After a turbulent beginning to Trump’s second term last year, 2026 has proven even more chaotic.
The year began with unrest in Minnesota, including the deaths of Renée Good and Alex Pretti in January at the hands of ICE—both garnering national attention and rare cross-party criticism. Protests, larger and more widespread than in the first Trump administration, gained momentum. With an unwanted, unexplained war in Iran, gas and diesel prices skyrocketed. Consumer sentiment has reached all-time lows since the University of Michigan started tracking it in 1960.
Likewise, Donald Trump’s approval has plummeted to its lowest level in either term, but a major question remains: Has voter behavior changed? How are Democrats faring in real elections in 2026 compared to 2018, when they made historic gains in Trump’s first midterms?
Special elections are often cited as proof of which way the electoral winds are blowing, and applying statistics makes it possible to reveal insights never previously reported.
Using a robust new framework, we previously identified seven distinct “periods” of special elections over the past decade—including the highest-performing period on record for Democrats, which went virtually unnoticed at the time.
Not only can we quantify the impact of events like the Supreme Court’s decision to overturn Roe v. Wade, but we can also see it in real time, providing extraordinary insight into whatever may be newly motivating voters.
Now we can go much further. By making adjustments to account for changes in presidential administrations and variations between states, we can compare all special elections across the entire previous decade.
And once we have, we can see that these special elections point to a political environment for Democrats heading into the midterms that is at least as good as it was in 2018—if not better.
Unlocking the potential of special elections
We often see cherry-picking of special elections to fit a certain narrative, whether doom and gloom or unfounded overconfidence. By holistically looking at all elections together in relation to one another, we avoid two common kinds of errors: false alarms, where we think something’s happening that isn’t, and missed signals, where we fail to notice something happening before our eyes.
Selectively highlighting strong Democratic performances while ignoring weak ones can lead to a false sense of an unusually good environment. On the other hand, the single best period of Democratic performance over the last decade largely passed without comment.
Our holistic approach—taking into account all special elections—ensures we aren’t cherry-picking. Still, though, there are other problems in the data itself that weaken the signal and render it misleading. We therefore have applied two corrections to create the cleanest possible dataset, giving ourselves the best chance of finding real patterns.
Adjusting for shifting presidential administrations
The Downballot has long compared special election results to the most recent presidential performance in the same districts. If Kamala Harris won a district by 2 points in 2024, and later a Democrat won a special election in the same district by 12, then we can say the candidate overperformed by 10 points.
This works well if we’re only looking at elections under a single presidential administration. But if we want to look across administrations—which we need to do to compare the environment going into the 2026 midterms with that in 2018—then we need to adjust our data accordingly. Here’s how we do that.
You can unlock this complete post by taking advantage of The Downballot’s special 23rd anniversary sale this week! To mark the occasion, we’re offering 23% off our regular rates, so you’ll pay just $7 a month or $58 a year to access our premium content and perks, and support our unique work as a small independent media outlet.