I’ll be giving a controversial paper at two conferences: the American Political Science Association (Sept 1-4, in Philadelphia), and the European Consortium on Political Research (Sept 7-10, in Prague).

My paper draws parallels between thought experiments in political theory and philosophy, and controlled experiments/comparisons in the natural and social sciences. Some of these parallels have been noticed before, by people like Frances Kamm, Tamar Gendler, and (in the book on political theory methods that I’m editing) Kimberley Brownlee and Zofia Stemplowska. But no one I’m aware of has taken advantage of the powerful toolkit that social and natural scientists have developed. I thus use ideas like internal and external validity, controlled comparison, omitted variable bias, interaction effects, spurious correlations, testable implications, and parsimony.

This helps us see better how to do thought experiments, and how much we can learn from them.

Of course, some readers will be more interested in my broader claims about the relationship between political theory and science. But note that I don’t equate the two: there are parallels, but also important differences. By contrast, I do argue elsewhere that some textual interpretation is essentially scientific: we often ask empirical questions (like what Locke meant by ‘rights’ or why he wrote what he wrote), and scientific ideas are the best tools we have yet developed for answering such questions. (See here for the most explicit version of the argument, and here for the most details account of what a scientific approach to textual intepretation involves.)

This isn’t really what’s going on in political theory thought experiments – which are, furthermore, only one part of political theory, and a part that many authors don’t use. Nonetheless, this casts some light on what some philosophers of science mean when they discuss ‘naturalism’, defined here as philosophy and science being ‘continuous’.

Although I’ve been thinking about and teaching some of these ideas for many years, my paper was written quite quickly, and needs more work. In particular, I cannot yet say how widespread the problems I discuss are.

The paper is here. Any comments and criticisms would be much appreciated!