they are claiming that proportion of black voters is predictive; that it's causal. i claim it's not, that it's a proxy for ideology. and, they're running through all of these other irrelevant things that are purely correlative, and at best proxies (but largely, not even). this is the basic error of correlation not implying causality. it doesn't. it never has. no matter how convenient pollsters think it is, it's still wrong. no matter how ubiquitous..
i am calling illinois for bernie because it is a blue, liberal state. i think he will win illinois big, and there is some polling that upholds this. missouri and ohio are purple states and should actually split - although i think bernie will win missouri and ohio is less clear (based on polling). so, ohio is the least likely because it is the most purple. and, frankly, the polling for florida (a blue-to-purple state) isn't making sense to me - but i don't have any valid argument to suggest he has a real chance, other than to question the modelling in the polling (and all that can do, at best, is take the margin down - but i said the same thing about michigan). i have claimed florida will be closer than expected.
we saw the black thing break down a little in michigan, but michigan is purple. illinois will collapse the whole thing altogether and make it clear that race is not the predictive variable, but merely a coincidental one.
but, it won't stop the media. this is what the media does. this is what the country is!
http://fivethirtyeight.com/features/can-bernie-sanders-pull-off-an-upset-in-ohio/