Professor Robert Shapiro suggests that pre-election polls, which differ from public opinion polls, still have a future:
Crucial state polls were significantly off once again, especially in Michigan, Wisconsin and Pennsylvania. Yes, Biden won these states. But he did so by thinner margins than were found in pre-election polling results, which steadily forecast him winning by 4 to 5 percentage points or more. Further, Trump defeated Biden in states that were allegedly close, like Florida and Texas, by handier margins than expected. In Arizona and Georgia, the polls were within sampling error margins. But they were way off on several congressional races. Maine’s Republican Sen. Susan Collins won handily, despite pre-election polls showing her opponent leading. And while some expected a “blue wave” election that increased Democrats’ control of the House, Republicans gained House seats. …
It will need to consider the same potential problems discussed during 2016. These include whether some respondents were “shy Trump voters,” reluctant to disclose their intentions, or whether Trump voters were less likely to respond to polls, which pollsters call “nonresponse bias.” This may not have just meant underestimating the number of Whites without college degrees who were likely to vote, but also underestimating likely voters in rural or small-town areas, who voted overwhelmingly for Trump. [WaPo]
Here’s the thing: vote recounts are being misapplied.
See, look at this like an engineer or applied scientist. The election is the experiment, as it were; the polls are the predictions about the results of the experiment. Now, after numerous decades of experience, you’d expect the pollsters, the agents the experimenters are measuring, to get it right. After all, they have margins of error, they work hard, and they’re the ones who have to blush under the bright lights if they get it wrong.
Given all that motivation, when the experiment goes wrong, such as in Texas, where it appears the nearly 6 point margin of error is outside the margin of error of the polls, or Florida, which might be just inside the margin of error at 4 points, or Wisconsin, where the results of a Biden victory of less than a point certainly didn’t reflect last few polls I saw, red flags are raised. At least for engineers. Even when the result is what you desire, when something unexpected occurs, a good engineer chases those anomalies, attempting to explain them.
Now, we apply recounts only in close races. From a governmental point of view, this makes sense. It enforces a sense of fair play by checking for mistakes, minor fraud, and exhausted poll workers.
But it ignores, or holds constant at zero, the potential contribution of widespread fraud. Sure, I sound like that idiot in the White House, but there it is: As an engineer, I’m baffled as to why we assume a collection of polls, run by people with professional reputations on the line and years of experience, that turn out to be at substantial variance with the final result are always assumed to be incompetently run.
Therefore, I’d like to suggest that at each Presidential election, the State that strays the farthest from expectations, and is outside the margin of error, be required to do a complete hand recount.
Georgia did that this year, and had a reassuring result that matched the results of their voting machine. (And confirmed Biden’s win. But that’s beside the point.) I was relieved, even as I expected that we might see some voting machine fraud. Didn’t happen, I’m happy.
But they only recounted because it was so close.
I’d like to see Wisconsin, Florida, or Texas recounted. Just to reassure the nation. And myself. It doesn’t even have to be advertised. Just go off and do it with a second team of poll workers – maybe drawn from across the nation – with full observation rights for the political parties.
And let’s see if the pollsters are really finding their job that hard – or if something else is going on, from either side of the political spectrum.
