Has there been a latest uptick within the proportion of girls amongst philosophy PhD recipients?
It’s tough to evaluate, says Eric Schwitzgebel (UC Riverside), within the following visitor publish.
[painting by María Freire]
Within the Seventies, girls obtained about 17% of PhDs in philosophy within the U.S. The proportion rose to about 27% within the Nineties, the place it stayed principally flat for the following 25 years. The newest knowledge counsel that the proportion is on the rise once more.
Right here’s a enjoyable chart (for user-relative values of “enjoyable”), exhibiting the 50-year pattern. Evaluation and methodological particulars to observe.
The info are drawn from the Nationwide Science Basis’s Survey of Earned Doctorates by way of 2022 (the latest accessible yr). The Survey of Earned Doctorates goals to gather knowledge on all PhD recipients from accredited universities in the US, usually drawing response charges over 90%. The SED asks one binary query for intercourse or gender: “Are you male or feminine?”, with response choices “Male” and “Feminine”. Fewer than 0.1% of respondents are categorised in neither class, stopping any significant statistical evaluation of nonbinary college students.
Two information are instantly apparent from this chart:
First, girls have persistently been underrepresented in philosophy in comparison with PhDs total.
Second, girls obtain fewer than 50% of PhDs total within the U.S. Because the early 2000s, the proportion of girls amongst PhD recipients throughout all fields has been about 46%. Though women have consistently been earning about 57-58% overall of Bachelor’s degrees because the early 2000s, disproportionately few of these girls go on to obtain a PhD.
The tough factor to evaluate is whether or not there was a latest uptick within the proportion of girls amongst Philosophy PhD recipients. The year-to-year variability of the philosophy knowledge (as a result of a pattern dimension of about 400-500 PhD recipients per yr lately) makes it unclear whether or not there’s any actual latest underlying enhance that isn’t simply as a result of noise. I’ve drawn a third-degree polynomial trendline by way of the information (the pink dots), however there’s a threat of overfitting.
In a 2017 article, Carolyn Dicey Jennings and I concluded that the perfect interpretation of the information by way of 2014 was that the proportion of girls philosophy PhD recipients hadn’t modified because the Nineties. The query is whether or not there’s now good statistical proof of a rise since then.
One easy strategy to the statistical query is to search for a correlation between yr and proportion of girls. For the complete set of information since 1973, there’s a robust correlation: r = .82, p < .001—impossible to be statistical likelihood. There’s additionally an excellent correlation if we have a look at the span 1990-2022: r = .49, p = .004.
Nonetheless, the chart seems fairly flat from about 1990 (24.3%) to about 2015 (25.7%). If many of the statistical work is being carried out by three excessive years close to the tip of the information (2016: 34.7%; 2019: 34.2%; 2021: 33.8%), the perfect mannequin may not be a linear enhance since 1990 however one thing nearer to flat for many of the Nineties and early 2000s, with the actual surge solely in the latest a number of years.
To drag extra statistical energy out of the information to look at a narrower time interval, I handled every PhD recipient as one commentary: yr of PhD and gender (1 = feminine, 0 = not feminine), then ran an individual-level correlation for the ten-year interval 2013-2022. The correlation was statistically vital: r = .032, p = .029. (Notice that r values for disaggregated analyses like it will appear low to folks used to decoding r values in different contexts. Eyeballing the chart is a greater intuitive evaluation of impact dimension. The vital factor is that the low p worth [under the conventional .05] means that the visually believable relationship between yr of PhD and gender within the 2013-2022 interval will not be as a result of likelihood.)
Since it is a post-hoc evaluation, and a p-value of .029 isn’t nice, so it is sensible to check for robustness. Does it matter that I chosen 2013 specifically as my begin date? Thankfully, we get related outcomes selecting 2012 or 2014 as the beginning years, although for 2014 the result’s solely marginally statistically vital (respectively, r = .037, p = .008; r = .026, p = .099).
One other strategy is to bin the information into five-year intervals, to easy out noise. If we create five-year bins for the previous twenty years, we see:
1993-1997: 27% girls (453/1687)
1998-2002: 27% (515/1941)
2003-2007: 27% (520/1893)
2008-2012: 28% (632/2242)
2013-2017: 29% (701/2435)
2018-2022: 31% (707/2279)
Evaluating all of the bins pairwise, 2018-2022 is a statistically considerably increased proportion of girls than the bins from 1993-2012 and statistically marginally increased than 2013-2017 (p values: .004, .001, .012, .037, .094, respectively). No different pairwise comparisons are vital.
I don’t suppose we may be assured. Put up-hoc analyses of this type are dangerous—one can see patterns within the noise, then unintentionally p-hack them into seeming actual. However the truth that the upward latest pattern comes throughout in two very completely different analyses of the information and passes a robustness examine inclines me to suppose the impact might be actual.
[1] “Philosophy” has been a “subfield” or “detailed area” within the SED knowledge from at the very least 1973. From 2012-2020, the SED additionally had a separate class for “Ethics”, with considerably fewer respondents than the “Philosophy” class. For this era, each “Ethics” and “Philosophy” are included within the evaluation above. Beginning in 2021, the SED launched a separate class for “Historical past / philosophy of science, expertise, and society”. Respondents on this class are usually not included within the evaluation above. Complete “Philosophy” PhD recipients dropped about 15% from 2019 and 2020 to 2021 and 2022, which could partly replicate a loss to this new class of some respondents who would in any other case have been categorised as “Philosophy” — however may additionally partly be noise, partly long-term developments, partly pandemic-related short-term developments.
