illinois
1) the nbc poll that puts clinton at 51 was taken from the 4th to the 10th. this is a polluted sample. but, i would suspect that sanders takes a huge lead in polling after the 8th. this poll likely overstates support for clinton by integrating data before last tuesday.
i want to be clear about what i'm doing here. i'm doing exactly the opposite of what the aggregates want to do. they want to say that you want to average polls over a longer distance, to "smooth" the data out. that works fine in some circumstances, like measuring responses to advertising (you can buy a sandwich the day of the polls, or the day after, or a month later, even). it's simply awful modelling for politics.
the aggregate sites will point to this poll as more accurate because it was taken over a longer period. but, i'm suggesting the longer period is just polluting the sample.
based on that poll alone, i would suggest a slightly better result than we saw in michigan.
2) there is another poll, with sanders ahead at 48-46 (i guess there's 6% undecided). on the one hand, this is an internet poll - and the predictive abilities of internet polls are widely variable (there is no margin of error!). on the other hand, it was conducted mar 9-10. despite not liking the methodology, this is entirely consistent with everything i'm saying.
---------
i don't have a good poll, but i don't want to just average them out, either. the reality is that the data for illinois actually looks better for sanders than the data for michigan did. plus, illinois is a very blue state. it is actually reasonable to predict a comfortable sanders win, if you consider that this is an open primary, the state is full of liberals and the phone polling is being done with democrats. but, the correct answer is that the data is inconclusive.
something to note, though, is that illinois will likely provide the data that is necessary to permanently throw away the racial theories of support. it's going to split on a left/moderate axis, and that will assert itself across racial lines.
ohio
we're looking at the same breakdown in the two polls - the phone poll is sampling over too long of a period, and over a defining pivot point, while the internet poll is solely after the defining event. and, sanders is doing better in the polling done solely after tuesday. there is again not a perfect poll, here, but there is reason to suggest a very close result [given that they're strictly polling democrats]. but, remember: ohio is completely purple. it doesn't have the liberal base that illinois does, and doesn't consistently lean democrat like michigan does.
florida
clinton is comfortably ahead here, and the event doesn't seem to have registered much. florida is the old folks' home of the nation, and these voters tend to be low-information and driven by routine. bernie needs to try and keep it under 60%. and, this is key - because florida is a very big state.
missouri
literally no polling. it's a purple state. i would expect it to split.
north carolina
literally no polling. obama won north carolina in 2008, but obama is a conservative. so, one must expect clinton to win north carolina very comfortably.
--
if he wins big in illinois, he could gain delegates on tuesday.
otherwise, he's going to be struggling to not get further behind.
but, remember: it was always about california.