That bit at the start with Biden ahead is before results started to come in. I would guess that it had a similar problem to GA, that the strength of Biden’s results in the suburbs was underestimated, possibly Miami-Dade has a bit to do with that. It wasn’t crazily wrong though, like out to 90s and then back to 86, which it stayed at consistently.
That is obviously a fuckup, but like I said, it’s a hard problem.
I’m considering building a projection model for next election, but it’s a ton of work for one night every four years. Seems like there’s a ton of money to be made trading it in-play though.
This was the basis of my take on Election Night. M-D came in and all the needles jumped, and M-D is one of the most unique counties in American politics with nothing in GA/NC that should correlate that strongly.
Of course when it was in the 80s weren’t a lot of NC results in at that point? 60-70% seems fair for the point when it initially shot up.
Yeah I mean GA should have been in the 55-65% range probably. Seems like there is an easy solution: reduce the impact that other results have until the sample size is bigger and make sure you don’t over correlate unique counties.
Depends which markets you have access to. Also while your edge would be high, how much higher would it be than just going off your own ability to analyze results given the softness of the markets?
I have no idea how the Upshot needles work, but my idea for a model is to do a regression to take various factors other than geographical proximity (education levels, population density, racial makeup, income etc etc) and predict how well swings in different counties will correlate with each other. Then on the night, as results come in, make predictions using those predicted correlations, but also re-run the regression and adjust the coefficients based on what is happening. Like after the first 50 county results come in, it might turn out that income is way less relevant in correlations than it has been previously. I feel like doing that, it might be possible to get pretty good evaluations of what is likely to happen in states where polls aren’t even closed yet.
Edit: And yeah you would have to have a “uniqueness factor” as well which would just be the amount that actual correlations with other counties nationwide have differed in other elections from what the regression would predict. I’d guess Miami-Dade would score high on that.
That makes sense, the problem is that you’re limited to a few states in early results and you can only apply it to a few states with polls still open. You crush it on an election with a surprise realignment (like 2016) but you don’t do much better than common sense in most elections. Tons of variance for a once every four years event.
Yeah I don’t think I’ll actually do it for that reason, it’s an unholy amount of work and there’s no guarantee either than I can get it to function well or that it provides a huge amount of value over and above eyeballing counties and figuring things out myself.
At some point this weekend all the deplorables figured out that the recount in GA was meaningless and was not going to change anything. They actually wanted to have the right to re-review every signature on the outside of every mail-in voting envelope and start tossing ballots all over the place. These people truly hate democracy.
I sent my GA mail-in ballot on 9/20, and was informed on 9/28 that my ballot was ‘accepted’ (ie no problems, will be counted). Now after their guy lost, they think there’s something fundamentally ok with pulling my ballot/envelope two weeks after the election and look at how I dotted the ‘i’ on my name and see whether my vote should still count. In any case, those envelopes were separated from the ballots in some cases weeks before election day.
Also, GA uses Dominion voting machines. When you cast your ballot in person, it prints a piece of paper that you can use to verify your votes, and then you scan that piece of paper (it tabulates via QR code). The hand recount is reviewing the human readable portion of this piece of paper. When the results of the hand recount match the original count, it should 100% dispel any myth about the machines “switching votes for Biden”, although we understand nothing is ever enough for these conspiratards.
Lol at trying to convince people who are divorced from reality of anything. No point in trying and the “recounts” are just a waste of time and money.
Y’all are going to be super shook when the “reasonable Republicans” who are just calling to wait and see how the process turns out and allowing all legal avenues to be pursued don’t like how that turns out and lets their freak flag fly.
Literally no way to do that. The votes are anonymous, because you have the right to vote anonymously. Once teh signature is accepted the ballots are all pooled together, there’s no way after the fact to challenge a signature and “find” that person’s ballot.
