Polling Uncertainty

Kind of what I been reading is what Infidel is writing about in this post of his. NYT has a detailed opinion article about the complexity of polling on the 24th (freebie). Even they see polling results as complex.

I believe Roe v Wade will still play a part in how women and others will vote. I am also hoping Alito will keep flapping his lips. It stirs the crowd up and reminds people he was the instigator in Roe V Wade. Six days off now to the election. “Vote!


Infidel753: “Polling uncertainty,” Infidel753 Blog.

I’ve never been one of those who dismiss polls as meaningless.  Political parties and campaigns, and the media, pay millions for polls during each election cycle.  If the polls weren’t giving them useful information, they’d stop doing so and save their money.  It’s true that pollsters rely on very small samples of the population, but by weighting various demographic categories, they can adjust their results to reflect the general population, or registered voters, or likely voters — whatever they are trying to analyze.

However, pollsters’ projections depend upon what assumptions they are making about voter turnout.  If turnout is substantially different from what the pollster expects, the actual election outcome will be correspondingly different from what the polls projected.  This year, turnout models are especially iffy because of the influence of a new factor (the end of Roe and subsequent wave of abortion bans in red states), for which nobody really knows how much impact it will have.  The only hard data point we have is the Kansas referendum, for which polls showed a very close result, whereas the actual result was not close (60% favoring abortion rights), while turnout was very high for an election of that type.  Obviously the pollsters misread how the sudden salience of the issue would affect turnout.  On the other hand, it was a referendum, and we don’t know whether the threat to abortion rights will affect candidate elections in the same way.  A referendum is on a single issue, while assessment of candidates turns on all the issues the voters expect an elected official to act on when in office.

Pollsters are also smarting from 2016, where they failed to anticipate how Trump’s unique appeal to citizens of a type who often don’t reply to polls would drive them to vote.  They may be over-compensating by over-estimating Republican turnout.  Trump is, after all, not on the ballot this year, so it’s unclear whether his loyalists will turn out as much as in 2016 or 2020.

So, basically, this year the accuracy of polling is more uncertain than usual.  The election could be close or it could be a landslide for either side.  We can’t be sure.  Be prepared for surprises — in either direction.  Democrats would be wise to fight as if they were behind, and to take nothing for granted.