The Sources and Magnitude of Partisan Bias in U.S. Federal Elections

• In Event: Racial and Partisan Gerrymandering: New Approaches for the Next Decade?

Abstract

Is the U.S. electoral system biased toward one party, how biased is it, how much has that changed over time, and what accounts for the bias? The literature has offered many answers to these questions, identifying incumbency, redistricting, and fundamental political geography to varying degrees.

We argue these accounts are incomplete because of flaws in the way that bias has been measured. We turn to a new measure of partisan bias in single-member district elections, the “efficiency gap,” that has gained traction and is the basis of a landmark legal case that will soon come before the U.S. Supreme Court. The efficiency gap does not require a counterfactual like the most common measure in the literature, symmetry. As such, it can be calculated for any redistricting plan and with either raw votes or vote shares, avoiding many problematic assumptions.

The measure allows us to calculate bias for both the U.S. House and U.S. Senate elections, and to break the sources of the bias into three groups: malapportionment, differential turnout, and differential majority size. We estimate the independent effects of incumbency, political geography, and redistricting on these trends, and extend the trends back to the dawn of popular elections for the U.S. Senate in 1916.

The results suggest that House elections have a growing Republican bias that is increasingly explained by redistricting, that incumbency was once important but no longer plays a significant role, and that neither malapportionment nor differential turnout have been important since the “one person, one vote” court decisions of the 1960s. Political geography also plays some role, but is not determinative. On the Senate side, bias has alternated more between the parties and has been much more a function of malapportionment. However, Republicans have still benefited more often than Democrats, especially in recent years.

These results suggest that, contrary to some arguments, political geography is important but is not destiny, and the relative contributions and consequences of other factors vary from one period to another.

Authors

• Eric McGhee, Public Policy Institute of California

• Nicholas Stephanopoulos, University of Chicago