Source: Daniel Jones, Randall Walsh, National Bureau of Economic Research (NBER), NBER Working Paper No. w22526, August 2016
From the abstract:
How does the partisan composition of an electorate impact the policies adopted by an elected representative? We take advantage of variation in the partisan composition of Congressional districts stemming from Census-initiated redistricting in the 1990’s, 2000’s, and 2010’s. Using this variation, we examine how an increase in Democrat share within a district impacts the district representative’s roll call voting. We find that an increase in Democrat share within a district causes more leftist roll call voting. This occurs because a Democrat is more likely to hold the seat, but also because – in contrast to existing empirical work – partisan composition has a direct effect on the roll call voting of individual representatives. This is true of both Democrats and Republicans. It is also true regardless of the nature of the redistricting (e.g., whether the redistricting was generated by a partisan or non-partisan process).