Monday, April 13, 2020

Investigating North Carolina's Absentee by Mail Vote Method

With the current discussion about voting by mail and its potential use for the 2020 elections, both for primary and general contests, and the possible partisan ramifications of this vote method, North Carolina has utilized absentee by mail voting as one of its voting methods for elections since the early 20th Century. Along with absentee onestop voting (which was introduced in 1977 and is in-person, no excuse voting during a window of time before an election) and Election Day in-person voting, absentee by mail voting is one of the top three forms of casting ballots for voters in the Old North State.

In researching North Carolina politics for nearly twenty years, I have data from the North Carolina State Board of Elections regarding both registered voter information and voter history (meaning, data on each voter who cast a ballot, along with their vote method--but not their vote choice). Recently, I presented research at the Citadel Symposium on Southern Politics regarding "A Matter of Electoral Convenience: Early Voting in North Carolina, 2004-2018" that prompts me to write this blog post, for public awareness and education on North Carolina's use of voting by mail.

Based on this research, I want to briefly summarize what political scientists know about "convenience voting," the term related to voting other than on Election Day (many people may refer to it as 'early voting,' 'voting by mail,' or 'absentee voting'). First, I'll cover some general observations about convenience voting and absentee by mail voting based on research by political scientists, and then an in-depth exploration of the data related to absentee voting, especially by mail, in North Carolina and in the 2016 presidential general election.

What is known about absentee/convenience voting? 



Five states--Colorado, Utah, Hawaii, Oregon, and Washington State--currently conduct all of their elections by mail, according to the National Conference of State Legislatures. Numerous other elections across the United States, as well as around the world, are conducted by mail, according to the NCSL.

One of the leading scholars regarding different early voting methods, Paul Gronke of Reed College, along with a number of colleagues, published a great summary about the research on "Convenience Voting" in the Annual Review of Political Science. In their summary, Gronke et al. find that different forms have different impacts in a number of areas, most notably whether voter turnout increases or not with the different forms.

As with most research, there are different findings and interpretations when it comes to convenience voting methods: while some studies show that in-person convenience voting may not have a statistically significant impact on increasing voter turnout (meaning, voters who cast ballots early would have done so on Election Day), voting by mail in several studies does show some impact on increasing voter turnout (anywhere from 5 to almost 20 percent). However, another study found that the use of voting by mail in general elections resulted in decreased voter turnout, but in local special elections, voter turnout increased.

Political scientists have engaged in a number of research questions regarding voting by mail as a form of convenience voting, and more research needs to be done in exploring and understanding the impact of convenience voting in a variety of areas, from turnout to the composition of who votes to campaign strategies by candidates with different vote methods. The remainder of this blog post focuses on North Carolina and its absentee voting methods, especially by mail.

What about North Carolina and voting by mail? 


First, a word about the following data: it comes from the North Carolina State Board of Elections via their FTP page and is from two data files: the voter registration data file (usually either the snapshot data file or the data file closely following the election) and the voter history data file, which contains each voter's registration number and county, along with information on the vote method used by that voter. In order to get the following demographic analysis, the two files must be combined, using the county description and voter registration number as the 'keys' to merging the data files. Other data files that add information to these two files are the absentee zip files for different elections, found here. For further information or specific questions on how I approach this data construction, please e-mail me.

In a presidential year, the three top forms of vote methods used by an overwhelming number of North Carolina voters are: absentee onestop (in-person), Election Day, and absentee by mail. Other forms (such as curbside, for both absentee and Election Day, along with provisional and transfer ballots) make up about 1 percent or less of the overall ballots cast.

In terms of absentee by mail voting, North Carolinians are required to go through a two-stage process for using absentee voting by mail: first, requesting an absentee by mail (ABM) ballot from their local county board of elections (the process of which has already begun for the 2020 general election), and upon successful completion and submission of the ABM application, an absentee by mail ballot is mailed to the voter, who completes both the ballot and the required submission information and returns it to their county board of election.

As was discovered in the 2018 general election concerning North Carolina's Ninth Congressional District general election, issues regarding ballot fraud (meaning, allegations of absentee by mail ballots were manipulated to the point of a new election being declared) brought about revisions to the ABM process, namely that only the voter or near relative can return a completed ABM ballot to the board of elections (more information on the requirements can be found here at the NC State Board of Elections website).

North Carolina has seen a dramatic swing in presidential vote methods since 2004: with the Obama ground game in 2008, 'early voting' (absentee voting) surged to become the majority of ballots cast, and each subsequent presidential election year has seen early voting (made up of both absentee voting by mail and absentee onestop/in-person) remain a favorite by voters.


In 2004, when George W. Bush won North Carolina by 13 percentage points, less than one third of votes cast came 'early,' meaning by absentee voting. In 2008, 61 percent of the ballots came early, with five percent coming absentee by mail. That percentage, for both overall absentee ballots and absentee by mail ballots, held in 2012, but in 2016, 66 percent of ballots cast came from absentee ballots, with 4 percent coming via mail absentee ballots.

Since 2008, voters who cast absentee by mail ballots have come from one political party:


In 2004, the two major parties were nearly identical in the use of absentee by mail ballots, but since 2008, registered Republicans were either a majority or plurality of absentee by mail ballots returned and accepted. What is interesting is that registered unaffiliated voters have grown in their percentage of using absentee by mail ballots, from 15 percent in 2004 to nearly 30 percent in 2016.

Next, a profile of those voters who cast absentee by mail ballots, by voter race. Overwhelmingly, absentee by mail ballots are utilized by white voters, who typically are 70 percent of registered voters in the past four presidential election cycles.


Finally, in terms of age of voters using different vote methods, absentee by mail ballots show the oldest mean age of voters of the three major vote methods. The following are histogram charts, which show the frequency, by voter age, of the use of each of the three forms of voting methods from the 2008, 2012, and 2016 general elections.

First, the 2008 general election with the age distributions for absentee by mail, absentee onestop, and Election Day ballots cast:

2008 General Election: Voters Who Cast Absentee Mail-In Ballots
Histogram of Age Distribution

2008 General Election: Voters Who Cast Absentee OneStop Ballots:
Histogram of Age Distribution

2008 General Election: Voters Who Cast Election Day In-Person Ballots:
Histogram of Age Distribution
Notice that in 2008's general election, absentee by mail voters had a mean age of 55 years old, while absentee onestop voters were nearly 51 years old, and Election Day voters were 47 years old. The unique peak on the left side of the absentee by mail age distribution is likely due to college students requesting those ballots.

For the 2012 presidential election, similar patterns in the mean age and distribution within the vote methods are found as in the 2008 election.

2012 General Election: Voters Who Cast Absentee Mail-In Ballots
Histogram of Age Distribution


2012 General Election: Voters Who Cast Absentee OneStop Ballots
Histogram of Age Distribution

2012 General Election: Voters Who Cast Election Day In-Person Ballots
Histogram of Age Distribution


For the 2016 general election, the pattern remains as in the previous two presidential elections: absentee by mail with a mean age of 55 years old; absentee onestop at 52 years old; and Election Day voters at 47 years old.

2016 General Election: Voters Who Cast Absentee Mail-In Ballots
Histogram of Age Distribution
2016 General Election: Votes Who Cast Absentee OneStop Ballots
Histogram of Age Distribution

2016 General Election: Votes Who Cast Election Day In-Person Ballots
Histogram of Age Distribution

In the past three presidential election cycles, distinct age differences are found in the three major voting methods in the Old North State.

What did North Carolina experience in terms of absentee by mail voting in the 2016 presidential election? 


In 2016, voting by mail ballots requests totaled slightly over 240,000 from the absentee voting data file. In comparison, the number of absentee onestop (in-person) ballots was 2.9 million ballots, meaning that 6 percent of the state's absentee ballots were by mail, while 94 percent were onestop.

Both types of absentee ballots must be 'accepted' for a legal vote. Typically there is a greater percentage of absentee by mail ballots that are rejected, for a variety of reasons, such as: 'spoiled' (meaning the voter incorrectly voted for their preferred candidate, or perhaps marked the ballot in a way that the intent could not be deciphered (such as bubbling in two candidates for one office that requires only one vote for a candidate)), not properly notarized, returned undeliverable or after deadline, the voter's signature is different from the voter's name or the signature is missing, or the witness information section is incomplete.

The data for both requested absentee by mail and absentee onestop ballots and their final status shows the following dynamics for both methods and the overall absentee voting method (combining both mail and onestop):


Among NC's absentee by mail ballots that were requested, returned, and reviewed, 80 percent were accepted as votes, while 9 percent were rejected due to at least one reason listed above. Of the total ballots requested, 11 percent were not returned.

If looking only at the absentee by mail ballots returned, the accepted rate was 90 percent, with the rejection rate being 10 percent.

Just because a voter failed to return an absentee ballot by mail or that mail ballot was rejected did not mean that the voter did not end up casting a ballot in the 2016 general election. In pulling together the data for voter registration, the absentee voting data, and the final vote method history, I found two interesting data aspects that voters who may have had an absentee by mail ballot rejected or not ultimately ended up casting a confirmed ballot for the 2016 general election.

An interesting dynamic in NC's absentee by mail data


First, in looking only at the absentee by mail ballots requested, there were a number of 'duplicate' files for individual voters, indicating that a voter may have submitted a mail-in ballot request, only to have that ballot spoiled (perhaps the voter incorrectly marked who they wanted to truly vote for a particular office, and thus requested another ballot), or the ballot was rejected or not returned.

Second, once the data was organized for analysis, it appears that four rounds of absentee by mail ballot requests were utilized by some voters in the 2016 election. This data file was developed by first sorting the absentee vote file for the 2016 election by three fields: the date requested for the ballot, the date sent by the county board of elections, and the date received by the board. In addition, a field is found in the data file that gives the ballot return status, and each request has information (or not) in that field to determine the last status of the ballot. Each request was then separated into a separate entry for an individual voter, and then the multiple requests were merged together to create one voter's history of ballot requests. This information was then merged into the voter registration and history data file to be analyzed.

The first round of requests totaled over 222,000, with a second round of nearly 8,000, a third round of nearly 400, and the final round of 32. The first round of accepted ballots totaled over 187,000 ballots, with the second round generating 3,860, the third round another 169, and the fourth round of 13 ballots accepted.

Next, I looked at both the requested and accepted ballots by a number of voter characteristics: by party registration; by voter race; by voter generation; by the region that the voter resides in (central city precinct; suburban within urban county precinct; surrounding suburban county precinct; and rural county precinct); and finally, by voter gender.

First, the requested ballots by the different rounds and the voter's party registration (the percentages below the different rounds on the X axis reflect that round's percentage of the total ballots requested/accepted):


The first two request rounds mirror the eventual final total, while the third round and the fourth round differ (again, both rounds count for about 0.02 percent of the total ballots requested).

Next, the requested ballots by different rounds and the voter's race:


Voters of color (black/African American, Asian, Native American, multi-racial, and other races) had a larger percentage in subsequent rounds, growing to 26 percent of the third request round.

Among generational cohorts, the rounds of absentee by mail ballot requests reflect the percentage of older voters, namely Boomer and the Silent/Greatest generations:


When looking at the four different regions of the state (central city voters; suburban but within an urban county voter; surrounding suburban county voters; and rural county voters), the first three rounds show a plurality of requests from suburban but within urban county voters (as does the total), with the final round (a very small percentage of the total) seeing surrounding suburban county voters the plurality.


Finally, a look at the rounds of absentee by mail ballot requests by voter gender:


Women made up the substantial majority of requested absentee by mail ballots at all rounds (the remaining 2 percent were from voters who did not identify their gender on their voter registration).

I analyzed, by the same method, the accepted ballots by the four rounds as well, with the same demographic factors as the requested ballots. First, for the accepted ballots by the voter party registration:


As noted in this chart, while the first round of accepted absentee by mail ballots were primarily from registered Republicans (as was the overall total), the second through fourth rounds of accepted ABMs saw different distributions by party. But again, these rounds (second through fourth) accounted for less than 2 percent of the total ballots ultimately accepted.

Next, accepted mail-in ballots based on voter race:


A similar to the requested ballots by each round, with the exception of the last round where a greater percentage of minority voters returned and had their ABM ballot accepted (again, note the percentage of the total at this round: 0.01 percent).

By generational cohorts, each subsequent round (beyond the first) had some unique distributions within it:


Millennials were a third of the third round, while the oldest voters (Silent/Greatest generations) spiked considerably--but again, there were very few ballots accepted at that point.

By the four regions, suburban voters within urban counties against were pluralities through the first three rounds of ballot acceptance, with surrounding suburban county voters a plurality in the final round.


Finally, in looking at the gender of voters, like with requested ballots, women tended to dominate the return ballots that were accepted:


The final piece of analysis sought to answer the question, did the voters whose absentee by mail ballots utilize any other vote method to cast their vote in the 2016 general election?

Thanks to the data that gives both the ABM ballot status and the voter's vote method, a cross-tabulation was conducted to see which voters who had requested ABM ballots either had their ABM accepted, used another vote method, or decided not to vote ultimately.


Even though an absentee by mail ballot may be requested or rejected, some voters (approximately 17,000) cast their ballot in-person, with 26,000 who requested, but did not return their ballot, deciding not to vote in 2016's general election.

As a faculty member in my graduate program was often prone to say: so what?


What this information hopefully portrays is that absentee by mail voting, in North Carolina at least, is used by older, white suburban voters, who are more likely to be registered Republicans. What 2020's general election could hold is anybody's guess at this point, but in a battleground state like North Carolina, the concerns regarding COVID-19 and the corona virus spreading during a fall campaign season could shift the dynamics of vote methods more heavily to vote by mail ballots.

Discussions have already started about how best to prepare for North Carolina's November election. While a prominent Republican state leader has indicated "zero trust that this process would be fair and transparent" in how the state board of elections is administered and what should be done to prepare for the November 2020 election, it is my professional opinion that this should not stop legislative, executive, and administrative officials--both at the federal, state, and local levels--from making critical policy decisions, both in terms of administrative implementation and funding, now for the 2020 general election.

Leaders should plan for the use of voting methods--whether it is voting by mail or voting in person--so that North Carolinians can achieve two distinct goals: to possibly protect their own health and well-being, while at the same time exercising one of their most important civic duties and rights come this November: voting.