Recent efforts at estimating bias and responsiveness in electoral systems typically proceed by assigning observations to subsamples according to which party controlled the redistricting process. We show this traditional procedure to introduce selection bias into the resulting estimates of bias and responsiveness and present an alternative strategy for estimating these parameters. Using data from the state legislatures, and employing two different measures of partisan control of redistricting, we obtain results that modestly differ from those obtained with the traditional approach. Measures of control of redistricting utilizing information about the partisan intent of redistricting commissions and tribunals are exogenous to the seats-votes relationship.