Nate Cohn

Nate Cohn @Nate_Cohn

Or put differently, I'm not drawing attention to the trajectory point to make a prediction, but instead to note that I think it's a valid argument in this case when I don't usually think it is one — PolitiTweet.org

Nate Cohn @Nate_Cohn

In general, there's not. To my mind, there's even evidence for a little mean reversion. But we also have a secular negative trend in Biden's approval rating, and I think it's straightforward to argue that there are deeper currents going on right now https://t.co/D4f6v81lyA — PolitiTweet.org

Nate Cohn @Nate_Cohn

On that 'trajectory' point, it's worth noting that we don't really have much data collected in the last week of the race. Most of the intervews were 10/20-27 or so It's hard not to wonder whether you can roll the trend forward, given how clear it was https://t.co/Cb4Lz2ufDX — PolitiTweet.org

Nate Cohn @Nate_Cohn

With that kind of split, you have to wonder about whether surveys represent non-readers in general. But it seems especially important to consider for online surveys--which literally require reading — PolitiTweet.org

Nate Cohn @Nate_Cohn

Among white voters without a degree under age 45: read a book in last year: Dem 49, GOP 40 no book in last year: Dem 25, GOP 67 — PolitiTweet.org

Nate Cohn @Nate_Cohn

In the Pew NPORS study, reading was very strongly correlated with Democratic support among white voters without a degree--especially under age 45. I think it merits research as a possible weighting parameter https://t.co/r0NH7dMazJ — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @ElectProject: Well, the university still approves of some things I do — PolitiTweet.org

Nate Cohn @Nate_Cohn

If you want to mentally adjust these Biden-based estimates using the polls, maybe you could put McAuliffe is at 59%. If you assume--as I tend to--that early voters are hardened partisans, than maybe it's 62 or whatever. That's quite close to a lot of the polling we've seen, imo — PolitiTweet.org

Nate Cohn @Nate_Cohn

(on that last point, i see a lot of people saying that this means that mcauliffe has a 300k lead or something. i have no hard data to address how mcauliffe, specifically, is doing in the early vote.) — PolitiTweet.org

Nate Cohn @Nate_Cohn

@JayCostTWS https://t.co/9GjlE89d0u — PolitiTweet.org

Nate Cohn @Nate_Cohn

With that in mind, I think that the estimate here--early vote thru Thurs at ~64% for Biden--is a fairly high confidence one. And these estimates have the additional benefit of calibration to 2020 results, unlike our pre-election ests. A key caveat, ofc: McAuliffe isn't Biden — PolitiTweet.org

Nate Cohn @Nate_Cohn

Here's all of our pre-election 2020 estimates for absentee/election day vote, by county, v. actual result. (since estimates were calibrated to pre-election polls, basically all of the estimates--but esp election day--are biased toward biden, like the polls) https://t.co/WLjBWVx50Y — PolitiTweet.org

Nate Cohn @Nate_Cohn

That uncertainty goes down to vote share, since the estimate for election day vote is conditional on turnout. IE: if your prior is that GA is probably going to be close, regardless of turnout, your estimate for election day vote share will have to be very different at 1.5 than 1 — PolitiTweet.org

Nate Cohn @Nate_Cohn

IE: if turnout could realistically be between 4 and 4.5 million in GA runoff, but 3 million have voted early, now the election day turnout range--between 1 and 1.5--just contains a huge amount of uncertainty (the latter being 50% higher turnout) — PolitiTweet.org

Nate Cohn @Nate_Cohn

Think about it like this: if advance voting doesn't really reduce overall uncertainty about the composition of the electorate, but locks in much of the electorate, then basically all of the uncertainty is channelled into the election day vote — PolitiTweet.org

Nate Cohn @Nate_Cohn

This really shouldn't be too surprising, given all the data for assessing early/absentee voters. But there's more going on than simply being better at the method for which we have more data. — PolitiTweet.org

Nate Cohn @Nate_Cohn

I got curious and looked back over all five states where I did pre-election estimates for vote by method (FL, GA, NC, PA, GA runoff). Basically, the estimates for the advanced vote--calculated in the same way as this--were really good. Estimates for election day? Pretty messy — PolitiTweet.org

Nate Cohn @Nate_Cohn

Depending on the apparent level of Election Day turnout, it should be pretty straight forward to infer Youngkin's target vote share among Election Day voters--and then evaluate whether he's hitting it. — PolitiTweet.org

Nate Cohn @Nate_Cohn

No, I don't think this offers us very much information on the eventual outcome. More than anything, it probably just helps us interpret the results as it arrives on Election Night. — PolitiTweet.org

Nate Cohn @Nate_Cohn

More than a million people have voted early so far in Virginia. As of yesterday, those voters appeared to back Biden by nearly a 2 to 1 margin, based on a combination of voter file data from L2 and about 4k interviews of Virginia voters from earlier cycles — PolitiTweet.org

Nate Cohn @Nate_Cohn

RT @clairecm: After the most restrictive abortion law in the country, abortions in Texas clinics fell by 50%. Half of patients who come in… — PolitiTweet.org

Nate Cohn @Nate_Cohn

@PatrickRuffini by that you do you mean that RVs are youngkin+3, as well? — PolitiTweet.org

Nate Cohn @Nate_Cohn

Anyway, I think it's something to keep in mind when you interpret this race. Youngkin's TV campaign is doing the kind of things that you'd think a Republican would want to do in order to compete in VA, certainly much more than the Twitter Trump/cultural war stuff makes it seem — PolitiTweet.org

Nate Cohn @Nate_Cohn

Meanwhile, the only Youngkin ad I've seen here on Twitter is the testimonial from the woman who wants to ban books or whatever, which I have not seen on TV — PolitiTweet.org

Nate Cohn @Nate_Cohn

The Youngkin ads, that i've seen, critique mcaullife on parental choice statement, and then Youngkin--speaking to camera--goes on to support for teacher raises and the largest education budget in VA history — PolitiTweet.org

Nate Cohn @Nate_Cohn

Twitter seems focused on two narratives--a backlash against CRT in schools vs. a Republican who supports Trump and flirts with insurrectionists. — PolitiTweet.org

Nate Cohn @Nate_Cohn

I happen to be in the DC media market and one thing I'll note is that there is a fairly large split between the campaign playing out in TV ads and the campaign as depicted on Twitter — PolitiTweet.org

Nate Cohn @Nate_Cohn

I'm getting dizzy https://t.co/W1AM6G9WdQ — PolitiTweet.org

Nate Cohn @Nate_Cohn

McAuliffe by 1 in the final Post/Scher poll, which now counts as decent news for Democrats in a state Biden won by 10 https://t.co/CJEUj1Wy1n — PolitiTweet.org

Nate Cohn @Nate_Cohn

I think the composition of the electorate--not revealed in the Fox poll--is very valuable. I think the crosstabs from a single poll aren't usually very useful, but I think they're worth publishing for transparency/accounting/showing your work basically https://t.co/43wK3QBJsF — PolitiTweet.org