say that the email with the survey explicitly stated the study was about biases in hiring. I’d also point out that the lifestyle description was, as far as I can tell, the biggest concrete detail provided in the summaries passed out to faculty. Otherwise, they are described in adjectives such as “likeable”, “powerhouse” and “imaginative”. I do not think that this effect alone could account for the staggering difference between male and female applicants in the results. But I do think it is important to consider.
All that said, the results are encouraging. Women do not appear to be discriminated against, with the possible exception of economics. The details are interesting – female faculty prefer divorced-with-kids women to married-with-kids men, male faculty the opposite, as an example. Overall, this is great! I’m really happy to see this. My knowledge of social science best-practices is limited, but the statistical analyses seem fairly robust. Systematic hiring biases are not as important as we thought!
And then. Then. The opening sentence of the discussion.
“Our experimental findings do not support omnipresent societal messages regarding the current inhospitability of the STEM professoriate for women at the point of applying for assistant professorships.”
No one, to my knowledge, has recently claimed that hiring bias is The One Big Obstacle for women. Calling this “omnipresent” is just weird. There is no single barrier like that. In fact, a much larger topic is not hiring women, but simply to make sure they are included in your hiring pool. Encouraging women to send in applications in the first place. That’s certainly been the discussion in our department, and supported by several different initiatives.
This study used applicants with identical qualifications. They did not use actual CVs, except for a small subset, but the CV summaries used made the male and female applicants appear to have the same expertise. All the CV summaries were for extraordinary people. And that is where things get hairy.
Women are cited less. Women are nominated for (and win) fewer awards. Recommendation letters are weaker. Women apply to fewer positions than men, and men apply for a wider variety of positions that they do not necessarily qualify for. Academic women hold fewer patents. Women are more likely to hold adjunct positions. Fields that perceive themselves as requiring “genius” are far more male-dominated. Women consistently report less mentoring. Fewer women hold tenured positions at universities. An even smaller number are in administrative positions. Much of this has been the case for decades, despite an increasing percentage of female graduate students.
Some of these are driven by women’s decisions, yes. Those that do, I think, are tied to societal expectations that women are “nurturing” and “good teachers” rather than “brilliant” (see interactive based on RateMyProfessor). There’s a reason that women are more successful than men once hired – probably because they had to be pretty extraordinary to overcome those obstacles and get hired in the first place. The candidates in this study were all very strong – results might have been different if the candidates were a little less superstar, and a little more typical.
Many of these hindrances are not based on “supply-side” decisions, as the paper calls the problem. Rather, they are a result of structural obstacles and biases within academia and society at large. The authors belaboring that “messages to the contrary [that it is a precipitous time to be a woman in STEM] may discourage women from applying” is misleading. I do not think that the scientific literature attributes the lack of women in STEM as driven by unfair hiring. I think the scientific literature is pretty explicit that there are a lot of things going on preventing women (or any under-represented group) from succeeding on the tenure track, possibly including unfair hiring. And, despite their previous claims (cited heavily in the new paper) that academic culture isn’t sexist, it’s pretty clear that gender biases and hostile workplaces are still a problem (evidence of which can be found in their own data).
Look, the experiments they did were valuable. And the results were encouraging, wonderful to hear. Everyone should know about them. But the simplistic, overly broad take-away – that this whole thing would be fixed if women applied to more jobs – jumps way beyond the scope of those results.