Research by Kira Flemke
Overview
In the 2024 general election, nearly 97% of voting-age citizens will live in a state that offers at least one option to vote before election day, such as early in-person voting or mail voting. These options have been widely popular among voters in the two most recent presidential elections, with 40% of votes cast before election day in 2016 and 69% in 2020. This focus brief details the distribution of votes cast before election day across states in 2016 and 2020.
Nationwide Use of Voting Before Election Day
Figure 1 displays how the nationwide increase in voting before election day from 2016 to 2020 was distributed across states.
In 2016, the share of votes cast before election day in the 50 states and Washington, D.C., ranged from just 3% to nearly 100% in some all-mail states, with a state average of 37%. Fifteen states saw more than half of all their votes cast using early in-person or mail voting, and 34 states plus Washington, D.C., saw more than 25%.
In 2020, the share of votes cast before election day ranged from 13% to nearly 100% in some all-mail states, with a state average of 63%, a notable increase from 2016. In 2020, 38 states had over half of all votes cast using these methods, and 47 states plus Washington, D.C. saw more than 25%.
Defining Terms.
This focus brief uses the phrase ‘voting before election day’ to refer to methods of voting that allow voters to obtain and cast ballots before election day. This term is inclusive of early in-person voting and mail ballots. “Early in-person voting” refers to voting that is available to all voters and conducted in person before the day of the election. This includes policies referred to by states as early voting, in-person no-excuse absentee voting, and advanced voting. Similarly, “mail voting” refers to voting where all voters are eligible to receive their ballot via mail, including policies referred to by states as all-mail voting and no-excuse absentee by mail.
Voting Before Election Day Across States
Under the U.S. Constitution, each state sets its own election administration procedures, including policies concerning options to vote before election day. While the details vary across states, they can each be sorted into one of three basic categories: (1) states that offer no early in-person voting to all voters and require an eligible reason to vote by mail;[1] (2) states that offer early in-person voting to all voters and require an eligible reason to vote by mail; and (3) states that offer both early in-person voting and mail ballots to all voters.
Table 1 shows these policy categories across states and the average share of votes cast before election day in each category for each general election. In 2016, the percentage of votes cast early in-person or by mail increases across policy groups as states offer more options. However, even in the most restrictive states, one-in-ten votes were cast using these methods in 2016. These trends are less clear in 2020, largely because temporary COVID-19 policy expansions left very few states in the more restrictive policy groups.[2]
Table 1. Overall Use of Voting Before Election Day, 2016 & 2020
While both early in-person and mail voting provide ballots to voters before election day, they differ in their administrative procedures, periods of availability, and—most notably for this focus brief—in their rates of use by voters in each election and jurisdiction. In states that offer both options, one is often more popular than the other.
Use of Early In-Person and Mail Voting Across States
Among states that offered early in-person voting to all voters in 2016, 27% of all votes were cast early in person. This value increased to 40% in the 2020 general election, during which all states except one offered some form of early in-person voting.[3][4]
Across states where mail voting was available to all voters in 2016, 34% of all general election votes were cast by mail. This value rose to 44% in 2020. Among states that required an eligible reason to vote by mail, mail voting made up 8% of all votes in 2016 and 12% in 2020.[5] Notably, the number of states with mail ballot restrictions decreased from 19 in 2016 to only five in 2020.
Figure 2 below displays the percentage of all votes in each state that were cast using early in-person voting and mail voting in the 2016 general election. Among the 34 states that reported vote totals for both early in-person and mail votes in 2016, a majority reported a greater number of early in-person votes. This includes 15 states that offered options for both early in-person and mail voting to all voters without restriction.[6] This implies that even though a greater overall number of votes nationally were cast by mail in 2016, early in-person voting appears to have been the more popular option in the majority of states where it was offered in 2016.[7]
Figure 3 below displays the percentage of all votes in each state that were cast using early in-person voting and mail voting in the 2020 general election. Among the 42 states that reported vote totals for both early in-person and mail votes in 2020, 28 states reported more mail votes than early in-person votes.[8] All 28 of these states offered both early in-person and mail voting to all voters that year. Only 14 states reported more votes cast early in person, including all four states that offered early in-person ballots to all voters but required an eligible reason to vote by mail. Various policy and social factors related to the COVID-19 emergency likely influenced the relative popularity of mail voting in the 2020 general election, and voters’ preferences in the 2024 general election remain to be seen.
This focus brief is part of CEIR’s ongoing research into voting before election day. To learn more, visit:
The Expansion of Voting Before Election Day, 2000-2024
CEIR Focus Brief: Dates for 2024 Early In-Person and Mail Voting
Data Dive: The Growth of Early In-Person and Mail Voting
Data Dive: 2024 Weekend Availability of Early In-Person Voting
Appendix: Data Sources & Quality
Data for this brief was initially collected from the 2016 and 2020 Election Administration and Voting Surveys (EAVS), which asked state jurisdictions to report (1) the total number of voters who cast a ballot that was counted; (2) the number of voters who cast a domestic civilian by-mail ballot, and whose ballots were counted; and (3) the number of voters who cast a ballot in person at an early voting location, and whose ballots were counted. The EAVS data for each local jurisdiction within states were aggregated to the state level, and any missing data for states or local jurisdictions within states was documented.
The EAVS data had two forms of missing data: category missingness, where no jurisdiction in a state reported a value for one of the above questions, and jurisdiction missingness, where one or more jurisdictions in a state failed to report a value for one of the above questions. In order to construct the most complete dataset possible, in a state-year where there was partial jurisdiction missingness or invalid category missingness, the EAVS data were replaced with the official statewide data for each category.[9][10] There are a few states in each year for which the most complete data could be constructed only by combining vote mode categories from EAVS and official state totals. For the handful of states that did not have complete EAVS data or official statewide totals broken out by vote mode, this focus brief includes their incomplete EAVS data, either alone, denoted as EAVS*, or in combination with official state totals, denoted as Mix * in Tables A1 and A2.
Table A1. Number of States by Year and Data Source
Table A2 contains information about the data sources of every state for which the complete data for a given category could not be collected from a single source, either EAVS or the official state totals. For states where incomplete EAVS data was used, this table presents the number of jurisdictions that reported values out of the total number of jurisdictions in the state.[11]
Table A2. Incomplete EAVS and Mixed Source Data
Click here to download the full data used in this focus brief.
Additional Notes and Resources
[1] This category includes all-mail states, wherein each active registered voter is automatically sent a mail ballot prior to an election which can then be returned in-person or by mail. In 2016, Colorado, Oregon, and Washington conducted all-mail elections. In 2020, California, Colorado, Hawaii, New Jersey, Nevada, Oregon, Utah, Vermont Washington, and Washington, D.C. conducted all-mail elections.
[2] Sixteen states expanded options to vote before election day due to the COVID-19 pandemic.
[3] This calculation excludes states that did not report mail ballot totals separately from early in-person totals. See Appendix for full list.
[4] Mississippi was the only state not to offer some form of early in-person voting during the 2020 general election. Despite Mississippi’s requirement of an eligible reason to use a mail ballot, 18% of ballots in Mississippi were cast by mail in 2020.
[5] This calculation excludes states that did not report mail ballot totals separately from early in-person totals. See Appendix for full list.
[6] In 2016, 17 states did not break out vote totals by mode, including six states that provided both early in-person and mail voting options.
[7] Nationally, in 2016 more ballots were cast by mail than early in person, partially due to the influence of all-mail states and states that did not offer early in-person voting to all voters in 2016.
[8] In 2020, nine states did not break out vote totals by mode.
[9] Official state totals were collected from each state’s official website for the secretary of state.
[10] “Invalid category missingness” refers to any states where no jurisdiction in a state reported a value for a vote mode category that their state policy permits. This excludes valid category missingness in states that do not offer early in-person voting and, as a result, do not have early in-person vote totals to report.
[11] Blank cells in the jurisdiction category indicate no partial missingness in that year for that category, because all jurisdictions reported values to EAVS or the official totals were used.