Good morning everyone, happy start of in-person early voting for Georgia Senate runoffs day
We already have a lot of data in GA: more than 1.2 million absentee ballot requests, including 260k votes.
It's hard to read too much into the data, but on balance it's hard to say that it augurs for a vastly different electorate than the general (in terms of partisanship)
It's hard to read too much into the data because voters didn't exactly have an equal opportunity to request absentee ballots for the runoff and the general election. The general election was coming all year; the runoff and its significance only became evident a month ago
Nonetheless, requests are at about 70% of general election levels, despite fewer opportunities (and evidence that GOPers are even less interested in participating by absentee).
Of these, about half were requested pre-11/3, by voters eligible to request for full cycle absentees
And even this is a tough comparison, as some folks requested ballots over the last three weeks. Today, requests are at 77% of where they were at this point before the general election
In terms of demographics and partisanship, the requests look quite a bit like the general election, tho a bit older, slightly whiter, and slightly more Dem.
Why?:
--fewer post-11/3, non-cyclelong requests, who are younger
--the non-cyclelong requests are more D than the general
In terms of party primary history (by our measure), the absentee requests so far are D45, R29; it was D41 R29 at this point in the general election.
Those who have requested since 11/3 are D49, R14; those who requested beforehand were D41, R42.
In the general election, the late requests--over the last three weeks--were a little bit GOP-leaning compared to those who requested August, for ex.. We'll see if that happens this time
Ultimately we're going to know a lot more once we add the early voting into the picture, which starts today, then we will from these absentee ballots. At the end of it, we could easily have 75% of the vote and it ought to be reasonable to compare to the general, and infer a swing
We can't do that yet. But at the outset, it's hard to say the absentee data is obviously great for either side and there's certainly no reason think this is an '08-type scenario, with a crash in turnout and a vastly different electorate.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Another big day of early voting in Georgia: 167k voted yesterday, about the same as Monday and up from 129k on the second day of early voting in October
No, I don't think this augurs for a higher turnout in the general--at least not yet. 98% of these voters turned out in the general--as long as that's true, this level of voting (with respect to the general) won't be sustainable
Demographically, yesterday's voters were a lot like the second day of early voters in the general--which was also true of day one's voters.
And also like day one, yesterday's turnout was at once much higher than the general and a tad more Republican than the same day.
Let's dive into some of these numbers. One important bit of context: day one of early voting is not usually representative, it's usually quite good for Democrats
In the general election, for example, our estimate was that Biden won the first two days of early voting, three of the four weekend days, and Trump won basically every other day of early voting
Yesterday's voters were, again, quite Democratic. By our measure of party primary history, it was Dem 47, Rep 36-- about the same or perhaps a little better for the GOP than day one of the general, which was Dem 49, Rep 34. Racial demographics quite similar too
Indeed, and so it's worth circling back on an element of the post-election discussion about Latino voters: the refrain that it's a heterogeneous group (which is undoubtedly true)
Every major demographic group is heterogeneous. Many frequently analyzed demographic groups, like women or young voters include people of every educational, racial or regional strata
Even racially/educationally homogenous groups (say, white working class voters) include huge variation: white no college voters in Mississippi and Vermont have... very little in common
One of the strangest parts about the Texas suit is that it originates in Texas. If there's any state GOP that ought to intuit that the election result is entirely sound, it's probably the Texas GOP
The state of Texas has strict voter ID laws and didn't expand absentee voting. As far as I know, Texas checked just about every box of the GOP voting rules wish list.
And yet Trump only won Texas by 5.3 points--a result that's pretty facially consistent with GOP defeat
Biden improved by 3.4 points over Clinton's performance in the state--a larger improvement than his gains in any of PA/MI/WI.
He did so despite suffering big losses among the state's Latino voters--a group that represents a sliver of the vote in PA/MI/WI
One simple way of thinking about the Biden coalition--and maybe not a very good one--is to note that it's the Kerry '04 states, plus NM/NV/CO/AZ in the southwest and GA/VA in the southeast.
If you look at long term change this way--in terms of state flips--it highlights the growing diversity of the electorate, which nudged (as many predicted at the time!) these states toward Dems. OTOH, it obscures wild swings among white voters by education and region
And in a certain sense, it is pretty remarkable that Trump wasn't able to flip any Kerry states, despite making such huge gains among whites without a degree--including states where >50% of votes fall into the category
I think there's really one case in which we'd learn something from the polls in Georgia: if they showed the GOP clearly ahead, indicating a shift in the national political environment to something like a Biden midterm
The polls don't really seem to show this so far, and therefore we're probably in the range where the polls won't help us much over the hard results from November. The changes in attitudes are too small to be precisely measured, and polls aren't good at measuring changing turnout
We could get some decent signal on changes in turnout with advance voting data, though tbh the absentee voting surge for the general will make it more difficult to parse this than you might think