of Americans will likely show up to the polls on Nov. 4 to find they
are no longer registered to vote. That's an estimate base on past
elections and the findings of two leading research groups that found
state-sanctioned voter purges are widely inaccurate.
Both the Brennan Center for Justice and the U.S. Public Interest
Research Group have independently called into question the methods
states use to clean up their voter rolls and the integrity of the
information those purges are based on. Roughly 13 million names were
purged from the rolls in 2005 and 2006, and while most of the removals
were legitimate, that still leaves thousands likely disenfranchised,
Washington state, which canceled the second-highest percentage of
voters, beefed up its voter-roll-cleanup efforts after a 2004
governor's race that determined 1,800 votes were cast by felons or on
behalf of deceased people, said Washington Secretary of State spokesman
"It really radica
lized us for cleaning up the rolls," Ammons said.
Sometimes, though, the purges get too radical. The Brennan Center report, "Voter Purges," shows
that flawed purge lists threaten election integrity some eight years
after the infamous felon purge lists used by Florida in 2000 wrongfully
dropped at least 12,000 voters from the rolls.
The lists used to delete voters are "riddled with inaccuracies,"
according to the report. The bottom line, writes Myrna Perez, the
report's author: "States maintain voter rolls in an inconsistent and
unaccountable manner. Officials strike voters from the rolls through a
process that is shrouded in secrecy, prone to error, and vulnerable to
In Mississippi before the March 2008 primary, a county election official - from her home computer - secretly deleted 10,000 voters from the rolls.
During the 2008 New Mexico caucuses, thousands of registered Democrats found themselves improperly knocked off the rolls.
Ohio election integrity advocates point to massive voter purges before the 2004 election that totaled more than 300,000 voters.
In Michigan, the America Civil Liberties Union and the Advancement
Project filed an injunction Sept. 17 to reinstate more than 200,000
voters purged over the past year.
The Michigan lawsuit in U.S. District Court would
retroactively restore the voting rights before Election Day of an
estimated 180,000 voters purged annually through the state's sharing of
Department of Motor Vehicle records with nearby states. Other states
have agreements to share driver's license records, but Michigan was
singled out because it relies strictly on the DMV match to purge
voters. A hearing was held for preliminary injunction on Oct. 1.
"They don't know that the person, merely by getting a driver's
license, is registered to vote in that state," said Brad Heard,
Advancement Project attorney. "If you're a student who wants to retain
Michigan residency, since he's in Illinois, he might get a driver's
license and still intend to vote in Michigan."
Kelly Chesney, spokeswoman for the Michigan Secretary of State, did
not respond to a request for comment. The state has argued in court
filings that the numbers of affected voters are inflated.
The suit would also block Michigan's practice of purging new
registrations based on voter ID cards sent to voters that are returned
as undeliverable, even those sent out within 90-days of an election.
The state refused to process an estimated 30,000 registrations per year
for this reason, the American Civil Liberties Union claims in its
lawsuit. Similar efforts to verify new registrations through so-called
"no-match" provisions are being disputed in Florida and Wisconsin.
Two months ago, the Advancement Project and the Fair Election Legal
Defense Network accused Louisiana and Kansas, too, of illegally purging
voters based solely on driver's license records.
In Louisiana, 21,000 voters were taken off the rolls when they
picked up driver's licenses in nearby states after Hurricanes Katrina
States Misunderstand Federal Law
The groundwork for today's voter purges stem from the federal Help
America Vote Act, passed in 2002, which handed out $4 billion to states
for voter-roll maintenance and overall election system upgrades. Much
of that money went to software vendors to design the data management
systems to maintain state databases. A handful of these vendors
dominate the voting system industry, and their electronic voting
machines and optical scanners often have a track record of faulty, insecure performance.
In 2006, Washington used $6 million in federal funds to create a
centralized statewide database. In Colorado, a statewide database went
live in April. From January of this year to the end of July, Colorado
changed or canceled 45,658 voter registration records, spokesman Rich
One of the biggest problems in managing voter registration rolls,
says Gary Kalman of the Public Interest Research Group, is the
inconsistency from state to state, and county to county, in enforcing
Kalman's study, "Vanishing Votes," found
19 states had not outlawed purging within 90 days of the vote - a
prohibition found in the National Voter Registration Act. Four states -
Colorado, Ohio, Rhode Island and Connecticut - have laws with purging
deadlines explicitly within the 90-day period.
"If somebody dies within 90 days, becomes incarcerated or convicted
of a felony, those voters may be purged before the election," said
Coolidge, adding that no such statewide purge was planned though
counties may take them up individually.
"There is widespread unawareness of the law," Kalman said. "So many
states were saying they left it up to counties. Because you have rules
being interpreted and enforced differently by counties, literally
within one state, there is a difference in how they convey rules to
voters. It's a decentralized system that can actually impact how votes
The Brennan Center also found vast inconsistencies with the way
states and counties interpret certain provisions of the National Voter
States purge voters for several legitimate reasons looking for such
things as a voter's death, duplicate registrations, change of address
or felony convictions - which individual states treat differently.
Social security numbers and driver's license records may also be used.
The Brennan Center report noted that consistent inaccuracies can be
found in the data used to compile lists. For instance, according to the
report, "some states purge their voter list based on the Social
Security Administration's Death Master File, a database that even the
Social Security Administration admits includes people who are still
In most cases when voters are removed from lists, with the exception
for the reason of an address change, election officials don't attempt
to notify the voter, Perez wrote. Those who find themselves removed
from the list may cast a provisional ballot while their voting status is reviewed.
Aside from common typos and other clerical errors, voter roll
matching sometimes fails to account for the presence or lack of middle
initials or non-standard surnames. This was largely the problem behind
purge lists in Florida in 2000 and a pending purge in 2004 of 48,000
names that disproportionately affected African Americans that advocates
identified and blocked.
"Far too often what appears to be a 'match' will actually be the
records of two distinct registrants with similar identifying
information," according to the Brennan Center report. "States have
failed to implement protections to ensure that eligible voters are not
13 Million Purged in 39 States
Voter purges in the years following the 2004 election were strongly encouraged by the Justice Department, citing voter fraud.
The DOJ had the backing of provisions in HAVA as well the NVRA that
required states to vigilantly keep its voter rolls spotless.
In 2005 and 2006, 39 states and the District of Columbia cancelled roughly 13 million voters from the rolls, according to a report to Congress by the Election Assistance Commission released last year.
Topping the list of states taking names off their registered-voter
rolls in 2005 and 2006 were Colorado, with 16.9 percent of the roll
removed; Washington, 15.4 percent; New York, 14.1 percent; Nevada, 13.2
percent; and Missouri, 10.4 percent.
Spokesmen for Secretary of State offices in Colorado and Washington
- states with the two highest purge rates - said the overall number of
cancellations included those who may have cancelled and then
Estimates of how many people moved out of each state, based on U.S.
Census Data in 1995 and 2000, roughly match the number of voter purges
in each state, suggesting that a large portion of the cancellations may be justified.
While Florida purged close to 1 million voters in 2005 and 2006, an
estimated 1.3 million moved out of the state over a five-year period
between 1995 and 2000, according to the U.S. Census.
Regardless of how many people should be appropriately removed from
voter rolls, states still must follow the federal law meant to protect
voting rights, said Heard, Advancement Project attorney.
"Certainly we're not advocating that people should be registered in
multiple places," Heard said, but in some cases voters' rights are not