![]() |
| Sen. Susan Collins (R-ME) |
The Maine 2020 Senate race seems to have done this to a generation of Democrats, and there's a reason for that. Headed into Election Night 2020, the Democrats had every reason to assume the best. Joe Biden had been leading in the polls in most swing states, and the Democrats appeared posed to win not just the White House, but also to win back the US Senate after six years in the wilderness, and make gains headed into the House. It looked like Biden would end the long national nightmare that Trump had wielded over the Democrats for four years. And to a degree, that happened. Biden did win an impressive popular vote victory, though his slim margins in key swing states like Georgia, Arizona, & Wisconsin meant a second consecutive Trump term was a closer call than we had realized. But the congressional victories, while ultimately successful in getting Dems those majorities, came with a lot of headaches, with Republicans doing unexpectedly well down-ballot, no more so than in Maine. Headed into the night, Real Clear Politics had State House Speaker Sara Gideon at a +6.5 margin spread in aggregate polling, while 270 to Win had her at +3.8. It seemed probable that Sen. Susan Collins' vote to confirm Brett Kavanaugh was going to cost her greatly, and Gideon would become the first Democrat elected to the US Senate since 1988, but that didn't happen. Collins win margin of over 8-points was extra salt in the wound. Never mind that it was due in part to Maine's use of Ranked-Choice Voting, and Gideon probably would've gotten closer to 47-48% of the vote in the end had Collins not squeaked over 50% of the vote: it made Collins look invincible, and it made the polling look like a joke.
I'm seeing a lot of talk online, both in articles about the midterm elections in five weeks and on social media, asking the question of whether or not the Democrats are headed toward another bout of polling that could point to disaster like it did for them in Maine in 2020. This is partially because Democrats, as a species, are allergic to good news. Show me a Democrat who isn't panicking before an election, and I'll show you someone who is either on a great prescription drug or is secretly a Republican; Democrats always panic until the results are called, while Republicans can march over-confident to their doom if that's what it takes. A big reason for this is the scope of a 2026 win if you look at aggregate polling. If aggregate polling across platforms like 270 to Win, Decision Desk HQ, & Silver Bulletin were exactly right at this exact moment (I said exact twice for a reason...these metrics shift pretty quickly toward the end of election season so I don't want to be called out in the comments if suddenly a new poll changes one of these races narratives!) the Democrats would pickup six Senate seats in Alaska, Iowa, Maine, North Carolina, Ohio, & Texas, and would (depending on which site you go with) also be looking at pickups in Kansas & Nebraska. They also wouldn't be at risk of losing any of the races in Georgia, Minnesota, New Hampshire, or Michigan that are oftentimes listed as a potential risk, and the Republicans would be looking at their closest margin in South Carolina in decades. A 6-8 seat pickup for the Democrats, mainly in states that Donald Trump won in 2024 by double-digits is a bit more good news than the Democrats can believe without actual results.
So I thought it'd be worth looking at the aggregate polling sites for the past three cycles in the Senate to look at how many called the victor wrong. And when I say wrong, I mean the eventual winner, not the margin; there are a lot of situations like South Carolina in 2020 where aggregate polling showed Lindsey Graham in a much more competitive race than he actually was...but still showed him winning. The races I want to look at are specifically ones that the aggregate was truly wrong, that they called a winner that never was.
In that case, there's a limited list of options for the past three cycles. Susan Collins in 2020 is obviously one of them, but she's joined by only four other races: the aggregate polling missed the wins of Thom Tillis (R-NC) in 2020, as well as John Fetterman (D-PA) & Catherine Cortez Masto (D-NV) in 2022, and Dan McCormick (R-PA) in 2024. It's worth noting that not all aggregate polling sites weigh and factor the same polls, so both Arizona in 2022 (Democrat Mark Kelly) and Ohio in 2024 (Republican Bernie Moreno) both had some sites saying they were headed to defeat, while others thought polling showed they'd lose narrowly. Overall, 5-7 seats in total, no more than 3 in a night, and no less than two.
Extrapolating this out, if we assume that the polling aggregates don't change between now and November 3rd (a very big "if" given the number of undecideds in some polls, especially in South Carolina, Kansas & Nebraska), the Democrats would need a gargantuan polling miss, much worse than any of the last three cycles, to not win the majority. Even if three of the eight seats that currently look like flips were to go red...the Democrats would still be at 52, a healthy majority (one no one could've guessed with a straight face even on Memorial Day, much less last year).
It's also not out of the question that it's the Democrats who are under-polled here. Look at the above lists. All of them are uniform in each cycle of who got missed (2020 it was two Republicans, 2022 it was three Democrats, and in 2024 it was two Republicans). In all cases, the party that we didn't realize had quite as much strength as it did on election night ended up getting the gain, and that's not the same as who won the night (i.e. in 2020 the Republicans were stronger than realized on Election Night...even on a night that the Democrats ultimately won a trifecta). We will not know until election night overall which party was stronger than we realized (see 2022 for a case of the incumbent party doing better than expected, 2014 for a night where they did worse). It is assumed, given how sunny this polling is, that it's the Democrats that are being over-polled (also because getting the polls wrong when the Democrats end up winning more seems to be a sin that the national media is far more willing to forgive than the reverse). But with rising gas prices, Trump's gargantuan disapproval ratings, and the administration's inability to stick to a narrative (yesterday scaring countless communities reliant on higher education money by upending the rules on student loan approval), it's possible that the Democrats are the ones who might be under-polled, in which case those mixed bag aggregate states like Kansas & Nebraska (or even South Carolina) could be where the surprise win happens, and in that case it might involve a blue checkmark. We won't know until November 3rd, but right now, polling is pretty clear-the Democrats are the favorites to win back the US Senate in five weeks.









