Nate Cohn

Nate Cohn @Nate_Cohn

Now, we do our polls the way we do them for a reason. It lets us get the partisan balance right, which is a big source of noise (and possibly bias) in other polls. It lets us reach low turnout voters, since we know who has or hasn't voted before we call them. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Finally, I would note that these are poll results of registered voters. These differences are not about turnout models or whatever. If there is a real, persistent difference it is due to the way we're measuring the population — PolitiTweet.org

Nate Cohn @Nate_Cohn

We also make sure we have a representative number of low turnout voters, who are really hard to reach and basically can't be properly sampled by most other pollsters. All of this is expensive, but it gives us a far better shot to capture the people who get missed by surveys — PolitiTweet.org

Nate Cohn @Nate_Cohn

One simple one is that we have the right number of reg. Dems, Reps., etc. in each region of every state, by race. So, when you look at this PA result, this is a reg D+9 sample, just like IRL, and our rural vote has the right number of rural Dems, reps, etc. — PolitiTweet.org

Nate Cohn @Nate_Cohn

--All that said, there's no way I can promise you that we're right--no pollster can do that. What I can say is that we do some things that other people don't do, that are pretty important in our view and I think are worth an added level of confidence in our results — PolitiTweet.org

Nate Cohn @Nate_Cohn

I think that gap is fairly manageable. A 3 pt difference is pretty well within the range where we expect polls to differ. I wouldn't insist either side is right or wrong, etc. — PolitiTweet.org

Nate Cohn @Nate_Cohn

--I do think the EC v. battleground gap is real, but shouldn't cover the whole thing. I'd guess that Biden was up 6-9 nationally, based on our polls alone. I think that's close to other national polls, but it is short of the average (Biden+10 or so). — PolitiTweet.org

Nate Cohn @Nate_Cohn

@RyanDEnos i didn't say we don't have any. but like, if we're 8 points off it's probably not sampling bias — PolitiTweet.org

Nate Cohn @Nate_Cohn

The ABC/Post poll otoh is fairly small for a national poll (~850 RVs). Doesn't make it wrong. But the difference could be sampling error on their end; if they did it again, it could be much better for Trump. It's worse for him than others, including their own prior polls — PolitiTweet.org

Nate Cohn @Nate_Cohn

--Our polls, together, are a very large sample. That means that, at least on our end, I can say with some confidence that we would not have a radically different result if we did this again. IE: if we're wrong, I can't dismiss that very easily as due to sampling or something — PolitiTweet.org

Nate Cohn @Nate_Cohn

Lots of questions about NYT/Siena v. ABC/Post in my feed this morning. I don't have any magical way of reconciling the two, but I do a few thoughts worth consideration — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @KevinQ: Today's Daily, featuring @jesscheung12's interviewer recordings from Siena College call centers, captures the humanity of polli… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Consider this chart, for ex, where AZ goes from worst to best for Dems as turnout goes from midterm to 100% https://t.co/1JURQKvoak — PolitiTweet.org

Nate Cohn @Nate_Cohn

As we wrote over this summer, this is what you should expect with polls of RVs now (and for this same reason, the national RV-battleground gap is even greater than LVs) RVs includes a bunch of low turnout, D Latinos in AZ; white, working class northerners https://t.co/fqz5FVsqlR — PolitiTweet.org

Nate Cohn @Nate_Cohn

@connoramulhern I guess so, but his weakness in AZ/FL in our poll gives me a little pause about whether his national support is distrributed the same way — PolitiTweet.org

Nate Cohn @Nate_Cohn

I did a quick look--it's hard bc of compositional changes (many hard to reach groups at end) to report a straight answer. But, controlling for strata and what we weight on, we did have movement against Trump in h2h and impeachment https://t.co/pFVWoWT5sP — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nickgourevitch well it's definitely far from the post poll! but i'd say the post poll is pretty far from most others on that front. fox was warren+5, nbc was +8 — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'd personally argue that it's closer to the polling average than these Post polls, but then again I'd note that our surveys were conducted from 10/13-10/26, and the president's standing has been slipping in other polls throughout the period--all the way to this newest batch — PolitiTweet.org

Nate Cohn @Nate_Cohn

For what it's worth, I'd guess our polls rea consistent with something like Biden +6 to +9 nationally among RVs, or Warren +2 to +5. If that's right--and that's an extrapolation to be fair--it would be close to the current national polling average, tho better for Trump — PolitiTweet.org

Nate Cohn @Nate_Cohn

It should, at least in terms of education weighting. The ABC/Post poll is a really good one, and I think they do RDD polling (random dialing) about as well as anyone https://t.co/4G4raxXwo5 — PolitiTweet.org

Nate Cohn @Nate_Cohn

My reaction at the time was to dismiss our polls as outlier-ish. There were only two, after all, and it was a big change from our prior surveys (4 and 7 pt shifts towards Trump). But we have a lot of data for analyzing this, and concluded later we were on-track for right reasons — PolitiTweet.org

Nate Cohn @Nate_Cohn

My reaction at the time was to our polls as outlier-ish. There were only two, after all, and it was a big change from our prior surveys (4 and 7 pt shifts towards Trump) — PolitiTweet.org

Nate Cohn @Nate_Cohn

A major reason why we went big on our own polling in 2018 is because we really thought our coverage, and all coverage, would have been better if we had more high quality polling in the battlegrounds (Our final poll also had Trump up in NC) https://t.co/jGtfwYCHj5 — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @PollsAndVotes: This was a fun comparison of @MULawPoll and NYT/Siena Wisconsin polls. Independent samples, very similar results. https:… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@IChotiner should note that you're cutting out 1/3 of the sample in each state, which adds some noise, and these subsamples not weighted to the 2018 election, which also adds noise. but overall, pretty close i'd say — PolitiTweet.org

Nate Cohn @Nate_Cohn

@IChotiner And, on request, the Biden lead among validated 2018 voters by state--many of which look quite familiar... AZ: Biden+1.2 FL: Biden+0.1 MI: Biden+9.6 NC: Trump+1.7 PA: Biden +3.2 WI: Biden +0.4 — PolitiTweet.org

Nate Cohn @Nate_Cohn

@IChotiner A few bonus tabs, corresponding to some points in the article. In Mich., Biden led among 2018 voters by 10 points, 51-41, and trailed among 2018 non-voters, 34-50--biggest gap in our polling (could be noise, small sample) Warren led among nonwhite men by just a 49 to 33 margin. — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @hames_cole: This Chotiner interview is great all around but this by @Nate_Cohn is something https://t.co/zFtQ8UwK5R — PolitiTweet.org

Nate Cohn @Nate_Cohn

Not the most cogent I've ever been, but hopefully @IChotiner pulled some new thoughts out of me that you might find interesting https://t.co/OTTJx2Tuic — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @IChotiner: New Interview: I talked to @Nate_Cohn about the different groups of Biden voters who aren’t sold on Warren, what the numbers… — PolitiTweet.org