A black mailbox with its red arm raised up. The mailbox is open, and an absentee ballot is halfway in, as though it's just been slipped into the box.

Divided States of Mail Ballots

The MIT Election Data and Science Lab manages the Elections Performance Index (EPI), an objective, nonpartisan assessment tool evaluating U.S. election administration. Following the recent update to the index with data from 2016, we are dedicating a series of posts to exploring the EPI’s underpinnings. Our post today was penned by Christopher Mann, who studies voting by mail and early voting.

Mail ballots continue to be one of the most divisive elements of the Election Performance Index. While there is controversy among policy-makers and scholars about mail ballots, I mean divisive more literally: dividing states in the index according to their policies about mail ballots.

Policy-makers debate the pros and cons of mail ballots: some states mail a ballot to every voter, while other states maintain strict requirements limiting access to a mail ballot. Scholars debate whether the use of mail ballots has an effect on participation and whether concerns outweigh benefits. Settling these debates requires answering two types of questions:

  1. the effect of mail ballot policy choices, and
  2. the administration of mail ballots.

The Election Performance Index has tremendously valuable data on elections in the United States. Unfortunately, choosing to look at one part of this realm (e.g. policy choices) often obscures the other (e.g. administration). It is important to understand which questions the Election Performance Index is designed to answer, because it can only tell part of the story about mail ballots in U.S. election performance.


Dividing States by Access to Mail Ballots

Should we care about mail ballots? For many Americans, the answer to this question is probably obvious — but these obvious answers are opposites. Voters in some states would correctly say that mail ballots don’t have much impact on election performance. Voters in other states would be equally correct saying mail ballots are the central factor in election performance.

The policy choices states make divide them into different groups based on the rules under which voters can obtain a mail ballot. The table below outlines the four different state mail ballot policy classifications used by the Election Performance Index (the National Council of State Legislatures uses slightly different terms, but the classifications are the same).

These policies have massive effects on the proportion of voters who use mail ballots. In states with limited absentee voting, the percentage of voters using mail ballot use is typically in the single digits. In states with full vote-by-mail, every registered voter is sent a mail ballot. Clearly, the administration of mail ballots has a different impact on election performance if it affects every voter than when it impacts fewer than one-in-ten ballots.

Mail Voting Policy



Registered voters must provide a specific reason, often from a pre-established list (illness, disability, travel, etc.), when requesting a mail ballot.

No Excuse

Any registered voter may request a mail ballot without providing a reason.


No-excuse mail voting is permitted, and registered voters have the option of automatically receiving mail ballots for all future elections.

Full Vote-By-Mail

Ballots are distributed to all voters by mail.


Changing American Landscape of Mail Ballots

More than a decade ago, when Yale Law School Dean Heather Gerken proposed the idea that eventually became the Election Performance Index, mail ballots were still a relatively obscure corner of election administration. They could be found primarily in the Northwest corner of the United States (Washington and Oregon), which used full vote-by-mail systems, plus a scattering of Western states with permanent mail ballot systems (e.g. Arizona, California, and Colorado).

Since then, however, the prevalence of mail ballots has continued to grow. Several additional states shifted from allowing limited to no excuse access to mail ballots. States have adopted new technology to facilitate mail ballot use, such as online ballot request systems. Colorado began using a full vote-by-mail system. In states ranging from California to Nebraska, local jurisdictions are opting to use postal ballot delivery to send ballots to all registered voters.

Even where mail ballot rules have not changed, growing proportions of voters are requesting mail ballots. As a result of these trends, the number of mail ballots cast has risen sharply. The US Election Administration and Voting Survey (EAVS) reports that “absentee voters nearly doubled, from 14.7 million in 2004 to 24.8 million in 2016.” Mail voting saw an even greater increase, “more than tripling from 2.4 million in 2008 to 8.2 million in 2016.”


Challenge for the Election Performance Index

Finding a way to measure elections across these different mail ballot policy classifications is a significant challenge for the scholars and election officials designing the Election Performance Index. Do we want to measure the policy choice made by state legislatures? Or do we want to measure the implementation by election administrators within each state? Data useful to answering one question can obscure answers to the other question — and risk unfair or incorrect judgments about “election performance.”

The two measures about mail ballots in the index are designed to assess policy choices. Both “Mail Ballots Unreturned” and “Mail Ballots Rejected” are calculated as the percentage of all ballots. These measures are policy-focused because rising numbers of mail ballots leads to rising numbers of rejected and unreturned ballots. States with very few mail ballots will have few that can be unreturned or rejected. Thus, it makes sense that states with full vote-by-mail or permanent mail ballot systems and high mail ballot use generally have the highest rates of unreturned and rejected mail ballots as a percentage of all ballots.

Some observers point to these high rates of unreturned and rejected mail ballots as a reason to be concerned about policy choices that allow easier access to mail ballots. While the data in the Election Performance Index is clear that more mail ballot use means more unreturned and rejected mail ballots, we need to look at other data to fully assess whether mail ballots are causing more “lost votes” than other policy choices.


Missing Story of Mail Ballots

Before going on, it is important to point out that the Election Performance Index indicators for mail ballots are not suited to assessing the administration of mail ballots. We need different data if we want to look at administration instead of policy. Specifically, we should look at the odds that a mail ballot will be rejected or unreturned. These odds are calculated by using the number of mail ballots as the denominator, rather than all ballots. (My chapter in The Measure of American Elections uses this alternative calculation to assess this question of administration.) Since states seem to be steadily moving towards policies allowing easier access to mail ballots, and voters are increasingly using mail ballots, the question of how well mail ballots are administered seems at least as important as the policy question currently addressed in the Election Performance Index.

The key finding when looking at the administration of mail ballots is that states appear to get better at administering mail ballots when they have more mail ballot use. Why do states with more mail ballots have fewer problems? Election officials invest in handling mail ballots and learn ways to improve over time. Voters in places where mail ballots are common are also likely to learn based on their own repeated use, media attention, and informal communication. In states with lower mail ballot use, on the other hand, election officials dividing resources between in-person voting and mail ballots means fewer resources to invest and less time and attention for learning and improving mail ballot handling. Individual voters may learn from their experience with mail ballots, but this learning is less likely to spread to others in the community via conversations with friends and family or media coverage when mail ballots are rare.

This missing story about mail ballot administration is very different than the Election Performance Index’s policy choice story. In particular, jurisdictions with low but rising use of mail ballots are concerning. Often these are states that have recently made it easier for voters to access mail ballots. As mail ballot use is driven upwards by voter interest and the promotional activities of civic organizations, political campaigns, and sometimes election officials, these jurisdictions have relatively little experience with handling mail ballots and fewer resources to allocate to the task. The alternative approach to calculating mail ballot administrative performance indicates they should look to states with full vote-by-mail and permanent mail ballot polices for lessons on administering mail ballots.


Compared to what?

Is increasing mail ballot access bad policy? Simply put, we don’t have enough data to answer the question. The Election Performance Index’s data suggests rising numbers of mail ballots lead to more “lost votes” due to unreturned and rejected mail ballots. However, we need to compare mail ballots to in-person voting, the alternative choice for how to cast ballots. Despite a much longer history and more widespread use, we have very little data about “lost votes” due to problems with in-person voting. One initial attempt to consider the issue of lost votes through mail ballots in parallel with in-person voting is a 2010 article by Charles Stewart III, aptly entitled “Losing Votes by Mail.”

If data were available, the Election Performance Index would likely contain measures like: “Voters Leaving Due to Long Polling Place Lines,” “Voters Rejected Due to Inadequate ID at the Polls,” “Voters Appearing at the Wrong Polling Place,” “Voters Who Can’t Find Their Polling Place,” “Voters Who Can’t Appear on a Tuesday During Voting Hours,” and more. The Election Performance Index does include “Provisional Ballots Cast” and “Provisional Ballots Rejected,” but these are only partially equivalent to capturing the “loss” of votes with mail ballots.

The debate over mail ballot policy — among policy-makers as well as scholars — often feels like belief that mail ballots are an improvement versus empirical evidence about problems. Neither side stands on solid ground until they grapple with the unspoken part of the question: Is increasing mail ballot access bad policy compared to in-person voting? Until we have more equivalent data on in-person voting, we will continue to have two sides talking past each other.

The data in the Election Performance Index is a massive and important contribution to understanding elections and the impact of mail ballots on elections. But one of the most important lessons is that we need more data before we can confidently reach conclusions about one of the most important developments in elections in the 21st Century.

A white, middle-aged man in glasses, wearing a white shirt and tie with diagonal stripes. He's smiling at the camera.

Christopher Mann studies election administration and voting behavior, with a special focus on voting by mail and early voting. He is an assistant professor of political science at Skidmore College.

Topics Vote by mail and absentee voting