The bare outline of our procedure is not revolutionary:
- We downloaded the information for a bunch (over 200,000) papers indexed in pubmed
- We computationally inferred the gender of the authors based on their first names (this process is somewhat complicated - you can find the code and some more explanation here)
- We analyzed the results, splitting the data a couple of different ways, bootstrapping statisticts etc
- No surprise, women are under represented in computational biology, like they are everywhere else:
There are a couple of points that I think are particularly interesting though. The first is that, if the senior author of the paper is female, women are much better represented at all other positions. Computational biology is still worse than biology as a whole, but the bio representation jumps to nearly 50%, and the computational biology jumps to 40%.
Paradoxically, I think that the most encouraging news comes from a graph that shows the lowest female representation. Pubmed data only allowed us to compare biology and computational biology, but what about computer science? For this, we turned to the arXiv - a preprint server for quantitative fields. We can’t really compare this directly to the data from pubmed, but they do have a “quatitative biology” section.
There, quantitative biology has better representation than computer science. It’s still abysmal, don’t get me wrong, but it suggests that maybe, just maybe, biology might be used as an inroads to get more women into computational and quantitative techniques.
This gets at the question I’m most interested in - we know represenation is bad, but is there a way to improve it? These data aren’t conclussive by any means, but they suggest there’s a reason to try.
Now I just need to get a job where someone will let me experiment on (with?) undergrads…