So we all know going strictly off polling Sanders is going to lose OH/FL/IL/NC, so I’ll stipulate that still “is likely to happen.” But exploring a little deeper:
Bernie outperformed polling exceptionally in Michigan, but only did a little better (2 delegates out of 130 worth) than his “projected demographic performance” that FiveThirtyEight basis their projection tracker on.
-Florida is demographically better for Clinton than Michigan was, she was “supposed” to lose Michigan, but she’s supposed to win Florida by roughly an 18 delegate margin. Additionally, Florida is a “true” closed primary state, so anyone who isn’t a Democrat and wanted to vote for Bernie had to have filed paperwork to switch parties 29 days or more before 3/15. We know that this generally is something that a lot of people don’t do, which hurts Bernie’s chances of repeating his poll-busting performance in Florida.
-Ohio is demographically quite similar to Michigan, and Clinton’s lead in the Ohio polling is a little less than his Michigan polling was. If Michigan represents a trend of Sanders support that the polls are missing, Ohio is a likely place where we would see this manifested. In Ohio one does not register to vote with a political party, and thus the primaries are true open primaries. The Governor of Ohio is participating in the Republican primary on the same day, which may lead a lot of moderate Democrats (and a ton of them voted for Kasich for Governor, he won his last election with like 65% of the vote), into that primary to support Kasich, which hurts Clinton. Of the four 3/15 states I think Ohio is the most likely to repeat Michigan for Sanders.
In fact, I’d go so far as to say I have an “unscientific hunch” that this may happen. I’m not saying I know more than the polls, just saying I have a hunch about Ohio. If I was betting on elections I’d probably still bet with the aggregate polling as they have been right more than they’ve been wrong (it’s a case of selection bias that some Bernie fans think the polls have been wrong all year.)
The only negative for Bernie in Ohio is unlike Michigan Ohio has early voting and no-excuse absentee voting (i.e. anyone can get an absentee ballot without having to justify it.) I haven’t explored the reasons why this is, but pundits have claimed the lack of early voting and no-excuse absentee voting in Michigan hurt Clinton there and helped Sanders. It could be because there is a belief that Sanders’ heavy campaigning in Michigan swung a lot of late voters his way, and his ability to do that is lessened somewhat by the Ohio voters who may have started voting as early as mid-February for this election, and who may have already mailed in their absentee ballots (early voting in Ohio is both in person and also absentee mail in.)
-Illinois I think Clinton wins for a few reasons. One is that her lead there is even bigger than in Michigan, so we’d need an even more historic polling error for her to lose. I think it’s quite likely, for reasons similar to Michigan and Ohio, that she doesn’t win by the large margin she’s expected to; but Michigan was also a stronger demographic state for Bernie. Demographics wise Hillary is “supposed” to win Illinois, unlike Michigan and Ohio. Chicago also is a hotbed of pro-Obama sentiment, and that may count for something here, as Hillary is essentially the Obama third term candidate. Illinois also has a reputation for having a well organized and highly influential/active Democratic political establishment, which also helps Hillary.
-North Carolina is listed as having a closed or semi-closed primary, but it’s functionally an open primary state since you can switch registration on the day of the election. Hillary’s polling lead here is a lot smaller than in any of the other three states. Having another “hunch” I think Bernie may win North Carolina. It’s a slightly red purple state (with a far red legislature as a sad result of what can happen when a small majority of the state is Republican), it’s a Southern state with a contingent of conservative white Democrats and a larger than average black population. But it’s also a state that has grown a lot, especially the “research triangle”, and it’s grown with lots and lots of college educated young whites who are politically very liberal (they are what is turning North Carolina from red to purple, and likely from purple to blue in a couple election cycles.) I think they will come out very hard for Bernie, plus aside from the young 20/30 something college educated white professionals, North Carolina has a ton of major universities who will also have a lot of enthusiastic Bernie voters. They also are hard to poll–automated polling cannot dial cell phones, most young people exclusively have cell phones. So those polls have to model the expected vote among cell phone users, an error prone process. Manual dialing pollsters can hit cell phones, but they run it based on North Carolina area codes and exchanges, a lot of college students have out of state area codes so will also be missed by this process. So there is good reason to believe the polling in North Carolina is off.