| Ryuu96 said: And one of the best pollsters in Michigan has Harris ahead.
And the most accurate pollster for PA in 22 has Harris ahead.
Doesn't mean they'll be right but it's easy to find very, very reputable pollsters with Harris leading. Atlas Intel overestimated Marine Le Pen's vote share in the French elections by 15 points but because they got the 2020 elections correct, they're heavily weighted. They also missed the Chile Presidential elections by 10 points. Personally I think Atlas is too new to consider them highly reliable, 2020 could have just been them getting lucky. They aren't even good in their home country (Brazil) where they got Mayoral races way off. Look at their crosstabs for their recent poll on the 31st. For Wisconsin they have Trump winning 62% of the Black vote...Trump winning 40% of the Black vote in Michigan but Biden exit polls were 92% of the Black vote in Michigan in 2020. They have Trump at 24% of the Black vote in Pennsylvania while Biden exit polls had it at 92%...Yeah, I have big doubts here that they're getting Trump's black voter % correct. |
They also say Trump will win among women in Arizona, will win 25% of the black vote in Georgia, will win 49.7% of the Asian vote there, 48.5% of Urban voters in Michigan, 40% of the black vote there, and that Kamala Harris is barely winning among women.
The problem with fivethirtyeight is that it there's just not enough time. Say, 10 pollsters start up in the 2020 cycle. 5 of them are dead wrong. 2 of them are right because they are good pollsters. 3 of them are right because of pure luck.
Now, the 3 that were right based on luck can release polls every other day like Atlas Intel is doing and can singlehandedly skew the polls by at least half a point. And even though they're putting out things that clearly don't make sense. And they could to a large extent shape the narrative.
