Nate Cohn Profile picture
I write for The New York Times at @UpshotNYT. I cover elections, polling, and demographics. Northwest expat.
eDo Profile picture Nat Welch Profile picture Thomas Wood 🌊 Profile picture white wine vigilante Profile picture unload Profile picture 16 added to My Authors
10 Jan
There's less confidence in the electoral system because people tried to erode confidence in it, not because of the way the election was administered
No 'blue-ribbon' plan could have gotten us out of the mess we found ourselves in
Go through that call between Trump and GA SOS, and think about how many of his assertions would be fixed by, say, a ban on mail absentee voting and a strict photo identification requirement
The answer is: not much of it
You can still assert that someone shredded ballots, or pulled a box from under a table, or that someone wasn't watching, or that there were 'dead' voters based on a file match, or that there isn't a perfect match with poll books, or lie that there are 'more votes than people' etc
Read 4 tweets
10 Jan
The needle is only a subject of angst because it's had bad news for Democrats. It's often this good (or better)
Here's Florida, just a few months ago
Here's North Carolina, just a few months ago (and the hick up in the estimate around 11PM result was induced by an IRL irregularity in the NC results, not the needle).
Read 5 tweets
9 Jan
Here's how Senate control would have changed over the last decade if DC had been a state:
2010: D (IRL) --> D (with DC)
2012: D --> D
2014: R --> R
2016: R --> R
2018: R --> R
2020: D --> D
Here's how Senate control would have changed over the last decade if DC and Puerto Rico had been states:
2010: D (IRL) --> D (with DC+PR)
2012: D --> D
2014: R --> R
2016: R --> R
2018: R --> R
2020: D --> D
*ALSEN in 2017 is an interesting side-story
If PR/AL were states, then the Doug Jones race in 2017 would have flipped Senate control (which the GOP would win back in 2018), though there's a distinct possibility that Jones wouldn't have won if Senate control was on the line
Read 5 tweets
7 Jan
Republican Senate candidates won the Georgia vote in November. Democrats won it on Tuesday.
The reason: a superior Democratic, and especially Black, turnout
nytimes.com/2021/01/07/ups…
We won't have an authoritative account for a bit, but based on what I see, there's basically no evidence of significant, net-Democratic vote switching since November.
Instead, turnout held stronger in Democratic areas than GOP areas
One fun way to check the proposition that turnout was decisive: if you take the Ossoff/Perdue tallies in November, and use precinct data to infer what proportion of Biden/Trump voters returned (which is not a safe assumption!) you get Ossoff +.4 with no switching
Read 9 tweets
6 Jan
Ossoff and Warnock are both on track for victory with a greater than 95% chance to win, according to our estimates.
This is not a projection, but the remaining vote--including another 18k DeKalb early votes and nearly 100k absentee votes--overwhelmingly favors the Democrats Image
Ossoff's lead is still just slim enough that you do want to make sure that some of these late absentees and provisionals really materialize to the extent we expect. I'd think we could see some projections in the Warnock race tonight
But the remaining early in-person votes in DeKalb, alone, will give Ossoff the kind of lead that Biden had in the final count, and there's a lot more for him beyond that. So there may not be a call there tonight, but it's not serious doubt
Read 5 tweets
6 Jan
I'm getting a lot of questions about the DeKalb County vote that remains. Apparently an election official said they only had 130k votes left, not 170k in-person early votes as they had previously said.
First and foremost, I think it's quite clear that it's 170k votes.
The state reports the early vote in great detail; and there were definitely 170k votes cast in-person in early voting in DeKalb County.
Now one possibility is that there's a reporting error--say, they uploaded 30k and called it in-person. I see no signs of that. The votes are left.
Read 5 tweets
6 Jan
Results (for Ossoff) so far v. our pre-election expectations:
Absentee: D+4.1 better than we expected
Early: D+.3
Election day: D+1.3
--on turnout--
Early: 100% of expected
Absentee 98.4% (expect the remainder tmro/later)
Election day: 106% (good for GOP; not enough so far)
To circle back to a pre-election question, Ossoff is on track to net-400k out of the advance vote with 56.4%. Perdue on track to get net-370k out of election day, by our estimate, with 64.5%
And the overall turnout estimate is at 4.415 million at the moment, though the public tea leafs on election day ATL-area turnout suggest that will be a bit low
Read 4 tweets
6 Jan
We're starting to get some votes and most of it is early and absentee votes, where Democrats were expected to fare well. In our view, Dems faring a hair better than expected in the absentee vote, but the big question mark is the Election Day vote and we still have very little
So far, Dems are faring net-4.5 pts better than we expected in the absentee vote and exactly as expected in the early vote.
We only have 2600 election day votes, so we'll wait on that but what little we have is in line with where the needle started the night
One easy way to track: our update-by-update log. So far, few surprises. Image
Read 4 tweets
6 Jan
The polls are closed and the needle is operational
nytimes.com/interactive/20…
One new feature:
In testing, I've got to tell you that this won't always seem 'rational'--you can't see the turnout half of the equation, to take one obvious thing that was actually pretty important in the way the needle moved in November. But it's enjoyable imo
Read 7 tweets
5 Jan
A few thoughts as we head into the polls closing. Let me start with the very most obvious one: this is not a regularly scheduled election, and that always makes it harder to predict
If this was regularly scheduled, we'd probably just call the clear GOP favorite. We had a 'perfect' poll of 5 million Georgians in November, and they favored Perdue. It's very hard to makeup a 2 point deficit on turnout, especially for the Democrats, in a high turnout race
But this is not a regularly scheduled election, and you can make up a 2 point deficit on turnout in this context. You can get an AL Senate special result that just wouldn't have happened if it had been scheduled in Nov 2018
Read 10 tweets
3 Jan
Let's talk about what the GOP, and more specifically David Perdue, needs to do on Election Day in order to prevail. It's nothing impossible, in my view. But it's worth thinking through where the GOP burden stands after the strong Democratic early vote in Georgia
Let's stipulate for this whole discussion that absolutely no one has changed their vote since the election.
If so, then our estimate is that Ossoff will amass a lead of around 350k out of the advance vote, including what he'll net out of ~100k absentee votes still to arrive
There's some fuzziness on any estimate like this, of course. But the fundamentals here are pretty straightforward, and clear to anyone with a voter file, based on the partisan makeup of the vote and the vigorous Black turnout. It's confirmed in our NYT/Siena poll data as well
Read 11 tweets
3 Jan
This is fairly typical of the allegations raised by the Trump legal team (as opposed to his twitter team) over the last month: ok, fine more folks voted for Biden, but the state didn't follow their laws to our liking, so give the election to Trump
This version, however, is a relatively bold one--as everything cited here was known before the election, and neither he nor anyone else argued that we should disenfranchise Pennsylvania's electorate until he saw that Trump lost
As with the other allegations, they then go further: they not only aspire to invalidate the election but hand it to the loser, even though no one even attempts to argue Trump won these states (as opposed to asserting there were imperfections in election administration)
Read 4 tweets
2 Jan
How big is the turnout going to be in Georgia? I have no idea, but here are a few things that loom in my thinking
One obvious thing: there's no good empirical basis for modeling the turnout in this election. But I do think that the level of spending, the stakes, and the type of contest (Senate) are above even the very highest-profile midterm contests
In 2018, the gubernatorial race attracted 4 million votes and voter registration's up 14%. Leaving aside the very important distinction that this is a runoff election, we would expect well over 4 million votes for a race like this
Read 9 tweets
1 Jan
In-person early voting wrapped up in Georgia yesterday, and over all the Democrats enjoyed a more favorable turnout than they did in the general.
Of course, the Dems needed and expected to do better in the early vote, given holidays and fewer mail absentees. And anything can happen on Election Day.
But still, this was a prerequisite to a stronger Democratic turnout and that's probably prerequisite to a Dem win
As an aside, a huge proportion of my replies in recent days are about age, and I just have to insist that it shouldn't weigh much on your thinking. Age doesn't matter a *ton* in general and certainly not in comparison to, or after controlling for, party and race
Read 7 tweets
31 Dec 20
Today's the last day of in-person early voting in most of Georgia (several counties wrapped up yesterday). Yesterday saw the highest turnout so far, with few changes in the turnout patterns that we've seen throughout early voting
This is typically a strong period for GOP voting and so it was yesterday as well. But as we've seen throughout early voting, Democrats outpaced their standing at a similar point ahead of the general election thanks to strong Black turnout
That said, the daily comparison at this point is not as clear as it once was. As you can see on the chart, the last day of early voting is a pretty good one for Democrats. A few counties, including Fulton, did that yesterday. And today will be the last in the rest of the state.
Read 5 tweets
30 Dec 20
Early voting kept humming along in Georgia yesterday, matching turnout from the equivalent date ahead of the general as we head into the last days of in person voting
The final Tuesday of early voting was the GOP's best day ahead of the general and that was true yesterday, as well, though they trailed the general election pace yet again
A relatively higher Black turnout continues to be the difference. The Black share of the daily electorate reached its nadir yesterday, as it did at this point in the general, but it was still 4 pts higher than at this point before the nov. election
Read 7 tweets
29 Dec 20
One question I keep getting about the Georgia early voting is about age: isn't the electorate older, and how much does it hurt the Democrats?
So far the answer is 'not really' and 'not at all.'
The first question is easy enough. As of today, youth turnout is basically keeping pace with the general, controlling for the slightly reduced opportunities to vote. This augurs for an unusually young electorate
The second question is more interesting: are the Democrats hurt by lower youth turnout? So far the answer is no, and there are two reasons.
One reason: there's not a *huge* gen. gap. Maybe young voters are D+20 while >65 are R+15. You need a big gap for modest changes to matter.
Read 9 tweets
29 Dec 20
Georgia early voting took a few days off over the holidays, like most everything else, but it was back near general election levels on Monday
As we've expected from the start, this creates a challenge for comparing where we stand now v. the general election.
Day 11-13 were fairly good days for the GOP in Oct.; they run farther behind without those days, but we have no reason to think that's for a meaningful reason
I'll offer two ways of trying to account for this. It's not perfect, but there are many circumstances in which it still permits useful analysis. 1) Remove those same days of in-person voting (not mail) from the advanced vote in October; 2) A direct day-to-day comparison
Read 11 tweets
23 Dec 20
Democrats continue to inch ahead in the Georgia advanced voting, thanks to persistently strong Black turnout, as we enter the last day of voting before the holidays
These comparisons are going to get a bit more complicated after today. The GOP is basically going to lose two days of early voting which were pretty decent for them in the general election. So we'll just take those days out of the 2020 early voting for comparison, I think.
At this point though, I don't think the comparison will be a problem. Since day four of early voting, Black voters have represented a much larger share of in-person early voting than it did in the general. There's no sign of that changing yet
Read 5 tweets
21 Dec 20
A lot for Democrats to like in the early voting data over the last few days, as they are now running ahead of where they stood at a similar point of the general election--thanks to a stronger Black turnout
The Black share of in-person early voters was basically identical to the general election over the first few days of voting, but that began to diverge late last week--and even more so over the weekend, which is typically a strong period for Black turnout in Georgia
As most of you know, the Black turnout in Georgia (and nationwide) was relatively weak v. other groups in the general election. That's been a matter of some consternation for progressives who wish it weren't so, but it's also the big source of upside for Democrats in this runoff
Read 11 tweets
18 Dec 20
Another day of strong early voting in Georgia, with another 154k voters turning out in person yesterday. That's similar to but slightly behind the fourth day of early voting in the general, which was at 164k
Cumulatively, we're still running ahead of the general election on in-person early voting, but I don't think anyone expects that to last--and there's a very long time left for that flip to happen
As I've said, the big takeaway is that the turnout looks like it will be healthy and high, and it's not obvious who it will help v. the general. As I've mentioned, there are crosscutting patterns at play that make it difficult to sort out where we'll end up. Let's look at a few
Read 11 tweets