This wasn’t supposed to be a good Senate year for Democrats. They’re competing in states Trump won, sometimes by double digits, and Maine looked like their best pickup opportunity. But if you believe the RealClearPolitics averages, we’re going to see a blue Senate. Democrats currently lead by 0.7 points in Iowa and Maine, 1.6 in Minnesota, and 1.8 in Alaska, while Kansas Republican Roger Marshall leads by only one. Recent polling is even worse for Republicans: Marist has Troy Jackson up four in Maine and Josh Turek up eight in Iowa, while a New York Times/Siena poll has Mary Peltola up seven in Alaska and Kansas tied.
My hypothesis is that polling is simply getting worse. It’s harder to reach people, fewer people are willing to participate, and the public averages are increasingly flooded with internal polling or something that looks like it. An interest group aligned with a candidate releases a poll, it enters the bloodstream, and it shifts the aggregate. At the same time, reputable pollsters may be overcorrecting based on whom they missed in the previous presidential election.
I took five of the most competitive Senate races from 2014, 2018, and 2022 and compared the final RealClearPolitics averages with the actual results. In 2014, the average error across those races was 2.31 points. North Carolina polling had Democrat Kay Hagan ahead of Thom Tillis by 0.7, but Tillis won by roughly 1.5. The biggest miss came in Iowa, where Joni Ernst was expected to win by 2.3 and instead won by six.
The misses got worse in 2018, when the average error reached about 4.5 points. Polling had Rick Scott losing in Florida, Martha McSally beating Kyrsten Sinema in Arizona, Claire McCaskill and Josh Hawley essentially tied in Missouri, and a Democrat winning in Indiana. Hawley and the Indiana Republican both won by nearly six points, while Democrat Jacky Rosen won Nevada by five. When those polls missed, they missed horrendously.
In 2022, the average error across the five races was 3.92 points. Polling had Pennsylvania essentially tied, with Mehmet Oz slightly ahead, before John Fetterman won by about five. It had Herschel Walker beating Raphael Warnock, who ultimately won the race, and Adam Laxalt leading Catherine Cortez Masto by 3.5 points in Nevada before Cortez Masto won by nearly one. These weren’t all misses in the same partisan direction.
The polling right now unquestionably shows a blue wave. To believe otherwise, you have to convince yourself that unreachable rural Republicans will come home even though they don’t want to talk to pollsters. I’m skeptical of rosy Democratic numbers in deep-red states like Iowa and Kansas, but in Ohio, Michigan, and perhaps Pennsylvania, the polls may actually be undercounting Democratic support. A three- or four-point miss is entirely plausible and would change many of these races. The problem for anyone relying on that possibility is that there’s no guarantee which way the miss goes.
Chapters
00:00:00 - Intro
00:02:48 - Polling
00:15:11 - Update
00:16:00 - Trump Ads
00:18:28 - Midterm Spending
00:20:31 - Supreme Court Climate Change
00:22:57 - Interview with JD Durkin
00:57:45 - Wrap-up
