Nate Cohn

Nate Cohn @Nate_Cohn

@Redistrict @great_northland hyde park tho — PolitiTweet.org

Nate Cohn @Nate_Cohn

@MetaAdamJohnson because, alternately, we're not reaching the most trump-leaning demographic: old white men. but we'll put it to the test and see. perhaps it'll be larger than i think — PolitiTweet.org

Nate Cohn @Nate_Cohn

@en_bp should work, though i don't have parameters ready for that off hand — PolitiTweet.org

Nate Cohn @Nate_Cohn

In the past, we've targeted this on the front-end, so we were already calling 59% women among 18-29. But since we used '18 data for this last round of polling, we were behind the curve: as <40 landline % from small to zero, the number of calls to women you'd need goes up — PolitiTweet.org

Nate Cohn @Nate_Cohn

There are ways around this--whether it's to weight or stratify on age x gender. I'd guess there's a negligible effect on the topline, but there would certainly be a result on the generation gap in the tabs. — PolitiTweet.org

Nate Cohn @Nate_Cohn

As a result, our young respondents are very male while our older respondents are more balanced. Weighting by gender gets women, over all, back to the right place, but without regard for this pattern. So you end up with a very female group of 65+, and relatively male young voters. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Basically in our polls--and, it turns out, other polls--the cell phone respondents are heavily male, at all age groups. But cells make up nearly all young voters, and less than half of >65 respondents. — PolitiTweet.org

Nate Cohn @Nate_Cohn

I've been doing some fairly deep poll methodology dives in recent weeks--more later--and one thing I've learned (that I don't think I'm first to piece together) is that there's a very interesting response by gender by cell phone and, as a consequence, by age — PolitiTweet.org

Nate Cohn @Nate_Cohn

@RonBrownstein @PollsAndVotes @WisVoter our poll had biden+3 back in october, with white no college women at biden+4. so, it would seem he could easily win while losing non-col women. i know you're not sold on this, but exit-to-poll do not work well! https://t.co/0wUsSvJxIz — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @kwcollins: If folks on the left are concerned that Trump has a very real chance of being re-elected, don't look to UK election results;… — PolitiTweet.org

Nate Cohn @Nate_Cohn

How wrong would the exit poll need to be for conservatives to fall short of a majority? Well, it would need to be the worst result since the first BBC exit poll And fwiw they've been doing quite well recently https://t.co/6lFyhOf19M — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @MULawPoll: The net approval-disapproval of 47-50 matches Trump’s best rating since taking office, that from Oct. 24-28, 2018. — PolitiTweet.org

Nate Cohn @Nate_Cohn

In Wisconsin... https://t.co/6GDat9M6zf — PolitiTweet.org

Nate Cohn @Nate_Cohn

@JHagner and then a big michigan poll... well you can do the math — PolitiTweet.org

Nate Cohn @Nate_Cohn

@JHagner i kind of get that. but it's not just hard--based on the data we collected in october, doing michigan the way we would need to do it might cost like 3x a typical state poll — PolitiTweet.org

Nate Cohn @Nate_Cohn

@JHagner do you think that michigan is particularly interesting right now, or do you just think we don't know anything — PolitiTweet.org

Nate Cohn @Nate_Cohn

@JHagner ok would you rather have 1 high quality 1200 person survey of Michigan, or N=600 surveys of each of PA, AZ, WI, FL — PolitiTweet.org

Nate Cohn @Nate_Cohn

Would you prefer two state polls at N=600, or one N=1200 state poll? (assuming fairly equal levels of competitiveness/significance) — PolitiTweet.org

Nate Cohn @Nate_Cohn

Do you prefer two state polls at N=600, or an N=1200 survey of the six closest states carried by Trump (PA/MI/WI/FL/NC/AZ) — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @jennymedina: Bookmark this: "Perhaps the best thing Mr. Biden has going for him now on Super Tuesday is that California takes a long ti… — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @HirokoTabuchi: To the naked eye, there was nothing out of the ordinary at the oil and gas sites @jonah_kessel and I visited in West Tex… — PolitiTweet.org

Nate Cohn @Nate_Cohn

Black share of the electorate on Super Tuesday 2016: 31% 2020: 21% Percent of delegates awarded by caucus-- 2016: 14% 2020: 3% (1% after IA/NV) https://t.co/m5ZPa1oeYB — PolitiTweet.org

Nate Cohn @Nate_Cohn

Over all, I'd say the changes bring the 2020 calendar to something more like a national primary. It could be a bit less interesting. IMO, IA/NH remain as the one big quirk of the calendar with the potential to offer a huge advantage to a certain kind of candidate — PolitiTweet.org

Nate Cohn @Nate_Cohn

The two biggest and related shifts: --Big primary nights, like Super Tuesday, are more representative. S.T. is no longer amazing for a candidate with backing from black voters, for instance --The caucuses are all but gone, and so is a familiar string of activist-friendly contests — PolitiTweet.org

Nate Cohn @Nate_Cohn

The Democratic primary calendar is a bit different this year. Not hugely different. But enough that your old conventional wisdom from 2016 may not apply https://t.co/m5ZPa1oeYB — PolitiTweet.org

Nate Cohn @Nate_Cohn

Now this could be part of a long-term tendency for polls to understate Dems in those states, but on its face the CNN results don't augur for a wider EC-PV shift — PolitiTweet.org

Nate Cohn @Nate_Cohn

It is, though in the CNN polling Trump holds up about as well in TX as he is in their national numbers, and he does quite well in CA compared to their national numbers https://t.co/Lz9KjQ23KA — PolitiTweet.org

Nate Cohn @Nate_Cohn

New CNN/SSRS poll has Trump+1 over Biden in Texas and his approval at 48/47. TBH, a bit better than I thought it would be for Trump (his approval was 47/48 in the Oct '18 CNN/SSRS, poll, which was Cruz+7 w LV/+5 w RV) https://t.co/PjAWgfAYjq — PolitiTweet.org

Nate Cohn @Nate_Cohn

(in Texas, my bad) — PolitiTweet.org

Nate Cohn @Nate_Cohn

New CNN/SSRS poll has Trump+1 over Biden and his approval at 48/47. TBH, a bit better than I thought it would be for Trump (his approval was 47/48 in the Oct '18 CNN/SSRS, poll, which was Cruz+7 w LV/+5 w RV) https://t.co/PjAWgfAYjq — PolitiTweet.org