On academic political science

Stephen Hayward writes what is in part a very standard sort of piece making fun of the lower tail of paper names from academic conferences. That's all good fun but not very deep and not very original.

The other part of Hayward's piece laments the attempts of political scientists to actually act like scientists, by developing theories and testing them against empirical data. This, he argues, makes them irrelevant, which apparently means that it reduces their chances of showing up on a Sunday morning spin show, or on the sale table at an airport book store, and thus gaining the attention of beltway insiders of the more intellectual sort.

My problem with Hayward's plea is that he seems to confuse political science departments with schools and departments of public policy. They have different names because, within the academic division of labor, they do different things. I would have thought, indeed, that it is also in part the task of institutions such as Hayward's employer, the American Enterprise Institute, to translate academic output into formats that beltway types can digest and use. Such translation is a very useful activity, just not one that is necessarily the best use of time for most academic political scientists.

Asking for "political insights" - does that mean things like "senator so-and-so's mistress is ready to go public for $200K"? - from political science is a bit like asking for tips on how to economize at the grocery story from academic economics. It misses the point.

Old and new

The paper presentation I found most interesting at the PolMeth meetings is this one by Gary King and co-authors. One way to think about it, which is not how it was presented, is that it seeks to codify the cell matching plus weighted regression schemes used in some of the ill-fated evaluations of CETA (= the Comprehensive Employment and Training Act, the predecessor to JTPA) back in the 1970s. Those evaluations are ably surveyed in Burt Barnow's (1987) Journal of Human Resources piece.

The basic idea is that you create cells using discrete variables plus discretized (or coarsened, in the language of the paper) continuous variables as a first step in a two step process, the second step of which is some parametric model that aims to catch any remaining bias. The method is marketed as simplifying the imposition of the common support - it is just those cells with both treated and untreated units - and also as avoiding the necessity of balancing tests as the (coarsened) covariates are necessarily balanced within cells.

The trick of course comes with choosing just how to coarsen. In the langauge of non-parametrics, this presents a k-dimensional bandwidth choice where the optimal choice for each bandwidth (which corresponds to the amount of coarsening in each of the k individual variables) depends on the choice of all the other bandwidths. In any given finite sample, institutional knowledge may suggest particular cuts - think of how you would group years of schooling into categories like < 10, 10-11, 12, 13-15 and so on) - but it is less clear how to think about the asymptotics. Also, once you rely on institutional knowledge and intution to guide the coarsening, rather than a more or less automated procedure like cross-validation, the opportunities for disagreement across researchers, and for misbehavior by researchers looking for particular answers, increase.

Nonetheless, the presentation was lively and got me thinking. I look forward to reading the (apparently already revised!) paper more closely.

PolMeth XXV

Last Friday and Saturday I attened the 25th annual meeting of the Society for Political Methodology here in Ann Arbor. This is the analogue of the Econometric Society for political scientists, but without the theorists. Michigan is very strong in quantitative political science so I thought this would be a good chance to get to know some of my colleagues better and also to see what was going on in this literature. Political methodologists often read both the economics and the statistics literatures which gives them an advantage of sorts over most economists and most statisticians, who read only in their own discipline.

Some observations:

1. Political science is going through the same non-parametric revolution now that economics went through about 15 years ago which produces a sort of deja vu as assumptions are tossed overboard in a (hopeless?) quest for purely data driven inference.

2. Political methodologists drink more than labor economists.

3. Political methodologists interrupt in presentations, which makes the visiting economist feel at home. They can even get a bit testy at times.

4. Political methodologists also fret about who to cite for the potential outcomes framework.

5. Not many, but a few, political methodologists surf the internet on their laptops during presentations.

6. I have never been to a conference that set aside so much time for discussion. Basically there are four papers a day in 90 minutes slot with a long lunch and a reception and dinners and with much of the 90 minutes set aside for the discussant and for audience comments. The time is well used; I have not been to a conference where the discussions (including lunch and dinner) had as a high a substance / gossip ratio in a while.

7. The triumphalist panel that ended the conference and celebrated the growth of the society over the past 25 years reminded me a lot so similar triumphal speeches at the Society of Labor Economists meetings.