The reapportionment revolution and bias in US congressional elections

We develop a formal model of the redistricting process that highlights the importance of two factors: first, partisan or bipartisan control of the red istricting process; second, the nature of the reversionary outcome, should the state legislature and governor fail to agree on a new districting plan. Using this model, we predict the levels of partisan bias and responsivenes s that should he observed under districting plans adopted under various con stellations of partisan control of state government and reversionary outcom es, testing our predictions on postwar(1946-70) U.S. House electoral data. We find strong evidence that both partisan control and reversionary outcome s systematically affect the nature of a redistricting plan and the subseque nt elections held under it. Further, we show that the well-known disappeara nce circa 1966 of what had been a long-time pro-Republican bias of about 6 percent in nonsouthern congressional elections can be explained largely by the changing composition of northern districting plans.