How accurate are American elections?


At a time of growing frustration about gerrymandering and how our elections fail to represent voters of all stripes, the Houses of Power data visualization seeks to help answer a crucial question: How accurate are American elections?

In the main “50 state grid” you will see the hallmarks of gerrymandering throughout many states: an overall party preference in the state (the upper right of each state’s box) that is significantly different than the party control of the state’s legislature or US Congressional delegation (the bottom row of each state’s box). But you may notice a wealth of other inaccuracies that go underreported: For example, 40% of Rhode Island votes Republican, but Republicans hold only about 10% of the seats in their state senate and 15% in the state house. Similarly, 30% of Hawaii votes Republican, but fully 0% of their state senate is Republican. And in Utah, a state always cited for being “deep red,” 1 in every 3 voters prefer Democrats — but Democrats hold only 17% of seats in the state legislature. The list goes on.

These inaccuracies aren’t always due to intentional gerrymandering, as voters with similar ideologies often live close together and naturally “sort” themselves into districts that create lopsided representation. And it’s also important to remember that just because legislative representation may look “accurate” doesn’t mean that gerrymandering hasn’t been happening (consider “sweetheart gerrymandering,” where incumbents of both parties work together to protect their own seats).

It’s important to note that viewers should take the percentage numbers offered with a grain of salt — the simplification required to create this type of data visualization has its limitations (see Methodology, below). The project does not mean to suggest that all representatives of one party are interchangeable, or that third parties or independent voters are unimportant. In fact, the opposite:

Overall, the Houses of Power visualization underscores what voters have long felt: In many ways, our elections don’t work well for Americans of either major party — and perhaps especially for voters who prefer neither party.

In this sense, Houses of Power is also designed to help us question whether our single-member, winner-take-all elections are the best way to decide who represents us. Much recent discussion (on both sides of the aisle) has explored how we could make practical changes to our election systems that would dramatically improve the accuracy of ideological representation — all while retaining district-based representation and without switching to a full-fledged proportional system.

Finally, Houses of Power is designed to remind us of the importance of state legislatures. As astute observers often note, state legislatures wield enormous power over most aspects of our daily lives. This holds true even with so much focus on national politics in 2018 — and it can reasonably be said that the dysfunction now so common in Congress means that the impact of state legislatures on our lives is even greater than before. And, perhaps most importantly, it is often state legislatures that draw (or gerrymander) the voting district maps for themselves and Congress.


American elections are not simple, but designing a practical visualization about the accuracy of our elections requires some simplification: While third parties and independents enjoy marginal support in the United States and the percentage of Americans who decline to associate with either party is trending upwards, our system is still dominated by the Democratic and Republican parties. Given this reality and the need to make this visualization accessible, Houses of Power focuses on displaying the balance of control between these two parties. The voting records and other actions of independent and nonpartisan state legislators were examined in order to determine if they were functionally part of either party’s voting bloc. If so, these representatives are colored in lighter shades of red or blue, and representatives who appear significantly independent are colored light purple. The few Libertarian party representatives are colored orange, and Vermont’s Progressive Party is colored dark blue.

Finally, while there are many ways to determine the overall party preference of a state’s electorate (shown in the top right corner of each state’s box), we reach a rough approximation of this preference by using a weighted average of the two-party vote share in each state's 2016 (weighted at 75%) and 2012 (weighted at 25%) Presidential election.

Data found at Ballotpedia and the National Conference of State Legislatures were consulted to build the dataset underlying Houses of Power, and these sites were indispensable resources.


Released in October 2018, the Houses of Power project is in beta mode — if you have a suggestion for improving it or see an inaccuracy, please send an email to.

The project will be updated after the November 2018 elections to reflect all changes.


Houses of Power is a project by David Kaplan. David is a data analyst in the NBA and, on the side, enjoys creating projects to help users visualize issues in today's society. Email David at.