Stop me if you have heard this one before…. there is a lack of high quality polling for the 2022 election. What polling we do have is mostly party internals, mostly from the party in power, pointing to how things are looking rosier even as their President’s approval is mired in the 30’s and low 40’s. Numerous internals and party aligned pollsters have found Democrats ahead in Ohio’s and Georgia’s Senate races, dominating in Pennsylvania and Arizona, and deadlocked in North Carolina and Nevada.
More importantly, the generic ballot has evened out for Democrats after Dobbs overturned Roe vs. Wade and two special elections (NB and MN) seem to indicate Democrats can mobilize their base with cultural appeals on abortion. But, are the polls really right about this?
During the 2016 election, and especially after, many analysts and journalists wondered what had happened. How had Trump prevailed and rocked the political landscape? An effort was made to not just poll but talk to these voters in person. Fast forward to the 2018 Democratic wave and that talk vanished. Once again, it seemed the political norm had returned.
But even in the midst of the 2018 election, we saw reminders of 2016. Pollsters vastly underestimated GOP support in Florida, Ohio and Iowa statewide contests. Move forward two years and we saw a gold standard WashPo/ABC poll come out two weeks before the election in Wisconsin showing Joe Biden up by 17 points (he won by less than a point). Nationally, the generic ballot, which had Democrats up by about 7 points, saw them by win by 3 points and lose a dozen individual seats.
Now, this wouldn’t be so bad if pollsters adjusted their tactics or models, but once again, despite misses in 2014, 2016, 2018 and 2020, they did not. Analytical gurus such as Nate Silver have argued on average pollsters have been pretty close to hitting the mark in high profile contests (except when they are not). Honest pollsters freely admit they have trouble reaching rural or Republican voters who don’t trust them. This has led to pollsters having to weight smaller samples which has led to its own problems.
Worse, many of the election forecast models still rely heavily on polling (538, Decision Desk, etc). If the polls are off the models are off simply because the closer an election gets the more heavily polls are weighted in the model.
This is not meant to be a diatribe against the polls. Caveats are in order. Most pollsters have not switched over to a Likely Voter Model as of yet so it is possible as they do so the GOP will gain in the generic ballot. Then again, the LV generic ballot average didn’t see a change when this happened in 2020. Perhaps the Dobbs ruling really did change the landscape.
But it is also important to remember all the fundamentals at the moment point to a solid November for the GOP. This is why Roll Call and the Cook Political Report have not adjusted their forecasts despite improving ballot projections for Democrats. As Dave Wasserman for Cook put it, “we are in the middle of the maybe Democrats can save their majority,” part of the midterm.
This brings up another point which harkens back to 2016 aftermath, the horse-race narrative vs. the fundamentals. In a nutshell, the horse race narrative focuses non – stop on polling, public perception and candidate differences. Sure, major party candidates have major differences, but ironically, both the Left and Right’s focus on populism tends to align them more than not on economic issues. Even cultural issues such as gay marriage seem to have fallen to the wayside.
So what is to be believed? The horse – race narrative which shows an environment toxic for Democrats but because Republicans nominated some “kooks” could cost them, especially in the Senate? Or the fundamentals which state record inflation, supply shortages, and the like will lead to gloom for the majority party despite their legislative achievements in recent weeks.
Well, historically, the answer would be obvious. Doing some digging, Cook’s found in 2014 (the last time we had a midterm with a Democratic President), Senate candidate only were able to outrun Barack Obama’s approval by five points in their state. Obviously, this meant some Senators survived in deep blue states but in purple states such as Colorado it spelled doom. What we see now is Democratic Senators running well ahead of that five points in sparse public and partisan polling of Registered Voters.
Finally, much has been made of recent special elections where Democrats have improved on their 2020 showing. Of course, we need to be careful in reading into special elections. First – off, Republicans lost four special elections in 2010 and gained 63 seats that November. Secondly, special elections historically benefit the party out of power IN THE DISTRICT. Thirdly, they almost always have local factors at play which don’t gel at the national level. I’m not downplaying the results in Nebraska’s 2nd (where Republicans won by six points compared to 11 in 2020) and MN’s 1st (where Republicans won by four points compared to nine in 2020) but caution is in order. MN – 1 has historically been more competitive down – ballot even as it shifts rightward at the Presidential level. NB – 2 featured a weird scenario where the boundary lines for this election are different than that for this November.
So, I come out with this. Everything points to a solid GOP November. Candidate quality in gubernatorial races such as Michigan could cost the party (though those races are less defined by national factors) but for the most part the GOP remains favored to take the House and (sorry 538 and Decision Desk) the Senate. Those models are plagued by the same problem any model would be in this era. If the polls are off and your model relies on them, your model is off.
Note: I don’t believe the the raid on Trump’s home or statements from the GOP or Democrats will dramatically impact November. Again, fundamentals matter more.