To examine this hypothesis, we adapted a computer algorithm that we recently introduced in the Quarterly Journal of Political Science. It allows us to draw thousands of alternative, nonpartisan redistricting plans and assess the partisan advantage built into each plan. First we created a large number of districting plans (as many as 1,000) for each of 49 states. Then we predicted the probability that a Democrat or Republican would win each simulated district based on the results of the 2008 presidential election and tallied the expected Republican seats associated with each simulated plan.
The results were not encouraging for reform advocates. In the vast majority of states, our nonpartisan simulations produced Republican seat shares that were not much different from the actual numbers in the last election. This was true even in some states, like Indiana and Missouri, with heavy Republican influence over redistricting. Both of these states were hotly contested and leaned only slightly Republican over all, but of the 17 seats between them, only four were won by Democrats (in St. Louis, Kansas City, Gary and Indianapolis). While some of our simulations generated an additional Democratic seat around St. Louis or Indianapolis, most of them did not, and in any case, a vanishingly small number of simulations gave Democrats a congressional seat share commensurate with their overall support in these states.
The problem for Democrats is that they have overwhelming majorities not only in the dense, poor urban centers, but also in isolated, far-flung college towns, historical mining areas and 19th-century manufacturing towns that are surrounded by and ultimately overwhelmed by rural Republicans.