Voices & Choices

What FiveThirtyEight's projections tell us about US House structural dysfunction

What FiveThirtyEight's projections tell us about US House structural dysfunction

FiveThirtyEight recently released its long awaited congressional projections, creating a projection model with an interesting visual style and fascinating interactive features that mostly reinforced the conventional wisdom that Democrats have about a 3-in-4 chance of taking the US House this November.

In building this model, FiveThirtyEight founder Nate Silver incorporated a wide array of weighted polls, projections extrapolated from polls of demographically similar districts, and expert handicaps from Cook Political Report, Inside Elections and Larry Sabato’s Crystal Ball. Unsurprisingly, with this vast trove of data, sophisticated model, and the expert opinions of much of the election analysis community, you can come up with remarkably accurate predictions. What is surprising is that as Silver admits in one of his tables, you can do about as well as his model in predicting the results using only past election results and public fundraising disclosures.

chart from 538 blog

FairVote has also projected the results of congressional elections in our monopoly politics series  since 1992 using only past election results, and has achieved remarkable accuracy in doing so, typically around the 95 percent of Silver’s fundamentals only model.

Americans generally want our elections to be competitive and unpredictable, this is part of the appeal of the much-derided horserace framing, but it's also why more people show up to vote when they think elections are going to be close. So why are 95 percent of congressional election outcomes predictable with just fundamentals and the generic ballot?

Gerrymandering, campaign finance, and extreme polarization all play important roles, but a severe constraint on the level of competition most voters experience is the single winner district itself. Under the Fair Representation Act (FRA), roughly 56 percent of the population would live in districts with a competitive U.S. House seat, compared to 16 percent estimated by FiveThirtyEight under the current map in an earlier article. Under their current methodology, FiveThirtyEight views only 14 seats (about 3 percent of seats) as true tossups. With a larger and more balanced battleground, the exact balance of the next Congress would be more interesting and difficult to predict. At the same time, the FRA would almost entirely eliminate the structural bias against Democrats in the House, meaning that any flip in the national popular vote would likely lead to the opposition taking control. Not just the flips of 7 to 11 points towards Democrats that most election analysts think Democrats need under the current system.

To read more information on this proposal, read the Fair Representation Act report, or explore our sample map click here.

For more information on FairVote’s Monopoly Politics series, visit our new interactive page.


Photo illustration by Mikhaila Markham

Join Us Today to Help Create a More Perfect Union