This paper compares two common methods used to measure representation in empirical research: “congruence,” whether a roll-call vote is aligned with constituent opinion; and “responsiveness,” the correlation between roll-call votes and constituent opinion. While sometimes viewed as different ways of capturing the same underlying concept of “representation,” theoretically the measures are not connected. This paper assesses to what extent responsiveness and congruence capture the same thing empirically. Using 4,192 roll-call votes on 33 state laws that were subsequently challenged in referendums, I calculate both congruence and responsiveness for the same set of votes. The main finding is that responsiveness is an unreliable predictor of congruence, and therefore cannot be taken as a proxy for congruence.
John G. Matsusaka (Thu,) studied this question.