Nate Cohn

Nate Cohn @Nate_Cohn

Now I do think, fwiw, that the few 'good' polls we do have--like the recent Monmouth state polls in IA/GA--feel more consistent with Biden+8 or whatever than Biden+12. So it very well may be that the next 'good polls' show tightening as well. But we don't have that data yet — PolitiTweet.org

Nate Cohn @Nate_Cohn

Basically, the 'good polls' have been better for Biden than average (which hasn't always been true and may not always be!). When we go without those polls, Biden's lead shrinks as the more prevalent polls, somewhat better for Trump, control the average. https://t.co/DrapQWzacT — PolitiTweet.org

Nate Cohn @Nate_Cohn

This does have some consequences for poll aggregation, because right now pollster quality is closely correlated with pro-Biden house effects--a reversal from earlier in the year. So right now, these polling droughts shrink Biden's lead — PolitiTweet.org

Nate Cohn @Nate_Cohn

And that's why we find ourselves with the major news organizations conducting polls at roughly the same times: they all have similar incentives and want to make sure they hit many of the same points of the race https://t.co/5NHWy1tGmx — PolitiTweet.org

Nate Cohn @Nate_Cohn

If you lay out the calendar, most organizations are going to want a poll ahead of the conventions/after the VP pick--and that means you probably avoid the late July period. https://t.co/EVHfKY2XNi — PolitiTweet.org

Nate Cohn @Nate_Cohn

If you lay out the calendar, I think many organizations would want something ahead of the conventions in mid-August, which means you probably choose not to poll in late July — PolitiTweet.org

Nate Cohn @Nate_Cohn

It's a murky moment for evaluating the state of the race, since we've had very few high-quality polls in the last three weeks. But I certainly agree with this characterization of what we do have https://t.co/JwXo6y6tFQ — PolitiTweet.org

Nate Cohn @Nate_Cohn

@nickahamed yeah that's a real case — PolitiTweet.org

Nate Cohn @Nate_Cohn

@EsotericCD personally, i would never drop PA — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'm a little surprised by the dominance so far of the PA-WI-MI combo. Maybe I shouldn't be, after 2016. But it's the least diverse of the bunch (can't you figure out MI from PA/WI?; how do you guess NC/AZ?) and the least useful if the election leans Biden — PolitiTweet.org

Nate Cohn @Nate_Cohn

Alright let's eliminate PA, WI, MI, AZ and add one from the comments — PolitiTweet.org

Nate Cohn @Nate_Cohn

As you know, lots of people think about six key battleground states: PA, WI, MI, AZ, FL, NC. But imagine that it's the day before the election, you don't know the state of the race, and you only got polls of four states. Which would you want? — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @Phil_Mattingly: Schumer describes the meeting with the Postmaster General as “heated.” — PolitiTweet.org

Nate Cohn @Nate_Cohn

And since my tweet isn't 100% clear: 'move last debate to early September' proposal means eliminating the current final debate and adding one in early Sept, not that a new debate in early Sept is somehow 'final' — PolitiTweet.org

Nate Cohn @Nate_Cohn

And finally, if you accept the argument that the debates have more risk for Biden than Trump--which I think is reasonable--then moving that debate farther from the election may be in Biden's interest — PolitiTweet.org

Nate Cohn @Nate_Cohn

It is quite clear that new registration numbers are low this year, likely to the disadvantage of Biden. I'd guess that's in part because people are less likely to move right now. But others could use an early prompt — PolitiTweet.org

Nate Cohn @Nate_Cohn

Another argument: debates engage many low interest voters, prompting people to consider updating registration or applying for mail ballots well ahead of deadlines, while late Sept is against deadlines in many states — PolitiTweet.org

Nate Cohn @Nate_Cohn

One possible argument: the last debate is just ten days or so before the election. Some people won't make up their minds until after the last debate. If you're worried about mail ballot returns being too late, then you want them deciding--and mailing in ballots--earlier. — PolitiTweet.org

Nate Cohn @Nate_Cohn

There's definitely an argument that the debates are too late this cycle, and I actually can imagine the Biden folks even thinking it makes sense to move the last debate to early September https://t.co/Q3Z0ATJMHd — PolitiTweet.org

Nate Cohn @Nate_Cohn

There's definitely an argument that the debates are too late this cycle, and I actually can imagine them thinking it makes sense to move the last debate to early September https://t.co/Q3Z0ATJMHd — PolitiTweet.org

Nate Cohn @Nate_Cohn

With respect to the folks at Marist, they have not made an argument against the need to weight by education. They only said that it's insufficient, which is true but irrelevant. No one wouldn't weight by race because it's not enough without education! https://t.co/47gChYNq3z — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @juliaioffe: This is crazy footage from #Beirut. https://t.co/yFAvG7igEQ — PolitiTweet.org

Nate Cohn @Nate_Cohn

And this line is about as clear as it can get: "Failing to adjust for survey respondents’ education level is a disqualifying shortfall in present-day battleground and national polls." — PolitiTweet.org

Nate Cohn @Nate_Cohn

A good set of thoughts on how to think about polling from Courtney Kennedy. One thing I learned: Pew's weighting on *12* variables now! I thought we were a lot. https://t.co/Akfwx4TBfQ https://t.co/9aTL8tYL6U — PolitiTweet.org

Nate Cohn @Nate_Cohn

Dear crowd, what are all of the online national (or state) polls with 'probability samples,' like Pew ATP, the AP/NORC AmeriSpeak panel, or Ipsos Knowledge panel. Anything else? — PolitiTweet.org

Nate Cohn @Nate_Cohn

Interestingly, the LV screen works Biden's way in this poll and the race is tied in a 'low turnout LV' universe. Keep in mind that these LV screens are super noisy, but it is also possible that Biden could have at least a relative turnout edge in relatively white states — PolitiTweet.org

Nate Cohn @Nate_Cohn

Trump by 3 in Iowa, 48 to 45, per Monmouth https://t.co/8HFmNS4MRt — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik I mean I can conjure up an argument for Ernst being likelier to actually win as a result of elasticity or something, but in terms of predicted vote share I would have to think quite hard before betting that Ernst outruns Cornyn by a significant amount — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik Anyway, I've been wrong or not seen something coming many times before and perhaps this will be the next occasion for it! But this one just doesn't quite add up for me — PolitiTweet.org

Nate Cohn @Nate_Cohn

@kkondik McSally's not a typical incumbent, so that makes some sense to me. I do find Tillis odd, as well, but it's not like we're calling that race 'lean Dem' despite routine 7-9 pt leads for Cunningham. It's fundamentally in the same category as the presidential race, i'd think — PolitiTweet.org