Early voting is over. Here's what the numbers can tell you before Tuesday — and what they can't

HARRIS COUNTY, Texas (KTRK) -- Early voting is over, and Harris County did something it has never done in a midterm. More than a million people voted before Election Day even got here. That's 1,081,010 of us, nearly four in ten registered voters, and it's almost as many ballots as the county cast in the entire 2022 election, Election Day included.

Friday was the busiest day of the 12. It was the only one that topped the opening-day crowd. Fort Bend and Montgomery counties set their own records, and the statewide count had already passed 6.8 million before most counties posted their final day. Here's a number to sit with: that incomplete total is already more than 80% of every vote cast in 2018, the biggest midterm Texas has ever had.

So, the question. What does it mean? Last week we showed you that Texas polls have missed the final result by three to five points in each of the last three statewide elections. This week we have something the polls never had: a count of people who actually voted. It is tempting to read that count like a scoreboard. It isn't one. Here's what it can tell you about Tuesday, and what it can't.

PREVIOUS REPORT: Harris County broke early-voting records in 2018, too. Democrats still lost.

The surge is everywhere, not in one kind of neighborhood. We matched the county's early-vote rosters to the precinct-by-precinct results from 2022 and 2024, and the pattern is boring in the best way. In the neighborhoods that voted hardest for Ted Cruz in 2024, early voters alone already equal about two-thirds of everything those precincts cast in a presidential year. In the neighborhoods that voted hardest for Colin Allred, it's a little over 70%. Everyone else lands in between. Measured against the last midterm, the bluest precincts have already passed their entire 2022 turnout. The reddest aren't far behind.

Put another way: the mix of neighborhoods that showed up early looks a shade bluer than 2022 and almost identical to 2024. That's a real finding, and a modest one. A few points of difference between the bluest and reddest precincts is not a wave. It's a county where nearly everybody is voting more.

Early voting in Harris County has been a faithful preview of the county. In 2022, the people who voted early in the governor's race split 54–45 for Beto O'Rourke. The full county, once Election Day was counted, split 54–44. In 2024, early voters went 52–46 for Kamala Harris, and so did the county. Early voters here are not some separate electorate that shows up before the real one. They are most of the real one.

RELATED: What do the early voting numbers actually mean?

Turnout is up in every big county, blue and red. Compared with the same point in 2018, Harris County's early vote is up about a quarter. Dallas, Tarrant, and Bexar are all up by double digits. The suburbs, Fort Bend, Montgomery, Williamson, and Denton, are up even more, and Hidalgo County in the Valley is up nearly half. If you're looking for a corner of Texas that sat this one out, you will not find it.

What it won't tell you

Who those seven million people voted for. Texas doesn't register voters by party. The state's daily files record that you voted and where. That's it. Every chart you've seen this week that colors the early vote blue or red is an estimate built on something other than the ballots. Be careful what you read on the internet.

Whether early voters lean one way is the heart of the "blue early vote" idea, and the exit polls say it's wrong in Texas. AP VoteCast, the Associated Press survey of voters that ABC13 uses on election night, asks people when they voted. In each of the last three statewide elections, Democratic and Republican voters cast early ballots at nearly the same rate, within a point or three of each other. A record early vote is a record turnout. It is not a party's turnout.

Harris County is not Texas. We hold about one in seven of the state's registered voters. In 2022, O'Rourke carried Harris County by nearly 10 points and lost Texas by 11. In 2024, Allred carried it by 11 and lost by 8.5. Our early vote tells you a lot about us. It tells you very little about the six in seven Texans who live somewhere else, and that is where the last three statewide races were decided.

VoteCast also asks voters which party they identify with, and the answer is the whole ballgame. In 2018, the closest statewide race in a decade, Texans who voted identified Republican over Democrat by about eight points, and Cruz still won. In 2022 and 2024, the gap was closer to 20. For Tuesday to be close, the people who voted need to look more like 2018 than 2022. A turnout count can't confirm that, because a turnout count doesn't ask.

Election Day was about a third of the Texas vote in 2022 and under a fifth in 2024. Small, but not nothing. In Harris County, it has nudged the result both directions: a bit more Democratic in the 2022 governor's race and the 2024 Senate race, a bit less in the 2024 presidential. About 1.66 million registered voters here haven't voted yet. Some will on Tuesday, and the data can't tell you how many or which way.

VoteCast found that somewhere between one in 12 and one in 16 Texas voters made up their minds in the last few days, or later. That sounds small until you remember that nearly every poll of the Senate race this month has been inside the historical miss. In a race that polls as a tie or a two-point lead, the people who decide in the car on the way to vote are bigger than the margin.

The first numbers after 7 p.m. will be the early vote, and in Harris County that number has looked almost exactly like the final number two elections running. If our early vote lands near the roughly 54–44 split of the last governor's and Senate races, the county is doing what it does. The statewide question is different: whether the rest of Texas, reported county by county through the night, moves off its 2022 and 2024 pattern. That's the one number the early vote could never tell you. It's also the only one that decides the race.