In the following guest post, Carolyn Dicey Jennings, associate professor of philosophy at UC Merced, shares some new data about graduate programs in philosophy that she and her team at Academic Philosophy Data and Analysis (APDA) have collected and analyzed.
New Data about Philosophy Graduate Programs
by Carolyn Dicey Jennings
Academic Philosophy Data and Analysis (APDA) has been running in some form since late 2011. It collects and analyzes data about academic philosophy, with a special focus on PhD graduates and their employment outcomes. Its blog, which has been running since 2017, has had tens of thousands of visitors, and its most popular post is Best PhD Programs in Philosophy. This is an update to that post, based on this year’s data collection efforts.
An important caveat: this is not a ranking of PhD programs. I have been engaged in public conversations about rankings in philosophy for years now and have come to the conclusion that it is more useful to have sortable lists based on different criteria. The table on the homepage of APDA’s website is intended to provide just that. The table is currently sorted by overall student rating (for graduates 2012 and later), but can be sorted instead by average student rating of the program’s climate, teaching preparation, research preparation, financial support, the number of total 2012-2021 graduates, permanent academic placement rate for those graduates, placement rate into PhD-granting programs, temporary academic jobs, nonacademic jobs, or the primary nonacademic sectors for these graduates. This data can be examined in more detail by clicking on the name of each program.
Ok, so which programs are the best? I will focus in this post on two metrics: student ratings and employment, ending the post with some comments on overall employment trends. (Below, “*” is used to indicate programs that show up on both lists.)
Let’s start with the programs that have the highest recommendation from their graduate students, focusing on current students and graduates from 2012 and later. These are divided by topical cluster. In a paper forthcoming in Metaphilosophy, Pablo Contreras Kallens, Dan Hicks, and I describe the method of clustering programs:
In machine learning, cluster analysis is any method that arranges units of analysis into subsets, i.e. clusters, based on some measure of similarity between the variables that characterize them (James, Witten, Hastie, & Tibshirani 2013, 10.3). In the current project, the units of analysis are philosophy PhD programs, and the variables that characterize them are aggregated from (1) the areas of specialization (AOS) of their PhD graduates and (2) the ‘keyword’ survey responses.
For this post I will use just three clusters, which were the best supported in our research: 1) analytic philosophy, 2) philosophy of science, and 3) historical, continental, and applied philosophy.
The following programs were given an average rating of “definitely would recommend”:
Rutgers University* (n=26; 4.73)
Massachusetts Institute of Technology* (n=19; 4.70)
Australian National University (n=20; 4.69)
University of Illinois at Chicago (n=6; 4.67)
University of North Carolina at Chapel Hill* (n=30; 4.65)
University of California, Berkeley (n=20; 4.60)
University of Southern California* (n=15; 4.60)
Yale University* (n=18; 4.53)
University of Michigan* (n=18; 4.50)
Philosophy of Science
Carnegie Mellon University (n=11; 4.55)
University of Cambridge, HPS (n=7; 4.50)
Historical, Continental, and Applied philosophy
University of California, Riverside* (n=14; 4.68)
The Catholic University of America* (n=6; 4.67)
Saint Louis University (n=5; 4.60)
Some of these also show up with “very satisfied” ratings in the other domains, such as:
climate – Rutgers, ANU, CUA (other universities with “very satisfied” ratings in climate are Rice, Sheffield, KCL, St Andrews/Stirling, UConn, and UVA);
preparation for research – USC, Rutgers, MIT, Riverside (other universities with this rating include Pittsburgh HPS, Bowling Green, UCLA, Cornell, UPenn, Irvine LPS, Toronto, and Penn State)
financial support – USC and Yale (other universities with this rating include Emory, Notre Dame, NYU, Baylor, Vanderbilt, Rice, and Pittsburgh).
Unfortunately, none of these programs have average ratings of “very satisfied” on preparation for teaching, but programs that do have this rating are: Kansas, Villanova, Bowling Green, Baylor, Georgetown, Fordham, and University of Washington.
Permanent academic placement rate just means the proportion of graduates whose most recent employment is in a permanent academic job. Because most graduates (around 90%) prefer academic employment, and because permanent academic employment is preferred over temporary employment, this metric is the standard for our project as a measure of successful placement. Beginning with this measure, then, the following programs have permanent academic placement rates that are at least one standard deviation above the mean:
Yale University* (52 graduates; 71%)
University of Southern California* (49 graduates; 69%)
University of Virginia (38 graduates; 61%)
Massachusetts Institute of Technology* (45 graduates; 60%)
University of Florida (5 graduates; 60%)
University of Michigan* (47 graduates; 60%)
New York University (54 graduates; 59%)
Harvard University (54 graduates; 56%)
Rutgers University* (63 graduates; 54%)
University of North Carolina at Chapel Hill* (56 graduates; 54%)
Philosophy of Science
University of Cincinnati (16 graduates; 69%)
University of California, Irvine LPS (24 graduates; 54%)
University of Pittsburgh, HPS (39 graduates; 54%)
Historical, Continental, and Applied philosophy
Baylor University (41 graduates; 63%)
Vanderbilt University (40 graduates; 63%)
Boston University (40 graduates; 58%)
University of Oregon (37 graduates; 57%)
University of California, Riverside* (30 graduates; 57%)
The Catholic University of America* (65 graduates; 55%)
Pennsylvania State University (50 graduates; 54%)
Many of these programs are also at least one standard deviation above the mean for permanent placement into PhD-granting programs (Irvine LPS, Yale, MIT, NYU, Michigan, Pittsburgh HPS, Harvard, Rutgers, and Penn State; other programs not on this list that have this quality include Cambridge HPS, Salzburg, Berkeley, Chicago CHSS, Carnegie Mellon, Princeton, Stanford, Oxford, LSE, Arizona, Cambridge, and Wash U).
Finally, nearly 20 programs are at least one standard deviation above the mean proportion of graduates with nonacademic employment, with proportion and primary nonacademic sectors listed:
University of Waterloo (43%: health; consultancy; education)
University of Guelph (41%: consultancy; arts; publishing)
University of California, Santa Cruz (41%: arts; consultancy)
Victoria University of Wellington (38%)
The University of Melbourne (36%: government; education)
University of Iowa (36%)
Fordham University (36%: education; tech; religion)
University of Otago (33%: education; government)
University of Rochester (32%: tech)
Western University (30%: consultancy; education; health; government; arts; publishing)
Brown University (30%: education; tech; law; consultancy)
Florida State University (30%: health; government; consultancy)
University of Dallas (29%: religion)
University of Arkansas (29%)
University of Kentucky (29%: education)
University of Toronto, IHST (27%: government)
University of Georgia (27%: education; non-profit/NGO)
University of Kansas (27%: tech; education; non-profit/NGO)
Georgetown University (26%: education; government; non-profit/NGO; tech; health)
As reported at APDA, we can get a bird’s eye view of employment trends through a Sankey chart, with the number of graduates in different categories on the left, and the number of graduates employed in different types of jobs on the right. In this case graduates are split into three groups depending on the permanent academic placement rate of their PhD program, with those in programs of the highest placement rates in Group 1, middle rates in Group 2, and lowest rates in Group 3.
Comparing the thickness of the bars connecting the left to the right, you can see that Group 1 prefers hiring from Group 1 over Group 2 and 3, and that Group 2 prefers hiring from either Group 1 or 2 over Group 3. This may reflect the prestige bias reported by De Cruz (2018) and in Contreras Kallens, Hicks, and Jennings (forthcoming):
we used two separate methods to establish that prestige plays an important role in the hiring of job candidates into philosophy PhD programs, finding a significant gap in prestige between those graduate programs that hire from all other programs (low-prestige) and those graduate programs that tend to only hire from a select group.
The chart also gives a better sense of where all other graduates are employed, with the largest group in temporary academic positions that are not fellowships or postdocs.
Finally, as reported in Jennings and Dayer (2022), the proportion who find permanent academic jobs in their first year post-graduation does not seem to have changed in the pandemic and associated recession, as can be seen with the black dashed line in the graph below. That is, whereas the permanent academic placement rate does drop for more recent graduates, this is largely because it takes time for most graduates to find this type of position—recent graduates have had less time, and so are less likely to have a permanent academic placement. In contrast, there is no drop off when we focus on those graduates who found a permanent academic position in the same year they graduated. This indicates that the proportion of those finding permanent academic jobs has held steady, at least for those who find such jobs relatively quickly.
If we look at the first-listed areas of specialization for those 2012-2021 graduates in permanent academic employment, only three areas are at least one standard deviation above the mean permanent academic placement rate for all AOS’s: Comparative (9 graduates; 89%), Asian (38 graduates; 68%), and American (including Latin American: 31 graduates; 52%). Another three are one standard deviation below the mean: Economics (15 graduates; 27%), Aesthetics (153 graduates; 25%), and Action (69 graduates; 25%).
Trends are difficult to assess. I used indexed placement rates for each AOS, such that the placement rate of a specific AOS in a specific year is divided by the placement rate for all AOS’s that year. Using linear regression over these values reveals that some AOS’s have a positive slope over this time period: Technology, Asian, Aesthetics, Medieval/Renaissance, Biology including Environmental, Social/Political, Gender/Race/Sexuality/Disability Studies, Metaphysics, and Math have a 3% or greater slope for those years with at least 5 graduates. Other AOS’s have a negative slope: Action, Mind, Law, American including Latin American, Logic, Ethics, Epistemology, Applied Ethics including Bio and Medical, Value General, Meta-Ethics, and Decision Theory have a -3% or less slope for those years with at least 5 graduates. But this should be taken with a grain of salt since the limitation of 5 graduates leads to gaps in the data. Technology, for instance, has the highest slope but only three years worth of data (.36, 1.19, and 1.64 for 2018, 2019, and 2020), whereas Decision Theory has the lowest slope but only two years worth (1.15 in 2016 and 0.64 in 2019).
If you have ideas for how the project can improve, or other studies we should run, we would love to hear from you in the comments or over email: [email protected].