When Do Legislators Represent Their Constituents? Evidence from Roll-Call and Referendum Votes

This paper develops a measure of representation that uses referendum returns to capture constituent opinion, and applies it to 4,192 roll-call votes on 33 laws in 10 states. Roll-call votes were congruent with majority/median opinion in a district 66 percent of the time. Roll-call votes can be explained primarily by legislator ideology, with constituent opinion of secondary importance. The data do not show a reliable connection between congruence and competitive elections, term limits, media attention, and party pressure. The evidence generally supports the predictions of representation theories that emphasize selection of legislators that share constituent ideology, and provides little support for theories that emphasize re-election incentives.