Skip Navigation
Analysis

Calls for More Purges Rest on Shaky Data

We examined the claims of groups pushing for aggressive purging. They didn’t hold up.

  • Jonathan Brater
October 19, 2018

In July, a Bren­nan Center report revealed that voter purges are on the rise nation­wide, rais­ing the threat that signi­fic­ant numbers of eligible voters could be disen­fran­chised. The report also found that the rise in purges has coin­cided with several activ­ist groups lead­ing campaigns to pres­sure elec­tion offi­cials to purge more people from the voter rolls. This push has included, at times, lawsuits seek­ing the imple­ment­a­tion of aggress­ive purge prac­tices.

Now, the Bren­nan Center has looked more closely at the activ­it­ies of the four organ­iz­a­tions lead­ing the charge for more purges — Judi­cial Watch, the Amer­ican Civil Rights Union (ACRU), True the Vote, and Public Interest Legal Found­a­tion (PILF). These four groups have collect­ively sent letters to more than 450 juris­dic­tions since 2012, claim­ing that they’re viol­at­ing federal voting law by not purging aggress­ively enough. 

We found:

  • These groups base their calls for more purges on claims that regis­tra­tion rates are inflated by ineligible voters. But when look­ing at some of the recent asser­tions by PILF in partic­u­lar, the data suggests that some of the claims are exag­ger­ated and even flatly contra­dicted by the same sources of evid­ence on which they are based. 
  • These groups as a whole are more likely to sue juris­dic­tions with higher minor­ity popu­la­tions, even if there is no appar­ent racial pattern in the areas they initially contact. 

No one objects to respons­ible voter roll main­ten­ance. There is agree­ment that elec­tion offi­cials should clean the rolls when people move, die, or other­wise become ineligible. But when offi­cials don’t use appro­pri­ate safe­guards or trans­par­ency, eligible voters are at risk of being taken off the rolls erro­neously or without notice. 

PILF’s False Claims About Ineligible Voters

PILF has been the most active of the four groups we looked at, contact­ing more than 250 juris­dic­tions just since last year. It also has repres­en­ted ACRU in several lawsuits. PILF typic­ally claims a county has more voters on the regis­tra­tion rolls than eligible voters. Though they typic­ally do not provide specific numbers, they say the claim is based on Census popu­la­tion figures and voter regis­tra­tion data collec­ted by the U.S. Elec­tion Assist­ance Commis­sion (EAC). 

In response to these claims, the Bren­nan Center worked with part­ner organ­iz­a­tions Demos and the Lawyers’ Commit­tee for Civil Rights Under Lawto contactall juris­dic­tions to which PILF sent letters in 2017 and request public records regard­ing the counties’ list main­ten­ance prac­tices. The Bren­nan Center compared that inform­a­tion, along with all publicly avail­able data, to PILF’s asser­tions.

The three counties below are examples of juris­dic­tions PILF claimed had, in its words, more voters “on the regis­tra­tion rolls than eligible, living voters.” In each case, a compar­ison in the relev­ant time period between regis­tra­tion numbers provided by the counties to the EAC and what the Amer­ican Community Survey estim­ates the citizen voting age popu­la­tion (CVAP, eligible voters) to be reveals that the claim appears to be false:

  • Plumas County, Cali­for­nia

o  Registered voters: 13,230

o  Eligible voters (estim­ated): 15,240

  • Edwards County, Texas 

o  Registered voters: 1,398

o  Eligible Voters (estim­ated): 1,480

  • Kenedy County, Texas

o  Registered voters: 265

o  Eligible Voters (estim­ated): 315

We found other meth­od­o­lo­gical flaws. In general, PILF did not take the time to analyze why the figures upon which they relied did not accur­ately meas­ure either the number of active voters or the extent of the eligible popu­la­tions. Some­times, PILF included in “registered” counts “inact­ive” voters — those whom juris­dic­tions have already iden­ti­fied as poten­tially ineligible and who may already be on the path to be removed. Some­times they even included “non-active” records — most of which repres­ent voters who have already been canceled but for sound reas­ons had not been wiped entirely from the voter regis­tra­tion data­base. 

On the other side of the equa­tion, PILF did not consider that for some counties, census data might not accur­ately meas­ure the eligible voting popu­la­tion. This is partic­u­larly true for counties with low popu­la­tions, seasonal resid­ents, or indi­vidu­als who might be tempor­ar­ily absent for school, work, or milit­ary service but still eligible to vote. A few illus­trat­ive examples of some of these flaws are below:

  • Roberts County, Texas, respon­ded to PILF by explain­ing it has many part-time resid­ents who are coun­ted by the Census else­where but are eligible to vote in the county. Thus, it appeared to have 692 registered voters (when the total number of eligible voters as repor­ted by the Amer­ican Community Survey was 697), because the Census number under-repres­en­ted the actual eligible popu­la­tion. 
     
  • Clark County, Wash­ing­ton, claims that the EAC report on registered voters wrongly included about 200,000 voter records on account of a confla­tion of data­base termin­o­logy (in this state, persons with “non-active” records are not able to vote, whereas voters in other categor­ies like “inact­ive” can still vote). Fewer than one sixth of these 200,000 voters were actu­ally still on the rolls, and thus PILF’s numbers are based on an over­es­tim­a­tion of the registered popu­la­tion by about 170,000.
     
  • Simil­arly, San Diego County retains all of its records in its voter data file, and PILF presumed that there were 2.5 million voters on the rolls compared to an adult popu­la­tion of 2.6 million, perhaps due to another misun­der­stand­ing of termin­o­logy). But the number of relev­ant (active) voters is only 1.67 million. 
     
  • Color­ado’s response to PILF further illus­trates how big the differ­ence can be when not compar­ing the right categor­ies. Color­ado provided a break­down of the differ­ence in these figures among counties targeted by PILF: 

Color­ado Counties

 “Regis­tra­tion” Rate

 “Active Regis­tra­tion” Rate*

Archu­leta 

100.4

80.5

Baca 

99.6

89.1

Boulder 

92.7

77.5

Chey­enne 

103.2

95.8

Clear Creek 

100.3

83.8

Dolores 

103.3

91.3

Gilpin 

100.0

81.1

Hinsdale 

119.1

109.4

Jack­son 

107.6

85.7

Mineral 

132.2

121.2

Moffat 

100.3

73.6

Ouray 

109.4

90.2

Park 

98.8

83.2

Phil­lips 

97.7

80.0

Pitkin

100.5

80.0

San Juan 

117.7

92.3

Summit 

104.4

72.7

*Bren­nan Center for Justice chart adap­ted from figures provided by Color­ado. Color­ado used the Amer­ican Community Survey 2016 popu­la­tion estim­ates to calcu­late counties’ active regis­tra­tion rate. The citizen voting age popu­la­tion estim­ates are similar for most counties. Hinsdale and Mineral, small counties, have regis­tra­tion numbers above their Census-estim­ated popu­la­tions (as well as Dolores and San Juan, when using CVAP). The Color­ado secret­ary of state’s office said this is because “census projec­tions don’t adequately account for popu­la­tions in these counties where many seasonal resid­ents may be absent during census visits,” as was the case with Roberts County, Texas. 

Lawsuits Target Minor­it­ies

There are no discern­able trends in the demo­graph­ics of the 450 counties that received letters from these groups in total pres­sur­ing them to conduct purges. But there are excep­tions within certain states: All 12 counties in Alabama that received letters were major­ity black, along with 17 of 19 counties in Missis­sippi. 

State

Juris­dic­tion

Years Suits Were Brought

Group(s) Involved in Suit

Cali­for­nia

Los Angeles

2017

Judi­cial Watch, Elec­tion Integ­rity Project

Cali­for­nia

State of Cali­for­nia

2017

Judi­cial Watch, Elec­tion Integ­rity Project

Color­ado

Gilpin County

2014, 2015, 2017

TTV

Color­ado

Mineral County

2014, 2015, 2017

TTV

Flor­ida

Broward County

2016

ACRU, PILF

Indi­ana

State of Indi­ana

2012

Judi­cial Watch, TTV, Elec­tion Law Center

Kentucky 

State of Kentucky

2017

Judi­cial Watch

Missis­sippi

Clarke County

2014

ACRU

Missis­sippi

Jeffer­son Davis County

2013, 2014

ACRU, Elec­tion Law Center

Missis­sippi

Noxu­bee County

2014, 2015

ACRU, PILF

Missis­sippi

Walthall County

2013, 2014, 2015

ACRU, Elec­tion Law Center

North Caro­lina

State of North Caro­lina

2017

ACRU, PILF

North Caro­lina

Wake County

2016

Voter Integ­rity Project NC, PILF

Ohio

State of Ohio

2012

Judi­cial Watch, TTV, Elec­tion Law Center

Texas

Starr County

2015

ACRU, PILF

Texas

Terrell County

2013, 2015

ACRU

Texas

Zavala County

2014, 2015

ACRU, Elec­tion Law Center

Virginia

Alex­an­dria

2016

Virginia Voter’s Alli­ance, Elec­tion Law Center

 

And when look­ing at the smal­ler number of juris­dic­tions that were actu­ally sued, a differ­ent picture emerges: Minor­ity communit­ies were singled out for lawsuits. We found:

  • Nine out of the 13 juris­dic­tions sued by Judi­cial Watch in Alabama and Texas were major­ity non-Hispanic white (“non-white”). 
  • The lone lawsuit filed in Pennsylvania was brought against Phil­adelphia, whose non-white popu­la­tion is 64.7 percent.
  • The only Flor­ida suit was brought in Broward County, whose non-white popu­la­tion is 63.5 percent.
  • In Cali­for­nia, Judi­cial Watch sued two juris­dic­tions: the entire state and Los Angeles County, whose non-white popu­la­tion is 73.8 percent. 
  • Most recently, ACRU announced a settle­ment in Starr County, Texas, whose non-white popu­la­tion is 96.7 percent.

Good voter list main­ten­ance bene­fits us all, and juris­dic­tions should strive to have top-flight prac­tices. Elec­tion offi­cials should be wary, however, of exag­ger­ated or false claims used to demand more purging. 

Kevin Morris and Rebecca Ayala provided research for this article.

(Image: Jessica McGowan/Getty)