Quote:
Originally Posted by dino7c
lol okay guy, it takes two weeks to figure it out when we see the results
|
It takes two weeks to know if there's a correlation this year.
It takes much longer to know if this correlation is strong enough to have predictive value
As for polls becoming more even, while it might be real movement, it could also be just the results of polling fatigue. 538 mentioned weeks ago that response rates to polls are really low and have been dropping as we get closer to the election. (This obviously isn't somethint pollsters are eager to talk about, since they get paid to do those polls, no matter the response rates.)
I have no idea how much low response rates affect the polls, but I would imagine that the fewer respondees there are, the more likely it is that there are significant hidden differences between people who answer polls and voters in general.
Since the doesn't-answer-to-pollsters demographic is currently almost all the voters (latest number I heard was 97% and it's likely dropped further since that), they don't have to differ much from the will-answer-to-pollsters demographic to make the polls borderline meaningless.
Which is why I think it's extremely valid to look at other data and consider it as good or better evidence.
(It is obviously possible to still have good polls with low response rates. My point is mostly that really it's impossible to know. To me this feels like an election where the polls could be way off, but that's just vibes.)