Abstract
Scholars studying the determinants of congressional voting behavior have always faced the problem of how to measure the effects of constituency influence. In place of random surveys in congressional districts, which are too costly to conduct nationwide, scholars have employed a variety of proxy variables to measure constituency attitudes. In this research note we review these proxies and test the reliability and validity of the most popular, that is, presidential election returns at the state and district level. The use of this proxy rests upon two assumptions: (a) short-term factors affecting the vote have a more or less uniform effect across subnational constituencies and (b) constituency ideology is the sole long-term factor affecting the vote. Both of these assumptions prove to be empirically questionable, leading us to conclude that presiden tial election returns should be used as a proxy for constituency ideology only with great caution.
Get full access to this article
View all access options for this article.
