A Direct Measure of Representation through Districts and Parties

• In Event: Representation and the U.S. Congress

Abstract

Do legislatures represent their constituents? Existing empirical research tend to examine only one particular instance of a representative-constituent relationship at a time, leaving surprisingly few systematic and direct assessments that answer this question involving democratic theory and empirical political science. In this paper, we paint a broad picture of representation in the U.S. Congress. Our key innovation is a new measure of issue representation that is widely applicable, straightforward, and direct (rather than latent). Accompanying the measure are over 3 million survey responses from the Cooperative Congressional Election Study (CCES) on specific Congressional votes on key issues, which we then link to the corresponding floor vote. This publicly available dataset covers 181 key House and Senate votes, 309,956 voters from all states and congressional districts, and the six contemporary Congresses (109th - 114th) which experienced frequent changes in party control. In addition to its breadth, our data is dyadic at the individual-level rather than at the coarser district-level, and collects information on each individual constituents’ districts, partisan identification, and degree of ideological coherence. Such granularity enables the flexible measurement of legislator’s issue congruence with any theoretically relevant constituency, e.g. her district electorate or national party-base. Our findings cover the three central institutions discussed in the representation literature. (1) At the level of Congress as a whole enacting legislation through simple and super-majorities, we find that both the House and the Senate represent the national majority’s preference more often than not (institutional representation). (2) At the level of parties, we compute normative benchmarks by constructing hypothetical voting schedules as stipulated by democratic theory. We find that Republican and Democratic legislators differ starkly in the representational strategies they pursue, even after accounting for differences in their districts’ distribution of preferences. Republican House Members and Senators, but not Democrats, appear to vote as if they voted according to the majority opinion among nation-wide co-partisan voters (representation through parties) more than their district electorates (geographic representation). (3) At the level of districts, we transport the concept of wasted votes in the re-districting literature to issue representation, and asses how the packing of issue publics, in addition to racial groups, shapes patterns of representation. Throughout, we assess Congressional representation with direct dyads and with clear benchmarks derived from democratic theory. In doing so, this paper proposes a new theoretical framework of legislation, one that puts the exercise of representing different constituencies front and center.

Authors

• Stephen D. Ansolabehere, Harvard University

• Shiro Kuriwaki, Harvard University