The study of representation requires being able to measure representation. The concept to be measured, in the abstract, is the distance or "congruence" between legislators and their constituents, but data limitations often preclude measuring congruence directly. A popular alternative is to regress legislator roll call votes on a proxy for constituent preferences, with the coefficient labeled "responsiveness." Previous research has shown that there is no theoretical connection between the responsiveness coefficient and congruence. This paper investigates if there is an empirical connection between the responsiveness coefficient and congruence. I study 3,242 roll call votes on state laws that were subsequently challenged in a referendum; for such laws, one can construct both congruence and the responsiveness coefficient. The main finding is that the responsiveness coefficient has no consistent connection to congruence, meaning that it does not provide a reliable measure of representation.The study of representation requires being able to measure representation. The concept to be measured, in the abstract, is the distance or "congruence" between legislators and their constituents, but data limitations often preclude measuring congruence directly. A popular alternative is to regress legislator roll call votes on a proxy for constituent preferences, with the coefficient labeled "responsiveness." Previous research has shown that there is no theoretical connection between the responsiveness coefficient and congruence. This paper investigates if there is an empirical connection between the responsiveness coefficient and congruence. I study 3,242 roll call votes on state laws that were subsequently challenged in a referendum; for such laws, one can construct both congruence and the responsiveness coefficient. The main finding is that the responsiveness coefficient has no consistent connection to congruence, meaning that it does not provide a reliable measure of representation.June 2015 * Comments, corrections, and suggestions are welcome: matsusak@usc.edu. I thank USC for support.