Guys. I am not a quant person but even I know that pushing a sub sample of a larger phone poll from early in June with a 0.8% response rate, where no one who ignores unlabeled numbers on their cellphone responded, tells you nothing about the electorate. Stop letting bad polls drive your politics.
The “weighted to account for response rates” is hiding a lot of bullshit here because if this is the poll I’m thinking of like 80% of the respondents were over 65, so they’re inflating the like 5% of people under 30 who answered into 35% of the poll response
Whenever I have seen this language in academic survey work it means they have an assumption of what the ideal response rate is per demographic and then they arbitrarily apply multiples to the actual responses to match that assumption.
It is a red flag that the survey vibes, not science.
