so, what's my post-mortem on my analysis?
what i said was that the democrats would win back their states in the north (and they did, all three of them) and that they would be disappointed and lose in the south (and they lost texas, florida and north carolina). that said, i refrained from predicting outcomes in nevada (due to low hispanic turnout, and maybe even a swing in the hispanic vote to the republicans), iowa (it wasn't clear to me if iowa would act more like wisconsin or more like missouri) and north carolina (which is currently a real swing state) under the argument that it didn't matter. while the outcome will not be clear for some time still, and despite a disappointing night overall in the south, the democrats are currently holding to very small leads in both georgia and arizona.
the reason i argued that the polls were wrong in the south was due to over-estimating the swing to the democrats amongst educated voters. so, i made an explicit claim about a modeling error and made a prediction surrounding it.
while i insist my critique is both valid and upheld, i seem to have overshot the correction by the smallest of margins. if biden wins georgia and arizona by .01% each, which is looking to be about right, then all i had to do was overshoot the correction by a point. certainly, the models had biden doing better in both of those states, so he did underperform his predictions - it's a question of by how much, and those small differences might end up flipping both those states.
i'm posting now to move on, because it sort of doesn't matter. as i stated in my projection, the south doesn't matter because they won back the north. but, regardless of the outcome, i am right in pointing out that a polling bias existed - what i did wrong was overestimate the correction, and by what is going to end up being the very slightest amount possible.
should i have penciled in arizona and georgia as grey, then? well, i penciled in arizona as light red, indicating i was convinced that the counter-correction would still make it very close. arizona is being absorbed by california, and i actually agree that the trends are that it's becoming more democrat. but, i actually think the voter breakdown in the future will be liberal, white, wealthy democrats v. desperately poor socially conservative hispanic republicans. because abortion. georgia, i think is more of a fluke - i think biden got a helping hand from a lot of white evangelicals and that's not likely to last. so, the short answer is "no" because the modeling arguments being presented to me were not convincing. i made an explicit argument, and i think my position was grounded.
i'm not willing to concede either state, yet, and won't be until after recounts. but, it largely doesn't matter in terms of the post-mortem - regardless of the eventual winner, my counter-correction was correct, but i over-counter-corrected and it may leave me short by a hair.
and, my projection was more accurate, overall.