(iPoster) Does Gerrymandering Increase Partisan Polarization in the US House?

• In Event: Poster Sessions: Legislative Studies & Authoritarian Demands, Compliance, and the Politics of Protest
  In Poster Session: Poster Session: Legislative Studies

Abstract

It is widely argued that gerrymandering is a primary cause of polarization in the US Congress. The academic literature on this matter is more divided, with some (Grainger 2010; Stephanopoulos 2018) arguing that it is a major factor while others (Mann 2006; Masket, Winburn, and Wright 2006; McCarty, Poole, and Rosenthal 2009) claim that redistricting has little effect on polarization in legislative bodies. Using US House election data from 1872 to 2022 and applying a fixed effects model as well as several other statistical approaches, my findings are consistent with the latter group of scholars: gerrymandering has some impact on partisan polarization in the House of Representatives, but that impact is small.

But why is this the case? I argue that a limitation in both the public debate and the current literature is that they do not distinguish between different types of gerrymandering. Packing and cracking would have a different impact on partisan polarization. The argument behind the gerrymandering-polarization thesis is that gerrymandering causes districts to be more partisan, and the increased number of very partisan districts causes the legislative body as a whole to be more polarized. However, cracking, or when the government deliberately divides a partisan group across multiple districts so that its votes are diluted across all these districts, has the opposite impact. Cracking makes the overall vote in districts less partisan, not more. Packing, on the other hand, concentrates supporters of one party into a single district so that the party wins that one district by an overwhelming level of support while losing all the surrounding districts. That type of overwhelming support in a packed district is exactly what could increase the level of polarization in a legislative body.

My analysis therefore focuses on the relationship between packed districts and partisan polarization. To accomplish this, I employ a new measure called the Reverse Gerrymandering Index (RGIx), which highlights how packed a district is. The mainstay of RGIx is that at a mathematical level, gerrymandering can be conceptualized as swapping voters across adjoining districts. In other words, when state governments gerrymander districts, those districts are redrawn in ways that keep the population numbers in each district the same, but certain voters are switched over to other districts. With packed districts, a large percent of supporters of one party are siphoned into one particular district from all the surrounding districts, while members of the other/opposing party are siphoned away from that one particular district into the other surrounding districts. This leads to the first party winning that particular district by a large margin but losing all the other surrounding districts.

RGIx works by reverse-engineering this process: it estimates what the vote would have been if the packing of each district were to be undone. It does this by approximating a process in which each voter is randomly moved to either their original district or to one of the adjacent districts. Such a process smoothens the vote across neighboring districts, and most importantly, it simulates what the vote might have been without gerrymandering. These smoothened votes act as an ‘expected value’ for each district, and the RGIx score for each district is then calculated to be the actual vote minus its expected value. In this way, each district is given an RGIx score that indicates how packed it is.

As I will show in this paper, the relationship between RGIx scores and DW-Nominate scores indicate that while members of the US House of Representatives from packed districts are more likely to have extreme voting records than members of Congress from districts that aren’t packed, that difference is significant but not substantial. One reason is that many of the districts that vote most heavily for one party are that way simply because of geographic variation, not district-line manipulation. In terms of the impact of gerrymandering on the overall polarization of Congress, another factor is that the number of packed districts nationally is relatively small, only around 10% of House districts nationally after the 2020 reapportionment. This is partially because many states do not gerrymander House districts (including because some states are apportioned few House seats) but also because packing requires siphoning off votes from adjacent districts. For the strategy to work, those adjacent districts have to have a two-party vote that is less extreme in its advantage for one of the parties than the packed district. Finally, by 2012, around 70% of packed districts were designed by Republican-led state governments and therefore heavily packed with Democratic voters, especially in southern states. Gerrymandering therefore cannot explain why DW-Nominate scores have shifted so far to the right by the Republican Party.

Author

• Bernard Tamas, Valdosta State University