Will your absentee ballot be counted? The risk depends on where you live.

Summary: Even absentee ballots that arrive on time can legally be rejected and not counted. In the April election, Wisconsin officials reported they rejected 1.8% of the ballots they received on time. Officials blame voter error (most commonly, missing signatures). Voting-rights advocates are trying to reduce the problem with voter education.
However, rejection rates varied widely among municipalities in a way that indicates local administrative practices play a big role in determining each ballot’s risk of rejection that is, absentee voters in some municipalities face a much greater risk of disenfranchisement than voters in others.
This post explains some of the reasons why rejection rates vary and what could be done to make them more consistent in November and in future elections. It argues that we MUST start paying attention to administrative practices and stop placing the blame wholly on voters. It argues that the easy, critical starting point is for the Wisconsin Elections Commission to routinely calculate local rejection rates following each election.

* * *

I’ve been studying absentee-ballot processing in Wisconsin and in other states for several months now and our state’s increasing reliance on it scares me to my bones.

Here’s why: Absentee voting is necessary for a few voters, but any ballot submitted in an envelope faces a gauntlet of risks that don’t endanger ballots that are marked and cast at the polling place. It is unavoidable. While polling-place ballots go directly from the voter’s hand into the tabulator, absentee ballots are not cast until days or even weeks after they leave the voter’s hand. During that time, those ballots face multiple risks. It is inevitable that some will never be counted.

But how many? And whose?

Voters who are very careful and diligent can reduce some of these risks, but that doesn’t help the rest of the voters. Other threats are beyond any voter’s control.

If someone was taking bets now, I’d put my money on something a little north of a 3% loss rate (from all causes) for absentee votes in Wisconsin next November. That is, for every 100 early or absentee ballots marked, the votes on 3 of them will never be counted. Three percent might seem insignificant to some (not you, I hope), but if 80% of Wisconsin’s voters submit absentee ballots in November as expected, tens of thousands of votes will be lost.

That sort of shrinkage could alter the outcome if the electorate is fairly evenly divided; if the voters for the more popular candidate lose 3% of their votes by putting their ballots in envelopes; and if the voters for the less popular candidate lose none of theirs, by going to the polls and putting their ballots directly into the tabulator with their own hands. That’s the situation shaping up in Wisconsin as I write this. I can give myself nightmares if I let myself imagine the chaos and recriminations that will ensue if uncounted absentee ballots determine who gets Wisconsin’s electoral votes.

What are the risks for absentee ballots? They start the moment an unmarked ballot arrives in a voter’s home. Some absentee ballots get lost in the home before they’re even marked; some get misplaced or misdirected in transit or after they arrive at the elections office; some arrive too late; and yes, some could be lost to undetected fraud. No matter how badly we wish otherwise, fraud is a risk whenever marked ballots leave the voters’ hands long before they are counted. My Illinois friends have conniption fits when they learn Wisconsin has no measures in place that could detect the problem if a municipal clerk was to go in after hours, remove some ballot envelopes submitted by voters, and replace them with ballot envelopes containing ballots he or she had prepared. (Illinois poll workers compare the signatures on the ballot envelopes with on-file signatures just before they cast the ballots, which deters substitution.) Wisconsin voters, however, are proud of their capacity for trust, and they are not going to listen to the Illinois voters’ worries about insider fraud. So there’s no point in me talking any more about fraud.

Instead, this blog post will focus on only one risk: legal rejection of on-time absentee ballots. Based on my study of 14 municipalities, I estimate that this one problem alone will disenfranchise more than 40,000 Wisconsin voters in November if absentee voting reaches 80%. My estimate is based on 3 million ballots cast, 80% of them absentee, and a 1.7% rejection rate for those ballots — the April rejection rate I found in those 14 municipalities (8 larger cities and 6 smaller ones). (Note: the official figure is 1.8%, but I’ll use my sample just to be consistent.)

Legal rejection of on-time ballots could disenfranchise more than 40,000 Wisconsin absentee voters in November —- 1 in every 59.

Voters might be surprised to learn that so many absentee ballots are rejected, but election officials already know. The Wisconsin Elections Commission has the data and calculated the statewide rejection rate. What the Commission hasn’t done — and does not intend to do, at this writing —– is to calculate the individual municipalities’ rejection rates.

As a result, Commissioners will be as surprised as anyone to learn that individual municipalities reject ballots at rates ranging from 0% (that is, none) to at least 3.7% (that’s 1 in every 27 on-time absentee ballots!). The true top end of that range is higher unless the strictest municipality in the state luckily made it into my small sample.

In addition, Commissioners will be surprised to learn that even within a single city, rates can vary dramatically among the wards. An individual ward can have a rejection rate three times larger than that of other wards— and again, the range is probably even bigger unless I fortuitously picked the worst city in the state for my sample.

Madison rejects 1 in every 126 on-time absentee ballots. The City of Racine rejects 1 in every 27.

From my reading of studies done in other states, I knew that variable local rejection rates are a recognized problem. So when I noticed that a grant application from five of Wisconsin’s largest cities contained raw data on rejection rates (not percentages), I calculated their rates. I was shocked to discover that an on-time absentee ballot in Racine was 4.7 times more likely to be rejected than an absentee ballot in Madison!

I assumed the Wisconsin Elections Commission would be similarly concerned. I called and asked for the rejection rates for all 1,850 municipalities. I learned the WEC has the data to calculate the rejection rates, but has not done that and has no plans to do it.

So I requested the data files. Volunteers with Wisconsin Election Integrity could do the definitive statewide analysis ourselves. WEC’s response: “Hand over $24,000 and the data are yours.”

So … what you are reading right now is an analysis done at my kitchen counter with data I collected directly from only 14 municipalities by email and phone. That’s the best monitoring of Wisconsin’s absentee-ballot rejection rates you can find on this planet. None better.

That is not how it should be. Without knowing comparative rejection rates among Wisconsin’s municipalities, even the local officials cannot know which municipalities are out of line, and no one can identify realistic targets for improvement.

So if anyone from the Wisconsin Elections Commission wants to argue with my observations or estimates, I challenge them — no, I beg them — to do their own calculations of municipal and ward-by-ward rejection rates and release the results.

If the Wisconsin Elections Commission has any question or problem with the information in this blog post, they should DO THEIR OWN analysis and release the results. Prove me wrong; you have the data.

For readers who believe Wisconsin elections deserve better: The Wisconsin Election Commissioners can be reached at elections@wi.gov. Put “Message for the Commissioners” in the subject line, and tell them you think their staff should be monitoring local rejection rates more diligently than an old lady in Waunakee.

Anyway, here’s what I found out …

* * *

How and why are on-time absentee ballots rejected? Talk to any election official and you will be told voter error is the sole cause of all legal rejections. But the extreme variation in local rejection rates within even my small sample indicates that administrative practices play a big role in deciding which absentee ballots are counted, and which rejected. City of Spooner voters are simply not 8.5 times more intelligent or more careful than voters in the City of Green Bay. Something else is going on that causes Green Bay’s ballots to be rejected at more than eight times the rate of Spooner’s.

City of Spooner voters are not 8.5 times more intelligent or careful than Green Bay voters. Something else is causing Green Bay’s ballots to be rejected at more than eight times the rate of Spooner’s.

After absentee ballots arrive at the local election offices, clerks are required to review each to make sure it fulfills the requirements for a count-able ballot. First, of course, the name and address must be that of a registered voter. Some requirements are unambiguous, including those that cause most rejections: Both the voter and a witness must have signed the envelope and provided their addresses. When that information is missing, the ballot can be rejected.

But even unambiguous requirements can be applied either strictly or leniently. For example, the witness might have written their address in the wrong spot, or smeared the ink. Some municipalities, some reviewers, might reject those ballots; others might accept them.

And a high rate of voter errors cannot be blamed entirely on the voters: the root cause of poor voter compliance can be administrative. Municipalities are free to design their own ballot envelopes, as long as they contain the required elements. They are free to write their own instructions. They can pre-print more or less of the required information on the envelope. Good forms design and instructions minimize voter error; bad design and instructions encourage it. More pre-printed information fosters fewer voter errors. If I was doing a serious study of the causes of rejection-rate variation, this is where I’d look first.

Inconsistent interpretation of more subjective requirements could be another cause. Here’s an example where you can be the judge: Officials are to reject envelopes that show signs of tampering. Consider an envelope that’s been unsealed and then taped shut. Would you accept it, assuming the voter sealed the envelope before inserting the ballot, realized the mistake, and fixed it? Or would you reject it on the grounds that the envelope shows signs of tampering? Clerks and election workers are sure to will make both decisions while reviewing ballots in November.

Other things are completely outside the voters’ control. If the ballot envelope was damaged in transit in a way that could be interpreted as a sign of tampering, it can be rejected. If the glue fails and the envelope comes open, it will usually be rejected. (But before you double-seal your ballot with tape, remember tape can be interpreted as a sign of tampering.)

Clerks’ errors can also invalidate an absentee ballot. Those who observed during the 2016 Presidential recount know that absentee ballots can be rejected if the clerk neglected to write his or her initials on the envelope, but often are not. Savvy voters know to double-check for the clerks’ initials before they submit an absentee ballot, but most voters are unaware of that risk.

The next step is deciding whether, when, and how to contact the voter if a ballot is reject-able. If the voter fixes the problem before polls close, the ballot is called ‘cured’ and the votes are counted. I did not find any clerk who keeps a record of how many ballots were cured, so we do not know how many absentee voters are merely inconvenienced, rather than disenfranchised, by absentee-ballot problems.

Contrary to popular assumption, Wisconsin clerks are not required to contact the voter when they determine a ballot is flawed. Clerks sometimes just reject the ballot without telling the voter. Other times they just fix it. For example, if the witness address is missing but the signature is legible, some clerks will look up the witness’s address on other city records and fill it in without notifying the voter. I suspect it’s practices like these that explain much of the very low rejection rate in rural communities. My sample included six municipalities with populations around 2,500, four of which rejected no absentee ballots. The other two both had rejection rates under 0.6% (1 in every 167 ballots). In contrast, the bigger cities in my sample rejected 1.7% of their on-time absentee ballots (1 in every 59).

Most clerks, however, try to contact most voters who submit reject-able ballots. Again, municipal clerks have several choices:

  • Putting the reject-able ballot in an envelope and mailing it back to the voter, with instructions to correct the problem and resubmit it;
  • Sending the voter a letter telling them of the problem and inviting them to come in to the office or go to the polls on Election Day to correct the problem;
  • Invalidating the ballot and sending the voter a replacement;
  • Looking up the voter’s phone number or email on the voter-registration record (not always there), and telling the voter about the options for curing the ballot; and
  • Making only one or several attempts to contact the voter.

You can see how these practices would have different effects on the ultimate rejection rate. They offer the voters more or less time to fix the problem, and they place more or less additional workload on the voter — and one municipality will do it differently than another.

By now, I hope you’re starting to see not just how ballots in different cities face different levels of risk, but how ballots within one city could easily be treated differently, depending on the voter. Studies in other states consistently find higher rejection rates for low-income, young, old, and minority absentee voters. Election workers don’t have to make deliberately biased decisions. If they are normal humans with a normal amount of implicit bias, they have to deliberately try not to.

Janesville Municipal Clerk Dave Godek helped me understand that the problem might not be just that low-income, minority, and younger voters make more mistakes, or that election workers are biased against ballots that come from the ‘wrong’ neighborhoods. He explained that his city’s Ward 3 (7% rejection rate, or 1 in every 14 ballots) is a younger, poorer neighborhood where voters may not have the time to cure their ballots even when notified of problems. In contrast, Ward 15 voters (2.31% rejection rate) are older and more affluent, and more likely to be able to make the additional effort to cure their ballots when notified of a problem.

Another variation is how often and when the clerks perform the reviews. All the clerks I spoke with said they try to review every ballot on the day it arrives. However, if others delay review until Election Day, voters will not be able to cure their ballots.

In addition, every clerk said that even if the ballot passes its first review, it will be reviewed again later, such as when the ballots are sorted by precinct or alphabetized, and might be rejected then. If the poll worker who is opening the envelopes on Election Day notices a problem, the ballot will be rejected then.

Finally, when we allow variation to go unmonitored, we create circumstances that allow for partisan advantage. Consider this: The City of Milwaukee reliably contributes a large proportion of the state’s Democratic votes, and Wisconsin’s Republican Party is reliably front-and-center promoting strict enforcement of voting requirements. If the Republican Party does nothing more than send observers to monitor the actions of Milwaukee’s election workers to make sure that every reject-able ballot is rejected, while no one makes any similar effort in the Republican suburbs, it is likely that the same ballot will receive harsher scrutiny if it is submitted by a city voter than by a suburban voter.   


A functioning democracy cannot tolerate rejection rates as high or as varied as these. And the variation is almost certainly more extreme than I could observe. The likelihood that my sample (only 14 out of 1,850 municipalities) captured the most extreme examples is practically nil.

So what can be done? The problem cannot be solved by telling WEC to issue guidance in absentee-ballot review practices. They already do that. If municipal clerks reliably followed their guidance, rejection rates would not show the variation they do. The fact is that we can never expect all municipalities to honor any guidance that WEC is unable to enforce.

What we need is transparency.

1) The Wisconsin Elections Commission needs to crunch its data. Given what I know of the work ethic of most local election officials, the WEC needs only to give them accurate metrics about their performance, and most local clerks will seek to improve it. So even if the Commission does nothing more than calculate rejection rates by municipality and ward, they will help our municipal clerks to manage those rates.

After each election, the WEC should routinely provide municipalities with metrics about local rejection rates around the state. Our clerks need to know whether their own performance is normal or extreme.

2) If the Commission will not do this, the state Democratic Party should purchase the data; do the analysis themselves; and intervene before November with the municipal clerks who had the highest rejection rates in August. Considering their vigorous promotion of early and absentee voting, I believe it’s the Democrats’ moral obligation to take steps to protect the ballots of the voters they’ve convinced to forego the option of going to the polls. And given what that party is spending to get out the vote, $24,000 is a small price tag to protect the votes that they get out.

3) In addition, voting-rights groups and civic organizations must realize the problem cannot be solved with voter education alone. They should be in the lead on this issue, working to ensure that WEC provides municipal clerks with the management information, and that individual voters can know their local situation in order to make an informed choice about whether to vote absentee or at the polls.

4) Finally, individual voters can contact their municipal clerks — soon! — and ask about August’s local rejection rate for on-time absentee ballots. If the clerk does not know or cannot produce the data, the voter should recognize a trouble sign of inattentive management and assume the rate is high. If the clerk can provide the information, and the rate is over 1.7%, the voter should inform the clerk that the local rate is above the statewide average. This will likely be news to the clerk. Voters should then discuss with the clerk what will be done to bring the rejection rate into line with the average.

If you try this and do not get a cooperative response from your municipal clerk, go to your municipality’s executive, the governing body, or the newspaper. Votes are at stake here, and if on November 4, Wisconsin sees a upset victory margin that was less than the number of rejected absentee ballots, you do not want to know you sat on your hands in September.

And again: The Wisconsin Elections Commissioners do not now realize that extreme variation in local rejection rates exists, because their staff have never calculated those rates. The Commission has the data; they could analyze it. Their email address is elections@wi.gov; put “Message for the Commissioners” in the subject line.

Sitting here at my kitchen counter in Waunakee, I cannot develop any better recommendations than those. I do know this: To protect future elections, WEC’s monitoring of local absentee-ballot rejection rates must become routine. When WEC makes the information public, people and groups who have more influence and insider connections than I do must examine local absentee-ballot processing practices; objectively identify the causes of high, localized rejection rates; and develop corrective measures.