Nate Cohn

Nate Cohn @Nate_Cohn

We've already released one interesting bit about the sample in these states yesterday: self-reported party ID remains at D+1, as in October (better than the exits in '16), with a 2016 recalled vote of Trump+2.5 among validated '16 voters (v. 1.5 IRL) — PolitiTweet.org

Nate Cohn @Nate_Cohn

Tomorrow AM, we'll release polls in the big six battlegrounds: PA, WI, MI, AZ, FL, NC In October, we found Biden+2 on average across these six states — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @ForecasterEnten: The state polling is entirely consistent with a 10 pt lead for Biden nationally... I think for many folks... Their gut… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@RachelBitecofer all of these results are among RVs and not subject to choices in turnout modeling. it's just the industry-standard cps figures. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Biden+9 in Wisconsin (51-42 w leaners), per Marquette, would stand out under any circumstances but it is particularly noteworthy v. the consistently close race in their prior six surveys: Biden+3, +3, Even, +4, +1, Trump+3 — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @MULawPoll: New Marquette Law School Poll finds Biden widening his lead over Trump in WI. Results put Biden at 49%, Trump at 41%. In ear… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Should be interesting https://t.co/imFOhFGaPQ — PolitiTweet.org

Nate Cohn @Nate_Cohn

Perhaps one small bit of consolation for the president is that his deficit is only 13 points using the unrounded margins (49.7 to 36.3) — PolitiTweet.org

Nate Cohn @Nate_Cohn

It's also worth noting that the president's 41-56 approval rating is quite consistent with other polls and, if anything, somewhat better than what else we're seeing this AM over appx. the same field period https://t.co/09zyFznN1k — PolitiTweet.org

Nate Cohn @Nate_Cohn

@RyanDEnos no i'm not set up to do the unserious ones quickly lol — PolitiTweet.org

Nate Cohn @Nate_Cohn

We could go deep on the numbers here, but bottom line is that the votes just weren't there for Trump. That could change, but you can't weight or LV screen your way out of being barely ahead among white voters and down 20 with independents or whatever — PolitiTweet.org

Nate Cohn @Nate_Cohn

We'll talk more about turnout later, but 2016 voters (even better for Trump than 2016 turnout, as that would exclude, say, everyone 18-22 or everyone who voted in the Dem primary that didn't vote in '16) is Biden+10 https://t.co/501osgQ2j5 — PolitiTweet.org

Nate Cohn @Nate_Cohn

Just out of academic curiosity, I spent some time yesterday seeing whether there were really any serious choices in weighting or design that might have led to even a modestly better picture for the president, and I've got nothing — PolitiTweet.org

Nate Cohn @Nate_Cohn

Joe Biden holds a commanding 14 point lead in the first NYT/Siena national survey, 50 to 36 percent https://t.co/fnGaZUHo0F — PolitiTweet.org

Nate Cohn @Nate_Cohn

Joe Biden holds a commanding 14 point lead in the first NYT/Siena poll, 50 to 36 percent https://t.co/fnGaZUHo0F — PolitiTweet.org

Nate Cohn @Nate_Cohn

@rp_griffin refusals/Dk — PolitiTweet.org

Nate Cohn @Nate_Cohn

@ElectProject both — PolitiTweet.org

Nate Cohn @Nate_Cohn

@ElectProject we have the full file and draw the numbers ourselves, and we do so within some deeply interacted strata (described in the method) to adjust for coverage — PolitiTweet.org

Nate Cohn @Nate_Cohn

@ArmenianBacon @Bouje99 i can imagine a case where we would do so but it is not necessary here; design effect is reported — PolitiTweet.org

Nate Cohn @Nate_Cohn

Yup https://t.co/cSxHp7oS9k — PolitiTweet.org

Nate Cohn @Nate_Cohn

@BenPiggot right; often times the difference was really staggering. and tbh if i didn't think that was the answer, we would have made changes or we just wouldn't poll the states. — PolitiTweet.org

Nate Cohn @Nate_Cohn

@Bouje99 we'd give those 10 people much more weight, such that they represented 60 percent of our estimate even though they were 10 percent of the sample. hopefully such a dramatic weight would not be necessary IRL! — PolitiTweet.org

Nate Cohn @Nate_Cohn

@BenPiggot In 2018 we definitely underestimated Democrats in heavily Latino districts, but in our post election analysis it was pretty clearly due to underestimation of turnout not inability to reach Hispanic voters. And after all, our best poll for Biden last October was AZ. — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kabir_here self report — PolitiTweet.org

Nate Cohn @Nate_Cohn

Looking at that exact question, here it seems our validated 2016 voters in Arizona are R+13 by party reg, while 2016 voters on the file are R+6. Either way, the sample winds up at R+2 or 3 or whatever, but our voters appear to be more R and nonvoters too D. — PolitiTweet.org

Nate Cohn @Nate_Cohn

IDK. It was true to some extent in Oct. as well (Trump 45, Clinton 37 in AZ). Keep that the poll is not designed to be representative of the 2016 electorate. For ex: we have the right reg Dems/Reps, but our 2016 voters could be too R and nonvoters too D https://t.co/3tvDyL4ftR — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @15c3PO: @Nate_Cohn I see this wasn't an idle threat https://t.co/CuppTC3Fbu — PolitiTweet.org

Nate Cohn @Nate_Cohn

The method also includes a full description of the nature of the sample, including fun things like this https://t.co/26EgCva4l6 https://t.co/KFHGhBl76K — PolitiTweet.org

Nate Cohn @Nate_Cohn

Tomorrow, NYT/Siena national survey Thursday, NYT/Siena surveys in MI, PA, WI, FL, NC, AZ Today, methodology https://t.co/OF0jk3Qb0C — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @WilcoxNMP: Whoa. *Happiness falls to record 50-year-low.* 👇🏽 New COVID Study from @NORCNews https://t.co/G69nPWEBg7 https://t.co/RpxFn… — PolitiTweet.org