Tea leaves to read for the 12/6/22 GA Runoff

 Deja vu all over again, as we go into Sen. Raphael Warnock’s 4th general election in <2 years…

I predict this election will be callable by 9pm Eastern (polls close at 7), but it’s possible that it will be so close that it will take longer and/or go into recount territory.

Today (Mon. 11/28) is the official start to early voting and the last day to request an absentee ballot; several counties allowed early voting over this past weekend, and 182k ballots have already been received. The GA Sec.State website has some good visualizations as does amateur analyst Ryan Anderson.

9p 11/28 update -- record breaking 320k early votes today sounds impressive, bringing early vote total to >400k. BUT, at this rate total early vote will at best hit 1.8m, compared to 2.1m in 1/21 and 2.3m on 11/8/22 -- Warnock & Walker will need to bring another 500k voters out on eDay than have voted in each of the past 3 General Senate elections (11/8/22, 1/5/21, 11/3/20 general) to have get north of 3.5m votes

9a 11/29 update -- GA Early vote to date is 38% Black (non-hispanic); in '21 it was 31% (and it was 29% early this month); BUT weekend early vote was 46% Black, and Monday down to 34%, Warnock's GOTV is definitely performing well.

9a 11/30 update -- Early vote will be low; eDay turnout will make difference. Nothing in data suggests any new / different dynamic than an exceptionally close race Tues. was a 2nd record-breaking early vote day (b/c EV period is much shorter for this election than any other). BUT, even at this rate, total early vote will likely be <1.9mil. (vs. 3.1m in '21runoff and 2.5m in Nov. general).

Total early vote now 35% Black (non-Hispanic) vs. 31% in '21runoff and 29% in Nov. general). At current trend, EV turnout should end ~33% Black a good sign for Warnock My current estimate suggests mail vote will be 80% of the Nov. election and Early in-person 75% of Nov. In Jan. '21, the fall-off was slightly less: 85% mail and 78% in person. And, in '21 the Day-of voting increased by 38% over the Nov. '20 general.

Currently 3.6% of early voters didn't vote on 11/8 (similar to '21runoff), age breakdown of early vote is skewing older vs. '21

Election night tips:

Assuming the NY Times does their prediction needle, that’ll be the easiest way to figure out who’s likely to win starting at 7:30p when results come in. The NYT needle essentially looks at actual results from a few precincts, and then extrapolates those results across the rest of the state.  This type of election lends itself to prediction models like the needle, since Warnock just ran against Walker a month ago, and after a few results from a few different counties, it will be possible to understand how results are trending (total turnout in different types of areas, whether the Warnock / Walker split has changed, etc.) 

I’ll likely be doing some predicting on election night as well, just to stay busy, but assuming the needle doesn’t go down (they took it off line for awhile during midterm election night, and I seem to recall a few glitches during the Jan. 21 runoff) that’s all one should need.

The harder thing to do is predict the election based on current data (including polling).  So for those want to see all the tea leaves I looked at before deciding a prediction based on data was not defensible here’s what I’ve been watching:

Turnout:

It’s a little early to conclude anything from the past few days of early voting, especially since it was primarily in heavily D-counties. Unsurprisingly, there’s a lot of interest in the race and a substantial amount of early voting already…

Here are some basic numbers to think about:

Total votes: 11/20: 4.886k; 1/21: 4,485k; 11/22: 3,936k  Note the 8% fall-off from the ‘20 General to the 21 Runoff (also, per the state voter file, 3.2% of the voters in the 1/5/21 special did not vote in the Nov. 20 election).

A few things to remember about the 2020 elections vs. this cycle’s runoff:

1. The November election for the seat Warnock won (a special election to fill the seat for 2 more years, and to which Kelly Loeffler had been recently appointed) had a large number of contenders on both sides, and in the end there were two Republicans battling it out in the final months of the Nov. election – Doug Collins (who received 980k votes) and Kelly Leoffler (1,273k); combined they did better than Warnock’s 1,617k votes in the November general.
   Thus in the Special election a number of people had to first decide if they were going to still vote (given their candidate wasn’t in the run-off) and then pick between Leoffler and Warnock;
   In the special Loeffler v. Warnock ended at 2,196k v. 2,289k votes 

2. There were two runoff elections in GA on 1/5/2021 Purdue v. Ossoff and Loeffler v. Warnock, but in the end results were remarkably similar (Warnock got 2,289k votes,19k more votes than fell D-Ossoff received); and Warnock won by 93k votes (out of 4.5mil) and Ossoff by 65k votes; I don’t think two elections vs. one will mean anything on turnout for this time around;

3. Arguably the Senate majority isn’t at stake (although there’s a big difference between the current 50/50 power sharing arrangement and what a 51-49 majority would mean in terms of rules, committee assignments, etc.); I believe this may decrease turnout, especially from R’s who don’t like Walker;

4. In ‘20,  Trump was fuming about election infrastructure, and some folks think he may have depressed turnout somewhat in the Jan. runoff

5. In the Nov. ‘22 election, Walker received 203k fewer votes than Gov. Ri-Kemp (my analysis suggests about 130k of those votes went to Warnock, the rest to the Libertarian or to no vote in the race).

6. There’s a much shorter early voting window, since this run-off is 4 weeks earlier than the one 2 years ago, and most of the Georgia counties (totalling about half the population) will not have any Saturday early voting.  In the ‘21 election, 71% of ballots were cast ahead of time (59% in person, 12% by mail) compared to the ‘20 General (79% before election day);  This past November (the midterms) 68% of votes were cast in advance (51% in person / 17% mail). 

Given the narrow early vote window, lack of weekend early voting, and limited time (< 3 weeks) to have gotten an absentee ballot application in, I’m guessing early vote will be about 50% of the final turnout (vs. 71% in the last special election).  BUT, I’m optimistic that >$50million Warnock has raised (and high numbers for Walker as well) will lead to an unprecedented Get Out The Vote operation next Tuesday.

Predicting turnout is never easy, my current (11/28) range guess is 3.2m-3.6m votes.  I’m guessing higher turnout favors Walker, on the assumption that the Warnock base is much more motivated (and a much more solid base) so the real question is whether Rs who didn’t vote for Walker in Nov (choosing Libertarian candidate or no vote) stay home.  It’s also reasonable to assume many of the ~130k voters who voted Kemp & Warnock will not come out in the runoff.

By mid-day Sat. 12/3 we should have most of the in-person early vote data available by age and race/ethnicity to possible push the prediction one way or another

A few other data points that don’t really help with predictions but some may find interesting:

Racial breakdown of ‘20 General, ‘21 Runoff and ‘22 General elections: 
White (non-Hispanic) 58% / 55% / 57% 

Black (non-Hispanic) 27% / 30%/ 26%

Whites as a percent of all votes decreased from the General to the runoff two years ago; Blacks increased their share of total votes from the General to the runoff.

I’ll likely do a few more updates later this week or over the weekend in advance of Tuesday.

More resources:

Georgia-specific Election folks to follow on Twitter (I’m guessing they’re also on Mastodon for those who are trying to re-platform)

In general #GApol and #GAsen are the best hashtags for the race

@stphnfwlr (Stephen Fowler, best GA-based political reporter)

@bluestein (Greg Bluestein, Atlanta Journal & Constitution reporter)

@gtryan (amateur GA-based numbers cruncher)