Ah ha! Pollsters base models partly on what happened in previous elections...

[Chichiri]

...and in the previous election, only Hillary was on the ballot!

Michigan so far is the only state that has seen more Democratic votes in 2016 than in 2008, with about 1.2 million people turning out Tuesday. Trying to predict what that turnout would look like based on half as many voters in a weird election eight years ago means there was a good chance the polls would be off.

Which, we now know, they were.

(...)

Numbers marked in light blue were demographic groups that were underestimated relative to actual turnout; the ones in red were overestimated. On gender and age, both polls overestimated the Clinton-friendliness of the electorate. Mitchell/Fox 2 vastly overestimated how old the electorate would be — and that’s a group that heavily favored Clinton.

On race, the polls slightly underestimated how many black voters would turn out — which, as we noted earlier, was a problem by itself. The black vote in Michigan was much less friendly to Clinton than it has been in other states; had she seen the support from black voters that she got in Mississippi on Tuesday, for example, she’d have likely won.

From WaPo: https://www.washingtonpost.com/news/the-fix/wp/2016/03/09/why-were-the-polls-in-michigan-so-far-off/

If this analysis is correct, it means that Michigan likely was a fluke after all.

Having said that, let's continue to act like all polls are wrong, wrong, wrong -- and GOTV like mad!

[LonePirate]

1. The 2008 MI Dem primary numbers are useless because Obama and Edwards were not on the ballot.

MI and the national party were squabbling so Obama and Edwards skipped the state and ballot. Had they both been on the ballot, then turnout number comparisons would be more viable.

[yardwork]

2. Oh, I see! Then MI really is an outlier. Maybe.

[Chichiri]

3. Maybe. Let's act like it isn't. nt

[Treant]

4. Maybe.

But they screwed up just as badly in MS, it was simply in Clinton's (the winner's) direction.

Surprise upsets get discussed. Surprise big wins in an expected win state don't.

Primary polling isn't that great, there's not a vast amount of money spent on it, and we all know that. There have been significant errors in every Presidential primary.

[LisaM]

6. I remember, and the usual suspects were squabbling that the delegates shouldn't count

towards the convention. IIRC, they didn't. I can't quite remember what Obama's beef was - was it that Michigan had moved up its primary date? But Hillary campaigned there anyway, she won, and the national party (you know, the ones that do nothing but her bidding) wouldn't add the total to the delegate count. I think there was another state involved in this, too. It was ugly at the time, one of the things that ultimately drove me off DU for a good long time.

[Lucinda]

5. That makes a lot of sense!

Thanks much!