GA: Insider Advantage: Sen. Warnock (D) +2 (rounded: +3)

[wbrocks67]

LMAO they're literally only returning to herd so that they don't get an F rating next time on 538.

[wbrocks67]

Like, these 18-39, White, Independent, Black numbers are all still ridiculous. I feel like they essentially do the same thing that Traflagar "does" - have a topline # in mind that they want to get to, and then fudge where they can to make it work.

They can't seriously still be having Democrats losing 18-39 year olds after all of their messy polls in the midterms... come on now.

[wbrocks67]

A few comments from a Georgia numbers guy:

https://twitter.com/stphnfwlr/status/1599868147191848960

I think it's more just a sign that the poll is fake.

Or just a symptom of low response rates leading to the same sort of systemically biased crosstab issues we see in many polls.

Although I guess in some sense that makes a poll "fake," but in that sense basically all polls are fake, because they basically all have these problems (especially Rs doing better with supposed minorities and young voters than in reality).

Eh, I'm starting to wonder if places like Landmark, IA, Trafalgar do really just cook the books sometimes. They all have the *same* issues (Reps always doing better with Blacks, young voters, etc.) There's no way it's uniform across all those outfits consistently unless something was seriously off.

[wbrocks67]

Eh, I'm starting to wonder if places like Landmark, IA, Trafalgar do really just cook the books sometimes. They all have the *same* issues (Reps always doing better with Blacks, young voters, etc.) There's no way it's uniform across all those outfits consistently unless something was seriously off.

Those specific issues with those specific crosstabs have been a pattern for non-hack pollsters also though.

True, but not to the extent of these ones though. It's just like IA, who consistently, poll after poll, had the Republican winning "other" (hispanic/asian) race like 75-15 in a ton of polls. It was so ridiculously consistent that there's just no way you could come up with that # over and over again unless you were fudging things to come out to the desired overall result.