Nate Cohn

Nate Cohn @Nate_Cohn

But when you go from Biden plus 5 or 6, like in April or March on average, to 8 or so--where we've been in high quality polls in May/June--that's, for me, where you go from 'this race is probably still pretty close' to 'Biden's probably got a comfortable lead.' — PolitiTweet.org

Nate Cohn @Nate_Cohn

It is worth remembering, of course, that "Biden+X" among registered voters nationwide is probably not as good for him as it looks. There's a probable LV-RV advantage for Trump--unusually large in this poll, fwiw. There's the battleground v. national issue. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Biden up 10 among RVs in the new ABC/Post poll. It's been a pretty sparse two months of polling, but the high quality stuff--however few and far between--has been strong for Biden https://t.co/PQUHuOKXOw — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @IChotiner: New Interview: I talked to political scientist Omar Wasow about how different kinds of political protest can impact election… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@virginialearns welcome back! — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @buitengebieden_: 22 oversized teddy bears in a rollercoaster. It really looks like they are coming to life.. 🧸 Walibi Amusement Park,… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik @jon_m_rob @AlanIAbramowitz when i look back at our old '16 polls, and see who is no longer registered, the 'departing the electorate' group sure seems to lean Dem in terms of self-reported vote. lots of mobile young folks. again, maybe it's only a half point of bias. but this stuff can add up. — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik @jon_m_rob @AlanIAbramowitz and you've got an issue with the target too. do we really know the vote choice of people who remain registered in WI, factoring in moving, deaths, etc. — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik @jon_m_rob @AlanIAbramowitz you've also got other pollsters getting huge numbers of 'some other candidate,' and then downweighting them from like 13 to 4 percent of the electorate. probably not appropriate. — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik @jon_m_rob @AlanIAbramowitz there's typically a lot. sure seems like there's less. but in our october polling, the 4% of validated '16 voters who said they 'don't know' backed biden by at two-to-one margin, so not sure what you do that. maybe it's only a point or two of bias but it could be there — PolitiTweet.org

Nate Cohn @Nate_Cohn

@PatrickRuffini for me the higher level of crossover support says most of what i need to know, given what we see in election results — PolitiTweet.org

Nate Cohn @Nate_Cohn

No matter how you cut the data, Biden's lead among black voters appears diminished since '16. It is less clear to me whether it's because of a narrow and deep decline in Biden share among young black voters, or a perhaps broader but less dramatic uptick in Trump support — PolitiTweet.org

Nate Cohn @Nate_Cohn

A slightly different picture on the comparison with 2016 if you use national telephone surveys rather than the CCES. By that measure, Clinton led 72 to 5 percent among 18-29 year old black RVs, or not vastly different than Biden's apparent standing https://t.co/ljpqID0nme — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @ByMikeBaker: NEW: Coronavirus deaths in the United States have now surpassed 100,000. Deaths by date they were reported: Feb 27: 0 M… — PolitiTweet.org

Nate Cohn @Nate_Cohn

neither not getting nearly enough love — PolitiTweet.org

Nate Cohn @Nate_Cohn

If you're the Biden campaign, would you make meaningful initial investments in Texas, Ohio, both, or neither? — PolitiTweet.org

Nate Cohn @Nate_Cohn

quite possible (and true in our polls from last october that you're getting tired of hearing about) https://t.co/N3wo9kFE8I — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jon_m_rob @A_agadjanian @cwarshaw perhaps there will be rbs national polling to facilitate it — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @nytimes: Defense lawyers in California are reviewing criminal cases in which Tara Reade served as an expert witness on domestic violenc… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Whatever their strengths and weaknesses, one interesting thing about Quinnipiac and Fox is that they're the only pollsters who have semi-recently conducted methodologically consistent, nearly simultaneous surveys both nationally and in the most crucial states. https://t.co/GLBxw9knUo — PolitiTweet.org

Nate Cohn @Nate_Cohn

@PollsAndVotes @MULawPoll what's that in pct? biden margin among almost certain/very likely v. rest? (or whatever your standard screen is?) — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @IChotiner: New Interview: I talked to Ben Stiller about the life and legacy of his father Jerry, the origins of Frank Costanza, and why… — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @NYTnickc: fwiw: The RNC has been mailing voters in Pennsylvania absentee ballot applications. They include applications and envelopes p… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@doug_rivers i'm comparing MRP-based estimate, using pre-election national RDD polls from 10/2016, to an RCP-style unweighted average of state polls over the final weeks — PolitiTweet.org

Nate Cohn @Nate_Cohn

@jbview i could say the same thing for the nation. but my point isn't about utility, it's whether they're dismissively bad. i have no reason to think a battleground poll can't represent battlegrounds, even if there's an interpretive challenge. i don't think CNN's subsample gets that far — PolitiTweet.org

Nate Cohn @Nate_Cohn

@ArmandoNDK lol right, the campaigns are all morons too. and surely it's an improvement over a national poll, which i assume you don't consider 'truly moronic' — PolitiTweet.org

Nate Cohn @Nate_Cohn

@cwarshaw depends on the survey design to some extent, but subnational weighting would improve it for sure — PolitiTweet.org

Nate Cohn @Nate_Cohn

If CNN did a real battleground poll of a 1200 people or whatever designed to represent the battlegrounds, just as their national polls are designed to represent the nation, I'd totally pay attention to that — PolitiTweet.org

Nate Cohn @Nate_Cohn

As an example: if you do a real 'battleground poll,' you weight your sample to match the characteristics of the battleground. If you do a national poll, you weight it to the country; the battleground subsample was not weighted to ensure it's representative of the battleground — PolitiTweet.org

Nate Cohn @Nate_Cohn

As an aside, my issue with the CNN battleground subsample wasn't that it was a battleground poll. I'd guess I like battleground polls more than most. My issue is that it's not a battleground poll; it's a subsample, and it was getting treated like a poll — PolitiTweet.org