Thread by @gtryan

We've made it through the full early voting period, with just a few more days of mail votes to count. 3,032,066 votes have been cast so far, and today we'll try to read some tea leaves.

#gapol #gapolitics

georgiavotes.com

As we go through the rest of this it's important to remember that the Election Day margins for the 3 major races in GA were:

Trump +217,250
Perdue +215,208
R's in the Special +200,108

That's what came from almost 1M Election Day voters.

First, how do these votes look if you take the county-by-county vote totals for mail and in-person voting and apply the November election results by voting method to them?

We have to solve few more problems to answer that well.

First, what do we do with the third party vote? I chose to throw it out completely. All percentages are based just on R and D vote totals.

Second, how should we classify them for mail or in-person? This seems like an easy question, just put people where they voted, right? Nah.

You see, the November results are based on how that population voted. Which means if you voted by mail in November and in-person in the runoff, the mail number is more representative of your likely vote. And a couple hundred thousand people moved between the two voting methods.

If you voted early, I chose to use your November early voting method to classify you as mail or in-person. If you did not early vote in November then I classify you as the voting method you used in the runoff.

The final problem: which November breakdown do you use to estimate votes, since there were multiple statewide elections? I took the Presidential results, both Senate results, and an average of the 3 to have a range of results.

So, with that said, what would the numbers say:

Democrats have a lead of between 170,409 (using Ossoff/Perdue) and 244,819 (using Biden/Trump results), so in the range of 6-8% lead for the Democrats.

But this is just one data point. We can look at some more.

But counties are big, even in a state like Georgia with 159 of them. So I've also been tracking 4 large counties that I feel like are bellwethers for what may be changing in-county: Cobb, Gwinnett, Cherokee, and Forsyth. A mix of R and D counties, although all northern burbs.

If I go through the same analysis as above but at the precinct level, those counties providing ~4k more net votes for Democrats than the county-wide averages would suggest. Again, not perfect, but as these represent ~20% they may show further D strength.

Now let's try looking at something completely different. Turnout by what party ballot people used in the 2020 primary election. We had record turnout for a primary with over 2.3 million party ballots cast, which gives us maybe our best simple estimate of voter partisanship.

So far, 959,743 (77.7%) people who voted in the Democratic primary have already voted in the runoff elections.

695,610 people (65.3%) who voted in the Republican primary have already voted in the runoff elections.

That's a net D lead of 264,133, with 1,376,713 other voters.

Of those voters, here is what method they used to vote in November

D Primary Voters
Early - 926,438 (96.4%)
E-Day - 19,589 (2.0%)
No Vote - 14,643 (1.5%)

R Primary Voters
Early - 665,514 (95.5%)
E-Day - 22,416 (3.2%)
No Vote - 8,584 (1.2%)

So based on the primary behavior, Republicans have actually cannibalized MORE of their Election Day voters despite having 27.5% fewer overall votes.

This is also significantly different than behavior from Nov. There the Ds turned out 1,078,410 (87.3%) of their primary voters in early voting, compared to 860,379 (80.8%) for Rs.

So the D margin has grown from 218,031 to 264,133, and the % difference is up from 6.5% to 12.3%.

Interesting side note to that 218,031 number, Biden's lead coming out of the early voting period was 229,920. Not saying it's predictive, but there it is.

There are also those millions of people who voted early that didn't vote in the primary. So I think we should take a look at how those voters breakdown by race to see any changes.

The next tweet looks at % of the non-primary voting electorate by race.

Race - General - Runoff
White - 57.6% - 55.4%
Black - 24.8% - 27.7%
Hispanic - 3.8% - 3.2%
Asian - 3.8% - 3.7%
Other/Unk - 10.0% - 10.0%

So we see a drop of ~2% among white voters and growth of nearly 3% among Black voters.

I'll do some more number crunching tomorrow, maybe tonight as well. Let me know if there's anything specific that you're interested in seeing more about.