This article was originally published by Votebeat, a nonprofit news organization covering local election administration and voting access.
By Nathaniel Rakich
Votebeat
Votebeat is a nonprofit news organization reporting on voting access and election administration across the U.S.
This news analysis was originally distributed in Votebeat’s free weekly newsletter. Sign up to get future editions, including the latest reporting from Votebeat bureaus and curated news from other publications, delivered to your inbox every Saturday.
Election officials, start your engines.
The 2026 midterm elections officially kicked off last week with primary elections in Arkansas, North Carolina, and Texas. It was the first big test for hundreds of candidates who are vying to take up residence in Washington, D.C., and state capitals next year — but it was also a dress rehearsal for election officials in a year expected to pose unusual and daunting administrative challenges.
And things were a little rocky.
While the election went smoothly for a majority of voters and jurisdictions, thousands of voters were affected by problems in Texas in particular.
Some of those — like a poll-worker shortage in South Texas or malfunctioning electronic pollbooks in northeastern North Carolina and El Paso, Texas — were routine, inevitable in a country with thousands of election jurisdictions and just as many things that can go wrong.
But others resulted from Republican-driven changes to voting procedures that Republicans argued would make the elections more secure. And in that sense, the messy primary day was a reminder that the GOP’s proposals to suddenly and dramatically overhaul voting this November carry real risks of harm to voters, election officials, and anyone affected by the results.
Entering this year, motivated by a trumped-up belief that voting machines are not trustworthy, a handful of county Republican parties in Texas considered counting their primary ballots by hand. (In Texas, county political parties have the authority to decide how to run their own primaries, although they usually outsource the actual work to the government.) However, hand-counting thousands of ballots requires a ton of manpower, and two of the biggest county GOPs considering it — Dallas’ and Williamson’s — ultimately decided the idea wasn’t feasible.
Instead, party leaders in those two counties settled on a compromise: conduct Election Day voting at neighborhood-based precincts, which Republicans consider to be more secure than the counties’ usual setup of a smaller number of large vote centers where anyone in the county could vote. Crucially, under state law, the county GOPs’ decision forced Dallas and Williamson Democrats to switch to precinct-based voting as well.
The result was mass confusion, as at least several hundred, and potentially thousands of, voters showed up at former vote centers where they usually cast their ballots only to be told they had to go to a different location. According to Kardal Coleman, the chair of the Dallas County Democratic Party, at least 359 Democratic primary voters had been turned away from two major polling places as of 11:30 a.m. on Election Day. That number, of course, does not capture anyone who was turned away after 11:30 a.m., anyone who was turned away from the dozens of other polling places, and any Republicans who were turned away.
Ominously, the chaos ensued even after election officials and parties made a concerted effort to educate voters. Dallas County spent $1 million on mailing notices to voters and airing digital and TV ads to get the word out about the relocated polling places. Both the Dallas County Democrats and Republicans said they had publicized the changes to their voters. And Williamson County conducted interviews with local media and posted on social media and on their website about it.
It’s worth emphasizing that many people who were initially turned away from the polls probably eventually found their correct polling place and were able to cast a ballot. Despite midday fears on Election Day, turnout in the Democratic primary in Dallas and Williamson counties was in line with the statewide average. Based on unofficial results from Wednesday evening, the number of votes cast statewide in the Democratic primary for U.S. Senate was 48% of the number of Kamala Harris voters in the state in the 2024 general election. In Dallas County, that number was 54%; in Williamson, it was 44%.
Voters got some relief when a judge ordered polling places for the Democratic primary to stay open two hours later, until 9 p.m., in Dallas because of the issue. A different judge also ordered two precincts in Williamson County to stay open for both parties until 10 p.m. because of long lines. However, within hours, the Texas Supreme Court temporarily stayed both of those orders, creating doubt about whether ballots cast by voters who got in line after the regular 7 p.m. poll-closing time would count. (Those voters all cast provisional ballots, allowing them to be separated from the undisputed votes.)
As of Friday night, the Texas Supreme Court still had not issued its final rulings on whether those ballots should be counted. A significant number of votes hang in the balance: In Dallas County, 2,316 provisional ballots were cast in the Democratic primary, and 541 were cast across both primaries in Williamson — although those numbers also include any provisional ballots cast for other reasons, like the voter’s eligibility being unclear.
Meanwhile, Republicans in three counties — Calhoun, Eastland, and Gillespie — actually went through with their plans to rely on hand-counting. Under state law, this compelled them to use precincts instead of vote centers on Election Day, too, which created headaches, at least in Eastland County, which had previously used countywide sites. Poll workers at one site told Votebeat’s Natalia Contreras, who was on the ground in Eastland, that at least 89 voters had shown up to vote that day and learned it was the wrong location. At least one was ultimately unable to cast a ballot.
The sudden switch to a precinct-based system and separate primaries for Democrats and Republicans also meant that Eastland County did not have enough accessible voting equipment to put at each precinct — a violation of federal law. At least one voter showed up on Tuesday asking to use the equipment and was told it was unavailable, the county election administrator told Contreras.
And then the hand counts themselves, as they are wont to do, severely delayed the reporting of election results. Gillespie County Republicans did not report full results to the secretary of state until after 5 a.m. on Wednesday morning, long after most counties, as well as Gillespie County Democrats (who were still able to machine-count), had gone to bed. And that was actually relatively speedy: Eastland County didn’t submit its Republican primary results to the secretary of state until around 5:45 p.m. on Wednesday, and Calhoun County did not finish reporting results until Friday morning, after the county GOP failed to deliver its results within 24 hours of polls closing as required by state law.
Taking a long time to report results is annoying, of course, to those of us who tune in on election night hoping to learn who won the election. But more seriously, it can decrease public trust in the election — an argument, ironically, most often deployed by Republicans when complaining about the slow pace of vote-counting of mail ballots in states like California.
Worst of all, research has found that counting ballots by hand is more prone to error than doing it by machine. When Gillespie County Republicans hand-counted their ballots in the 2024 primary, they ended up needing to make corrections in 12 out of 13 precincts. We don’t know yet how accurate the 2026 hand counts in Texas were, but their spotty track record at least raises the specter of the worst-case scenario: that the election results won’t reflect how people actually voted.
In the face of some undeniable problems that disenfranchised at least a few voters, even some Republicans are asking themselves if their prescriptions were worse than the disease. That might be the kind of question it’s better to ask beforehand.
Nathaniel Rakich is Votebeat’s managing editor and is based in Washington, D.C. Contact Nathaniel at nrakich@votebeat.org.
Votebeat is a nonprofit news organization covering local election integrity and voting access. Sign up for their newsletters here.
