Making mail-in ballots universal would be “the greatest scam of all time,” President Trump said Monday in an interview on Fox News. “Dogs are getting them, okay?” he said. “People that have been dead for 25 years are getting them.”

Dead people do receive ballots in the mail. And that apparently is proof enough for Trump that mail-in voting is inefficient and illegitimate. But it’s likely that what Trump really fears is the ballots of the living. Dead people don’t vote. And there’s no evidence of fraud connected to “dead ballots.” Even if there were, those votes would not favor one party or change the outcome of an election.

Dead people get a lot of mail, not just ballots: They get magazine subscriptions, political ads, charitable solicitations, bills, veterans benefits, tax refunds and stimulus checks. This flow continues until someone notifies the sender of the recipient’s death, typically by returned mail, or the sender strikes the recipient from its mailing list. Every state has a record of all the people who were registered in the last election. That is their mailing list, and it is generally two years old.

By the time of the next election, however, a lot of those voters will have died. In 2018, 153 million Americans were registered to vote. Standard actuarial tables suggest that roughly 2.4 million of those voters died over the two-year election cycle. If no one has alerted an election board to those deaths, it is inevitable that at least two million dead people will be getting ballots in the mail if we adopt universal mail-in voting — which is the fault of neither the election boards nor the U.S. Postal Service.

The only way “dead ballots” would affect an election is if third parties were to obtain the ballots and use them to cast votes. The issue then would be voter fraud, not inefficiency. Since those dead ballots are mailed to family residences or returned to sender (election boards), the opportunity for collecting and misusing dead peoples’ ballots is exceedingly small.

The same is true of ballots mailed to people who have moved. Fourteen percent of households change residence every year, which means the number of ballots “incorrectly” directed could be as high as 40 million in any election cycle. But most people leave forwarding addresses when they move, so the ballots will follow them. Furthermore, most moves occur within the same state. Relocated voters who want to cast their ballots have the incentive and opportunity to alert election boards to their new address. Here again, one has to wonder how likely it is that a third party would obtain those movers’ ballots and use them to cast a fraudulent vote.

But forget about dead people and movers for a moment and focus on that question of fraud. Is there evidence of significant fraud in American elections? The answer is no. Oregon, where mail-in-only voting has been the norm since 2000, has confirmed only two cases of fraud out of 50 million mailed-in votes cast in the past 20 years. A recent MIT study observed that of the 250 million ballots mailed nationally in the past 20 years, there were only 143 cases of criminal fraud. The conservative Heritage Foundation came to the same conclusion; the fraud rate was roughly 0.00006 percent.

In view of the extraordinarily low incidence of voter fraud, what is the basis for the president’s concern? He appears to believe that voter fraud is more common among Democrats than Republicans and would reflect a bias against him. If that were the case, the Democrats would indeed gain an electoral advantage from voter fraud, but there is no evidence to support such a notion. And even if all of the documented voter fraud was committed by Democrats, the low incidence of it would make it highly unlikely to affect any election outcome.

Perhaps Trump is really more worried about participation rates than fraud. His campaign may fear that a large portion of the mailed ballots will come from groups that didn’t vote before and are more sympathetic to the Democrats (students, minorities, people living in low-income neighborhoods). The implied larger turnout would tip the table in the Democrats’ favor. A recent Stanford University study showed that such a fear would be unfounded. In the three states (California, Washington, Utah) the researchers looked at, there was no statistical effect on party-line votes from mail-in ballots for elections between 1996 and 2018. The share of Democratic votes from mail-ins differed by only minus 0.01 to plus 0.3 percent from in-person voting. Political scientists weren’t surprised; they say Republicans generally did a better job of facilitating mailed ballots from elder-care institutions that tended to favor conservative candidates. The fear of party bias from mail voting is unfounded.

Two other issues that arise with mail-in ballots are more legitimately concerning. With mailed ballots in hand, voters enjoy a “voting window” that is considerably wider than the traditional voting day. In some states they can vote as early as August. That means that people will be casting their votes before presidential debates and without taking into account the progression of the pandemic and the longer path of the economy, or any events that might later alter their preferences. They will certainly be less informed than voters who cast their votes on Election Day.

But is there a party bias here? Are Democrats more committed to their candidate than Republicans are to theirs? Is Trump’s base so uncertain about their allegiance that they won’t commit until Nov. 3? Political scientists don’t think so. Most experts and pundits point out that Trump has a fiercely loyal and unwavering base that supports everything he does and says. Aren’t they more likely to commit their votes to Trump early on than the Democrats who have been fighting among themselves over vice-presidential candidates and party platforms? If anything, early voting would seem to favor the Republicans.

Then there is the matter of late ballot counts. We want to know the winner by midnight on Nov. 3. That is unlikely, though. If a high percentage of the votes do come in the mail, there is no way the tally can be completed by that deadline. Counting of mailed ballots will take days, if not weeks. Maybe even longer if Trump succeeds in crippling the Postal Service. Unless victory margins are substantial on Nov. 3, we won’t be sure who won for a while. That will create a lot of uncertainty and probably another rash of accusations about voter fraud and accounting irregularities. That’s what happened in 2000, and the country somehow persevered. Americans can get through this again, if everyone recognizes that the problem isn’t dead people.