Predicting election "waves"
Updated: Jun 7, 2022
In a recent Washington Post column, Henry Olsen writes that it is too late for congressional Democrats to avoid big seat losses in November. It’s going to be a “wave” election for Republicans, he argues, adding that “recent political history shows that the course of a fall election is almost always set by Memorial Day” (a point also made by Sean Trende of Real Clear Politics). Is the Democrats’ electoral fate really sealed?
Olsen is correct that, as of right now, the traditional predictors of an election outcome indicate trouble for Democrats. As Seth Masket wrote here back in May, presidential popularity, the state of the economy, and the size of the Democratic majority in the House suggest a substantial gain in seats for Republicans.
But there are important caveats. For one thing, Masket notes that how the economy is measured is very important. Income growth suggests a big loss for Democrats, but economic growth suggests a small one.
It’s also possible for those variables to change in value. Predictive models for congressional elections usually draw from data taken on or around Labor Day. Presidential popularity and the economy could shift significantly over the next three months.
Finally, even if it’s a safe bet that Republicans will win seats in Congress, it’s far from clear that a “wave” is building for the GOP. Leaving aside the chronic misuse of the term “wave” (which I have written about before), electoral trends can change considerably between June and November.
One can see this by comparing the number of seats held by each party that are considered vulnerable (i.e. either toss-up seats or ones leaning or likely to flip to the other party) and looking at how that has changed over time. Using Cook Political Report data on individual congressional races since 2013, the charts pictured here show these data every three months, followed by what actually happened in November. (It’s an update from data in my October 2020 post on that year’s election.)
In no case did a political party that was expected by June to lose congressional seats in November actually gain seats. And a couple of elections did indeed fit Olsen’s profile of a building wave, namely the 2014 Senate elections (the blue line on the second chart) and the 2018 House elections (the grey line on the first chart).
Most electoral environments did not fit that profile, however. Some showed no real change over time (the 2018 Senate elections) or a more gradual trend that led to relatively small seat gains (the 2014 House elections). And several showed reversals: electoral conditions trended favorably for one party by mid-year, but that party ultimately did worse than anticipated. (In the 2016 and 2020 elections for both House and Senate, shown by the red and green lines, Democrats' prospects looked increasingly promising by June, but the party won fewer seats (or lost more) in November.)
To be sure, the electoral momentum right now is with Republicans. The yellow line, which shows the data for this year's House and Senate races, is clearly moving in the wrong direction for congressional Democrats. So November could indeed turn into a big win for GOP candidates.
But if history is any guide, it would wrong to assume that the trend towards Republicans will necessarily continue. The GOP may be a good bet to gain seats in Congress, but it’s far too soon to say how many, and especially if it will be a Republican wave election in either chamber.