Nate Cohn

Nate Cohn @Nate_Cohn

Buttigieg lead expands to 44 SDEs with 85% reporting, up from 28 SDEs at last update — PolitiTweet.org

Nate Cohn @Nate_Cohn

Buttigieg lead expands to 44 SDEs with 85% reporting, up from 28% in the last update — PolitiTweet.org

Nate Cohn @Nate_Cohn

@twitnitrick nope, always meh and noisy — PolitiTweet.org

Nate Cohn @Nate_Cohn

(it is just as possible, by the way, that the '16 entrance poll underestimated the youth share of the electorate as it is that the '20 one overestimated it) — PolitiTweet.org

Nate Cohn @Nate_Cohn

Now, turnout by precinct isn't a precise measurement, either! Turnout could increase among young voters without showing up in young precincts. It's possible. — PolitiTweet.org

Nate Cohn @Nate_Cohn

It's just a sample of precincts. Say, 30 to 50 (IDK). So, if in 2020 there's an extra college campus site in the sample precincts, that on its own could move the youth share of the electorate. They're just not designed for precise measurement here — PolitiTweet.org

Nate Cohn @Nate_Cohn

The entrance poll tells a different story, I know, but those are pretty imprecise instruments for measuring the composition of the electorate. For one, it's a poll of 1600 people. Margin of error? Yes, it applies. And it's even greater than usual because it's a cluster sample — PolitiTweet.org

Nate Cohn @Nate_Cohn

I should emphasize one caveat: I'm measuring change in turnout by raw numbers, not with a denominator of population or registered Democrats (I do not have denominators for 2016 by precinct) It is conceivable, for ex., that pop growth in suburbs could exacerbate this pattern a bit — PolitiTweet.org

Nate Cohn @Nate_Cohn

Similarly, turnout tended to increase more in Buttigieg precincts / those where Sanders struggled (well-educated, suburban) — PolitiTweet.org

Nate Cohn @Nate_Cohn

In perhaps a bad sign for the Sanders campaign: no greater increase in turnout in areas where 18 to 24 year olds represent an above average share of the population — PolitiTweet.org

Nate Cohn @Nate_Cohn

Early analysis of turnout by caucus site shows a very clear trend: lower turnout than '16 in less educated areas and an increase in well-educated areas, likely reflecting the Trump era realignment in Iowa and elsewhere — PolitiTweet.org

Nate Cohn @Nate_Cohn

Satellite caucuses still a big area of uncertainty in the model btw that may not be uncertain IRL at this point, based on twitter reports — PolitiTweet.org

Nate Cohn @Nate_Cohn

Buttigieg now with an 86 percent chance to win the most SDEs, according to our forecast model https://t.co/jgNuWuz1KD — PolitiTweet.org

Nate Cohn @Nate_Cohn

As a general proposition, our turnout data suggests that this kind of tight, non-probabalistic screen isn't the bes… https://t.co/JTJMPiQp9u — PolitiTweet.org

Nate Cohn @Nate_Cohn

The tightest screen--only Almost certain--would have been most accurate, with an average error of 1.2: Sanders 25.6, Buttigieg 20.2, Warren/Biden 15.0. Could be noise though, as we fall to N=323 here — PolitiTweet.org

Nate Cohn @Nate_Cohn

If we change our likely voter screen to the same as the Selzer poll (Almost certain, very likely as 100%; all else as zero), our result is Bernie 24.7, Buttigieg 18.9, Warren 15.2, Biden 16, which brings our average error down to 1.4, the exact same as the Selzer poll — PolitiTweet.org

Nate Cohn @Nate_Cohn

And it does seem that every firm did better on their last poll than their first, so there is a real advantage to going last. Voters who decided in the last few days appeared to bad for Sanders, per entrance polls. https://t.co/ftd3D8J263 — PolitiTweet.org

Nate Cohn @Nate_Cohn

@HotlineJosh what the entrance poll showed too — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'd guess he won by even more in the preference votes than in the SDEs, for the same reason Warren/Sanders are disa… https://t.co/y8hYC9ry8o — PolitiTweet.org

Nate Cohn @Nate_Cohn

Pretty happy with our result, especially given that we're farther before the election than the others and that the demographics seem dead-on in the entrance poll. And I'll give ourselves a bonus prize for being the only firm to have Sanders in 1st and Buttigieg in 2nd — PolitiTweet.org

Nate Cohn @Nate_Cohn

Selzer appears to have nailed it again, and that's without even crediting her with nailing the multi-day delay in r… https://t.co/GZkBe5wRm5 — PolitiTweet.org

Nate Cohn @Nate_Cohn

(the Iowa Republican caucus, for instance, uses a straight popular vote without any of the realignment pageantry) — PolitiTweet.org

Nate Cohn @Nate_Cohn

And since I keep talking about this IRL: there was no reason why the Iowa Democrats had to keep using SDEs to award… https://t.co/WT8vFmRMA3 — PolitiTweet.org

Nate Cohn @Nate_Cohn

Is there a worse outcome for the future of the Iowa caucus than this? --failure to report the vote --a second alig… https://t.co/F3qZDvggLD — PolitiTweet.org

Nate Cohn @Nate_Cohn

In the first alignment vote, we now think Sanders has a 94% chance to prevail — PolitiTweet.org

Nate Cohn @Nate_Cohn

For our purposes, the uncertainty is particularly noteworthy since: the number of state delegate equivalents for the satellite caucuses is not set in stone, unlike all other sites — PolitiTweet.org

Nate Cohn @Nate_Cohn

One major source of outstanding uncertainty is the satellite caucuses, which as far as I can tell are not included in any of the data being reported by the Iowa Democratic Party. — PolitiTweet.org

Nate Cohn @Nate_Cohn

Buttigieg now at 81% to win the most SDEs, according to our live forecasting model, even though his lead narrowed w… https://t.co/1ZalvYNuJn — PolitiTweet.org

Nate Cohn @Nate_Cohn

needle inbound; stand by — PolitiTweet.org

Nate Cohn @Nate_Cohn

New data... ...brings us to 71% of precincts reporting Buttigieg still holding a 25 SDE lead — PolitiTweet.org