Sun Dec 5, 2004

Academic Job Markets and Status Hierarchies

Over at Brian Leiter’s blog, there’s a debate going on about the role of publications in the hiring process. Keith DeRose is arguing that a graduate student’s publication record should be given a larger role than it often is:

[W]hich graduate school one gets into and what job one initially lands tragically does very much to determine how well one is likely to do, long-term. It often happens for instance, that extremely talented philosophers who deserve to do as well as those landing the great jobs instead end up at some low-prestige job with a heavy teaching load. Every now and then, one of them quite heroically overcomes the odds of having to write while teaching so much and puts out a bunch of excellent papers in really good journals (which at least often they’re able to do largely b/c the journals use blind review!). But, too often, they can’t get the people with the power in the profession (& who know that the candidate works at a low-prestige place) to take their work seriously. They lose out to candidates (the “chosen ones”) who, despite their very cushy teaching loads, publish little in good journals but who have something that all too often proves more valuable on a CV: a high-prestige institutional affiliation.

The data strongly back Keith up on this point, but they also suggest that the probability that things will change is not very high. Studies of academic disciplines show that by far the most important predictor of departmental prestige is the exchange of graduate students within hiring networks. These networks shouldn’t really be called job markets, incidentally, because they lack most of the features normally considered necessary for a market to exist.

A recent paper by Val Burris in the American Sociological Review, for instance, shows that in Sociology,

centrality within interdepartmental hiring networks explains 84 percent of the variance in departmental prestige. Similar findings are reported for history and political science. This alternative understanding of academic prestige helps clarify anomalies —- e.g., the variance in prestige unconnected to scholarly productivity, the strong association between department size and prestige, and the longterm stability of prestige rankings — encountered in research that is based on the more conventional view.

The “conventional view,” of course, is that departmental prestige is a function of the scholarly productivity of faculty members and graduate students, but this is not the case in practice. A related paper by Shin-Kap Han describes the structure of the exchange network in several disciplines. A I wrote when discussing it last year:

Job candidates in all disciplines are exchanged in a well-defined manner between three classes of departments. Class I departments, at the top, exchange students amongst themselves and supply lower-tier departments with students but do not hire from them. Class II departments are on the “semi-periphery,” generally exchanging candidates with each other (though there is a hierarchical element to this) and also sending students to Class III departments, which never place students outside of their class and usually do not hire students from within their class.

What does all this mean for individual graduate students on the market? In effect, it means they are not market actors at all. They’re the currency in a different exchange system altogether, namely the status hierarchy of departments. The data show that departmental status rankings are remarkably stable—in some cases almost unbelievably so. It’s very difficult to fall out of the Top Tier once you’re in it, despite the enormous turnover of faculty, the way departments age, changing academic fashions and so on. So if you’re thinking about grad school, remember that the prestige of your Ph.D-granting department is the best predictor of your chances on the job market. If you’re at a low ranking school it is highly unlikely that you’ll move up more than one tier when you’re exchanged.

Now, top departments turn out more graduate students than there are jobs in top departments, so many Tier I products will not find Tier I jobs. So it’s not enough just to go to a top school. Formal research is harder to do on this point, but it seems clear that within Tier I there will be further stratification by status, with some advisors and research groups being more powerful or popular than others, and so on. Choosing the right topic and being in the right networks will increase your chances of getting a good position. The upshot is that in some fields it’s possible to get hired on the strength of pure “promise”—but of course only people in the right structural positions will be seen has showing promise, because everyone starts from the position of not having written anything. In philosophy, uncertainly about whether candidates will pan out is perhaps higher than usual because there’s no empirical component to anyone’s project and a lot of work gets done interactively, in conversation. As a result, the field’s collective commitment to the ideal of the individual genius is much higher. This may also make it possible to survive for longer in a high-prestige position on very little output.

Besides the question of whether this is a sensible way for supposedly rational philosophers to organize their field, the network perspective suggests a further irony. As Ron Burt argues in the current American Journal of Sociology, innovation and new ideas tend to come from the edges of networks, not the center:

Opinion and behavior are more homogeneous within than between groups, so people connected across groups are more familiar with alternative ways of thinking and behaving. Brokerage across the structural holes[1] between groups provides a vision of options otherwise unseen, which is the mechanism by which brokerage becomes social capital. … Compensation, positive performance evaluations, promotions, and good ideas are disproportionately in the hands of people whose networks span structural holes. The between-group brokers are more likely to express ideas, less likely to have ideas dismissed, and more likely to have ideas evaluated as valuable.

In other words, it may be that the people in the structural position to get the best ideas are less likely to be hired. At the level of informal job-market gossip there are bits of conventional wisdom that remind people of this worry. For instance, people may say “X is just a clone of Prof Y,” or “X is just yet another person working on problem θ. Boring.” Burt’s argument suggests that in the long run, truly innovative actors should do better than the ordinary runs-of-the-mill, whereas Burris’s suggests that the stable center ought to prevail. I don’t know of anything that considers the structural stability of departmental rankings (and their dependence on exchanging students) alongside the idea that new ideas ought to be generated by unconventionally-networked individuals. It might turn out to be quite a complex dynamic.

My own intuition is that uncertainty about future productivity creates sufficient anxiety that departments tend to choose the safer candidates in the short run—i.e., the ones coming out of their peer-group departments. This might also account for the high propensity of academic exchange networks to herding behavior (where one candidate gets all the offers, initially). Indeed, according to Han’s paper, the discipline that does the most to take the market out of the market—to make it more like a hierarchical queueing procedure, instead, where the top departments get first pick of the candidates—is Economics. This is also a field where it’s conventional for candidates go on the market without any publications.

Notes

fn1. A structural hole is, roughly, a gap between two dense network clusters. If the two clusters have something to offer each other—resources, profit opportunities, information, ideas—then the individual who bridges the hole and serves as the connector between the two groups can benefit. The advantage to the bridging agent declines the more connections form between nodes in the two groups.