Tweet by @Nate_Cohn | PolitiTweet

Nate Cohn @Nate_Cohn

Marist appears to do two things: --they weight by region, which is helpful but insufficient --I *think* they stopped calling out of state cells, which is hard to defend but *could* cancel out not weighting by edu. The hope is two wrongs make a right? https://t.co/8rFBVn5Hb4 — PolitiTweet.org

@Nate_Cohn Can you speak a bit ut more to whatever Marist does that they claim makes the poll the same as weighting… https://t.co/7feSoQM9sU

Posted Sept. 9, 2020 Hibernated

Preceded By

Nate Cohn @Nate_Cohn

Whatever you think of the USC poll overall, I would certainly take its change over time seriously. It's a panel survey, so most changes aren't due to changes in the sample, and shift in USC predicted change in the average in '16 pretty well — PolitiTweet.org

Posted Sept. 9, 2020 Hibernated

Followed By

Nate Cohn @Nate_Cohn

Yes: random, unbiased sampling creates noisy subgroups that cancel out. Biased methodological choices, OTOH, may not cancel out. https://t.co/1emtygwg7A — PolitiTweet.org

@Nate_Cohn Is that really different then telling people not to look at cross tabs Bc they cancel out ?

Posted Sept. 9, 2020 Hibernated