Specialization and Status in Philosophy
I’ve been looking at some data from the Philosophical Gourmet Report, a well-known and widely-used reputational survey of philosophers in the U.S., Canada, the U.K. and Australasia. The survey asks philosophers to rate the overall reputations of graduate programs as well as their strength in various subfields. The ratings are endogenous, in the sense that the philosophers who produce them are members of the departments that are being rated. This gives the survey some interesting relational properties and allows for an analysis of the social structure of reputation in the field. I’ve written a working paper that analyses the data from this perspective. It’s still in in pretty rough shape: there’s not much in the way of theory or a framing argument yet, and it’s way short on citations to the relevant literature. (So don’t be too harsh on it.)
I’m not sure whether philosophers will like the paper. On the one hand, they tend to think of themselves as sensible individuals guided by common-sense and rational argument. This makes them resist thinking of themselves in sociological terms, subject to the influences of context, social relations and role constraints. On the other hand, in my experience they have an insatiable capacity for gossip. Within the limits of the data, the paper addresses three questions:
- (1) What contribution does prestige in particular subfields make to the overall reputation of philosophy departments?
- (2) Does being a specialist in a subfield affect the way philosophers assign prestige to other departments?
- (3) How much consensus is there about prestige in philosophy?
The answers are, in short, “It depends on the field”, “Yes, but only sometimes, and then only for high-status specialists”, and “A great deal.” Some quick findings:
- Subfields cluster into four groups: Continental philosophy; Medieval/Religion; a large “contemporary” group comprising Metaphysics, Mind, Language, Epistemology, Science, and Ethics; and a “historical” group comprised of Ancient, 17th Century, 18th Century and Kant/German Idealism. High ratings in the first two groups do not contribute to a department’s overall ranking. High ratings in the second two almost always matter, but to differing degrees.
- Metaphysics and Language make the largest contribution to overall reputation, though most other subfields also make a significant difference.
- High-status specialists in Philosophy of Language, Ethics and Contintental philosophy are more likely to assess departments harshly.
- British raters are substantially more generous than average.
Here are two visualizations of the field: the first is a blockmodel describing the relational structure of prestige amongst U.S. philosophy departments. The second is a segment plot showing the profile of departments across a range of different subfields. I think they’re both pretty cool, so read on.
A Blockmodel of Prestige
First, the blockmodel. You can get the full-size version of this graphic, or a higher-resolution PDF version, or the captioned version from the paper. This picture is a department by department matrix. Each colored cell represents the average “vote” by a department in the row for a department in the column. Departments are sorted in the same order in the rows and columns, according to an algorithm that groups them by how similar their voting patterns. Purple and blue cells represent high rankings; green cells represent middling rankings; brown and yellow cells represent low rankings. (The captioned version provides a reference scale.) The main diagonal is blank because departments are not allowed vote for themselves. In a high-consensus field, we’d expect each column to be the same color all the way down: that is, everyone agrees on how good a particular department is. In a low-consensus field, we’d expect more heterogeneity, with disagreement on the quality of particular programs. The data suggest that—at least according to the respondents to the PGR —philosophy is a very high-consensus discipline.
To help the interpretation, we can further group departments into “blocks” based on their similarity: members of the same blocks will stand in similar structural relations to other departments. In this case, I’ve generated a model with 5 blocks. Blocks are set off by thicker lines that project out into the margin. Block 1 is made up of the just first four departments, so the first four rows and columns show Block 1’s assessment of itself, for example. The four Block 1 departments enjoy the highest prestige and the greatest degree of consensus about their quality. Looking down the first four columns lets you see what everyone else thinks of Block 1—almost everyone agrees they’re the best, as you can see by the almost unbroken strip of purple and dark blue. Looking across the first four rows lets you see what Block 1 thinks of everyone else. Thus, focusing on the intersections of the graph created by the thicker horizontal lines lets you see how different blocks relate to one another (and themselves). For instance, the bottom right corner of the figure shows what the lowest-status block, Block 5, thinks of itself, so to speak. It turns out that it agrees with everyone else’s assessment of its relatively low quality. In fact, as I show in the paper, Block 5 thinks a little better of Block 1 than Block 1 thinks of itself, and thinks a little worse of itself than Block 1 thinks of it. In other words, the lowest-prestige block is slightly more committed to the hierarchy than the highest-prestige block. Although the mean scores awarded to blocks varies across blocks, there is complete agreement on the rank-ordering of blocks. So, for example, there’s no dissenting group that thinks itself better than everyone else believes.
Departmental Strength in Specialist Areas
This is a segment plot. Again, you can get a larger version of it, or a much nicer PDF version. For each department, the wedges of the plot represent the department’s reputation in a particular subfield. The bigger the wedge, the better the reputation. A department that was equally and maximally strong in all areas would look like the key at the bottom. To simplify the presentation I’ve grouped metaphysics, philosophy of mind and philosophy of language into a single group, “MML.” (This has a substantive justification, as strength in these areas is highly correlated.) The distribution of segments gives a nice picture of a each department’s profile: what it’s known for. I’ve also ordered the subfields clockwise from the left, roughly in order of their contribution to the overall reputation of a department. You’ll notice that Princeton and Oxford are the only departments in the Top 15 or so to have a roughly symmetric “fan-like” structure, indicating strength in a wide variety of areas. By contrast, NYU is very strong in MML, Ethics and History, but not ancient or continental. Rutgers’ profile looks like a chambered nautilus: it’s very strong in MML and Epistemology, and gets progressively weaker as you move around the half-circle. Yet NYU and Rutgers outscore Oxford and Princeton in terms of overall reputation. This is because—as the paper shows—not all areas contribute equally to the status of a department. Strength in MML is more important than strength in, say, Ancient philosophy or (especially) Continental philosophy. The segment plot scales all the specialties so that they have the same weight, but in reality this isn’t the case—so Oxford doesn’t capitalize on its strength in Ancient philosophy, for example. Michigan is an unusual case in that it ranks very highly despite lacking a strong reputation for MML and Epistemology. Conversely, strength in MML will only get you so far: MIT and the ANU excel in these areas, but probably won’t go any higher in the ratings without diversifying.
The plot shows some other features of departments and the field in general, too. Harvard’s relative weakness in MML is clear, for instance, as is Chicago’s strength in continental philosophy. As one moves down the rankings, the size of the wedges declines, of course, and departments with distinctive niches appear: the LSE is strong in the philosophy of science, Penn in modern history, Wisconsin in Science.
The data have a number of limitations, of course. For one thing, not all departments are present in the survey, and in most cases only one or two representatives of those departments were sampled. But it’s still a rich dataset. The draft paper has a fuller discussion of all this, together with a few other neat visualizations of the structure of the field. Comments are welcome, of course.