There's the leak in the pipeline

We're always talking about this curious phenomenon, that we see lots of women at the undergraduate and graduate level in biology, but large numbers of them leave science rather than rising through the ranks. Why is that? It seems that one answer is that elite male faculty in the life sciences employ fewer women, that is, the more prestigious, well-known labs headed by male faculty with great academic reputations tend not to hire women for the next level of training.

Women make up over one-half of all doctoral recipients in biology- related fields but are vastly underrepresented at the faculty level in the life sciences. To explore the current causes of women’s underrepresentation in biology, we collected publicly accessible data from university directories and faculty websites about the composition of biology laboratories at leading academic institutions in the United States. We found that male faculty members tended to employ fewer female graduate students and postdoctoral researchers (post-docs) than female faculty members did. Furthermore, elite male faculty—those whose research was funded by the Howard Hughes Medical Institute, who had been elected to the National Academy of Sciences, or who had won a major career award—trained significantly fewer women than other male faculty members. In contrast, elite female faculty did not exhibit a gender bias in employment patterns. New assistant professors at the institutions that we surveyed were largely comprised of postdoctoral researchers from these prominent laboratories, and correspondingly, the laboratories that produced assistant professors had an overabundance of male postdocs. Thus, one cause of the leaky pipeline in biomedical research may be the exclusion of women, or their self-selected absence, from certain high-achieving laboratories.

These statistics were obtained by sampling a large number of labs across the US. The leaky pipeline is rather obvious in this table: note that we have parity at the graduate student level, but that it falls off dramatically at the next level up.

surveytable

This is a problem. One (not the only one!) of the criteria used to select academic hires is the reputation of the lab they came from -- some labs are just really good at cranking out the data, publishing publishing publishing, and new graduates coming out of those labs are likely to continue that pattern. Coming out of a well-known lab provides a real leg-up for an academic career. But what this paper found is that women were less likely to find themselves in those labs.

We found that female trainees were much less likely to work for an elite PI, particularly at the post-doctoral level. Combining faculty of both genders, men were about 17% more likely to do their graduate training with a member of the NAS, 25% more likely to do their postdoctoral training with a member of the NAS, and 90% more likely to do their postdoctoral training with a Nobel Laureate. Thus, the gender skew in employment results in fewer women being trained in the laboratories of elite investigators.

Get with the program, Nobelists!

My first thought was that maybe this was a product of an older generation -- that more senior faculty are going to be much older and perhaps unfortunately traditionalist, so all we have to do is wait for them to die off and be replaced. No such luck. When the data are carefully dissected, the correlation isn't with age, but with elite status (as defined by membership in prestigious organizations). Young male investigators are just as unlikely as old male investigators to hire women.

As expected, among male faculty, elite status was negatively correlated with the percentage of female postdocs in a laboratory (P < 0.0001). This relationship remained true even when several other explanatory variables were added, including faculty rank, years since a faculty member had received his or her PhD, and total number of trainees in a laboratory. As a single independent variable, years since PhD was moderately negatively correlated with the percentage of female postdocs in laboratories with male faculty members (P < 0.045), but this effect disappeared when other variables were included in the model. This observation suggests that a faculty member’s age is not a significant determinant of the gender makeup of their laboratory, and both young and old elite professors employ few women. Laboratory size was also negatively correlated with the representation of female postdocs both as a single variable and in multivariable models. Regression against the percentage of female graduate students in each laboratory revealed similar, although less robust, results. In multivariable models, elite status was associated with a significantly lower percentage of female graduate students trained by male faculty. However, years since PhD correlated with an increasing representation of female graduate students, whereas laboratory size was not significantly correlated in either direction. Finally, we constructed equivalent linear models for female PIs, but we failed to find a single variable that was significantly associated with differential representation of female trainees in these laboratories.

The paper is careful to point out that they don't know the direct causes of the differences, whether it's exclusion, conscious or otherwise, by faculty men, or reluctance of women to apply to those labs. We should probably try to figure that one out, since that's how the problem gets fixed…but it's probably a combination of all of these factors.

Irrespective of the cause of the gender disparities in elite laboratories, its consequences significantly shape the academic ecosystem. Our data show that these laboratories function as gateways to the professoriate: new generations of faculty members are predominantly drawn from postdocs trained by high-achieving PIs. However, these feeder laboratories employ a disproportionate number of men. According to the theory of cumulative disadvantage, persistent inequalities in achievement can result from small differences in treatment over a prolonged goal-oriented process. In controlled studies, women in academia receive less favorable evaluations, receive lower salary offers, and are ignored by faculty more frequently than men. Access to training in certain laboratories may be another level at which women are disadvantaged. The absence or exclusion of female trainees from elite laboratories deprives them of the resources, visibility, networking opportunities, etc. that could facilitate their professional development. These differences may contribute to the leaky pipeline by shunting women toward laboratories that provide fewer opportunities for advancement in academic science.

I'm certainly not at one of those elite laboratories, so I can't do much at that level -- but I am training swarms of undergraduate women and stuffing them in at the base of the pipeline. One thing we can do here is encourage our graduates to be ambitious and push hard to get into the labs they really want…and to prepare them for the institutional biases that will get in the way.

More like this

Early last year when the whole Larry Summers saga broke out, I posted some data on gender and science that was floating in the public sphere. Here I've reposted some of this data. THEN I'll tell you some recent data from Harvard ... From a NY Times article Feb 22nd, 2005 Women in Physics Match Men…
Tim left this comment over at Uncertain Principles on Chad's post The Pipeline Problem: I thought the data was pretty clear about this: past high school, the [physics] pipeline is no more leaky for women than it is for men...here's the Report: Read it for yourselves. Examination of the academic "…
...if you're not a tenure-track PhD (and that will be most of you. Sorry). I'll have more to say about ScienceBlogling DrugMonkey's training post tomorrow, but one of the disturbing things in the comments of his post was the high numbers of people who viewed PhD training only in light of…
Mary Ann Mason has a column in this week's Chronicle of Higher Education describing the importance of role models and mentors for women graduate students. Though Zuska recently wrote a provocative post that argued that "the problem of motherhood" might be a red herring for those interested in…

I wonder whether some of the issue might be that "elite" labs are more likely (or considered more likely) to require very long hours of their postdocs, which may be a problem for women who are more likely to have family responsibilities.

By Alphagamma (not verified) on 08 Jul 2014 #permalink