In recent weeks, President Donald J. Trump has repeatedly called for eliminating mail-in voting (except under special circumstances.) The President believes that mail-in voting undermines the security and credibility of elections. A recent, small election in Delaware illustrates the continuing validity of such concerns.

Last month (August 5^th), a special election was held to determine who would represent District 20 in the Delaware House of Representatives, which includes much of Lewes and Milton, with Republican Nikki Miller competing against Democrat Alonna Berry. Absentee ballots played a pivotal role in this election. Miller garnered 4,342 votes at the polls (Election Day plus early voting), compared to 4,184 for Berry. But Berry’s 469 votes from absentee ballots swamped Miller’s 194, yielding Berry’s slim (1.3 percent) margin of victory.

To be clear, this lopsided outcome among absentee voters does not imply fraudulent voting. It illustrates the concept of “selectivity bias” — there is no reason to expect the (non-randomly drawn) absentee cohort to be representative of the voter population as a whole. In this case, it turned out to be highly non-representative.

However, an additional statistic included in the official tally from the election should give pause. Beginning in 2025, the Delaware Department of Elections began reporting the number of “undervotes” in an election—instances in which an individual “voted for less than the specified number of candidates a voter may vote for in a specific race.” In the District 20 election, 55 out of a total 721 absentee ballots, or about every 12^th ballot, were undervotes, meaning that they were returned blank. 

Wait, what? Two other elections have taken place in Delaware during 2025—special elections for the State Senate held in February—each of which were also single choice elections (where an undervote is equivalent to leaving the ballot blank.) In these two elections combined, just two out of the 1,136 absentee ballots received were undervotes. 

Why would so many absentee voters in the August election in District 20 have returned a blank ballot? No explanation has been forthcoming. 

Why does it matter? In the first place, this phenomenon points to a possible ballot harvesting scenario in which partisan operatives sought out absentee voters and collected absentee ballots. Unlike in most states, there are no explicit restrictions on ballot harvesting in Delaware. But such an extraordinarily odd outcome could arise from inappropriate or overly aggressive harvesting tactics, which should be a concern. (For example, pestering of elderly voters who are not interested in voting might prompt them to hand the ballot harvester a blank ballot.) 

Additionally, in banking, voting, or other contexts where integrity and trust are of paramount importance, statistical anomalies naturally merit investigative attention. To allay concerns, state officials should conduct a review of the absentee voting during this election 

Moreover, the officials should be transparent about what they may find. Lack of transparency is not conducive to trust in elections.

The facts that Delaware has no restrictions on ballot harvesting nor any signature verification process for absentee ballots do not help matters. In other words, lack of trust is compounded where the rules governing remote voting are lax.

The broader issues at stake here are not unique to this particular minor election. Undoubtedly, analogous outlier situations have arisen elsewhere. For example, as discussed in my recent book The Big Flip, on Election Day 2020 in Pennsylvania, over 200 provisional ballots were deposited in rural McKean County as replacements for mail-in or absentee ballots (see Tables 5.4 and 6.4 in the state’s official election report.)  Every one of the voters depositing these ballots came from outside McKean County (as indicated by the official tally showing zero provisional votes from county residents), an extreme outlier situation.

Thus, there are at least four important lessons one can draw for elections nationally: 

• Collection of expanded and detailed real-time data on mail-in voting activity should be prioritized in every state (it is only by chance that Delaware election officials began including the additional statistic on undervotes in 2025.) 
• Formal and systematic use of data and statistics to monitor mail-in and provisional voting for indicators of potential malfeasance should be the norm. 
• Officials should prioritize transparency when called on to review election outcomes or anomalies.
• Best practice rules and procedures to mitigate the integrity risks posed by mail-in voting, including restrictions on ballot harvesting, should be implemented consistently across states.

In short, the President is spot-on about the need for stronger guardrails and more trustworthy processes to ensure the integrity of the mail-in vote. The President should encourage a serious, non-partisan examination of how to best accomplish that, which may (or may not) be possible without abolishing remote voting entirely.