So, we're at that time of year. Election season, and democratic partisans (as in, people who are ride or die on everything democrat, not just leaners who hold their nose for them) are sticking their heads in the sand again.
Apparently polls don't real, and the latest part of their "denial" part of their stages of grief is appealing to Allen Lichtman's "keys" theory.
I've seen this theory hyped up every election in the past few cycles, and it always kinda struck me as electoral tarot. Basically the idea is that he has these thirteen metrics that he uses to judge the election, and if 7+ of the "keys" or metrics favor the incumbent party, they win. If they flip the other way, the opposing party wins. I admit, it's kind of been somewhat accurate as the metrics do track with things that tend to be correlated with electoral outcomes, but honestly, the theory isn't and has never been perfect.
While people act like he never got an election wrong...he actually did. In 2000 he predicted Al Gore would win...and then Bush won. Now now, people will make excuses saying "well he didnt predict the electoral college outcome, he predicted the popular vote", and uh...okay. Let's assume that.
In 2016 he predicted Trump would win. Trump won the electoral college, but not the popular vote. Now he says since 2000 he stopped caring about the popular vote and that now he just predicts elections.
Now we're going into 2024 and he's saying Biden is gonna win even though the polling is terrible for him. And people are ignoring the polls, because they want to huff this guy's copium.
Uh, honestly? Aren't we kinda doing the texas sharpshooter fallacy on this guy? he got one election wrong, then says he got it right because of the popular vote. Then he gets another wrong and says it's because he predicted the electoral college. I admit it is possible to change methodologies and the like, but which one is it? It comes off to me like people are just drawing the outcome around the data that fits his theory, and then saying he never got it wrong. Even if we say he got it half right both times, he still got it half wrong and his track record isn't perfect. And honestly? I'm not sure that I even agree with the outcome that Biden is gonna win. One "key" is "strong economy" and yes, the economy is "strong" by traditional metrics, but yet peoples' #1 issue is inflation. Maybe having a strong economy is a bit of a problem if its too strong, even if "the fundamentals" seem okay. No third party challenge is another key, yet last I looked RFK has 8% of the vote. Even if we ignore the 1-2% losers like the libertarians and the greens, yeah, we are facing a substantial third party challenge. He has that key as "leans true.""No social unrest", leans true according to Lichtman. Okay, so what about the free Palestine protests? That isn't social unrest? "Uncharismatic challenger", uh....of the two candidates, Trump is significantly more charismatic I'd say. He literally told people he doesn't care about them he just wants their votes and they cheered. This was during a 110 degree rally in Nevada where people were collapsing from the heat.
I mean, at point, this theory IS just electoral tarot and establishment liberals are basically just using it as an argument from authority while huffing some serious copium. And when I actually tell people that, i get told, "oh yeah? well how many predictions have you predicted and gotten right?" And the answer to that is three. 2008, 2012, 2020. 2016 I got wrong, but so did most people, and even then I was sounding the alarm bells in 2016 that yeah Clinton could lose this thing and people were ignoring me then too. And now it's happening in 2024, when all of the polling fundamentals tell me Biden is struggling. Not saying he can't win, but he isn't favored and if I HAD to predict it either way, my current prediction is Trump.Heck, we could even go back to that retroactive 2004 prediction I did, which would have confirmed Bush was gonna win. So let's face it, I'm basically 4 for 5 so far. And even then, I'm harsh on myself. I don't go with my 2020 forecast that oh i was off but i still got the right outcome and blah blah blah, I grossly overpredicted that one for Biden and I screwed up. And I owned up to that, even if I got the outcome right. Because let's face it, that revamped prediction was a lot more accurate compared to the real result.
Now, onto the actual poll denial crap, because a lot of people are saying polls don't matter and can be wrong. Yes, they can. BUT....I'll still base my predictions on them because they're the best we actually got. What polls are are people going out and asking random voters questions in a systematic way to simulate the voter as if it were election day. There are hurdles to this. people not picking up phones, not being able to reach certain people, reaching too many other people, etc. It happens. But these guys tend to be very educated in statistics, tend to be aware of these shortcomings, and try to design their methods in ways that work, and they do the best that they can. This polls denialism reminds me of the weirdos who say because Fauci wasnt right on LITERALLY EVERYTHING AT THE TIME in 2020, that we shouldnt listen to anything he says. These are experts. And honestly, their predictions come with a certain level of transparency, and a certain margin of error. You know polls are generally accurate to within 3-5 points for each of the candidates numbers? That means that anything within 6-10% can technically be accurate. This is why my own methodology tends to be centered around a 4 point margin of error and I mostly look at states within 8 points. I could make it 6 but that might be too narrow (see rust belt polling in recent cycles). i could make it 10, but that might be too broad. 8 seems to be a good number, and I just take polling averages and assume the results will be within 4 points of that outcome, meaning it can be 4 points low, or 4 points high. And from there, I calculate the probabilities assuming a normal bell curve. Most polls are geared toward a 95% confidence level, which means 97.5% applied to a one tailed model, and that's where you see me get around 97.7% for my confidence level. Because actually 95% confidence is technically like 1.96 standard deviations, not 2, but yeah. You get the idea.
And it works, mostly. Not saying the polls are always dead on, but they are about as predictive as they can be when we don't know for sure. No one knows for sure what the exact outcome is gonna be, but polls are generally correct in 47-48 states each election cycle, and unless it comes down to the wire, they're gonna accurately predict the outcome.
To be fair, Lichtman's model has been pretty accurate outside of "down to the wire" predictions, but at least I'm not acting like he was never ever wrong. he had been wrong a few times, but all in all, I would say he's not more accurate than polls. And if I had to choose between polls and his "keys" model, I'm going with polls. I feel like the keys model is flawed, the keys are being framed in a way that arent quite lining up with reality, and honestly, polls are actually predicting voter behavior, while this is based on things merely correlated with it. It doesn't necessary apply to every single possible situation, and in a year like 2024, I think it's straight up wrong.
The only way the polls are gonna be completely and utterly off is if they did something funky with their methodology like all go the trafalgar 2020 model of just adding points to trump to account for "shy trump voters". Even then that turned out to actually be correct, and I owned up to it. Which is something I do that centrist libs NEVER seem to do. They just live in their delusional little reality where they're popular and everyone loves them and you're an idiot if you disagree with them. Then they get shocked when the country votes for an orange con artist over their extremely unlikable candidate. Do we need to do this again after 2016? If they're that stupid, maybe, while I'm gonna be here to tell them I told them so again, I won't be as happy about it this time (I had serious schadenfreude on election night in 2016, not gonna lie).
No comments:
Post a Comment