Nate Cohn

Nate Cohn @Nate_Cohn

This leaves one major turnout question: did the Democrats enjoy a turnout advantage at the individual-level, perhaps due to higher turnout among well-educated voters? We're still aways from digging into that question — PolitiTweet.org

Nate Cohn @Nate_Cohn

The story in Pennsylvania is about the same. D+1.6 electorate (45.5-43.9) v. D+5.4 registration, thanks to 65% turnout among Democrats and 71% among Republicans on the L2 file — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @MarinersVintage: Nike (1996) Griffey in ‘96 campaign https://t.co/kjXZTm6DII — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob i didn't want to start a fight over whether RBS was prob lol — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob marquette/SSRS is prob internet. marist now has online panel/multi-modal stuff — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob though the prob stuff is mainly online now too — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 now, to your earlier thought -- yes, at the NYT it is *much* easier for me to apply an objective classification (prob/nonprb) with a very defensible analytical lens than for me (NYT) to declare polls good/bad subjectively and get into countless fights to defend it — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 i guess that would depend on your subjective gradations, so idk where we disagree, but on my subjective gradiations the story is basically the same whether it's a categorical or continuous scale — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 the central point of the article is that there is great variance among nonprob, not that they're all the same. that variance appears to be a function of their unifying trait -- no single sample frame --> crazy different samples -- which makes it appropriate to treat it together — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 my vague read of the room is that pollsters are more pesmisitic than they were when i wrote this, which i think is still my basic position here https://t.co/2h6LU3CXyN — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 i don't think it's misleading. almost everyone i talk to thinks there's real daylight between nonprob panels and online prob/phones. that's my independent read as well. i think what i wrote is consensus: nonprob can solid, but it's hard, can be terrible and it's hard to evaluate — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 yeah, i mean, that's the 'subjective classification' option. i don't think it's an option for me to make that kind of characterization without in-depth explanation, let alone on a chart — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 and unfortunately, i don't think we have anywhere near the level disclosure / understanding of best practices to be able to distinguish which nonprob polls are in which category -- which is part of why the spread in trump support has the potential to be so revelatory — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 in the interim, i think it remains entirely true that probability polls by rigorous firms weighted by standard demographics = presumptively pretty decent, while the same cannot be said for the nonprob polls (which can be good with great effort or absolute trash) — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nataliemj10 if you have a better way to show 'good' v. 'bad' polls on a chart like this, other than my subjective characterizations (which would call, say, YouGov Good but might call Suffolk BAD), I'm all ears — PolitiTweet.org

Nate Cohn @Nate_Cohn

Is Trump way up or way down? The polls are extraordinarily split, but the higher-quality surveys suggest he might be pretty weak. https://t.co/3ABGKid5VU https://t.co/ChO0jv7Gzc — PolitiTweet.org

Nate Cohn @Nate_Cohn

@tbonier that's not really how it works. i haven't done the analysis yet, but i fear the likelier explanation is that the sample was indeed deeply problematic and we were lucky that turnout and nonresponse biases largely cancelled out — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @LoganDobson: pro bowl punting from Mitch at least — PolitiTweet.org

Nate Cohn @Nate_Cohn

Do you regularly use chatGPT or another AI tool for work? — PolitiTweet.org

Nate Cohn @Nate_Cohn

@aaronstrauss @MichaelPodhorz2 yeah, absolutely. it's hard to turn this into a MAGA dummy. but clearly there were a series of competitive states where trump tried to overturn the last election and where MAGA types were running to do it for real, and dems did well in those states — PolitiTweet.org

Nate Cohn @Nate_Cohn

@aaronstrauss @MichaelPodhorz2 i do think national media discussion of crime (portland v. denver) could be the/a missing variable in the blue states where MAGA obviously wasn't at stake. but hard to model state effects on n=50 so i'm willing to accept an imperfect explanation — PolitiTweet.org

Nate Cohn @Nate_Cohn

@aaronstrauss @MichaelPodhorz2 i think there are very clear state effects downballot; i think they are loosely correlated with democracy/MAGA but it's not perfect (CO v. OR being an example) — PolitiTweet.org

Nate Cohn @Nate_Cohn

@Wertwhile didn't look at age — PolitiTweet.org

Nate Cohn @Nate_Cohn

We still don't have data from Pennsylvania, where I am interested to see whether we might get a different story. After that, the final/next question is whether there is a different story at the individual-level, perhaps due to educational polarization or something else — PolitiTweet.org

Nate Cohn @Nate_Cohn

We now have final turnout data for all of the jurisdictions with historic turnout data by party going back to at least 2004: NC, IA, LA, KY, Clark (NV), Pinellas (FL). Overall, the GOP enjoyed a turnout edge much like recent GOP midterms, with ~ R+5 electorate v. RVs https://t.co/ARZKpnukEl — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @Neil_Irwin: In the final three months of 2022, core inflation was running at a 3.1% annual rate, which is a little on the hot side but… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@baseballot and has anyone ever been disqualified for nondisclosure of funding? — PolitiTweet.org

Nate Cohn @Nate_Cohn

@baseballot (or nondisclosure of funding) — PolitiTweet.org

Nate Cohn @Nate_Cohn

@baseballot but no one is newly disqualified under it, i assume? is there a case where a pollster has been disqualified for nontransparency on weighting/mode/voter file source ( as opposed to non-methodological questions like release dates/call center location/sample size, etc) — PolitiTweet.org

Nate Cohn @Nate_Cohn

@baseballot are any prominent firms newly disqualified under this 'basic questions' standard? — PolitiTweet.org