Nate Cohn

Nate Cohn @Nate_Cohn

We've conducted two polls of Georgia, and we can join them up to the state absentee file to see who has voted and who hasn't yet. Here's how it breaks down: Mail: Biden 64, Trump 28 Early in-person: Trump 49, Biden 41 Hasn't voted yet: Trump 51, Biden 33 (emphasis on yet) — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @Redistrict: FL's early vote data tells us virtually nothing about the vote preferences of the final electorate. We do have plenty of ot… — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @adamrpearce: wanted to see how correlations between states differed between the 538 and Economist models so i built a little gizmo htt… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Major changes: all live caller; much more significant education weighting (maybe even too far), R+5 instead of R+6 on reg, and an increase in Latino share. — PolitiTweet.org

Nate Cohn @Nate_Cohn

FWIW, some very significant methodological changes under the hood on this poll, which basically rules out a direct comparison to their prior surveys as far as I'm concerned. But they're generally changes for the better https://t.co/JFztqgZ1Ty — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @OHPredictive: #AZPOP #POLL Biden has a marginal lead over Trump in Arizona. #BattlegroundAZ #2020Election Biden (D): 49% (+3) Trump (R… — PolitiTweet.org

Nate Cohn @Nate_Cohn

I was looking for this, but couldn't find it. If it's still their practice, that'll explain the difference https://t.co/5GhfGNthYb — PolitiTweet.org

Nate Cohn @Nate_Cohn

We stratify our sample on voter file race x party, and we weight on voter file race (not self report), so we don't wind up needing to upweight our Hispanic subsamples https://t.co/QtBf4W7mux — PolitiTweet.org

Nate Cohn @Nate_Cohn

All of our polls this year (skewed quite a bit toward purple/red states, so highly unrepresentative of Latinos nationwide, but for ex.): Self-reported Hispanic 53-32 Voter file Hispanic: 59-29 Both: 63-24 — PolitiTweet.org

Nate Cohn @Nate_Cohn

There's an entirely separate way you can define it, too: based on a model of Hispanic ethncity, based on surnames/geography. If Latino Decisions or other firms use that, and then screen on self-report after that, you get a *very* Dem group of Latino vts — PolitiTweet.org

Nate Cohn @Nate_Cohn

There's an entirely separate way you can define it, too: based on a model of Hispanic ethncity, based on surnames/geography, or based on self-report. If Latino Decisions or other firms use that, and then screen on self-report after that, you get a *very* Dem group of Latino vts — PolitiTweet.org

Nate Cohn @Nate_Cohn

@Taniel @NickRiccardi @ThePlumLineGS I'd push back on this year, fwiw. I know everyone's fixated on our TX poll, but our polls have been way better than average for Dems in Arizona — PolitiTweet.org

Nate Cohn @Nate_Cohn

And I'd keep in mind here that this isn't just a difficult group to poll, it's also a tough group to define. Question-wording matters, and different ways of defining the group can yield different results for both the size of the Latino vote and its vote choice — PolitiTweet.org

Nate Cohn @Nate_Cohn

Leaving methodological debates aside, I think it is difficult to evaluate which is 'right.' The exit polls wind up closer to the standard national polls, but LD criticizes the exit polls for a number of reasons, many of which are credible but, to me, don't fully explain the gap — PolitiTweet.org

Nate Cohn @Nate_Cohn

Leaving methodological debates aside, I think it is very difficult to evaluate which is 'right.' The exit polls wind up closer to the standard national polls, but LD criticizes the exit polls for a number of reasons, many of which are credible but don't explain the gap — PolitiTweet.org

Nate Cohn @Nate_Cohn

The difference in the *level* of Democratic support between Latino Decisions and the national live interview 'gold standard' type surveys has been true for a while. I haven't looked at the LD method in a long while (and can't find it quickly), so I'm going to refrain on 'why' — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'd start with the agreement: basically everyone has Trump's doing better than '16. LD, for ex., had 79-18 in '16; now it's 68-25. https://t.co/sS7KkJw92a — PolitiTweet.org

Nate Cohn @Nate_Cohn

As with national surveys and our other polls, Trump does hold up show well enough among Hispanics, with Biden up only 59-30. But Biden's doing better than Clinton among white voters, and even in this particularly diverse state that's enough for a comfortable edge — PolitiTweet.org

Nate Cohn @Nate_Cohn

The number of undecided voters is declining in our polls, including in the polls still out in the field. Here in Nevada, we're down to 4% undecided with another 5% on a minor party candidate or saying they wouldn't vote in the presidential race. — PolitiTweet.org

Nate Cohn @Nate_Cohn

This is our first survey after the final debate and no evidence of a shift. Post-RNC: 46-42 Post-debate one: 48-42 Post-debate two: 49-43 Voters narrowly thought Biden won, 38-35 No meaningful change in candidate favs. Trump approval down a tick, at 43-51 — PolitiTweet.org

Nate Cohn @Nate_Cohn

Joe Biden leads in Nevada, according to the final Times/Siena survey of the state. Biden 49, Trump 43 https://t.co/tazhwMwSM1 — PolitiTweet.org

Nate Cohn @Nate_Cohn

Well this is our first post-debate poll, so you'll have to wait and see. Our prior NV polls were Post-RNC: 46-42 Post-debate 1: 48-42 https://t.co/nO65X71Oge — PolitiTweet.org

Nate Cohn @Nate_Cohn

@DanRosenheck we do ask whether they voted — PolitiTweet.org

Nate Cohn @Nate_Cohn

@DanRosenheck overreporting turnout is certainly part of it, but because can vote in another state and not be captured here for ex — PolitiTweet.org

Nate Cohn @Nate_Cohn

I don't know, but do keep in mind that there are lots of different ways to cut this data. Among those who say they voted in '16, it's even. Among voters with a validated record of voting in '16 in NV, it's Clinton+2. https://t.co/8shdOISlqe — PolitiTweet.org

Nate Cohn @Nate_Cohn

Even; party ID and reg are both D+4 https://t.co/21bNfXrmMm — PolitiTweet.org

Nate Cohn @Nate_Cohn

Today at 1PM, NYT/Siena in Nevada Tomorrow, Michigan https://t.co/EWzv48tO2c https://t.co/LwzVmshPpe — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'd guess they can't play serious defense--and Bloomberg could always reraise anyway--which makes it all the more dangerous https://t.co/utYnxlxRlc — PolitiTweet.org

Nate Cohn @Nate_Cohn

In our poll, Biden's up 7 among early voters. If you need to try and overcome that kind of a deficit on Election Day, in on the assumption that you can match the turnout we see in early voting, I don't think you want a bunch of negative ads at the last second — PolitiTweet.org

Nate Cohn @Nate_Cohn

There are also a handful of undecided and minor party voters who you might hope to lure over. In our poll, they disapproved of the president. So there's some upside there too. And that's leaving aside the unusual and huge uncertainty about what the turnout looks like — PolitiTweet.org