about errors in polling.

Have you met America?

It'll sometimes take some time, but we see our mistakes and tend to correct them after a while.

We live in an intensely, pathologically divided nation. It is the nature of America at this point in history. That doesn't leave room for a 10 point victory. 3% would be amazing. I don't know why people have to pretend it will be a blow out. All we need is a win.

We need a landslide as a bare minimum for survival of our country, and quite possibly our species.

I can revisit the numbers with you and tell you why I think this 10 points is possible and even likely, but it will be easier just to wait and let you see it for yourself.

We'll find out in just over a week.

True – it’s called the Electoral College.

Just a guess, could be more or (God forbid) less.

I have talked to several of my friends about phone polls and many of us are not even answering the phone or responding to them in other ways. I hope there are lots more of us out there holding out because it would be one hellova surprise.

…but if Trump supporters are equally reluctant or not. If there is no particular partisan bias in willingness to participate in polling, the effect is a wash.

Good modeling can also compensate to a degree for a partisan bias in poll participation.

Everything now is speculation, so what's the point?

You don't have to be capable of perfect prediction to want that. For most people speculation is an automatic brain function. The only choice part about that speculation is if you engage others in a conversation about it or not. I choose to engage.

And given the TOS here, it’s not worth the potential blowback to discuss what’s happening on the ground, and that’s a shame it can’t be mentioned. But too many itchy fingers.

That's a 1.5 margin. My guess is she wins 51-47 - maybe that 47 rounded to 48. So, 3-4 point margin of victory in the popular vote.

That would put the polls, on average, 2.5 off at best. I think that's reasonable and within the MOE of most polls.

Very little chance Kamala wins by more than that.

My reason for a little more optimism isn't in the accuracy of the polling itself, but the turnout motivation that can only be roughly estimated based on previous elections.

As for MOE, that's actually less important the more polls you have to average together, and/or have a running track record of. MOE captures the completely random part of polling error, related to the luck-of-the-draw of who randomly pick to poll, nothing more and nothing less.

The more you aggregate, the more the MOEs tend to cancel each other out. Average together 10 polls each with an MOE of 3% and the average will have an MOE of only around 1%.

MOE doesn't capture modeling errors or participation bias at all.