A Deep Dive into Absentee Ballot Rejection in the 2020 General Election
The November 2020 general election saw a surge of mail-in/absentee voting, mainly in response to safety concerns surrounding voting in-person during the COVID-19 pandemic. As a result, the overall total of mail ballots cast went from 28.8 million in 2016 to 66.4 million in 2020, a whopping 131 percent increase. With it likely that mail-in voting will continue to be prevalent, it is important to examine the nuances of the data more deeply, in order to improve the process for voters and election administrators alike. Today, we hone in on one particular facet of mail voting during the 2020 general election, absentee ballot rejection.
In this exploration, we see that absentee/mail-ballot rejection rates dropped significantly in 2020 compared to 2016, dropping the most in states that had previously erected high barriers to the use of mail ballots. Pre-processing laws and deadlines for the receipt of mail ballots didn’t seem to have much effect on rejection rates. Policy choices that did seem to matter were requiring multiple forms of identification with a returned ballot, which significantly increased rejection rates, and allowing ballots with administrative deficiencies to be “cured,” which significantly decreased rejection rates.
Now, on to the data.
State Level Absentee Rejection Rates
For this analysis, we define the absentee rejection rate as the total number of absentee ballots rejected out of the total absentee ballots returned. Interestingly, the dramatic increase in the raw number of absentee ballots cast was accompanied by a significant decrease in the overall absentee rejection rate for the country: from 0.96 percent in 2016 to 0.79 percent in 2020, according to the U.S. Election Assistance Commission’s Election Administration and Voting Survey (EAVS).
However, the overall absentee rejection rate was not evenly distributed amongst the states. A majority of states fall below the threshold of rejecting 1.0 percent of absentee ballots returned. Seventeen states rejected between 0 and 0.5 percent, while 20 states rejected from 0.5 to 1.0 percent. Nine states fell between the range of 1.0 and 2.5 percent. The key outliers were New Mexico (5.0 percent), Arkansas (4.1 percent), and New York (3.6 percent).
We can drill down to the county level to better inform our understanding of the state-level statistics. Notice that there is relatively low within-state variation between counties. In other words, in most states, most counties have similar absentee rejection rates. Florida and Oklahoma provide two distinct illustrations of this. Florida, which had low statewide rejection, appears almost monotone in color in the graph above. Oklahoma, which had a relatively high rejection rate overall, also saw relatively high rates in each county.
Taken together with the state map, this low within-state variation between counties implies that most counties deviate very little from their particular state’s overall rejection rate. This phenomenon speaks to the reality that major policies about mail-in voting that impact absentee ballot rejection are decided at the statewide level, not at the local level. There will of course be variation at the local level, because of demographic factors and matters of discretion. Still, state laws for the most part determine the rejection rate around which counties vary for these reasons.
The most prominent outlier to this observation in 2020 is New Mexico. Most counties in the state actually had very low absentee rejection rates. However, rejections were much higher—in the 5-to-10 percent range—in three of the four largest counties in the state, Bernalillo (1st), Dona Ana (2nd), and Sandoval (4th).
One of the biggest election administration stories in 2020 was the transition to high-volume vote-by-mail among so many states that had previously discouraged mail, or absentee, voting. Many of these states not only had barriers to voting by mail, but they also strictly scrutinized the ballots that were returned, and often did not even notify absentee voters that their ballot was rejected because of an administrative deficiency. This means that among the states new to large-scale voting by mail, there were very high rejection rates in the past. The question here is whether these high rejection rates continued into the future.
The accompanying graph helps to address this question. In this graph, we have divided states into three categories: those that had seen less than 25 percent of their ballots cast by mail in 2016 (red), those that previously saw more than 25 percent of their ballots cast by mail (blue), and a final category for states that had previously conducted all mail elections (gold). Indeed, in 2016, the states with lower rates of mail-ballot usage also had higher rejection rates on average. All types of states saw a decline in rejection rate in 2020, but the steepest decline occurred in states that previously had not seen much mail-ballot usage.
Explaining Absentee Rejection through a Policy Lens
Not only the rate of mail-ballot rejections, but also the reasons for rejections vary due to a number of factors; the most fundamental factor is state policy. And, among state policies, the most important is the method of absentee ballot verification.
Research by the Stanford/MIT Healthy Elections Project allows us to divide states into five categories of ballot verification, ranging from requiring both a signature and identification, to having no witness or signature verification on the returned ballot. The map below shows which states had what verification practices in 2020.
In the graph above, we display the reasons for ballot rejection, grouping states according to their verification procedures. The overall height of the bars indicates the overall rejection rate for states that use a particular practice. (You can hover your cursor over each bar to see the breakdowns of the reasons for rejections.)
What immediately jumps out in this graph is how states that required two forms of identification for verification had much higher rejection rates than those that only required one, or none at all. The rejection rate for “Signature + ID” states is especially high, about five times the U.S. average. This category had only one state in 2020, Arkansas, which required that a photocopy of valid identification (with a name and a picture) be sent in addition to the voter’s signature for their ballot to count. Understanding precisely why Arkansas’s rejection rate is so high is a challenge, because so many counties do not record—or at least report—why they rejected ballots. (Thirty-five percent of all absentee ballot rejections in Arkansas are marked “other” for the rejection reason in the EAVS dataset.) Even so, 17 percent of rejected ballots that were accounted for by reason were rejected because of the lack of an ID (0.70 percent of all absentee ballots returned), whereas another 8 percent of rejected ballots that were accounted for were rejected because of the lack of a signature, and another 3 percent because the signature failed to match the one on file. (As a percentage of all returned absentee ballots, these percentages were 0.31 percent and 0.11 percent, respectively.)
The other multi-identification states, Louisiana and Mississippi, required the signatures of both the voter and a witness. These states together had an absentee rejection rate that was more than two and a half times the U.S. average. In these states, the most common problem pertained to the witness signature, which accounted for 32 percent of all rejections, among rejections accounted for, or 0.62 percent of all absentee ballots returned for counting.
“Signature only” remained the most common type of verification method in 2020. The majority of states fell into this classification, and this explains why the rejection rate for this type of verification is just about level with the US average. These states were able to account for why the ballots were rejected in almost all cases. Here, the most common reason for rejections was the failure to match a signature, at 39 percent of all rejections and 0.30 percent of all returned absentee ballots. The lack of a signature altogether was much lower in these states, compared to the three states that required an additional form of identification.
States with a “Witness only” requirement for verification in 2020 saw the lowest overall rejection rate among the five state groupings. Unsurprisingly, the main reason for rejection, accounting for 39 percent of all rejections in these states, was the lack of, or a problem with, the witness signature (0.26 percent of all returned ballots). The next most common reason for rejection in witness-only verification states was lateness, making up 14 percent of all rejections.
Finally, we note the states with “No verification” requirements. “No verification” states still reject ballots for a variety of reasons, including but not limited to lateness, a missing voter signature, an unsealed ballot envelope, or if the voter is ineligible to vote by mail. (Note that although these states may not verify the signature on a returned absentee ballot, they still require a signature, usually attesting to the fact that the ballot was marked by the voter.) It is notable that among these states, there is little information about why the rejected ballots were rejected—70 percent were grouped in the “other” category. The reason that was most frequently reported was the lack of a signature (11 percent), followed by being late (10 percent).
Ballot Preprocessing and Mail-in Deadline
Two policies that attracted a lot of attention in 2020 were ballot preprocessing time and the deadline for the return of absentee ballots.
Mail-in ballot preprocessing is the practice of preparing absentee ballots for counting, by means of verifying them according to the state’s laws and sometimes opening ballots so they are ready for tabulation. Longer preprocessing times before Election Day give officials more time to contact voters and correct defects, such as missing signatures, that might render a ballot invalid. However, there is little, if any, evidence that preprocessing time influenced rejection rates. The rejection rate among the 7 states that did not allow pre-Election-Day preprocessing was 0.93 percent, compared to 0.72 percent for those that allowed preprocessing to start within the week before Election Day, and 0.79 percent for those that allowed preprocessing more than a week before Election Day. Preprocessing is an important tool to improve the speed and efficiency of election night reporting, but it has little-to-no effect on absentee ballot rejection as a general matter.
The other policy of interest is the deadline for the return of absentee ballots. Intuitively, allowing ballots to arrive after Election Day (but only if they were postmarked by Election Day) would seem to reduce rejections, especially for being late. However, this intuition is not borne out by the data. In the graph above on the right (dark red), we display each state’s absentee rejection rate specifically for ballots marked late. If we ignore the outlier in grey (Illinois), the trend line helps us observe a slight decline in late rejection rates as the deadline is extended further away from Election Day. However, the overall rejection rates were not lower among states that allowed the post-Election-Day arrival of mail ballots.
The final policy of note is signature curing. Signature curing is when a voter whose ballot has been challenged or found deficient (e.g., it is missing a signature) is contacted and given the opportunity to correct, or “cure,” the problem. Here, the policy choices matter a lot. The rejection rate among the states that allowed signature curing was 0.67 percent, compared to 1.10 percent among those that did not.
What do we make of this analysis? Two important takeaways seem key, related to voter verification and ballot curing.
- Adding layers of absentee voter verification increases rejections. In 2020, three states subjected voters to two layers of verification after the mail ballot was returned. (This was after, presumably, eligibility had already been established when the ballot was requested.) Arkansas required a signature match and an ID; Louisiana and Mississippi required a witness and a signature match. In sessions this past year, several states considered following one of these paths. The rejection rates in the three states with double verification ranged from twice to five times the national average. It is hard to believe that mail ballot fraud is so much greater in these three states than in the rest of the country—including several states that overwise have similar barriers to requesting and returning mail ballots as do Arkansas, Louisiana, and Mississippi. Thus, it is reasonable to conclude that this suspender-and-belt approach to absentee ballot verification is primarily a barrier against legitimate voters.
- Ballot curing systems help legitimate voters. The majority of states use signature verification as their only method of absentee voter verification. For these states, it is important to acknowledge that signature matching is susceptible to error on both the part of the voter and verifier. A clear ballot curing process is an obvious way to catch these inevitable errors. Only 18 states offered such a procedure in 2020, and they had lower rejection rates than non-curing states. Georgia is a good example of how instituting a uniform curing process can help voters. The Georgia General Assembly enacted a uniform notification and curing system following the 2016 election (House Bill 316) which was reinforced by 2020 court settlement. After doing so, Georgia saw its absentee rejection rate fall from 6.4 percent in 2016 to just 0.36 percent in 2020. Implementing a ballot curing system provides voters with a failsafe for fixing their ballot should a mistake in the signature-matching process occur, while exemplifying states’ commitment to the democratic principle that every vote matters.