I would really like to see the age breakdown some of these polls are getting for their voters of color respondents
if it’s hard to reach young people and it’s hard to reach Black people then potentially you’re getting an absurd amount of leverage off one respondent who says he’s both
like, basically what I want to know is if you change one single respondent’s answers, what’s the maximum amount your toplines move. and for some of these poll designs I’m worried the answer is “two and a half points”
wasn't there a rolling panel in 2016 that had wildly fluctuating numbers in MI depending on whether a single Trump supporting black voter in Detroit was on the panel that week?
