Monday, March 7, 2016

j reacts to snapshot polling right before the michigan primary

michigan prediction...

charting polling averages (whether through rcp or 538 or anything else) is a stupid way to try and analyze voting results or make voting predictions. you don't want to chart a trend. you want a good snapshot as close to the date as possible.

i'd like to see a poll taken after the debate last night, as i expect bernie will get a slight bump. not likely to happen. and, apparently, there's another one tonight [at fox. i'm sure that will be great.]. better, the polling released over the last few days is suggesting that it's tightening.

i expect clinton to win. but i don't expect her to get to 55%. and, that's a symbolic victory. what i said was two things:

(1) he needs to at least nearly split the delegates. so, it's gotta be at least close.
(2) the last poll i saw had her at 76% black support and 49% white support (to bernie's 48%). that is still awful - 27%. in fact, it's worse than the previous 84/61 (84-61 = 23%) difference. he needs to get that something closer to 10%.

she will win. but, it will be surprisingly close.