electionline Weekly

In Focus This Week

Midwest Voter Registration Data-Sharing Project Moves Forward
Kansas leads groups of states crosschecking information; Advocates voice concern

By Sean Greene
electionline.org

An effort by officials to share voter registration information across state lines has been growing in recent months 

A group of four Midwestern states that agreed to cross-check voter registration data to identify duplicate records has now grown to six, with a second group in the works including six more states, ranging from Arkansas to New Mexico.

At the forefront of the effort is Kansas, where Secretary of State Ron Thornburgh (R) said he is interested in finding both potential duplicate registrations as well as possible double voters.

The first group of states came together in December 2005 at the Midwest Election Officials Conference where secretaries of state from Iowa, Kansas, Missouri and Nebraska signed a memorandum of understanding (MOU) to share and crosscheck voter registration data.  

Brad Bryant, Kansas’ state election director, said sharing information yielded immediate positive results in safeguarding the voting system.  

“The thinking was threefold: with our Help America Vote Act voter registration databases we are now more alike than before and have a tool for crosschecking; we were all conducting intrastate duplicate checks, and it made sense to expand to interstate checks; and we had found three Kansas City residents who had voted in both Missouri and Kansas a few years ago. We got convictions on all three,” he said.

Initial testing started in early 2006. Later that year, Minnesota joined the other states, followed by South Dakota in 2007. Two more states have expressed interest in joining in 2008.

Bryant added that the checks are not part of any new, vendor-created voter registration database, but rather performed by in-house information technology staff paid for no extra cost.

Currently, the Midwest program is being hosted by the information technology department of the Iowa Secretary of State’s office. Klint Belz, director for voter registration systems, said data matching is a hands-on job.

“It is a manual process that uses SQL queries to identify if corresponding fields contain matching information,” he said. “In short the data we receive from other states is loaded into tables so that we have all of the records in corresponding fields in each table. Each state’s table is compared field by field to identify matching information in the fields.” 

The data provided by each state includes first name, middle name, last name, date of birth, a number used to identify voters, county name and date of registration. Of these, first name, middle name, last name and date of birth are used to match records – records are considered a potential match when all four of these match exactly.

The possible matches are compiled into a table which is then sent on to the appropriate state. The states then process the information according to their state laws or regulations.

In Kansas, the first crosscheck run in July-August 2006 produced 15,171 possible matches. The second and most recent crosscheck in August-September 2007 produced 11,205 possible matches. According to Bryant approximately 20 percent of the matches in 2007 were identical listings from a year earlier.

These matches were then sent on to Kansas county election officials who investigated further and canceled a voter’s registration only if, “the county election official is certain the records represent the same person and the Kansas record is the older record, meaning the record in the other state has a newer registration date,” according to a September 2007 state election newsletter.

The state does not currently have data on how many voters were removed due to the duplicate crosscheck. Counties are asked but not required to supply that information back to the state. Overall in 2006, Kansas removed 27,629 registrations from the rolls and in 2007 they removed 66,483.

Brian Newby, election commissioner for Johnson County, Kan., said the process worked well in his jurisdiction.

“We are a growing county on a state line. We don’t want someone voting twice and it’s also really important to know from an election-planning perspective how many registered voters we have,” he said.

Other states have found thousands of potential duplicates as well. More than 16,000 duplicates were identified in Iowa in 2006 and 5,753 in 2007. Nearly 2,800 matches were detected in South Dakota in 2007. In both states, data for how many voters were then removed was not available at the state level as, like in Kansas, the final processing was done at the county level and the state does not have the collected data.

South Dakota Secretary of State Chris Nelson (R) praised the data checking between states.

“We have an interest in keeping our voter list as clean as possible,” he said. “It’s that simple. As I scanned the list of matches from the county where I live, I found eight names of people I know who had moved to other states and in every case the system accurately identified the state which they had moved to. Without this system, these names would have stayed on the active voter list for at least another four years and perhaps much longer than that. That only invites fraud.” 

While election officials are pleased with the interstate data sharing, some advocates have concerns about data produced by states’ matching processes – both about how that data is used and the potential for voters being erroneously removed from voter rolls.

“There’s nothing wrong with creating the information itself as long as it’s used appropriately,” said Justin Levitt, a lawyer from the Brennan Center for Justice at New York University’s School of Law.

“But over and over again, we’ve seen similar information used inappropriately, by citizens or policymakers who either don’t understand the limits of the information or who succumb to political pressure to use the information when the end use is unwarranted,” he said. “Using voter-caging materials to challenge voters is one example of the misuse of limited, unreliable information. There are many other examples and that history gives us reason for substantial concern.”

A recent paper by Levitt and Michael McDonald, an associate professor of government and politics at George Mason University, cited another issue termed “the birthdate problem,” where statistically speaking, it is possible for two different people to have the same first name, last name and birth date in a large population pool.

And data-sharing has faced criticism in other regions that have performed crosschecks. In Kentucky concerns were raised in 2006, including by the state’s attorney general, when it shared registration data with Tennessee and South Carolina leading to the purging of more than 8,000 voters from the state’s voter list. Of these 8,000, 196 showed up to vote on primary election day in May 2006 and were allowed to vote after showing proper identification. Later in the year, a judge restored all 8,000 to the inactive voter list, although the court agreed with the state that database matching was a good idea.

Louisiana faced similar criticism and a lawsuit from the NAACP (which was eventually dismissed) earlier this year when it crosschecked registration information with other states and jurisdictions including San Diego, Calif., Colorado, Florida, Georgia, Las Vegas, Nev., New York, N.Y., Tennessee and Texas and subsequently purged nearly 20,000 voters from the rolls.

Bryant said Kansas state and local election officials are quite aware of these concerns.

“We and our counties are very sensitive to the fact that we are dealing with people’s constitutional rights. We are more concerned with not canceling a voter who should not be canceled than with leaving someone on the list who is a duplicate and should be removed,” he stated.

He said he was not aware of any cases where a voter has said they were wrongly removed from the rolls due to the duplicate check. He added that the state defers to localities, who are the most familiar with their voters and their address files and have the local knowledge to determine identities.

And the Midwestern model continues to expand. Another MOU is being circulated to create a new group led and hosted by Kansas including Arizona, Arkansas, Colorado, New Mexico, Oklahoma and Texas. The hope is to undertake the first cross check in most of these states sometime in the first half of 2008.

The language of the new agreement includes an option to include voter history in the data exchanges to uncover possible double votes. Kansas did that in its 2007 check and found a few potential double votes that the state is currently researching.

Bryant says his state plans to continue the crosschecks at least once a year and thinks other states can look the Midwest agreements and Kansas if they want to start crosschecking data as well.

“The model we in Kansas developed that we think will work on a national level is for states to form regional agreements like ours and decide what data fields to include, frequency, etc.,” Bryant said. “They can use our format if they wish. We’re also working on XML which we would share with anyone who wants it to facilitate the exchange of data.”