Skip to main content
ABC News
Special Elections Foreshadowed A Blue Wave In 2018. What Are They Saying Now?

Special elections in 2017 were one of the first signs that Democrats were going to have a really good 2018. The party’s candidates were consistently outperforming the baseline partisanship of their districts in both federal and state-level special elections, something that tends to foreshadow the national political environment in the next congressional election. But the signal from special elections so far in 2019 is a lot weaker.

So far in the 2020 cycle, there haven’t been any special congressional elections like the high-profile 2017 races for the Georgia 6th District and a U.S. Senate seat in Alabama.1 But as of April 15, there have been dozens of special state-legislative elections across 15 states, including 27 elections that we think can be used to measure which party is performing better than we might expect based on partisanship alone. (This includes both traditional Republican-versus-Democrat races and all-party primaries but excludes nonpartisan elections and uncontested races. The exact criteria for determining which elections made the cut are in the footnote.2)

It’s not enough to look only at which party won these special elections — it’s also important to account for how red or blue a district normally is to understand whether a swing is happening. To do this, we compared the vote share margin in each special election with the district’s FiveThirtyEight partisan lean.3

The difference, or “margin swing,” tells us how much more Democratic or Republican the district was in the special election than it “should” have been in a neutral political environment. Historically, an average of these margin swings has proved predictive of the national mood in the next regularly scheduled federal election. But as you can see in the table below, special election results in 2019 have been all over the place, making it difficult to ascertain a clear pattern of overperformance for one party or the other.

Lots of noise in special-election results so far

How much special state-legislative elections in the 2020 cycle have swung toward Democrats or Republicans, based on the seat’s partisan lean and the special-election vote margin

Date Seat Partisan Lean Special Margin Margin Swing
April 2 Pennsylvania SD-37 R+11 D+4 D+15
April 2 Maine HD-52 D+25 D+33 D+8
March 30 Louisiana HD-18 R+14 D+37 D+51
March 26 South Carolina SD-06 R+37 R+11 D+25
March 26 California SD-33* D+58 D+57 R+1
March 26 California SD-01* R+21 R+39 R+18
March 19 Minnesota HD-11B R+23 R+37 R+14
March 19 Iowa SD-30 D+3 D+15 D+12
March 12 Texas HD-125 D+19 D+17 R+3
March 12 Tennessee SD-32 R+45 R+68 R+23
March 12 Pennsylvania HD-114 R+2 D+25 D+27
March 12 Maine HD-124 D+17 D+30 D+13
March 5 Kentucky SD-31 R+53 R+5 D+47
Feb. 26 Connecticut HD-99 R+5 R+8 R+3
Feb. 26 Connecticut SD-06 D+22 R+6 R+29
Feb. 26 Connecticut SD-05 D+25 D+29 D+4
Feb. 26 Connecticut SD-03 D+19 D+18 R+2
Feb. 23 Louisiana HD-27 R+68 R+84 R+16
Feb. 19 Virginia HD-86 D+26 D+25 R+1
Feb. 12 Georgia HD-176* R+44 R+71 R+27
Feb. 5 Minnesota SD-11 R+7 R+6 D+0
Jan. 29 Texas HD-145* D+27 D+49 D+22
Jan. 29 Texas HD-79* D+32 D+60 D+28
Jan. 8 Virginia SD-33 D+23 D+40 D+17
Jan. 8 Georgia HD-05* R+66 R+91 R+24
Dec. 18 Virginia HD-24 R+35 R+19 D+16
Dec. 11 Texas SD-06* D+36 D+54 D+18
Average D+5

Results are as of April 9. In some cases, results are unofficial.

* Election featured multiple candidates of the same party; the vote margin is calculated using the total vote share of all Democratic and all Republican candidates.

“Partisan lean” is how much more Republican- or Democratic-leaning a district is than the country as a whole, based on a weighted average of 2016 presidential results and 2012 presidential results, with an adjustment for state-legislative results.

Sources: Ballotpedia, Daily Kos Elections, state election results

Democrats have had some massive overperformances, such as in South Carolina Senate District 6, where their candidate lost by only 11 points in an R+37 district. But Republicans have had some big-time successes as well, like when they won Connecticut Senate District 6, a D+22 seat, by 6 percentage points. Still other results barely deviated from the district’s partisan lean.4

Out of the 27 special elections in the table above, Democrats outperformed partisan lean in 15 of them, while Republicans outperformed it in 12. Compare that with the much more lopsided ratio in special-election performance through this point in the 2018 election cycle (April 15, 2017): Democrats had overperformed in nine out of 11 races.5 Similarly, the average margin swing at this point last cycle was much more Democratic-leaning than it is this cycle. So far, this cycle’s special elections have had an average 5-point margin swing toward Democrats relative to partisan lean, but at this point in 2017, Democrats were overperforming by 11 points on average.

This might suggest that 2020 will be a slightly Democratic-leaning year — just maybe not on the same level as 2018. Then again, with 2019’s margin swings so spread out (from +29 for Republicans to +51 for Democrats), I’m not sure it tells us much yet.

Other early indicators of the national political environment are similarly inconclusive. Democrats lead most early polls of the generic congressional ballot by single digits, but almost all of those polls have been conducted by one pollster, HarrisX, which FiveThirtyEight gave a C+ pollster rating in 2018. So if their numbers are off, our whole picture might be biased. In 2018, Republican retirements from Congress — and robust candidate recruiting by Democrats — were also early signs of Democratic strength, but it’s too early to draw any conclusions from those factors for the 2020 cycle. Only two Republicans and two Democrats have announced that they are retiring from Congress and not running for higher office, and both parties have landed solid recruits in other races that are expected to be closely fought.

Taking everything into account, 2020 looks a little better for Democrats than for Republicans right now, but we probably shouldn’t read too much into so little data. We’ll just have to wait until we’re further along in the cycle.

Midterms 2018: FiveThirtyEight reacts to election results


  1. We’re treating the 2020 cycle as having begun on Dec. 9, 2018, which was the day after the last 2018 midterm election runoff took place. The first congressional special general election is scheduled for May 21 in the Pennsylvania 12th District.

  2. To be considered in my analysis, an election must have featured at least one Republican and at least one Democrat. We further excluded elections in which any single minor-party candidate got at least 10 percent of the vote, because the candidate might have drawn disproportionately from another party, skewing our numbers. In California, Georgia, Louisiana and Texas, special elections consist of an all-party primary followed, when necessary, by a runoff. We used the runoff results if the runoff was between one Democrat and one Republican, but if the runoff was a one-party affair (or simply wasn’t necessary), we used the primary results. In these cases, the Democratic performance is the combined vote share of all the Democratic candidates in the race, and the Republican performance is the combined vote share of all the Republicans.

  3. How much more Republican- or Democratic-leaning the district is than the country as a whole, based on a weighted average of 2016 presidential results and 2012 presidential results, with an adjustment for state-legislative results. Technically, these partisan leans are from the 2018 cycle; we haven’t calculated FiveThirtyEight partisan lean for 2020 yet. Presidential results by legislative district are from Daily Kos Elections. Note that Daily Kos has not calculated 2012 presidential results for Georgia House District 176 because of mid-decade redistricting, so its partisan lean is based solely on 2016 results and the legislative adjustment.

  4. It wasn’t a special election, but probably the highest-profile state election so far this year was the April 2 race for Wisconsin Supreme Court. The conservative candidate beat out the liberal candidate by half a percentage point — a close match for Wisconsin’s R+1 partisan lean.

  5. This count is culled from a total of 17 special elections that had taken place through April 15, 2017. In addition to four races that were excluded because they did not feature candidates of both parties, we excluded two races fought under unusual circumstances: Louisiana House District 92, where the Democrat dropped out at the last minute but his name remained on the ballot, and Pennsylvania House District 197, where the Democrat was not on the ballot but waged a write-in campaign instead.

Nathaniel Rakich is a senior editor and senior elections analyst at FiveThirtyEight.