Skip to main content
How Much Was Incumbency Worth In 2018?

For decades, running as an incumbent was undoubtedly a huge advantage in electoral politics. As recently as 20 years ago, holding office added an average of 8 percentage points to a candidate’s margin. But in this century, experts say, the incumbency advantage has significantly diminished. Now the verdict is in for the 2018 election: According to our method of calculating it (which is different from other researchers’, so keep in mind that these numbers can’t be compared directly to those from previous years), the electoral benefit of already being a member of Congress this year was down to less than 3 points.

We’ve already shown how partisanship, adjusted for the national popular vote, explained the 2018 election results pretty well in the Senate and very well in the House. But to better account for district-by-district variations, we wanted to quantify the one major thing that varies from district to district: the candidates.

A very basic model for predicting congressional election results might look something like:

Election Result = Partisan Lean + (Popular Vote x Elasticity) + Net Incumbency Advantage

Where “Election Result” is the incumbent’s margin of victory or defeat, “Partisan Lean” is how much more Republican- or Democratic-leaning a state or district is than the country as a whole,1 “Popular Vote” is the national popular vote for the House of Representatives and “Elasticity” is a measure of how sensitive the state or district is to changes in the national mood.2 The last variable, “Net Incumbency Advantage,” is a rough proxy for things like a candidate’s experience, personal appeal and fundraising, and it shows how many percentage points stronger the incumbent is than the challenger at the ballot box.3

Before an election, we really only know what partisan lean and elasticity look like,4 although we can guess at what the popular vote might be thanks to the the generic congressional ballot polling average. After an election, we know the popular vote and the election result, so we can roughly calculate what the net incumbency advantage might have been. For example, Democrats won the national popular vote by 8.3 percentage points,5 but Republican Rep. Chris Collins of the New York 27th District won his election by 0.6 points in a district with a partisan lean of R+22.9 and an elasticity score of 1.10. That means Collins performed about 13.2 points worse than we would have expected based on partisanship and the popular vote alone — a net incumbency disadvantage.6 Given that Collins was indicted on insider trading charges earlier this year and Republicans initially tried to replace him on the ballot, that dramatic negative number makes sense.

We did these calculations for every contested Senate and House election with an incumbent in 2018.7 In the House, net incumbency advantage ranged from +28.4 points for Democratic Rep. Dan Lipinski (who ran against a Nazi sympathizer) to -13.2 points for Collins. In the Senate, it spanned +25.2 points (for Democratic Sen. Joe Manchin in West Virginia) to -14.6 points (for Democratic Sen. Elizabeth Warren in Massachusetts, a potential red flag for her rumored 2020 presidential campaign). The average U.S. representative had a net incumbency advantage of 2.7 points; U.S. senators got an average boost of 2.6 points over their challengers.

What the incumbency advantage was worth in 2018

Average net incumbency advantage in the 2018 midterms, by type of incumbent

House incumbents Net Incumbency Advantage
All representatives +2.7
1- to 2-term +2.8
3- to 4-term +2.5
5- to 9-term +2.8
10+-term +2.8
Senate incumbents Net Incumbency Advantage
All senators +2.6
In small states +1.9
In big states -1.3
Appointed +0.0
1-term +2.0
2-term +5.2
3-term -2.3

Small states are the eight states that have one or two congressional districts and had a Senate race this year. Big states are the eight states that have 12 or more congressional districts and had a Senate race this year.

Sources: ABC News, Associated Press, New Jersey Department of State

If you slice and dice the data by different kinds of incumbents, you start to see small differences in the net incumbency advantage. For example, senators from small states8 enjoyed a larger net incumbency advantage (1.9 points on average) than senators from big states9 (-1.3 points).

While you might expect that the incumbency advantage would be larger for a politician who has been in Congress for longer, in 2018, that seemed to be true only in the Senate, and the correlation wasn’t very strong. Dovetailing with existing research that says being an appointed incumbent isn’t nearly as powerful as being an elected one, the two appointed senators on the 2018 ballot (Republican Cindy Hyde-Smith of Mississippi10 and Democrat Tina Smith of Minnesota) both had a net incumbency advantage of just about 0 points. One-term senators had an average net incumbency advantage of 2.0 points; that increased to 5.2 points for two-term senators. Interestingly, however, three-term senators (of whom there were only four on the ballot this year) underperformed their challengers by an average of 2.3 points. Meanwhile, in the House, there was no clear relationship between the number of terms an incumbent had served and his or her net incumbency advantage.


  1. In FiveThirtyEight’s partisan lean metric, 2016 presidential election results are weighted 50 percent, 2012 presidential election results are weighted 25 percent and results from elections for the state legislature are weighted 25 percent.

  2. FiveThirtyEight’s elasticity scores are coefficients where 1 is average. They represent how many percentage points the state or district would be expected to shift if the national popular vote shifted by 1 point.

  3. In open-seat elections, this isn’t net incumbency advantage per se, but rather the difference in candidate quality between the two major-party candidates.

  4. And even those can be off if an election turns out to be a realignment, or even a recalibration like 2016 may have been.

  5. The election results we used for this calculation and all the calculations below are up-to-date as of Dec. 3 at 11:30 a.m. Eastern.

  6. Our formula is 0.6 = 22.9 + (-8.3 x 1.10) + x, where 0.6 is the election result, 22.9 is the partisan lean, -8.3 is the popular vote and 1.10 is the elasticity. The reason the popular vote is negative is that this year’s popular vote favored Democrats and Collins is a Republican, so the popular vote was working against him.

  7. We excluded open-seat elections and elections that did not feature both a Republican and Democratic candidate. We also threw out the Louisiana 3rd District, where incumbent Rep. Clay Higgins faced both Democratic and Republican opposition thanks to Louisiana’s unusual jungle-primary system, and the Pennsylvania 17th District, where both candidates were incumbents thanks to court-ordered redistricting. Finally, we also did not calculate the net incumbency advantage for independent candidates.

  8. Which we’re defining as the eight states that have one or two congressional districts and had a Senate race this year.

  9. The eight states with 12 or more congressional districts that also had a Senate race this year.

  10. We used the results from the Nov. 27 runoff, not the Nov. 6 jungle primary.

Nathaniel Rakich is FiveThirtyEight’s elections analyst.