Nonwhite Voters Face High Risk of Being Dropped From Arizona’s Mail Ballot List

The past year has been char­ac­ter­ized by wide­spread assaults on voting rights across the coun­try. This has been espe­cially true in swing states where the 2020 pres­id­en­tial elec­tion saw narrow margins. As our colleague Will Wilder explainsArizona is the focal point of the fight for voting rights, even among these states.

Here, we look at one provi­sion of Arizon­a’s Senate Bill 1485, which was passed into law in May 2021 but won’t be in effect for the midterms. The bill turns Arizon­a’s extremely popu­lar “Perman­ent Early Voting List” into an “Active Early Voting List.” Voters on the list are auto­mat­ic­ally mailed a ballot before every elec­tion. Under the new rules, Arizon­ans will be removed from the list if they go four years without cast­ing a ballot by mail. Even if a voter parti­cip­ates but only does so in person, they will fall off the list.

Like in much of the West­ern United States, Arizon­ans have a robust history of cast­ing their votes by mail. Since 2010, an aver­age of two-thirds of the elect­or­ate has voted by mail. Like four other states, Arizona has allowed voters to sign up to receive a ballot in the mail for all elec­tions, obvi­at­ing the need to fill out a new absentee ballot request form for each and every contest. Accord­ing to voter file data from L2 Polit­ical, some 3 million Arizon­ans (roughly 75 percent of the elect­or­ate) were on the list in April of 2021, shortly before S.B. 1485 was passed.

To analyze the expec­ted effects of the law, we looked at voter file data from L2 Polit­ical, which recOrds whether and how each voter parti­cip­ated in an elec­tion. Although voters won’t be removed from the list fOr a few years, we wanted to know who would be impacted based on the list as it was when the legis­lature was consid­er­ing this change. To do so, we looked at which voters on the list cast a mail ballot in any elec­tion in 2017, 2018, 2019, or 2020. If they voted by mail in any of these elec­tions, they are considered “not at risk” of being dropped; they are considered “at risk” if they did not cast a mail ballot.

In Arizona, voters do not self-identify their race or ethni­city when they register to vote. To estim­ate each voter’s race or ethni­city, we use a tech­nique common among polit­ical scient­ists called Bayesian Improved Surname Geocod­ing, which predicts race/ethni­city based on each voter’s last name and the demo­graph­ics of the census block in which they live. This method is widely used by academ­ics and voting rights litig­at­ors alike to predict racial/ethnic demo­graph­ics. While it cannot perfectly estim­ate any indi­vidu­al’s race or ethni­city, it has very low error rates at the popu­la­tion level. (This method considers “Latino” to be mutu­ally exclus­ive from other races and ethni­cit­ies — put differ­ently, a voter is considered either white or Black or Latino.)

To be clear, being dropped from the perman­ent early voting list does­n’t mean that a voter will be disen­fran­chised. They’ll receive a notice from the state indic­at­ing that they’ve been removed and can then either reen­list or vote in person. However, many voters might miss the noti­fic­a­tion in the mail — and mail deliv­ery is not always reli­able. As such, some people might not end up voting if they don’t receive their expec­ted mail ballot. And for all voters, the require­ment to rejoin the list of mail voters imposes the sort of small costs that can add up to major burdens on the right to vote — a prob­lem under any circum­stances, and doubly so when the costs have such clear racial dispar­it­ies as the Arizona bill.

Below is a chart showing the percentage of each racial/ethnic groups that are at risk of being removed from the early voting lists. About 11.6 percent of voters — or more than 340,000 Arizon­ans — are at risk of being removed, but there are clear racial dispar­it­ies. Although 8.4 percent voted by mail four years ago for white voters, this was not the case for more Latino voters.

This adds up to big discrep­an­cies in the over­all number of voters who could be dropped when broken down by race and ethni­city. These voters, although they make up 31 percent of the voters on this list, are not white. HalfThere are many people at risk of being dismissed. All nonwhite racial groups are overrep­res­en­ted among those at risk of being removed, but none as much as Latino voters.

Latino voters make up just 17 percent of the perman­ent early voters who aren’t at risk of being dropped but more than a third of those who might fall off the list. The figure below shows the relat­ive risk ratio that each racial group faces of being removed from the list. If a group’s share of the at-risk popu­la­tion is equal to its share of the total popu­la­tion, the group’s risk ratio is 1. A risk ratio above 1 indic­ates that a given group is overrep­res­en­ted in the at-risk popu­la­tion. As you can see, each racial/ethnic group except for white Arizon­ans is above 1 (although the risk ratio for Asian Amer­ican voters is very close to 1).

Under­min­ing the perman­ent early voting list also spells danger for Arizon­a’s Native Amer­ican communit­ies. Because Native Amer­ic­ans make up a relat­ively small share of the popu­la­tion, it is often diffi­cult to identify them using tech­niques such as surname geocod­ing. As an altern­at­ive, we used the geocoded voter file and census shapefiles to identify voters who live on Native Amer­ican tribal lands. To be sure, this misses Native Amer­ic­ans who live off reser­va­tions. Never­the­less, it is help­ful to know whether indi­vidu­als living on reser­va­tions are dispro­por­tion­ately harmed by the new law.

The differ­ence is stark: about 23 percent of voters on the list who reside on tribal lands are at risk of removal, compared to only 11.6 percent of perman­ent early voters statewide. This means that voters who live on tribal lands are twice the likely to face removal.

These results show that S.B. 1485, just like so many other policies that restrict access to voting, will have a dispro­por­tion­ate effect on voters of color and Native Amer­ican voters.