my hawaii argument was based on the idea that hawaii would vote similarly to other pacific island states, but it was never presented with much confidence. there wasn't any polling done, and the demographic modelling is anti-scientific and should be rejected. i always said that this idea was preliminary and subject to modification by incoming data. all anybody could have given you was an inference (that is, a guess). one could have argued that sanders should have won because it was a caucus, but hawaii had a secret ballot so the logic of peer pressure is kind of funny. nor is hawaii a white state. i think the better analysis is likely to invert my argument. i had essentially suggested that clinton should be favoured because of greater name recognition, in what is really a distant colony (and that was the pattern in the other pacific island states). what was probably true in the end was that the name recognition worked against her - which is to say it was probably less about a vote for sanders and more about a vote against clinton. turnout does not seem to have been low, either. for substantive policy: how about the asia pivot, and concern about clinton starting a war in the pacific? there's no polling, so all one can do is make stuff up.
either way, that buffers the loss in arizona a little. even still, it looks like he's about 25 delegates behind where i said he should be. but, i also suggested that he only really needs to split wisconsin. he now needs to have wisconsin act more like minnesota than illinois in order to make up the difference. he should still be aiming for 200 going into new york. and, he absolutely must win new york.
i would not advise taking early polls seriously. but, it is a closed primary. and, the results in massachusetts and illinois would suggest a split is more likely than a win.