Percentage of Women at APA Meetings, 1955, 1975, 1995, 2015

Last spring, I posted a gender analysis of the program of the Pacific Division meeting of the American Philosophical Association, broken down by ethics vs. non-ethics and by role in the program. I've been coding some other APAs along similar lines. For a broader picture over time, I have now examined gender data for all three divisional meetings of the APA for 1955, 1975, 1995, and the 2014-2015 academic year [note 1].

Gender was coded by first name and/or by personal or professional knowledge as either "female", "male", or "other/indeterminable". [note 2] I coded the main program only, and I excluded sessions organized by special committees (and all other symposia listed at session end rather than session beginning). My idea in doing this was to capture the mainstream research sessions rather than sessions on the state of the profession, teaching, the status of different ethnic groups, etc.

As expected, the majority of philosophers on the APA main program are men, but the gender ratios are less skewed now than they were a few decades ago. Overall, the proportion of women on the APA main program has increased from about one sixth in 1975 to about one third in 2015.

Merging all three divisions, here is the gender breakdown by year:

1955: 6% women (7/121, excl. 5 indeterminable)
1975: 16% women (62/397, excl. 20)
1995: 25% women (220/896, excl. 38)
2014-2015: 32% women (481/1526, excl. 177 [note 2])

All three of the 20-year-interval increases are statistically significant, considered individually (two-tailed z tests, p < .001). Differences between divisions were not statistically significant.

Recent estimates of the percentage of women in philosophy in the United States are typically in the low 20%'s, with 21% the most commonly cited number. Interestingly, at 32% women, the 2014-2015 program data are significantly higher than women's overall representation in the profession (481/1526 vs. 21%, p < .001). Possible explanations: younger philosophers more likely to be women and more likely to attend conferences; non-U.S. participants who are more gender balanced; the gender-indeterminate category ("Chris", foreign names) being disproportionately male; women having more interest in participating in APA sessions; and/or the program committees working to reach out to women.

I also divided sessions into "ethics", "non-ethics", and "excluded". "Ethics" was construed broadly to include social and political philosophy. Philosophy of action and philosophy of religion were excluded as borderline, unless they were on ethical topics in those sub-areas. My hypothesis was that within philosophy, a larger percentage of women specialize in ethics than in other areas. The results:

1955 ethics: 5/32 (14% women)
1955 non-ethics: 2/64 (3% women)

1975 ethics: 16/110 (15% women)
1975 non-ethics: 41/249 (17% women)

1995 ethics: 101/275 (37% women)
1995 non-ethics: 105/531 (20% women)

2014-2015 ethics: 206/500 (41% women)
2014-2015 non-ethics: 217/824 (26% women)

The numbers are too small in 1955 and 1975 to draw firm conclusions. However, in both 1995 and 2014-2015 the predicted effect is large and statistically significant (p < .001 for both). Since at least 1995, ethics sessions at APA meetings have been much closer to gender balanced than non-ethics sessions.

Here are the numbers in a graph with 95% confidence intervals:

I also examined role in the program, to see if women were more or less likely to serve in roles that are typically regarded as more or less prestigious. I divided roles into five types: (1.) Presidential or named lecture / author in author-meets-critics / symposium speaker with at least one commentator on just her paper. (2.) Symposium speaker not in category 1, or AMC critic. (3.) Symposium commentator or introductory remarks for named lecture. (4.) Presenter or commentator in colloquium. (5.) Chair (timekeeper/moderator) in any session.

The program role results are a bit difficult to interpret, with women more likely to appear as ordinary symposium speakers (role 2) and as session chairs (role 5) and perhaps least likely to appear in colloquia spots (role 4). The trend is evident both in the aggregate data and when only 2014-2015 is considered (for all other years, the individual-year analysis is underpowered). Here's the breakdown for the 2014-2015 data:

Cat 1 (most prestigious): 27% (27/99)
Cat 2 (ordinary symposium speaker): 37% (117/314)
Cat 3 (symposium commentator): 30% (29/96)
Cat 4 (colloq speaker/commentator): 26% (155/597)
Cat 5 (chair): 36% (153/420)

This is statistically significant variation (chi-square [DF 4] = 18.9, p = .001). Overall, I'd say that this tends to disconfirm the hypothesis that women are disproportionately likely to appear in lower-prestige program roles, but beyond that I hesitate to speculate.

Carolyn Dicey Jennings and I are at work on other analyses of the percentages of women in ethics vs. other areas of philosophy. If other measures also suggest that ethics contains a higher percentage of women than do other areas of philosophy, then at least two conclusions appear to follow for those who wish to help steer philosophy closer to gender parity:

First, in an ethics context, a proportion of women that is representative of philosophy as a whole might still constitute an underrepresentation of women relative to the available pool.

Second, the situation outside of ethics might be even more unbalanced than one would guess from looking at philosophy as a whole.

However, the long-term upward trends both within and outside of ethics are encouraging.

---------------------------------------------

Note 1: The Eastern Division did not meet in 2015, shifting from a December to a January schedule, so I use the December 2014 data.

Note 2: I coded as indeterminable: gender-ambiguous Anglophone names ("Pat", "Robin"), mere initials ("C."), and non-Anglophone names whose gender associations were unknown to me ("Asya", "Lijun"). I allowed personal knowledge to resolve ambiguities (e.g., "Pat Churchland" as female). Impressionistically, the higher rate of indeterminable in 2014-2015 (10% of participants, up from 4-5% in 1975 and 1995) was due to more participants with non-Anglophone names. If women are substantially more or less common among the indeterminable names than among the remainder, that might skew the results by a few percent either direction. Still, the overall trend remains clear.

---------------------------------------------

Thanks to Mara Garza for help with coding some of the data. Thanks to Roger Giner-Sorolla for catching an error in the labeling of the Y axis, which has now been corrected.