So, we're coming to election time, and we're getting to the point where we are all started to develop predictions of what will happen. 538 had a model with a 52% 50% chance that Biden will win. The Hill thinks there's a 56% chance Trump will win. The economist has a 2 out of 3 chance Trump will win the election. My own latest full prediction has Trump at a 72% chance of winning, and yes, before people ask, I still consider that prediction GENERALLY valid. PA is still at 2.3% ahead for Trump, it's still the swing state, and while a lot of little details have changed that I'm going to want to discuss, I feel like if I update now, it will screw up my timing for my next update, which I want to be on July 12th, right after Trump's sentencing. But yeah, the point is, everyone has an opinion. And some opinions are more well founded than others.
My own predictions come entirely from polling. People often question this, and I always get people online who are like "HAHAHA YOU THINK POLLS ARE ACCURATE? BUT WHAT ABOUT X?" while condescending to me about "the fundamentals" or their idea of how X state is ironclad and will never flip. What do I have to say to these people?
Well, quite frankly, I think you're all idiots. I hate to get all elitist and credentialist here as it's kinda argument from authorityish, but I literally studied polling and polling methodologies in college. I trust the experts to actually go out and measure how voters think. Will they be dead on? No, and that's okay. Polling isn't ever going to be exact. They always come with a margin of error, and a range of reasonable outcomes that align with the prediction, and so far, the polling hasn't really been SO WRONG that it's outside of the margin of error. You gotta understand, you can be accurate, but that doesn't mean you're precise. Polling is ACCURATE, but people rip it for not being precise. If you assume a 4 point margin of error, any poll within 8 points is technically accurate. But if the poll was off by 6 points, it wasn't very precise, now was it? It's technically accurate. It's within the expected range of results, but that doesn't mean it's dead on.
There has been controversy about polling recently, and how in the modern era, it seems to be off. There are reasons for this. I mean, are shy trump voters a thing? Do certain demographics not pick up the phone? How do we make the polls representative of the general population? And that's what polling experts have to figure out. How to get a result that aligns with the voting behavior of the general population. They don't always get it dead on, but they get it generally accurate, and even stuff like 2016, 2020, and 2022 mid terms, which constantly come under fire, are GENERALLY accurate.
The problem is, my dear reader, is that if you don't agree with the polls because "they're often wrong", then you don't understand probability and statistics, and your opinion isn't worth dealing with.
So why don't I go by "the fundamentals"? Because fundamentals can be wrong, too. They assume a bunch of things too. You can look at the demographics of a state, think something is gonna happen, but then it doesn't. You can assume certain levels of reasonableness in an electorate that doesn't value reason. I think this is especially relevant to 2016 and our post 2016 environment. So many establishment democrats, for example, think that because "the economy" is "good" that they're gonna win. But then people don't FEEL like it's good and vote the other way. Dems need to learn to understand peoples' pain and resentment and adjust themselves to such things. But they often don't. And I always get these really snotty neolibs online who think they're so much smarter than everyone else (I know, I know, pot meet kettle moment) who are like "HAHAHA YOU LISTEN TO POLLS? *CLOWN EMOJI* and then treat me like a stupid trumper simply for thinking Trump is gonna win. I mean, dude, I don't even like Trump, and even though I hate you and your elitist guts as an anti establishment progressive, I'm actually on your side. So please shut the heck up. You're embarrassing yourself.
Honestly, I do recall trying to play with things like fundamentals in early elections like 2008 and 2012, and I actually got a lot of stuff wrong. I assumed virginia would go red in 2012 for example. I admit the polling was an either way thing, but what tipped me to romney was the fact that it was historically a red state up to that point. Same with north carolina in 2008. Heck looking at it now, its suburbanization hasnt really made it reliable for dems.
And honestly, this election cycle seems to really lean toward Trump. I'm seeing stuff I never thought I'd see. But let's be honest. Is anything I'm seeing impossible? No, it's just what happens when you shift the entire electoral map around 4-6 points to the right of where it was in 2020. You start seeing states that you deem safe, like virginia which HAS gone hard blue in recent years, kinda shift back to swing state territory. You start seeing states like Florida start to not really be swing states at all, or be borderline swing.
I actually had someone laugh at me because I said virginia was more in play with florida. The idea broke their brain. But florida has gone hard red in recent years, and we're, again, in an environment where we're a good 4-6 points even redder than 2020, so 8 points isn't inaccurate. In 2008 and 2012, it was super swingy. In 2016, i started becoming more reliably dem. 2020 it was solid dem. But again, environment is now 4-6 points more blue. So now it's a swing state again. What is so unbelievable about that?
Here's the thing. Everyone has a mental model in their head about how things SHOULD be, and then there's what the data is actually telling us. Your mental model can be like HAHAHA THATS A SAFE STATE THAT CAN'T FLIP. Uh....yeah you realize the electoral map is constantly changing, right? In the 1980s it went hard red every time. I mean, under Reagan and Bush Sr., we saw the GOP SWEEP the entire electoral college, getting 400-500+ electoral votes. It was a BLOWOUT. Popular vote was around +20R. Then in the 90s, things started evening out. The dems won around 300-400ish, and a lot of the south ended up going blue. By 2000, this stopped, which was why Al Gore lost narrowly. 2000 and 2004 had similar maps. Then 2008 was more of a clintonesque map, but with no southern states, and the rust belt being the big strength for dems. 2012 had a similar outcome, but narrower. 2016 was expected to follow similar trends, but the rust belt went red. That wasn't expected by anyone. People thought clinton had that locked down. If anything my analysis of polls gave us the first indication something was up. I couldnt predict the rust belt defections exactly, but they were...again, within the realm of realistic possible outcomes. 2020, even with my prediction being "off" due to my weird methodology quirks that time, it was still within the realm of possibility, and if I had predicted it PROPERLY, i would've been a lot closer.
So yeah. Let's dispense with this nonsense that polls are wrong. They're not wrong, you're just not reading them properly and understanding their limitations. They never advertise themselves as perfect. They all come with margins of error, and certain probability ranges. Anyone who expects a poll to be 100% dead on accurate doesn't understand polling. And I'm tired of arguing with such people. Update your understanding of reality by looking at the evidence. Don't just assume reality to conform to what you think.
No comments:
Post a Comment