I get it. We got burned in 2016, and then (very nearly) in 2020 by taking the polls at face value. The fundamentals (we are told) are against us, and the president’s party almost always loses seats (at least in the House) in the midterm. In the senate, of course, the most recent counterexample is… 4 years ago.
But I digress. You can’t swing a cat without hitting someone claiming that the “momentum” has shifted and now the Democrats are doomed. Doooomed!! But are we?
To some degree, I think the press is being played by rightwing pollsters. An overwhelming number of the recent senate polls in 538 are dominated by:
- Trafalgar — a republican pollster working for places like the Daily Wire, and whose polls in 2018, 2020, and in the off-cycle polls have been off by about 2% in favor of the Republicans. Yes, even in the 2020 election.
- InsiderAdvantage — Less of a track record in the final weeks of elections, but is the pollster of choice of “American Greatness” (not surprisingly, a rightwing outlet).
- Cygnal — Clients include the conservative, “Iowans for Tax Releif” and “The John Lock Foundation.”
Every time they post (which is practically every day at this point), they skew the averages rightward. And so, the narrative goes, Dems are blowing it. But what happens when we look at how pollsters have trended from August or September until now?
I’ve downloaded all of the governor and senate polls from the 538 database, excluding straight-up internal polls and looked at every pollster who has put out a margin in October, and also some time August 1 or later. And asked how they trend. The complete spreadsheet is here, but by and large, within any given pollster (especially these high output pollsters), there’s very little movement over the last couple of months.
Here are the most competitive races (all numbers are D-R margins):
-
Arizona
- Trafalgar went from 1.2 to 1.0 in the last month
- InsiderAdvantage went from 6 to 4.4
- OnMessage (by far the biggest movement) went from 10 to 3
- CenterStreet went from 20 to 17. Almost certainly an outlier.
Verdict: Probably some actual movement (Trafalgar and InsiderAdvantage, the two non-outliers, average to about a point toward the R), but others like OH Predictive Insights and CNN have the race currently at +13 and +10, and most have it around +6.
- Georgia
- InsiderAdvantage went from 3.6 to 2.9 (-0.7) since October 4, but -3 to 2.9 (+5.9) since September. Depends on your horizon scale, but at worst Warnock is stagnant in this poll, and at best, he’s improving dramatically.
- Trafalgar went from -0.8 to 1.5 (+2.3)
- Quinnipiac went from 6 to 7. (+6)
- Emerson went from -1.7 to 2.2 (+3.9)
- InsiderAdvantage went from -3 to 3.6 (+6.6)
- U Georgia went from -1.6 to 3 (4.6)
Verdict: As 538 points out there seems to have been some movement toward Warnock from the latest Walker scandals. My estimate is that the race has shifted by 3.9 points.
- Nevada
- Suffolk went from 4.4 (in august) to -2.
Verdict: Almost no polling, but the little evidence we have is that we have, indeed, lost ground.
- Pennsylvania
- Trafalgar went from 1.8 to 2.4.
Nobody is polling this! So why is the narrative that Fetterman is losing ground?
So, in the 4 closest states, 1 clear gain, 1 clear loss, and 2 ¯\_(ツ)_/¯. And this is to say nothing of places like Iowa or governors races in South Dakota and Oklahoma which have seen huge swings toward the Dems.
Now, I will make one caveat. You don’t necessarily have to compare only within a single pollster to get a trend line. The presumption in polling has always been that pollsters are more or less looking at the same thing and such whether it’s Monmouth or Siena or Quinnipiac, as long as someone is polling a state, the differences will come out in the wash. It may well be that Fetterman is losing ground in PA but that since almost nobody has polled it twice we can’t do a comparison. However (they’d argue) you can compare the Fabrizio (D+2) result from last week to the Muhlenberg result (D+5) a month ago, and if you average enough similar results you get a trend — and that trend, roughly, is that Fetterman has lost ground.
I honestly don’t know that that’s the case, but the press is sure pushing the hell out of that narrative.