A State-Level U.S. House Election Forecast Model for 2022: Modeling the Potential Effects of Gerrymandering

Abstract

This model was developed specifically for the Midterm Election Forecasting Roundtable at the 2022 APSA Annual Meeting in Montréal. While most House forecast models generate forecasts either at the district level, or the aggregate number of seats won by a party, this model is different. It models the number of House seats in each state won by the Democrats and generates a forecast at the end of August. With 2022 being the first post-redistricting election of the decade, the main motivation behind this model was to attempt to capture the potential impact that gerrymandering would have on state-level outcomes. Gerrymandering is most likely to occur in a state under two conditions: (1) when reapportionment leads to a change in the number of seats apportioned to the state, and (2) when the state’s redistricting process is entirely controlled by one political party. Given that, this model includes a simple dummy variable for Gerrymander Potential, which is simply 1 if both of those conditions exist in a state in the first post-redistricting election of each decade. It is also party adjusted, taking on a negative value if the state’s redistricting process is controlled by Republicans, and positive if it is controlled by Democrats. If a state’s redistricting process was subject to divided party control, handled by an independent redistricting commission, or where the maps were drawn by state courts, I assigned this variable a value of 0.