Apparently my state gave Bernie Sanders the most unexpected of wins last night, winning a close race that only one poll even got close (but still predicted a narrow Hillary win). Most polls predicted she’d win it by 20%.

I’m going to spend the day looking at the results. But here’s what I understand to have happened:

• Hillary won African-Americans, but not by as much as she had in the south. That’s probably consistent with MA’s results, the exit poll for which didn’t break down people of color, but as far as I know no one did the granular analysis of the MA results to figure that out.
• MI’s Muslims voted for the elderly Polish Jew, by wide margins. That’s not a huge population, but it is big enough to have accounted for much of the differential between Hillary and Bernie. Keith Ellison, one of Sanders’ few Congressional backers, has close ties to the Muslim community here in MI and did some events for him in the last week.
• Independents turned out in big numbers, and they voted for Sanders by big margins. MI has a completely open voting process, which means independents — who might normally be considered less likely primary voters — can (and apparently did) turn out in big numbers. I suspect, though haven’t yet confirmed it, that support for Bernie outside of SE MI was much higher in terms of numbers; usually when people analyze Democratic contests in this state they just blow off the areas outside of SE MI, because the latter is such a population center of Democratic voters. That may be what pollsters missed. Update: Adding, the Democratic party in my part of the state is pretty dysfunctional, even though there’s a population that should be turning increasingly Democratic. This is where Bernie picked up the numbers to offset Hillary’s SE MI numbers. That suggests he basically created a Democratic infrastructure that doesn’t exist (as Chris Savage pointed out, Bernie set up offices here and in other cities so had more field organizers throughout the state, save Detroit, than Hillary).
• Students voted, and not just in Ann Arbor. Because of the way they suppress student turnout, University of Michigan is the school at which it is hardest for college students to vote (because they’re more likely to be out of state and/or from all over the state, which means if there’s a registration problem they’ll have harder time going “home” to vote). Bernie did events at EMU, MSU, GVSU, and only then U of M, and while those students are more likely to be dispersed than UM students, they appear to have turned out. Bernie continued to win young people by enormous margins.
• Hillary underperformed in the rich suburbs. Hillary only won Oakland County, which is the second largest county and is the kind of affluent county Hillary has done really well in this year, by about 4.6%. I thought she’d win by 5 to 7%. I’m not sure why that is — again, it’s always possible cross-over voting is affecting the Democratic side. But had she gotten the numbers in Oakland I would have expected she likely would have won the state. The same is true of Detroit, but turnout in Detroit is less reliable than in Oakland.

There was a lot of talk last night about Hillary’s cynical auto bailout attack misfiring. Apparently, some top UAW people got quite pissed about it and were communicating about it among themselves.

It may be that Michiganders realized that was a dishonest attack, but I think it just as likely they responded to Sanders’ comments about Hillary’s trade record (curiously, he never even hammered on KORUS, where her involvement is much more direct than some of the other trade agreements she has supported, and which has had a clear impact on MI jobs). I’ve been predicting for some time that Hillary’s record on trade would hurt her in the state. Exit polls showed that those who think trade agreements have cost MI jobs voted in fairly big numbers for Bernie.

Update: One more point. I’m hearing a lot of talk about Hillary voters crossing over to vote for Kasich or Rubio to slow Trump’s momentum. I doubt that explains Hillary’s loss. First, any talk of that is anecdotal: there was no big discussion about how to strategize a cross-over vote. Second, the Kasich and Rubio numbers just don’t show any big swing, except perhaps in Washtenaw (Ann Arbor), but the Democratic results are about what we’d expect there. Third, the only people I heard talk about this were Sanders voters, so to the extent it did happen, it likely happened, in small numbers, among both Bernie and Hillary voters. Finally, while crossover voting is quite common here, it’s less common among party insiders than more independent voters (in part, because what ballot you pull does get recorded, so it shows in the voting rolls), so you’d think fewer of Hillary’s supporters would even consider it.

Update 2: I know that exit polls showed 7% of Dems crossed over compared to 3% of Republicans. I actually suspect those numbers are lower than what happened in other states and mostly stems from Democrats voting for Trump. Indeed, it’s possible that those who wanted to vote for a populist in MI felt like Bernie was a viable choice here (which would be consistent with where he got his biggest margins) as compared to places in the south where Trump was a more attractive populist to vote for.