Committee Power, Structure-Induced Equilibria, and Roll Call Votes

Abstract

Listen

Translate article

Using the concept of structure-induced equilibrium, this essay develops a simple behavioral model of roll call voting. The main results are (1) If some committees are outliers relative to the legislature as a whole-that is, if the median ideal point of the committee members is different from the median ideal point of the whole legislature-then roll call data are likely to be artificially unidimensional. This is because outlier committees will often exercise their gatekeeping power, so that the set of bills on which roll calls are taken will not be a random sample of all possible bills, but instead there will be important selection bias. (2) The unidimensional bias generally persists even as the dimensionality of the policy space becomes large; in fact, when the number of outlier committees is large, the bias may actually increase as the number of policy dimensions increases, both in absolute and relative terms. (3) Even if committees are preference outliers, it may be impossible to discover this fact by looking only at roll call votes. That is, the roll call data might indicate that committees are representative samples of the full legislature even when they are not. This is again the result of a selection bias: the roll calls that would distinguish certain committee members from other legislators never occur.