Summary: The pandemic has created much giddy enthusiasm for absentee and mail-in voting, but little sober assessment of its risks and benefits. In fact, evidence of better turnout is unclear, while evidence of rejected ballots and uncounted votes is undeniable. Generally, we can predict that for every 100 voters we convince to give up polling-place voting in favor of absentee, we will lose at least two ballots. This article makes the case that putting your ballot in an envelope for an election official to cast at a later time is inherently less secure than casting your own ballot and that voters should vote absentee only when truly necessary.
This blog post is a condensed version of a more comprehensive paper, which provides additional in-depth explanation, references, and examples.
Let’s start with a quiz—just two easy questions.
1. Imagine a polling place where voters mark their ballots by hand but instead of inserting the ballots into a tabulator, they give them to a poll worker. The poll worker puts 1 in every 110 ballots into a reject pile and casts the rest.
Would you: a) Be happy with that practice, or b) Run screaming to the district attorney to report intolerable interference with voting rights?
2. The State of Washington conducts elections by mail and rejects 1 in every 110 submitted ballots. Rejection rates vary among cities, however, so a voter’s odds of being disenfranchised depend on where they attempt to vote.
Would you like your state’s elections to be like Washington’s? a) Yes. b) No.
If you answered “b” to both questions, you’re with me and the good folks of Washington State, who would like to reduce their ballot-rejection rate and make rejection practices more consistent.
There are two ways you can vote: you can cast your ballot yourself (in-person voting), or you can submit your ballot in an envelope, expecting it to be cast later by an election official (absentee voting.)
In-person voting is simple and transparent. You go to a neighborhood polling place on Election Day. A poll worker places a paper ballot directly into your hand. You mark the ballot privately, and with your own hand insert it into a tabulator. This is done in one visit, which typically takes less than ten minutes except at the busiest times of day, in the busiest elections. For what it’s worth, I timed my polling-place visit on April 7—it took me 3 minutes and 47 seconds between entering the building and casting my ballot. My son took 8 minutes because he also had to register.
Once the ballot is placed in your hand, any interference would be extraordinary, obvious, and illegal. If you spill coffee or mark too many candidates in one race, you can re-do your ballot. The risk that an in-person ballot will not be counted is close to zero.
Absentee voting is a dicier proposition. It is necessary for some voters but requires more steps over a longer time. More things can go wrong. The problems are harder to prevent; sometimes undetectable; and sometimes impossible to correct in time.
We can think of an absentee ballot’s travels, from voter request to ultimate processing by a computer tabulator, as a long pipeline. Charles Stewart III of the MIT Election Data and Science Lab estimated that for every 100 requests for absentee ballots that enter the pipeline, 79 ballots will ultimately be counted. Some of the lost 21 ballots will be those of voters who chose not to submit the absentee ballot. Many other ballots will be lost to accident, interference, or rejection. In Wisconsin’s April 2020 election, conducted during pandemic stay-at-home orders, for every 100 voters who requested an absentee ballot, almost 89 ballots were counted while more than 11 were not.
To begin to understand these leaks, let’s start near the end of the pipeline, with the marked ballots that have made their way back to election officials. Deliberate rejection is one of the leaks that is easiest to quantify because officials keep track of how many ballots they reject.
In Wisconsin, an absentee ballot envelope must contain the voter’s name, address, and signature, plus a witness’s signature and address, plus other information usually filled out by the clerk. You can view the form here. Absentee voters can be disenfranchised if any of this information is missing, incorrect, or illegible. Absentee ballots are rejected if the envelope glue has failed so that the envelope is no longer sealed when it comes time to count the ballot.
Understandably, this makes voters nervous. Washington State went to all-mail elections in 2005. Seven years later in 2012, their voters were among the least confident in the nation. Only 52% were willing to tell surveyors they felt “very confident” their votes would be counted.
Washington voters are perceptive, not paranoid. Their votes are the least likely to be counted. About 1 in every 110 Washington voters’ ballots were rejected—the highest rejection rate in the nation. Oregon was about the same—0.86%, and neither of the other two vote-by-mail states that year (Utah and Colorado) had much to brag about, either.
A missing signature is Wisconsin absentee ballots’ biggest problem, but there are other reasons for rejection. Records of Wisconsin’s 2016 recount contain evidence of frequent local differences of opinion over the grounds on which an absentee ballot could be rejected. The Wisconsin Elections Commission (WEC) did its best to answer the local clerks’ many questions but finally could do no better than to tell local officials they were on their own in many cases to figure out which absentee ballots to reject.
The danger of subjective inconsistency is bias and sadly, it shows up in the hard data. Researchers followed 2.6 million absentee ballots in Florida’s 2018 elections, 1.2% of which were rejected. They could see that minority voters’ absentee ballots were twice as likely to be rejected as those from non-minority voters. Researchers wrote: “younger voters and voters needing assistance are disproportionately likely to have their mailed ballots rejected. We also find disproportionately high rejection rates for out-of-state and military dependents.”
In addition to the thousands of disenfranchised voters, thousands more are inconvenienced. When election officials decide to reject an absentee ballot, they often try to contact the voters, who can sometimes with quick extra work correct the problem before the deadline. Those voters must run down to city hall to correct a flaw on an envelope or submit additional paperwork by mail—unless they just give up on having their votes counted.
But they don’t always get that chance. Pat Haukohl was a Waukesha County Board Supervisor for 18 years and is a regular voter. She voted absentee for the first time in April, and her ballot was rejected because the witness (her husband) did not include his address. She did not learn of the problem until after Election Day, from a reporter.
An absentee ballot is not yet safe once accepted. Here, again, records of Wisconsin’s 2016 Presidential recount contain revealing evidence. In Dane County alone, recounters found 66 valid absentee ballots that had been misplaced and left in their envelopes, accepted but not cast, in addition to 644 wrongly rejected absentee ballots.
At the very end of the pipeline, officials feed the ballots, now removed from their envelopes, into computer tabulators. These machines reject ballots they cannot read (e.g., smeared ink) or cannot understand (e.g., too many marked votes). Polling-place machines also reject flawed ballots, but polling-place voters are there to fix whatever’s wrong. Absentee ballots are not only more likely to be damaged (kitchen-counter coffee stains; wrinkles and ink transfer from being folded; and being torn when removed from the envelope), but absentee voters are unable to complete a replacement ballot.
Election workers are supposed to ‘remake’ the unreadable absentee ballots—that is, copy the votes onto machine-readable ballots and cast those instead, if they can tell what the voter intended. But elections are run by an army of temporary workers, lightly trained and supervised, who get no more than four days’ on-the-job experience every year. They don’t always know or remember all the rules. Very often, they instruct each other, and in doing so, make incorrect practices standard.
Again, the 2016 recount revealed the problems. Several Wisconsin counties discovered poll workers had on Election Day failed to remake thousands of rejected but human-readable absentee ballots. Instead, the poll workers had been pushing an override button to force the machine to count the absentee ballot even when it could not count the votes. In this way, poll workers saved themselves work but sacrificed the absentees’ votes. Election officials rarely perform routine quality-assurance reviews, so such problems go undetected and uncorrected in elections that are not recounted—that is, most of the time.
The problems I’ve described to this point affect only those ballots that successfully made it back to the election offices. But one big leak is right in the voters’ homes. Ballots might be mistaken for junk mail and thrown away. They might be misplaced; damaged; buried in a stack of bills; or set aside until too late. Among the states that report complete data on absentee voting, 36.3 percent of uncounted absentee ballots went uncounted because they arrived too late at the elections office.
Coercion may be the most intractable—and heartbreaking—cause of absentee disenfranchisement. Journalist Rebecca Solnit recounted a story like many she heard as she studied this issue. A campaign worker told her:
I can’t stop thinking about this woman I met while door-knocking for Beto O’Rourke in Dallas. She lived in a sprawling low-income apartment complex. After I knocked a couple of times, she answered the door with her husband just behind her. She looked petrified and her husband looked menacing behind her. When I made my pitch for O’Rourke, her husband yelled, ‘We’re not interested.’ She looked at me and silently mouthed, ‘I support Beto.’ Before I could respond, she quickly closed the door.”
When absentee ballots arrive in that household’s mailbox, what are the chances the wife is going to be able to mark her own ballot, in peace, with her own choices? According to the American Association for Marriage and Family Therapy, 20% of American marriages are affected by abuse, and 90% of the targets of controlling abuse are women. If those facts are true, mailing ballots to every registered voter could be undetectably disenfranchising thousands of women.
Fraud is more of an illicit tap into the pipeline than a leak—and it does exist. In a practice known as ballot harvesting, political operatives collect absentee ballots from voters and deliver them to election officials. Honest operatives may provide a genuine service to voters, such as reviewing the envelopes to make sure they are filled out correctly. But the practice can just as easily be corrupt. Ballot collectors can supervise the voters while they are completing their ballots to dictate, coerce, intimidate, or pay them for their votes. Once out of the voter’s sight, collectors can spoil the ballots to make them uncountable or dispose of them. In Wisconsin, simply opening the envelope is enough to get an absentee ballot rejected. Democrats who say fraud doesn’t happen are shooting themselves in the foot. It may already have cost them a congressional seat.
Another opportunity for fraud created by large-scale absentee voting is the possibility of providing voters with faulty or fraudulent materials. This has already happened twice to me. Before an election several years ago, I received an unsolicited blank ballot and return envelope—with the envelope addressed to the wrong municipality. Had I been inattentive or naive, I might have mailed it and skipped going to the polls on Election Day. Shortly before the April 7 election this year, I received a form for requesting an absentee ballot, also unsolicited. It was addressed to the correct municipality, but my preprinted name was incorrect: Kim, rather than McKim. In a recent Georgia incident, the Secretary of State mailed materials that gave absentee voters the incorrect date for a rescheduled election—the ballots said May 19 and the election was scheduled for June 9.
Those incidents might have been honest mistakes, rather than deliberate attempts to interfere. But accidents demonstrate things that could be done deliberately and may provide cover for them.
Earlier leaks in the pipeline prevent ballots from even reaching the voters. I will skim over this information because in May the Wisconsin Elections Commission issued an excellent 24-page report that described those leaks well, with solid data and description of real-life problems that affected absentee voters in the April 7 election. The report describes only the first segments of the pipeline because those are the only parts for which WEC has responsibility.
The April election marked Wisconsin’s first, sudden, foray into large-scale vote-by-mail, but many of the snarls illustrate problems that could crop up in any vote-by-mail election. WEC Technology Director Robert Kehoe explained at the Commission’s May 20 meeting that staff will correct the problems they can, but he would not promise no others would arise: “Until those things happened, they were unanticipated, so other unanticipated things can happen.”
Among the problems: The City of Milwaukee submitted a large batch of 8,607 approved absentee ballot requests to the WEC computer so that mailing labels could be printed. The batch was still running when state staff shut down the system after midnight for maintenance because they believed all batches submitted during the workday had been completed. This caused the system to indicate that the requests had been processed when they had not.
Those are some of the risks. So what about the benefits? Proponents believe early, mail-in, and absentee voting increases turnout. That belief persists even in the face of evidence to the contrary. Here, for example, is reporter Jesse Opoien, commenting soon after the November 2016 election: “The decline (in turnout) — down nearly four points from 2012 and three points from what state elections officials projected — was all the more stunning as it followed record-high early voting numbers.”
Reporters like Opoien don’t seem to realize the possibility that heavy early voting could be nothing more than dedicated eager voters doing their dedicated eager thing, while the less-engaged marginal voters (the ones whose conduct makes the difference between high and low turnout) are unexcited by the opportunity to submit a ballot a month before Election Day. Or worse, that the low-enthusiasm voters read the news about thousands of eager early voters; concluded the election was already decided; and chose not to vote.
Academic studies show mixed turnout results at best, and none find large effects in either direction. Unfortunately for researchers, states rarely make one change to their election laws at a time, so we cannot know which might have affected turnout. For example, Colorado’s vote-by-mail law was adopted in 2013 as part of a comprehensive package that included same-day voter registration at the polls, expanded in-person voting to several days; added provisions that reduced deactivation of registered voters who miss several elections and included a ‘less onerous’ Voter ID requirement.
The most frequently offered explanation of why absentee voting might increase turnout is convenience. The home page of the National Vote At Home Institute, an organization that promotes voting by mail, explains that voters:
don’t have to take time off work, drive to a polling place or stand in long lines. Vote-by-mail equally serves everyone from seniors and disabled voters, who might have trouble getting to the polls, to rural voters a long way from one, to a single parent working two jobs, a busy family, sick kids, or someone with an unexpected business trip.”
Absentee proponents cite other problems with in-person voting that could be solved in ways that don’t expose ballots to additional risks. For example, some vote-by-mail proponents say absentee voting gives them an opportunity to understand what races and candidates are on the ballot before they arrive at the polling place, although that opportunity is not denied to them now.
Some mail-in voters imagine absentee ballots are hand-counted. In fact, absentee ballots are counted by computers from the same few companies that provide polling-place equipment. Their software is just as vulnerable to hacking at the source or before it reaches the computers. Absentee voting does require hand-marked paper ballots, and that’s a good thing. But paper ballots’ security benefits are obtained only when the ballots are used in routine outcome-verifying audits of the type recommended by national election-security authorities—and Wisconsin officials still won’t do those.
Another touted benefit of absentee voting is avoiding long lines at polling places. However, the waiting-line problem is often exaggerated. In real life, few voters encounter Election-Day lines in excess of 10 minutes. And if voters demanded their election officials follow accepted best practices, even that could be improved. Long lines would form only on rare occasions such as power outages.
Knowing what we know about benefits and risks, how would large-scale absentee voting affect Wisconsin’s November 2020 election? Unfortunately, no signs point to good results in Wisconsin if voters rely on absentee ballots at the rate they did in April.
The law of diminishing returns presents Wisconsin with one enviable barrier to increased turnout. Wisconsin turnout is already reliably high compared to other states. In the 2016 Presidential election, Wisconsin voters turned out in greater numbers (69%) than those in three of the four states that conducted all-mail elections that year: Oregon (68%), Washington (66%), and Utah (58%). In fact, only four states had better turnout than Wisconsin, only one of which was a vote-by-mail state, Colorado (72%). None of the top three states—Minnesota (75%), Maine (73%), and New Hampshire (73%)—were vote-by-mail states.
Further, the states that claim positive effects all have absentee-voting procedures that are easier and simpler than Wisconsin’s. In particular, absentee voting amplifies the vote-suppressing effects of Wisconsin’s strict Voter ID law. Voters must not only possess the right kind of ID, but must also have a way to submit it electronically.
Things don’t look any better on the risk side of the equation. The WEC report described what can go wrong in the part of the pipeline managed by a professional, full-time election staff. The rest of the pipeline, in Wisconsin, is managed by 72 county clerks, 1,850 municipal clerks, and thousands of poll workers, without oversight by the WEC. It is unrealistic and unfair to expect them in a few months to develop and practice reliable procedures for processing a flood of absentee ballots.
Giving the United States Postal Service a key role in November adds another layer of risk. The USPS just isn’t set up to be a critical moving part of the elections machine: Witness the US Supreme Court’s ruling that ballots postmarked by Wisconsin’s April 7 Election Day could be counted, only to have it come to light that the USPS does not put postmarks on all the mail. And with the current administration’s stated intent toward both the USPS and voting by mail, reliable operation of the USPS is far from a sure bet for November.
We take the neighborhood polling place for granted. It is too plain, too obvious, too old-school to have many enthusiastic advocates. It has no novelty value. It doesn’t fit with our modern individualistic lifestyle. Even before the pandemic, we were shopping online instead of visiting local retailers. We were socializing on Facebook, not on the sidewalk. Voting from a kitchen counter is compatible with our e-lifestyle in a way that visiting with our local officials and neighbors at the town hall is not.
If absentee voting becomes standard, as I fear it will, there are a few things responsible voters can do, collectively and individually, to minimize the damage.
First, we must expect professional-caliber management from our local election officials and hold them accountable. Take a look at that May 2020 report from the Wisconsin Election Commission. That is what it looks like when officials hold themselves accountable: 1) The WEC collected solid data on performance; 2) They performed solid investigation into things that went wrong; 3) They made a public commitment to measurable improvement.
Expect the same of your local election officials. Give it a try: Call your municipal clerk and ask about the absentee-ballot rejection rate in the most recent election; what the performance target is for the next election, and what he or she is doing to meet it. A clerk who is actively managing his or her operation will know those things. If your clerk doesn’t, arrange a meeting between the clerk, you, and other local voters to figure out how to make absentee ballots more secure in your community.
And don’t encourage any other voter to vote absentee until you do.
As an individual, here’s a checklist of what you should be prepared to do if you plan to vote absentee. However, be aware that even if you follow each of these steps, some risks are outside your control.
Personally, I am not going to bother with any of the extra work required for absentee voting, nor am I going to accept the extra risks. I will look at my local ballot ahead of time and research my options. I will then visit my neighborhood polling place on Election Day and follow any public-health advice then in effect. I will say “Hi” to my neighbors and “thank you” to my municipal clerk and the poll workers; mark my ballot privately; and cast it with my own hand directly into the tabulator.
Simple. Easy. Convenient. Secure.
- Vote-by-mail has not yet been discussed in any meeting among participants in Wisconsin Election Integrity. This article reflects only the references cited and my own observations, not the consensus of the group.
- Thanks to reviewers Dr. Barry Burden, Director UW-Madison Elections Research Center; Charlotte Goska, Coalition of Voting Organizations of Brown County; my husband Keith Nelson; and Rebecca Alwin. Their suggestions and corrections were valuable; any remaining flaws are mine. The Wisconsin Elections Commission was also provided with a draft; this post will be updated if they offer any comments.