by JakeS
Thu Jan 28th, 2010 at 07:21:07 PM EST
One of the subthreads in this story turned towards a meta-discussion about gender culture on ET. This discussion has cropped up several times in the past, and I have usually avoided partaking in it, for a variety of reasons. One of those reasons is that it usually turns into a flame war, and my insurance stopped covering asbestos suit replacements after I began frequenting YouTube comment threads. Something about wilfully subjecting myself to avoidable risks, or somesuch insurance-speak.
However, to quote myself,
When you have a (set of) topic(s) that are overwhelmingly discussed by either gender, it does not seem irrelevant to ask why this is, and whether the reasons should prompt a change in site policy or culture.
[...]
If the entire community is composed mostly of one gender/skin colour/socioeconomic group, a little introspection might be in order.
It's not that homogeneity is necessarily evidence that something is wrong - there may be a variety of entirely acceptable reasons (or at least reasons that are beyond the power of the editors to change). But it should be cause for some introspection on a not too irregular basis.
It was correctly pointed out that a meta diary would be a suitable venue to have such a discussion. It was also pointed out that if I objected to the tone of the currently available example of said meta-diary, ET comes with a perfectly functional diary submission form.
This seems like a reasonable request, so here we go.
Since the prospect of having actual data in a discussion tends to focus the mind wonderfully, I decided to run some numbers on the comment thread that triggered the current iteration of the debate.
While my pidgin analysis1 can't tell whether the comment thread in question displayed genuine overrepresentation of male commenters,2 it is possible to draw some conclusions. Specifically, at least one of the following must be true:
- There is an overrepresentation of males in the active ET commenters.
- There is a subset of topics covered by ET which display a heavy overrepresentation of males relative to the overall distribution of commenters. (I only have one example on hand, but unequal representation tends to be like cockroaches - if you see one, you can bet the farm that there's more where it came from.)
In the case of a general underrepresentation in the commenters, I think it is worthwhile to ask ourselves as a community whether our culture, style and/or editorial policies make it unnecessarily difficult to recruit female commenters.
Of course it is also possible that the broader Euro-American culture, from which ET draws its membership, socialises women to participate less in wonkish discussions of economics and foreign policy than men. Which would be regrettable but unfortunately beyond the scope of the ET community to materially change.
In the case of underrepresentation in certain topics relative to the ET distribution, it is more difficult to imagine that broader social issues are to blame, because the ET'ers have already self-selected towards wonkish types interested in energy policy, economics and foreign policy. In that case, it is easier to imagine that something about the way the ET community handles the subjects in question is to blame. (Although spillover from the broader society is not wholly impossible - on some issues, the well may have been poisoned sufficiently in the public discourse as to discourage participation by some groups.)
[Edit: Poll added to try to get some more data. (Thanks to Fran for suggesting it.)]
- Jake
1Because I am lazy, I used a simple null hypothesis that commenters in a given thread are picked at random from all active contributors, and that the total number of active contributors is large compared to the number of commenters in any given thread (this latter assumption is the lazy one, because I didn't want to have to guesstimate the number of active ET'ers).
For any given gender ratio of ET regulars, this gives a simple binomial distribution with which to evaluate the likelihood of the null hypothesis.
Using a 5 % significance threshold, the thread in question gives reason to reject the null hypothesis if the male:female ratio of ET regulars is less than roughly 70:30 (if the meta-subthread is discounted, the threshold ratio goes to roughly 80:20).
Since I don't know what the gender distribution among active ET commenters is (though judging by the meetups, a 3:1 male:female ratio is not beyond the scope of the imagination), this doesn't by itself permit me to accept or reject the null hypothesis for the comment thread in question.
2As an aside, this illustrates one of the rhetorical advantages of having data in the discussion: "I neither wish to accuse the diary of gender bias nor exonerate it" is a cop-out. "I am unable to unambiguously retain or reject the null hypothesis based on the available data" is being careful ;-P
