Intellz Early Vote Desk: what the 2026 early vote actually says

Texas data through Fri 23 Oct 2026 (early-vote day 5). By Intellz Research (intellz.com).

1. Who has voted so far: 3.69 million. That is 19.4% of registered voters: 82% of the ballots cast by the same day in 2024, well ahead of the 2022 midterm (14.2%), and at or slightly above the 2018 midterm pace per registered voter in the 30 largest counties (19.75% against 19.80% in the state's file, 19.96% with the county offices' own counts).
2. Who the early voters are: R+0.2. By primary history (which party's primaries a voter has taken part in, a proxy for party and not a vote), early voters with a Republican primary history now match those with a Democratic one, a lean of 0.2 points Republican (11.7 points Republican at this point in 2024); the first days leaned Democratic, which is consistent with regular primary voters voting first, the usual shape of early voting, and Friday's new voters leaned 3.0 points Republican, against 5.5 points Democratic on Monday.
3. Why "Democrats are surging" headlines can mislead: 2.5 points. The 2024 early vote was the most Republican of any year shown, so against 2024 every state in our party-reporting table looks far more Democratic; measured against the 2022 midterm instead, those 22 states moved 2.5 points combined (D+12.2 then, D+14.7 now; 1.0 to 2.7 allowing one day of misalignment), and with Texas mail ballots down 24.8% from 2024, in-person "records" overstate the total. Neither year is a clean baseline (Republicans are being urged to vote early this year, and some states have changed their early-vote rules), and ballots by party say who voted, not how.
4. What the polls missed before: 4.9 points. Texas polls ran too Democratic in 2020 and 2024 (4.9 points in the 2024 Senate race); in 2022 the governor average was close (RealClearPolitics had Abbott up 10.4, he won by 10.9), though a simple mean of the listed polls ran 3.1 points too Democratic; in 2018 they ran too Republican by 3.4 points, so a miss can run either way.
5. What the fake-poll story is, and is not: 0.28 points. Four polling firms were exposed this year (two admitted inventing polls, one admitted moving respondents toward a candidate whose campaign, records show, had paid it, one went dark without admitting anything), but taking out the two that polled Texas moves no Texas average by more than 0.28 points.
6. Who is still left to vote: 688K to 351K. Of the 2.24 million Republican and 1.91 million Democratic primary voters who have not voted, the most reliable split 688K to 351K, a reservoir that can show up rather than votes in hand (a primary vote is not a November vote): the Republican count includes March primary voters who backed Cornyn against Paxton, both parties' March early voters have turned out at about the same rate (56.6% and 57.3%), and voters with no primary history are about half of the votes still to come (49% in the public model's middle case).
7. What to watch next week: 22.5%. Watch the smaller counties: outside the 25 largest they have cast 22.5% of the early vote, against 26.5% at the same point in 2024. In 2024 that share fell in week two (to 24.4% by the end of early voting), so a rise next week would be new.
8. What past elections say about the polls and the early vote: 2.0 points. Tested on past state races with only the early vote known 10 to 13 days out: the final polls missed toward Democrats by 2.0 points on average over 26 races. A turnout read built from early ballots beat the polls when each midterm was compared with the last midterm (3.7 points off against 4.0, 8 races), and in Texas the primary-history read missed by 1.95 over 4 cycles. The stat the media leans on most, the party-registration lead in the early vote, missed the result by 16.0 points on average over 18 races.

Method: every figure carries a grade (V opened at source, M computed by Intellz, R relayed, E estimate, I inference), a source link and an as-of date. Aggregate counts only; no voter names or IDs. Not affiliated with any candidate.