Women in Philosophy 1930-1979: What can it tell us about diversity today? (Updated Numbers, 5/29/16)

Eric Schwitzgebel alerted me to a post at the Leiter Reports blog on the work of Jonathan Strassfeld (University of Rochester), who has compiled a document with philosophers appointed at 11 doctoral programs in the United States between 1930 and 1979: Berkeley, Chicago, Columbia, Cornell, Harvard, Michigan, Princeton, Stanford, UCLA, U Penn, and Yale. I was curious whether appointments in this period could predict present day diversity for these programs. My prediction was that a higher percentage of women among those appointed in this period would predict a higher percentage of women among faculty and graduate students today. I also wondered, given my work with Eric Schwitzgebel, whether area of specialization would interact with this effect (in that work, women were shown to be more likely to specialize in Value Theory). Although this is not a formal analysis, it appears as though programs that appointed a higher percentage of women in this period do have a higher percentage of women and non-white graduates today, and that there is some interaction with area of specialization such that programs with more faculty in LEMM/analytic fields tend to correspond with lower percentages of women, and historical fields tend to correspond with higher percentages of women. Given this first pass look at Strassfeld’s data, I think it would be useful to attempt to collect this data for a larger set of programs, to more formally explore these connections. More details on my first pass look at Strassfeld's data below. (Numbers updated on 5/29/16 to reflect a change made to Strassfeld's data. Namely, I had incorrectly removed one woman faculty member from the analysis, which Strassfeld pointed out to me.)

To explore the data I assigned gender to Strassfeld’s list, corrected some errors in his list (see below), accounted for duplicates and triplicates in that list (those who moved between these 11 universities once or twice in this time period), and compiled the numbers and percentages of women faculty per program, comparing that to the number and percentage of women faculty overall for these 11 programs. I then created a correlation matrix, comparing these percentages to:

• The % of women faculty per program in 2014, using the Philosophical Gourmet faculty list, program websites, and personal websites

• The % of women graduates per program in 1973-2014 and 2004-2014, using the Survey of Earned Doctorates data, and 2012-2015, using the Academic Placement Data and Analysis data

• The % of non-white graduates per program in 1973-2014, using the Survey of Earned Doctorates data

• The % of faculty per program in each of four AOS categories in 2014 (LEMM, Value Theory, History and Traditions, and Science, Logic, and Math), using the Philosophical Gourmet faculty list, program websites, and personal websites

• The % of 2012-2015 graduates per program in each of four AOS categories (LEMM, Value Theory, History and Traditions, and Science, Logic, and Math), using the Academic Placement Data and Analysis data, in which most of the recorded graduates have placements

Doing so, something surprising stood out, given my initial hypotheses: although the percentage of women faculty appointed between 1930 and 1979 in these 11 programs was strongly positively correlated with the percentage of women graduates between 2004 and 2014 (.59) and the percentage of non-white graduates between 1973 and 2014 (.47), it was slightly negatively correlated with the percentage of women faculty teaching in these programs in 2014 (-.14). In exploring the data more, I found that one university was most responsible for this effect: Cornell appointed a relatively low percentage of women between 1930 and 1979 (1 of 41, or 2%), but had the highest percentages of women faculty in 2014 (37.5%). If I remove them from the analysis, then the strong positive correlations remain (.53, .37) but the negative correlation becomes a positive one (.24).

Another feature that stands out is the set of correlations between areas of specialization and diversity, with two especially striking trends:

The percentage of women among faculty and graduates is negatively correlated with both the percentage of faculty in LEMM in 2014 and the percentage of faculty in what Strassfeld calls “analytic” fields between 1930 and 1979 (he contrasts these fields with “historical” and “non-analytic” fields).

• On the first, the percentage of women faculty appointed 1930-1979 is negatively correlated with the percentage of faculty in LEMM in 2014 (-.27), and the percentage of faculty in LEMM in 2014 is negatively correlated with women faculty in 2014 (-.45), the percentage of women graduates in 1973-2014 (-.58, a strong effect), 2004-2014 (-.86, a very strong effect), and 2012-2015 (-.34).

• On the second, the percentage of faculty in “analytic” fields in 1930-1979 is negatively correlated with the percentage of women faculty appointed from 1930-1979 (-.48), the percentage of women graduates 2004-2014 and 2012-2015  (-.50 and -.53, both strong effects).

• This pattern remained even after removing each individual university from the analysis. Removing Cornell from the analysis, for example, yields an even stronger pattern with both recent LEMM specialization and past analytic specialization negatively correlated with every category for percentage of women among faculty and graduates (numbers above become -.30, -.61, -.58, -.90, -.37 and -.42, -.46, -.48, respectively).

The percentage of faculty in historical fields between 1930 and 1979 is positively correlated with the percentage of women faculty and graduates now and the percentage of women appointed between 1930 and 1979 is positively correlated with the percentage of both faculty and graduates specializing in historical fields now.

• On the first, the percentage of faculty in historical fields in 1930-1979 is positively correlated with the percentage of women faculty in 2014 (.36) and the percentage of women graduates 1973-2014 (.49), 2004-2014 (.25), and 2012-2015 (.52).

• On the second, the percentage of women faculty appointed 1930-1979 is positively correlated with only the percentage of faculty in History and Traditions in 2014 (.42) and is strongly positively correlated with the percentage of 2012-2015 graduates specializing in History and Traditions (.62).

• In contrast, the percentage of women faculty appointed 1930-1979 is negatively correlated with the percentage of faculty in Value Theory in 2014 (-.29) and strongly negatively correlated with the percentage of graduates in Value Theory 2012-2015 (-.58). Yet, the percentage of faculty in 2014 and the percentage of graduates in 2012-2015 specializing in Value Theory are both positively correlated with the percentage of women faculty and graduates in nearly every other category.

• This pattern remained even after removing each individual university from the analysis.

These results were surprising to me mainly because I expected to see Value Theory, rather than History and Traditions, corresponding with higher percentages of women. Here are some possible explanations for this:

• Women faculty tend to attract other women faculty and graduate students to their programs, and this set of women faculty happened to be more historically focused than other sets of faculty. In fact, the women appointed between 1930 and 1979 in these programs were more likely to specialize in historical fields (women versus men were 31% and 11% in historical, 14% and 28% in non-analytic, and 55% and 61% in analytic), whereas women faculty in these programs in 2014 were more likely to specialize in Value Theory fields (women versus men were 36% and 19% in Value Theory, 23% and 34% in LEMM, 25% and 26% in History and Traditions, and 17% and 21% in Science, Logic, and Math). Those appointed women who specialized in historical fields were also more likely to stay in philosophy at either these programs or other highly ranked programs than other women (89% of those in historical fields stayed, versus 73% of those in analytic fields and 50% of those in non-analytic fields).

• Historical fields tend to attract more women faculty and graduate students and the women in historical fields who were appointed at these programs in this time period included several especially prominent figures, who helped their programs become leaders in historical areas of philosophy. In fact, although there are a number of prominent women philosophers appointed at these programs in this time period from several fields of philosophy (e.g. Ruth Barcan Marcus, Mary Mothersill, Holly Smith, and Susan Wolf), several of the women in historical philosophy stand out as having had a major influence: Janet Broughton, Julia Ching, Gail Fine, Marilyn McCord Adams, Martha Nussbaum, and Margaret Wilson. These 6 women jointly worked at 8 of the 11 programs, and while the overall percentage of historical faculty appointed 1930-1979 for these 8 programs was 12.7%, compared to 10.2% for the other 3, the overall percentage of faculty in 2014 specializing in History and Traditions for these 8 programs was 30.0%, compared to 15.3% for the other 3.

• Finally, perhaps some of these programs simply happened to be better work places for both women and those working in historical fields, relative to other institutions at the time. For example, I noted that of the 6 women mentioned above, 4 were associated with Harvard either as graduate students or faculty (Broughton, Fine, Nussbaum, and Wilson). Harvard has an above average percentage of women appointments 1930-1979, women faculty 2014, women and non-white graduates 1973-2014, women graduates 2004-2014 and 2012-2015, faculty in historical fields 1930-1979, faculty in History and Traditions 2014, and graduate students in History and Traditions 2012-2015.

Overall, it appears as though the women appointed in this period likely did have an impact on present day diversity for these programs, but the effects are not exactly what I expected at the start. I applaud Strassfeld for making this data available, and hope that others will add to his efforts so that these trends can be investigated further, beyond these 11 programs. I have put up an Excel file with data here, and you will see any changes I made to Strassfeld's data in red in the second tab. Comments and suggestions welcome.