The Partisan Advantage Tracker is a tool to evaluate congressional district maps drawn for the U.S. House of Representatives. The Tracker measures how much these maps favor one political party over another, in each state, and nationwide.

For a given state, and for a given list of total number of votes for each of the two major parties (Democratic and Republican) in each precinct in the state, we calculate:

• First: the number of seats each party should win given these votes, according to a notion of fair representation.
• Second: the number of seats each party would win, with these same votes, according to the current map of districts.

The difference between the number of seats each party would win and the number of seats it should win, is the partisan advantage of the current map, given these votes. Adding up across all states, the Tracker also provides the total Partisan Advantage nationwide. By convention, positive numbers indicate a partisan advantage for the GOP (the GOP wins more seats than it should), and negative numbers indicate a partisan advantage for the Democratic Party. We repeat this computation for four different notions of fair representation, and the Tracker reports the results for each notion in a separate column.

The Partisan Advantage Tracker shows the extent to which the 2022 congressional redistricting map in each state favors the Democratic Party or the Republican Party, relative to the number of seats each party should win according to each of four different notions of fair representation, and always given the votes that each party received in recent statewide elections.

The organizations Politico and fivethirtyeight.com also publish redistricting trackers that show whether the 2022 maps favor either party compared to the 2012 maps. But if the case is that the 2012 maps were unfair, then this comparison to the older maps only tells us how the size of the partisan advantage has changed. It doesn’t necessarily tell us whether new maps, drawn after the 2020 U.S. Census, give one or the other party an advantage in seats, relative to the number of seats that should be won under any standard of fair representation.

Our approach is to take the number of votes each party received in each precinct or ward (or smallest geographic unit counting and reporting votes) in a recent statewide election. We then use a definition of fair representation to determine how many seats each party should win with these vote tallies, according to this definition of fairness. This is what we call the number of Fair Seats for a given list of vote tallies and a given notion of fairness.

We then look at how many seats the party would win if these exact same vote tallies in each precinct were added up into districts according to the current maps. We call this number of seats the Map Seats. The difference between the number of seats a party wins if we add up votes by district according to the map --in other words the Map Seats— and the number of fair seats a party should win according to the notion of fair representation, --in other words Fair Seats— is our measure of partisan advantage for these vote tallies and this notion of fairness.

PARTISAN ADVANTAGE = MAP SEATS - FAIR SEATS

Computing the partisan advantage in this manner for the vote tallies in different statewide elections in each state, and computing an average over the results obtained with each election data, we obtain our measure of Partisan Advantage for that state. We add up the Partisan Advantage measure across all 44 states and obtain the Partisan Advantage nationwide. In this way, the Tracker evaluates the congressional redistricting maps in 44 states with a combined 429 U.S. House seats. The other six states have only one congressional district each, and thus do not engage in congressional redistricting.

By convention, positive numbers denote political advantage for Republicans. Negative ones indicate political advantage for the Democratic Party.

In theory, our model allows for use of election data from any election, but in practice we rely on data produced by the Voting and Election Science Team (VEST) as compiled by Dave’s Redistricting Application DRA 2020 on presidential elections, U.S. Senate elections, and gubernatorial election from 2016 to 2020. For most states, the dataset contains either five or six elections, but the exact number varies by state. At present, this dataset doesn’t include U.S. House election tallies. The use of this data is regulated by the Creative Commons Attribution license (CC BY 4.0)

A downloadable Excel file (.xlsx) provides the results election by election.

There are different views as to what is fair representation in a democracy. Many countries have a principle of proportionality built into their constitutions: in these countries, the number of seats for each party is proportional to its number of votes. In contrast, the electoral system in the United States is by design non-proportional. Our model considers four notions of fairness that recognize the “pro-majority” feature of the U.S. electoral system. The first three of these notions determine the number of seats as a mathematical rule that depends on the party’s vote-share, defined as the share or fraction of votes cast for this party, among all votes cast for either of the two major parties. Similarly, we define a party’s seat-share as the share or fraction of seats held by this party.

These are the four understandings of fairness utilized in this Partisan Advantage Tracker:

• 1. “Efficency Gap” rule. The difference between the two parties’ seat-shares should be twice their vote-share difference i.e. if one party gets 60% of the vote and the other gets 40% (a 20% vote-share difference), then they should split seats 70%-30% (a 40% seat-share difference).
• 2. “Quadratic” rule. Seat-share should follow a mathematical square function. In this calculation of fairness, the minority party gets seat-share equal to twice the square of its vote-share, i.e. if a party gets 60% of the vote, and the other party gets 40%, the second party should get 2*(40%)*(40%)=2*(16%) = 32% of the seats and thus the majority one should get 68% of the seats.
• 3. “Cubic” rule. Seat-share should follow a more complicated cubic function of statewide vote-share, one that fits the historical pattern well. In particular, the ratio of seat shares (one party’s seat-share divided by other party’s seat-share) should be the cube of their ratio of vote-shares.
• 4. “Jurisdictional” rule. A party should win seats in proportion to the population in jurisdictions (such as counties and cities) in which that party won more votes than any other party.

[More about the four notions of fairness (.pdf)]