So, after doing some analysis on RCP vs 538, I have concluded, I will trust RCP for now. Despite claims of partisan lean, they've never been wrong more than anyone else. if anything, I was able to outperform 538's predictions largely because of them.
The last time we had this debate, it was 2020, and RCP had a claim of partisan bias there. 538 was deemed as more objective, and academically, I actually agreed with them. but they were wrong and I was wrong too because despite following their lead, I tried messing with RCP's averages to "unskew" the results, and it made me way off.
In 2016, 538 gets credit for being thinking Clinton had a 70% chance...I had a 56% chance with RCP. Who was more accurate? Well, Trump won so figure it out yourself.
In 2022, they were claimed to be wrong on the senate, but I got my data from there. I guessed a 49D-51R outcome based on their data...and so did 538, actually. Yeah the polls went in a bluer direction, but let's be honest, The PA and GA races were tossups and basically coin flips. Arizona was lean R and I got that wrong, but eh, if everyone else guessed 49D-51R, well, I doubt that's RCP's fault.
The fact is, despite accusations of partisan bias (and in their journalism they ARE biased AF), their polling has largely been as accurate as anyone else's, while 538 has actually had issues with providing copium for resistance liberals only for them to be freaked out on election day when reality doesn't conform with their models.
As such, idk. If anyone seems to actually be off more consistently, it's 538...in the dems' direction.
As such, who should I trust for polling? Well, much like 2020, I do have questions about RCP's methods. And I do understand their results are slightly more R leaning. They're not crazy R, as we can tell where it matters most differences are less than a point, but of the four forecasts I compiled, they are the most right leaning one. But....the real question is...who is right? Methodologically, I probably would, much like in 2020, be inclined to buy into 538 and them having more polls included, including liberal polls RCP doesnt include. But going back to 2020 predictions, a lot of those pollsters were also largely excluded from RCP's averages and they...ended up being more right.
So idk. It probably doesn't matter a ton, but I do need to take claims of bias seriously, and do want to explain why I do what I do for full transparency's sake. But I will use RCP for the rest of this election cycle. We can do a post mortem afterwards, but I don't want to change methods going into late October only for it to blow up in my face on election day like I did in 2020. I said at the start of this cycle, I would just report RCP's averages as is and make predictions based on them, and do that I shall. Whether it ends up being a smart decision remains to be seen.
To be fair, a one point difference is going to mean that vs 538 alone, it doesn't matter much. In the larger commuinity, it could make a difference. Race to the white house was even MORE D leaning, and despite them seeming quite academic up front, I think their forecast is TOO optimistic.
Meanwhile other sites are just incomplete, and the other one i did look at (the hill) was actually almost as R leaning as RCP. So...yeah. We can debate which poll aggregator is best, but they all seem to exist on a spectrum with partisan lean. And generally Im most interested in accuracy. And RCP has never actually let me down, even in recent years with all of their accusations of being trump leaning. If anything the more left leaning sites just end up being copium for resistance liberals who imagine themselves smarter than everyone else and they just want a set of facts that conforms with their take on things. And let's face is, those guys can be as delusional as the trumpers sometimes.
Like really...why does polling have to turn into cheerleading? The point is to try to figure out the truth, not to just deal copium for your own side. I mean, havent we learned from the republicans in 2012? They did that. And then Rachel Maddow had to deliver that awesome dose of reality to them on election night. Now resistance liberals do that with polling and uh...yeah. Remember how that could've worked out with us if we stuck with biden? We're in better shape now, but we're not great. Either way, listening to copium just deludes us. it doesnt actually help us come election day.
No comments:
Post a Comment