The leftward shifts in special elections are happening in higher turnout races, too

So far this year, there have been five special elections to the House of Representatives. Republicans have won three of those five races — not surprising, since they also won those three seats in last year’s general election.

What is surprising is that, in each of those five races, the Democratic candidate saw double-digit improvement on the final margin over both last year’s Democratic candidate and how Kamala Harris did in the district. On average, the races shifted 16 points to the Democratic House candidate and 17 points relative to Harris.

Those presidential results come from the essential site The Downballot, which has also tracked special election outcomes. Across those contests in 2025 (meaning, including state-level races), the average shift to the left has been about the same: 13 points.

The most recent contest, as you are probably aware, was in Tennessee’s 7th District on Tuesday. Again, the Republican Party held the seat, but much more narrowly than it or Donald Trump won it last year.

Final vote totals aren’t in yet, with the New York Times estimating that about 95 percent of the vote is in (as of the time I compiled this data). But it’s already clear that turnout in the Tennessee special was relatively high for a House special this year — and the Democrat still improved on last year’s margins by 13 percentage points.

It’s important to reinforce that there aren’t a lot of lessons to be drawn from a set of five elections. Everything that’s available reinforces the idea that Democrats will do well in next year’s midterms, certainly, but we should be cautious about using five unusual contests to assume that Democrats will overperform the GOP by 10 points next November.

But this is also why considerations of turnout are useful. One of the reasons that special elections are an imperfect predictor of future outcomes, of course, is that they usually involve fewer voters. If an election instead includes most or all of the vote total for a November contest, we might assume it’s a better indicator of how that district is likely to vote. There are a lot of squishy words there — all or most, better, etc. — but the idea holds.

In each of the five special House elections this year, the Democratic candidate received fewer votes than the candidate in last year’s general, just as the Republican candidate received fewer votes than last year’s Republican. But in each case, the 2025 Democratic candidates captured a larger share of their party’s 2024 House vote totals than Republican candidates did of theirs. On average, Democratic special election candidates got 52 percent of the 2024 Democratic candidate; the Republican special election candidates got 37 percent of the 2024 Republican.

In Tennessee, Democrat Aftyn Behn got 7 votes for every 10 the 2024 candidate for that seat got. Her opponent, Matt Van Epps, got about 5 votes for every 10 the Republican got last year.

There are a lot of reasons such a discrepancy might exist. Perhaps a lot of people who voted for the Republican last year backed the Democrat this year. Perhaps Democratic voters were more energized to turn out to vote. Perhaps it’s a function of the difference between last year’s specific candidates and this year’s.

Whatever the reason, the dual pattern — less drop-off for the Democrat and a shift to the left in the overall margin — are working against Trump’s party. Even in a race where Republicans managed to turn out some 100,000 voters for a special House election (as will likely be the case in Tennessee), those patterns hold true.

Comparing House races to House races seems like a better point of comparison than contrasting the House special elections with the 2024 presidential contest. But, if you’re curious, here’s that comparison.

In this case, the shift to the left in Tennessee was more modest than in the other four special elections. Republicans investing heavily in the idea that this means that higher relative turnout in 2026 will further erode the leftward shift in House races might scroll up a bit and notice that in Arizona’s 7th, where turnout was relatively modest compared to last year’s House race, the shift to the Democrats was about the same as in the relatively higher turnout in Tennessee.

We risk overreading all of this, of course. This is what happens, though: in the absence of concrete information about next year’s midterms, an election of enormous significance, we dig through whatever tea leaves litter the bottoms of our cups.

Above, you can see five leaves in two different cups. Have fun with your forecasting.

Photo: Tracking the results of the 1954 election. (National Archives)