These two articles, one from July 2021, one this morning, from Politico make the needed point that polling this time around reasonably may be as inaccurate in the 2022 mid-terms as it was in 2016 and even worse in 2020. I honestly had not remembered the 2020 polls being worse than 2016, probably because the presidential result didn't flip. These refreshed my memory that the Blue Wave in 2020 receded first into a Blue Ripple and ending in a Red Ripple in congressional results. Two big caveats to these articles' caveats: Trump is not literally on the ballot this time as he literally was not in 2018 when the polls were more accurate predictors.
The 2016 polling post-mortem consensus was that the polls had systematically missed Trump's high school educated voters and when they did reach them they claimed that they were undecided. For 2018 and 2020 the polling firms corrected by going to lengths to reach those voters and to weight the undecideds more for Trump. Their efforts made for much more accurate polls in 2018 but then came the Debacle of 2020 which overstated support for Joe Biden by 3.9% nationally and by 4.3% in state polling. There is no consensus in the 2020 post-mortem but the factor of most interest is Trump voters' refusal to answer questions from pollsters when they are called. That is, in 2016 Trump voters claimed they were undecided; in 2020 there was that faux undecided problem but in addition a new Delta variant of refusals.
That this unintentional bias is likely still occurring is captured in these two passages from today's article:
The current 2022 polling is wildly favorable for Democrats. FiveThirtyEight’s “lite” prediction model, which is based solely on the latest polling data, says Democrats have a 79 percent chance to retain control of the Senate. That probability clashes with the expectations of both parties and most independent handicappers, who consider the battle for the chamber to be closer to a coin flip.And the New York Times noted that some of Democrats’ strongest numbers are coming in the states that have seen the greatest polling misses over the past few elections.
The undersigned has ignored FiveThirtyEight since the Catastrophe of 11/8/16 and even the non-partisan Cook Report overstated Demos strength in 2020. For the 2022 mid-terms I have taken note (with irritation) that RealClearPolitics' projections have consistently had, and to this day have, Republicans making a net gain of two seats in the Senate, flipping Georgia, that is Herschel Walker beating Ralph Warnock, and Nevada, and holding Pennsylvania, that is Oz defeating Fetterman. The polls have been tight in Georgia which makes RCP's projection more understandable even if the projected Republican victor is a joke. In Pennsylvania however polling has Fetterman with a larger lead than Walker has in Georgia and the polling has been consistent in showing Fetterman the leader. Cook recently moved Pa. from toss-up to "lean Democratic." RCP though says "nah-ah" to that and calls the Commonwealth for Oz.
I read the NYT article when it came out. One of the states it profiled was Ohio. JD Vance, the Republican, and Tim Ryan, the Democrat, are neck-in-neck according to polling. I cry bullshit on that. We Demos have to write off Ohio as we have Iowa. Both states have moved decisively into Trump Country. In Florida too, although polling gives Marco Rubio a 4.5 point edge over Val Demings, there is no doubt in my mind that Rubio will win. Demos are competitive in Florida as they have been in all state-wide elections but have almost always (18 out of 20) been on the wrong side of the competition.
I think pollsters fears are warranted. I'm going to go against personal judgment and with RCP on the Senate and with them on the House too: Repubes will gain 20-30 seats. I think Blue Balls will be busted agayne.