Georgia Early Vote: Day 1 2024 Day 1: 30.0% Black 2026 Day 1: 36.6% Black Because Georgia doesn't have party registration, some of the most interesting data we can get from the state is race data. And today was a very, very strong day for Black turnout compared to Day 1 in 2024 (or to be honest, it was more of weak day for White turnout. Overall, turnout was not especially impressive). It's important to remember that Georgia's 2024 early vote was generally very good for Republicans, so this was a relatively low baseline for Democrats. Still, that baseline was exceeded by a substantial margin. This shouldn't be especially surprising given that in-person early voting has also been going much better for Democrats than in 2024 in Virginia (where we have a lot of data) and Ohio (where we have a little data). A relatively consistent story is beginning to emerge across these states. Black turnout was also very strong in primary elections this year. There are two big questions going forward. 1. How good was today actually for Democrats? Was this just high-propensity Black voters who were going to turn out anyway, or are we seeing unusually strong participation from lower-propensity voters, which is predictive of a bluer electorate in November? And how Black is that lower-propensity electorate? We'll get much more detailed data tomorrow through our partnership with L2, including voters' past party primary participation and turnout propensities. This will help us understand not just the racial composition of today's electorate, but what that means. Georgia's Day 1 early vote data will also be added to our early vote tracker at VoteHub tomorrow morning. 2. How much of this can stick over time? This is closely related to the first question, since understanding the propensity of today's voters gives us some clues about how "real" this surge in Black turnout is and how much it will stick through the final electorate. I don't think a Black share this high is at all sustainable throughout the early voting period, and I expect it to tick down in the coming days. But it's a very strong start that gives Democrats a lot of runway. Virginia offers a useful comparison. Even focusing on lower-propensity voters, the first couple of days were the high point for Democrats. Republicans then began picking up steam, which is also happening in Ohio right now. But in Virginia, that Republican improvement eventually stopped and even reversed slightly. We're now left with an early voting electorate that is tracking to be more Democratic than in 2025 and much, much more Democrats than in 2026. That's an important distinction--when the first days look like this Democrats don't necessarily need to maintain their opening-day advantage throughout early voting for the overall electorate to end up substantially more favorable to them than 2024. A note on the data: Right now, 9% of voters on Georgia's state website have their race or ethnicity listed as Other/Unknown. Many of those voters will be Black, too. Tomorrow, when we receive the more detailed data through our partnership with L2, we'll have modeled racial classifications for those voters as well. So the 36.6% figure isn't necessarily the final Black share of Day 1 voters. Once we account for the unknown voters, I suspect it could be closer to 39%. Georgia is 32% Black, so that's huge Black turnout. For now, I wanted to make an apples-to-apples comparison using the state's own data from 2024 and 2026. So I've calculated the Black share as the number of Black voters divided by the total number of voters with a racial classification, including Other/Unknown in both years. We'll have a much clearer picture in a week or two, but this is a very substantial opening-day increase in the Black share of Georgia's early electorate. p.s. I do think it's worrying for Republicans that Rick Jackson has spent $170 million on ads and this is what he got turnout-wise Day 1. Not a good sign for Paxton's $$.

Early voting can be highly useful and predictive if it reveals surprising turnout patterns that make you think there will be a different electorate that pollsters are generally assuming (like we saw in FL AZ and NV, and to a lesser extent in GA and NC in 2024). If pollsters are weighting AZ polls to have an R+6 electorate by registration but then it has an R+10 electorate that's obviously super notable and a huge source of polling error that in my opinion *really isn't the pollsters fault*. And EV is just people trying to sus out what the final electorate will look like. The issue is EV analysis on this platform is really all over the place and there isn't some cohesive agreement people have--If Michigan actually had a Trump+4.3 electorate that would be totally insane for example and super, super useful and mean Mike Rogers is favored. But that's not the same read of the numbers I have, as right now, I am still thinking a median scenario would be a Harris+2 or so electorate in the state, which is the same as I thought a month or two ago before we had any ballots. Yesterday I said: "AB so far this year is D+19 by past party primary voting in MI vs D+18 at this point in 2024. I have the current electorate as Harris+30 compared to Harris+27 last year 22 days before Election Day in 2024." So to recap, we have 7% (very little) of the final electorate already casting their vote and it's basically the same composition as it was at this point in 2024 (slightly more blue if anything). I'm not saying I'm always correct on this sort of stuff--there are a lot of other smart people on the platform who are going to notice stuff I miss (for example I'll always trust Chaz in VA)--but I don't know how anyone looks at this and decides MI is going to have a way more red electorate than 2024. And--states like OH, MI, WI, NE, PA I just think it's really hard to know what much when there's very little IPEV and so much ED vote. TX, GA, FL, NC you have more like 2/3 of the vote before election day and it's mostly IPEV that's way more useful. And then AZ, NV, CA, CO, etc. can be really useful too like we saw in 2024. Generally in states without party registration (OH, WI, MI, TX, GA) I find EV analysis less useful, but I'll take actually getting a lot of the vote over party reg for sure. So a rough ranking on how excited I am to see EV data/how much I would change my priors by state: A: FL, NC, NV, AZ, CA, CO B: TX, GA, AK C: IA, KS, PA, NE D: OH, WI, MI And then VA kind of stands along because it's (1) no party reg and (2) a good amount of Election Day vote but also @ChazNuttycombe has a proven track record of really knowing what to look for in VA and then it's also 45 days of in-person early voting so we have tons of VA data and basically no data from anywhere else right now. We don't put it in our main forecast, we do some separate internal modeling for our own sakes though. votehub.com/2026-early-vote-…