The case for and against the polls

Uh, I'm gonna do a first-- Well, I'm gonna share on my screen if you're listening, join, you know, find this recording on YouTube, um, because there's, uh, gonna be a lot of information, a lot of information there.

Um, and then, uh, so I'll do the case for underestimating Democrats first, and then I may explain why I think the polls might be underestimating Republicans.

And, uh, then try to synthesize that so that it's useful and not so conflicting.

Um, David, can you see the screen? Just checking.

I sure can, but how about you, chat?

chat?Chat can see the screen too.

It looks like I'm looking at what the screen looks like on the stream, so we're getting some hearts.

Good stuff.

Good stuff.

Okay.

So here is the case for why polls might be underestimating Democrats.

Um, I have one, two, three, four, five points here on my handy cheat sheet, uh, for why this might be the case, and this goes from most, sort of most accessible, easiest explanation to the most, uh, the most technical.

So the first is, and, and, and, and basically it may be a higher level thing here.

What-- I think what you'll come away from with this conversation is that basically there's just a lot of decisions pollsters make, right? When r- the, the problem is that the, you know, capital P problem in polling is that, um, we, we as pollsters need a certain sample to represent hundreds of thousands or millions of people, and you can only get a couple thousand on the phone at, at best, and a lot of people you call hang up, and the chance is that those people that don't answer might be different than the people who do.

So pollsters have to do a lot of, you know, statistical weighting decisions.

They do a lot of, uh, they make a lot of assumptions about who is and is not answering their polls.

And I, I think what, what I want you to come away from, from both of these sides of the conversation is that there's just a lot of those decisions that could be going wrong, basically.

So, um, before we really even get to those, though, uh, it's three weeks away, and the election can change a lot in three weeks.

So if you're watching along, you'll see on screen now, uh, this is a table from the 2018 exit polls that shows the vote intention for Democrats or Republican, and Republicans, depending on when respondents said they made up their mind to vote.

Um, you kind of made up your mind in the last couple of days or earlier than the last month, and a couple of options in between.

Um, this chart shows that people who say they made up their mind in the last few days of the election voted for Democrats by 12 percentage points, 53% to 41%, while those that made up their mind earlier were more split, giving Democrats just a five percentage point margin in their, uh, uh, in the National House popular vote.

Okay? Um, so that's a pretty big difference, and, uh, 8% of people made up their mind in the last couple of days in 2018, um, and another 8% made up their minds in the last week.

That, that group is much more split, um, at least in 2018.

It is not the case in 2022, although I did not pull that chart for you 'cause, you know, one table's enough.

You really need like two or three tables sometimes.