i haven't really poked the models in the eyes, yet. i refused to accept the argument that sanders had a better chance in ohio than he did in illinois due to racial breakdowns, and i suggested that michigan would be a split, but i've otherwise largely arrived at the same conclusions, via different metrics. and, you might expect that, really - as i'm arguing that race is a proxy variable, rather than a predictive one. as such, you'd expect it should work out more often than not, even if it's logically incoherent to suggest that racial breakdown is a causal factor.
alaska and hawaii are giving me an opportunity to push back a little, although i need to provide the caveat that i have not seen any polling and reserve the right to modify my analysis as a consequence of direct data appearing.
so, the models will tell you that there aren't any black people in hawaii or alaska, so bernie should win big. it's a lack of people predisposed to voting for clinton due to the colour of their skin. i take a different view.
rather, let's look at votes, so far, for places outside of the contiguous 48.
american samoa
clinton: 68
sanders: 26
northern marianas
clinton: 54
sanders: 34
this is a small sample, and direct polling will render it obsolete. but, i think it's a better measure in at least hawaii - if maybe less so in alaska.
we should not forget this:
democrats abroad
clinton: 31
sanders: 69
...but, it's not the same thing as a state or territorial primary. sanders has consistently outperformed clinton with people that are very liberal and people that are very educated. if you're going to vote in the democratic primary from singapore or something, chances are that you're both. nor are we dealing with indigenous inhabitants. so, that should be removed.
the takeaway is that i can't automatically favour sanders in either of these states. although, note that they are caucuses. and, you know what i say about caucuses...
without direct polling, i think clinton's tendency to do better in distant territories gives her an immediate advantage in hawaii - even if it's not a large one. that logic may be less applicable to alaska, but it's at least as useful as racial profiling.
how about washington? well, i need to learn from past experiences. it looks like she cheated in arizona and ohio and inflated the wins. it looks like she cheated in masachusetts and illinois and snatched victory away from defeat. and it looks like she cheated in michigan and lost anyways. why wouldn't i expect her to cheat in washington?
washington is an open primary, which is the same as michigan and illinois. so, he ought to demolish her, there. 70% should be a low bar. but, that just means she'll stuff ten times as many ballots and make it that much harder for people to vote.
so, this is my actual prediction for washington: if there is any remaining doubt about whether this process is rigged or not, it will be over by the time washington is tabulated. washington will just make it that much more obvious.
*washington is not an open primary, but an open caucus. that changes everything. it is easy to see that i would have suggested a large sanders win is most likely if i thought it was a caucus.