Many academic fields are staffed by a male-biased mix of faculty. But the existence and degree of faculty gender imbalance varies among fields. Further, those fields often are quite broadly defined in published datasets (e.g., “biology”), which can leave many people wondering how well published data apply to their own, narrower field (e.g., “ecology”). Gender balance of academic fields also changes over time, but only slowly. Published data therefore only give you an imperfect sense of the gender balance of recent hires in your field. And personal anecdotes and experiences provide only a very small sample. Every year there are many dozens of faculty hired in ecology and closely-allied fields, but nobody hears through the grapevine about the outcomes of more than a small fraction of those hires.
So I decided to quantify the gender balance of recently-hired ecology faculty at North American colleges and universities. I’m doing it by going through this very comprehensive list of all ecology & evolution faculty positions advertised in 2015-16, and checking the university websites to identify who was hired. This turns out to be really easy in many cases, and difficult or impossible in the remaining cases (I therefore remove from the dataset). To keep things manageable, I’m skipping positions outside North America, of which there are very few on the linked list. I’m also skipping non-ecology positions, of which there are many. So not, e.g., biology, anatomy & physiology, genomics, evolution, paleontology, museum curator, science education, etc., even though some of those positions might have been filled by ecologists. But I’m defining “ecology” pretty broadly so as to include fields in which people who self-identify as ecologists often apply for and obtain positions. “Ecology” for purposes of this exercise includes wildlife management, conservation, ecological genetics, ecological physiology, evolutionary ecology, microbial ecology, fisheries, etc. My judgments on what constitutes “ecology” obviously are somewhat subjective and arbitrary, but I don’t see why that would affect the results. To focus on new faculty, I’m only looking at assistant professor positions, so ignoring the (very few) ads for heads of department, program directors, endowed senior chairs, etc. See the footnote (*) at the end for more nitty-gritty details on my procedure. UPDATE: To be clear, I’m including positions at all types of institutions, not just R1 universities. And you should do the same when answering the poll below. I’ll present the results broken down by institution type for anyone who’s curious about that.
I’ll present the results in a future post, in a sufficiently-complete form that you can go back and reproduce my work if you wish.
But before I show the results, I’m very curious what you think I’ll find. So below is a little poll. What do you think is the gender balance of recently hired North American ecology faculty? (UPDATE Nov. 10: responses have slowed to a tiny trickle, so I’ve closed the poll so that I can start analyzing the results. We already have 468 respondents–thanks to everyone who responded!)
p.s. Obviously, these data won’t tell you whether the outcome of any particular search was fair, much less whether every individual applicant for every position was evaluated fairly. And I have no way to collect lots of contextual information that you might want in order to interpret the results, such as the gender mix of the applicant pool for every position. In that future post I’ll talk more about what I think we can and can’t learn from these data.
*Failed searches are among those for which I can’t tell who was hired, so they automatically get dropped from the dataset. The difficulty of identifying who was hired mostly has to do with departmental web page design, so I’m confident that the easily-identifiable hires are a random sample of the population with respect to gender balance. A couple of times, I’ve determined that a position that I thought was ecological wasn’t filled by an ecologist; I’m dropping those cases from the dataset. I’m being careful to remove duplicate ads from the list so that I don’t double-count anyone. I’m including some other recent (2015-16) hires that weren’t on the linked list. I learned about these either from colleagues, or by stumbling across them while checking on listed positions. A few of the hires I stumbled across might actually be 2014 hires, but I’m fine with that because those are still very recent hires. In every case so far, gender has been obvious from the person’s name and photo.
I’d also be interested if on top of total balance across the discipline of ecology if this study considered the type of school. For example research focused vs. liberal arts.
+1
I did the poll assuming R1 only. Shows my bias…
Oh dear. That wasn’t what I wanted people to do. Sorry if the post wasn’t clear on this. I’ll update the post to prevent anyone else from doing this.
Yeah, if I had read the comments before voting, I would have thought about the variety of academic jobs there are. Instead, I thought about what I’ve *seen*, which has all been in the context of R1s…
Okay, just voted again. (Vote early! Vote often!) That will at least help balance out my first vote a little…
Way ahead of you. I have those data, already planning to do this. 🙂
Indeed, I just finished the data collection, so already have a sense of the answer…
Thanks for doing this! A lot of us on the job market who contributed to that list will be interested in your results.
You’re welcome. It’s been hours of work (I got sucked in this weekend…). But of course, I’m piggybacking on the collective efforts of the folks who contributed to the spreadsheet. No way I could do this without that spreadsheet.
Echoing Julia’s sentiments! Thanks very much.
Thanks so much for this — I am the associate dean for faculty in our college, and having this kind of metric, however imperfect, is helpful for our search committees. Can’t wait for the next installment!
You’re welcome Marlene. I can tell you that the poll is getting a lot of responses, which I take as an indication that many people share your, and my, interest in seeing the data.
I’ve seen a lot of comments on gender bias that focus on gender ratios in full vs associate vs assistant vs postdocs vs grad students, which increase from low female representation in full profs to high in grad students. There is a clear “leaky pipeline” with women leaving for many reasons, often associated with bias. An additional reason is that gender representation is improving over time, but it takes many years to go from grad to full professor, so gender ratios lag. I suspect it’s probably a bit of both. Has anyone looked at this? This would basically be a gender-based population model of ecologists over time, with “births/graduation” being grad student ratios, and “deaths/exit academia” measuring the relative attrition for each gender.
Check out:
“Leaks in the pipeline: separating demographic inertia from ongoing gender differences in academia” by Allison Shaw and Daniel Stanton, Proc B 2012
http://rspb.royalsocietypublishing.org/content/279/1743/3736
Aha! Thank you Shan. Should’ve read all the comments before replying to Trevor’s earlier comment.
Wouldn’t be a difficult exercise. I’ve never sent it done but it may well have been. Whether available data are sufficient to precisely estimate the model parameters and how they’ve changed over time I don’t know.
Dear Jeremy, I’m collecting data on gender bias in ecology in Brazil, so your post is definitely very interesting to me. I found, as expected, a huge drop in women presence from grad student to professor position. We have a completely different hiring system: it’s a series of exams (written on specific ecology fields, a class simulation, an interview and the cv analysis).
I was not able to find the results of your research though.
Thank you for doing this.
Best,
Eugenia
Sorry, what I meant to say is that I could not find the results of the poll. Are they available?
Thank you!
No, I’ll reveal the results of the poll at the same time as I reveal the data I’ve compiled. The poll results would get skewed if new respondents could see or hear about the responses of previous respondents.
The paper linked to by Shane Kothari in an earlier comment is an excellent examination of this issue at all career stages, albeit based on US data. Have you written up the Brazilian data anywhere?
I haven’t posted my results yet, I’m saving them for a future post. I’ve finished the data collection, but have just started analyses. And I want to collect a large sample of guesses as to what the data will look like before I reveal the data.
I just started to collect the data on Brazilian Ecology Graduate programs and presented them to a meeting. I hope to publish them in a Brazilian Journal (written in English) soon.
All the literature I found so far is mostly from US institutions, some on European ones and basically none from South America.
I did not understand the “Number of individuals that applied” column in your dataset. The numbers seem very low (in some cases only 1 individual applied?). I must be missing something. Could you clarify?
Thanks,
Gustavo S. Betini
Anyone who reads the spreadsheet and applied for the job in question can choose to add themselves to that column. Which of course is ordinarily only a very small fraction of all the applicants.
A slight methodological critique – the ranges are not constant, with 25-39% being the broadest by far, which may definitely have an impact on voting, and the order of the list is a concern as well. Simply having people write the number they think most likely in a black box would be a more neutral way of asking this question.
The shape of the imposed distribution is symmetrical around 50%. 25-39% and 61-75% are equally broad. And neither is the broadest category. The broadest categories are >75% and <25%.
I doubt the ordering had any impact. And it would greatly inconvenience respondents to randomize the ordering, given the categories I chose.
None of this is to say that having people write in a number wouldn't have been fine as well. It would've given more resolution in the tails of the distribution.
In the follow-up post I'll briefly discuss whether I think the results are an artifact of my methodological choices in any important way. But the short version is no, I don't think I created any important methodological artifacts (I may have created one very minor one…)
Fair enough … it’s an interesting poll, I’m looking forward to the results.
Although, of course, we have learned today that we might have to re-think the whole concept of polling 😉
Too soon. Sorry.
Pingback: A bit of good news that most ecologists weren’t expecting: recent ecology hires are gender balanced | Dynamic Ecology