Read on Twitter
This will be an extremely long thread without a TL;DR, but if you're interested in what will happen in the Georgia runoffs tomorrow, I suggest you read it all. (1/25)
There will be multiple parts to this thread. The first is about the political geography of Georgia. I have composed this chart of what I consider to be the state's political regions: (2/25)
I have divided the state into six regions; four of which are large and two of which are smaller but meaningfully different from the other four. (3/25)
The first region is what I will refer to as the "Inner Atlanta metro," composed of Clayton, DeKalb, and Fulton counties. These three counties are the most urban counties in the Atlanta metro, and all are plurality Black. This region is by far the bluest in the state. (4/25)
The second region is the next region out in the Atlanta metro -- what I will refer to as the "Atlanta mid-ring suburbs." It is composed of Cobb and Gwinnett counties; both suburban counties that flipped to Clinton in 2016, and has continued to trend --> Democrats since. (5/25)
The third region is the Atlanta exurbs. Defining "exurbs" is difficult, so to avoid arguments I am using the Census Bureau's Atlanta MSA minus the five counties previously mentioned. Strongly Republican, but Trump's margin was cut by 9 points in the last 4 years. (6/25)
The fourth region is not really a "region," but it includes 5 similar counties: Bibb, Chatham, Clarke, Muscogee, and Richmond, home to Macon, Savannah, Athens, Columbus, and Augusta, respectively. These mid-size cities are quite blue but not very elastic. (7/25)
Then fifth region is the Black Belt. Here I define it as the cluster of 40%+ Black counties that are not in any of the above regions. These counties are extremely inelastic, with Black voters being overwhelmingly D and White voters being overwhelmingly R. (8/25)
The sixth and final region is the rest of the state -- the rural, heavily white counties that includes about 30% of Georgia voters, votes Republican by about 40-45 points, and forms the base of the Georgia GOP. (9/25)
I am spending so many tweets simply describing GA's political regions b/c it's very politically diverse; unlike, for instance, NH, which is white, educated, rural, and purple everywhere. Understanding the differences between these regions is key. (10/25)
The Atlanta suburbs and exurbs are heavily persuasion-driven. The rest of the state includes virtually no swing voters and comes down to turnout. (11/25)
However, turnout has not seemed to make much of a difference recently: only the Black Belt has seen a meaningful percentage chance in relative turnout in the last 4 years -- and this includes the 2018 midterms. (12/25)
The region dropped by about 10% in relative turnout from 4.5% of the Georgia electorate to 4.1%. While that sort of change could matter in a one-sided region, the region voted for Clinton by 5 points in 2016, Biden by 4 points in 2020, and Abrams by 4 points in 2018. (13/25)
This drop-off in turnout cost Biden <0.2% statewide. Obviously all votes matter, particularly in a close state like Georgia, but this isn't very big. HOWEVER, there is reason to believe that these runoffs may be different. (14/25)
First, early Black turnout is incredibly high. I don't buy the theory that a high-profile Black candidate automatically increases Black turnout; in '18 w Abrams at the top of the ticket there was essentially no change in relative turnout among Blacks vs Whites vs 2016. (15/25)
Nonetheless, whatever the motive -- which may indeed be Warnock being Black -- there has been a huge surge in Black turnout. If Ossoff and Warnock equal Biden's numbers by demo but get the Black Belt up to 2016 levels for relative turnout, they would both win by 0.4 pts. (16/25)
Furthermore, Republicans might be disillusioned about the voting process after Trump's baseless claims of fraud. One recent poll showed that 55% of "very conservative" Georgians will not vote because they fear the process will be rigged vs 0% of liberals. (17/25)
It is hard to imagine a huge drop-off in turnout from this, but each 1 percentage point more of Trump supporters who stay home than Biden supporters is worth an additional 0.5 pts for Ossoff's/Warnock's margins. (18/25)
These are small effects. The race largely depends on the Atlanta suburbs/exurbs. In 2016 and 2018 the mid-ring suburbs and exurbs were 38% of the Georgia electorate; in 2020 they were 39%. A 1-point swing here is worth almost twice Biden's statewide margin. (19/25)
And it's easy to imagine more than a 1-point swing. The outer Atlanta metro has swung about 10 points from 2016 to 2020, including a 4-point swing from Kemp 2018 to Biden 2020. (20/25)
Ossoff and Warnock both lead by 2 points per the 538 averages. It's hard to see much of the swing from the general election -- where Biden won by 0.2 -- coming from anywhere other than the outer Atlanta metro. That indicates a 3ish-point swing towards Ds in the area. (21/25)
A 3-point swing in 2 months is a lot, but I buy it given all of the events of the past two months. Even if it doesn't materialize, and there is no change in relative turnout, Loeffler and Perdue still need to find about half-a-point on Trump in the outer Atlanta metro. (22/25)
A 2-point lead the day before the election with high pre-election turnout with no other information is in itself indicative of about a 70% chance to win. By no means certain, but a whole lot better than being down by 2. (23/25)
It's hard to think of anything in the demography/fundamentals that disputes that 2-point lead. If anything, it may be understating effects of high D turnout. Furthermore, polling error has had a distinctive geographic pattern lately, and it hasn't included Georgia. (24/25)
I'm not going to get much further into the weeds than I've already gotten in terms of how I'm getting to these numbers. But the bottom line is that, considering everything, I think @ReverendWarnock has a ~75% chance of winning and @ossoff has a ~70% chance of winning. (25/25)
