The ASL statement on women in logic

Life got in the way of this sequence of blog posts, but now I'm back, and I have the last analysis ready. Today I'm going to consider the ASL's statement on women in logic, which was adopted at the Annual Meeting in 2012:

Logic benefits when it draws from the largest and most diverse possible pool of available talent. We at the ASL would therefore like to add our voice to the growing list of initiatives launched by organizations in the various science, technology, engineering and mathematics fields aimed at correcting the gender imbalance in those fields.

Female students and young researchers may be concerned about entering logic, where few senior women occupy visible roles. The atmosphere in classes and seminars can feel unwelcoming, and many young women have practical questions about managing a career and personal interests.

The ASL therefore states in the strongest possible terms that it welcomes the participation of women in logic and in particular in the activities of the Association. Accordingly, the ASL Council has adopted a statement urging those responsible for appointments and conference programs to pay attention to gender balance.

It's Women's History Month, and it seems appropriate to ask whether the proportion of women who are invited to speak at ASL meetings has noticeably changed since this statement was issued.

To figure this out, I'm comparing the proportions of female invited speakers in each of the four main conference series in the few years since this statement was issued to the proportions of female invited speakers in the corresponding time span before it.

	Annual Meetings		Logic Colloquium
	Men	Women	Men	Women
2009-2012	29	7	57	9
2013-2016	23	8	43	9

	AMS/ASL		APA/ASL
	Men	Women	Men	Women
2010-2012	17	4	21	1
2014-2016	14	6	19	3

As you can see, I considered different sets of years for my pre- and post-statement counts depending on the conference series. I made these decisions based on whether the speakers for the 2013 meetings would have been invited by the time the statement was issued. The speakers for the 2013 Logic Colloquium and Annual Meeting would have been invited after the 2012 Annual Meeting, so I compared the 2009-2012 and 2013-2016 intervals for these, and the speakers for the 2013 AMS/ASL and APA/ASL meetings might already have been invited by the time of the 2012 Annual Meeting, so I considered the 2010-2012 and 2014-2016 intervals for each of those. In each of the latter two cases, it was easy to sidestep the question of what to do about 2013 itself: the tallies for the AMS/ASL meetings are identical for 2010-2012 and 2011-2013, and there were no invited talks at the 2013 APA/ASL meeting.

I used a chi-square test to determine whether the proportion of female speakers in the few years before the ASL statement was adopted is statistically distinguishable from the proportion of female speakers since then.

When I carried out this analysis, I found that the p-value for the Annual Meetings was 0.7422 and the p-value for the Logic Colloquium was 0.7697. This gives us no statistical evidence suggesting that the ASL's statement has made any difference.

However, there aren't enough female plenary speakers in these time spans to carry out this analysis for the AMS/ASL and APA/ASL meetings. A reliable chi-square analysis can't be done unless there are at least 5 people in each category. Only 4 women gave plenary talks at the AMS/ASL meetings during 2010-2012, while only 1 woman gave an invited talk at the APA/ASL meeting during 2010-2012 and only 3 did during 2014-2016. While these numbers are very low, we should also recall that the time intervals for these series are one year shorter than the intervals for the others. However, the (relatively unreliable) p-values for the AMS/ASL and APA/ASL meetings are 0.6509 and 0.6, respectively, which suggests that adding one year to the pre- and post-statement intervals is unlikely to result in any meaningful statistical difference.

Summary: In the conference series in which enough women have given talks in these time intervals, there is no statistically significant difference in the rates at which women have spoken in the years immediately before and the years after the ASL statement was adopted, suggesting that the statement has not had an effect on speaker selection. However, this analysis is only possible in two of the four series.