So, just wanted to get a couple more thoughts out there, but I kind of am walking back my bashing of RCP and praise of 38 at least somewhat/ 538 often doesnt just do straight polling averages, they often weight things as they see fit and that throws averages off (like they did in 2020). Sure i might like that they include a broader range of polls than RCP, but thir weighting sucks and is often more D leaning than elections actually end up being.
Also, analyzing my own forecast, I kind of realize what's making my RCP based one so right leaning is actually Michigan. That one randomly flipped like 2 points in 1 day and its thrown off my forecast ever since. If Michigan wasnt such a weird outlier where it started polling like sun belt states do, I would be getting 45-55 with PA being the deciding state, which is a bit more moderate an outcome than 41-59. I admit that 41-59 and 45-55 arent that much different, but neither is 49-51 from 45-55. And neither is 56-44 from 49-51. But 56-44 and 41-59 are much different. Because it's a spectrum. ANd as we can see on the map, its the difference between pale red and pale blue. So...yeah. Quite frankly, Michigan's erratic behavior is RCP's fault entirely, no other aggregator has it that red, but yeah, that one state is currently skewing my entire forecast.
It could just be a temporary artifact that resolves itself as we get more polling, or it could be RCP dropping the ball hard. We'll see how things turn out in 2.5 weeks.
No comments:
Post a Comment