Will your vote count in 2020? Will your neighbors’ vote count? That might depend on whether you vote in person or try to vote by mail. In states with a significant vote by mail component, hijacking mail-in ballots is the tool of choice for those trying to steal elections.

In 18 states, a photo identification is needed to vote on Election Day, with another 17 states requiring at least a non-photo ID. In the 2016 General Election, about a quarter of all the votes cast were by mail. But unlike when voting in person, voting by mail almost never requires an ID.

Voting by mail, while convenient for the voter, lacks the protections that voting in person provides. The ballot can get lost in the mail. The mail-in ballot application can be completed by others, sometimes with false information. Ballots, when they arrive in the mail, can be intercepted by political operators. And, in some cases, largescale fraud can take place where, through nominal gifts of food or alcohol, voter intimidation, or deception, a professional ballot harvester can simply substitute his vote for the voter’s and turn in the ballot.

Ballot harvesting got so egregious in 2018 in North Carolina—where it’s illegal (as it is in Texas)—that Republican Congressional candidate Mark Harris saw his 905-vote victory annulled in early 2019, with a redo ordered.

Mail-in ballot fraud is so pervasive in some parts of Texas that they have a local name for those who broker election victories: Politiqueros. Unlike in North Carolina, there are rarely consequences for ballot harvesters or the campaigns who hire them in the Lone Star State.

To be clear, campaigns do have a legitimate reason to turn out their voters. Hiring people or engaging volunteers to help and encourage people to vote, whether early and in person, by mail, or on Election Day, is a critical part of a winning strategy.

But ballot harvesters can get aggressive, they can cut corners, and they can guarantee votes for the politician or group that hired them by taking physical possession of ballots and voting in place of the legal voter. And in Texas, vote harvesting is illegal, at least to the extent where state law prohibits a voter’s absentee ballot being completed and mailed by someone other than a close relative.

The practice is more widespread and problematic than most people believe.

In the 2017 special session of the Texas Legislature, lawmakers were concerned enough by illegal ballot harvesting that they passed SB 5. The law, in effect for the 2018 election cycle, tightened mail-in ballot rules and increased criminal penalties for ballot fraud.

Instructively, as Texas tightened its mail-in ballot laws in 2017, California weakened them by the same measure. This unleashed a massive deployment of ballot harvesters on a statewide and partisan scale, resulting in the largest historical victories for California’s Democrats since the blowout Watergate aftermath elections in 1974.

But just because a law was passed in Texas doesn’t mean the practice of illegal ballot harvesting has ended or has even been curtailed.

The 2018 General Election in Texas was highly competitive, shaped largely by the most-costly U.S. Senate race in American history between incumbent Senator Ted Cruz and then-U.S. Representative Robert (Beto) O’Rourke. Thus, typical electoral losses for the party controlling the White House were amplified by a larger-than-usual turnout. In the aftermath, Republicans lost two congressional seats, two state senate seats, and 11 state house seats, and had very close calls in many others.

Analyzing turnout and mail-in ballot data from the 2018 election sheds light on the extent to which mail-in ballots might have played a leading role in the competitive electoral landscape.

Texas has strict rules on mail-in balloting, limiting it to people aged 65 or older, the disabled, and people who will truly be out of town during the election. As a result, the share of mail-in ballots cast in each county should closely track the share of that county’s population 65 and above. But not exactly. For instance, in Harris County—essentially Houston—about 32,000 more votes were cast by mail than would be expected based simply on the age of voters in the county. This amounted to 0.7% of the countywide turnout. In Hays County, just to the south of Austin, some 1,400 additional mail-in ballots were cast above what statistics would predict—about 0.6% of votes in that county.

Comparing the level of mail-in ballots across Texas in 2018 to what demographics would predict (where there were highly competitive state or federal elections) shows an extremely high correlation. (For the math nerds, a regression analysis shows an Adjusted R Square of 0.026 between larger than expected use of mail-in ballots and competitive races.) Now, much of this heightened mail-in ballot activity may well have been perfectly ethical and legal. But some of it might not have been.

For instance, in Hays County, where mail-in ballots were heavily used in 2018, campaign workers noticed a far larger turnout than usual at Texas State University at San Marcos, with precincts in and around the campus generating up to five times the usual number of votes. Were any of these votes by mail-in ballots? If so, how many of the requests for a mail-in ballot made by improperly checking the “disabled” box? Given that this has been a common problem in the Rio Grande Valley, why not on a campus flooded with dollars and organizers paid for by California billionaire Tom Steyer as he was laying the groundwork for this 2020 presidential campaign?

Some 165 miles to the east in Houston, national labor unions have been spending millions of dollars a year to organize in America’s fourth-largest city. Given the key role unions took in California’s 2018 electoral blowout for the Democrats, might fraudulent mail-in ballot harvesting have been part of those efforts? Some 32,000 additional mail-in ballots in excess of turnout and demographic numbers in Harris County suggest that possibility.

The statistical evidence tying mail-in balloting to competitive Texas races in 2018 suggests that campaign operatives might have been testing out tactics—some potentially illegal—to boost the mail-in vote. If Texas is seen as a battleground again in 2020, how many votes might be purchased for the ballot harvesters to reap? It behooves election officials, party volunteers, campaign workers, and law enforcement to be on the lookout for improper harvesting of mail-in ballots.