heard an interesting theory about polling and I think it's probably true. Pollsters use the census to weight their polls and the 2020 census had 14 states with over/under counts and notable dips in Black, Latino, Indigenous, and Asian participation. Then throw in the partisan lean of Covid death.
Then throw in a drop in polling responses in the past few decades from over 30% to 6% and it ads up to the way we do polling in the United States getting increasingly worse as the decade progresses. The data for weighing it is all wrong, Covid 'removed from the voting rolls' a lot of conservatives
Nonresponse plays a big roll but probably not enough to tell the whole story and the weighing from the census being off and affecting weighting in these constituencies which are normally very left leaning constituencies, and you've got polls that just seem to get less accurate every year.
100.
The scary part tho is that the GOP knows all of this too — and remembers how much it rigged the 2020 census (and where), and so is targeting those districts where they know the polls are wrong.
This will mean last-mile fake newspapers, skewed local TV, and hyper-targeted social ads.
It used to be 30%? That seems reasonable to get good data but 6% is tough and i don't think that is the floor either. I saw one poll a few days ago that had a 1.6%.
