Election Law @ Moritz


Recent Opinion Articles

Edward B. Foley
Free & Fair is a collection of writings by Edward B. Foley, one of the nation's preeminent experts on election law.

Weekly Comment

Statewide Registration Lists: The Most Important Election Reform?

Print Page

March 21, 2006

Effective January 1, 2006, every state in the country was required to have in place computerized statewide voter registration list, commonly known as a "statewide registration database." Although this requirement has received less public attention than those in the area of voting technology, statewide registration databases may well be the most significant change in federal law contained in the Help America Vote Act of 2002 ("HAVA"). Many states are not in compliance with this requirement. Worse still, the available evidence suggests that some states that are in compliance have some serious problems with their database standards, that may result in eligible voters wrongly being stricken from the rolls. This could mean that voters will show up at their polling place on election day in 2006, only to find that their names aren't on the list.

This comment summarizes HAVA's requirements with respect to statewide registration databases, as well as the progress of states in meeting this requirement. It also discusses some difficulties that may arise in the implementation of this requirement, and considers next steps that might be taken toward the improvement of statewide registration databases.

HAVA's Database Requirements

When Congress enacted HAVA in 2002, it recognized that the antiquated machines that used to cast votes were only one of the problems afflicting our system of election administration. Numerous studies conducted after the 2000 presidential election showed that the problems with the country's registration lists were at least as serious. For example, the influential report of the Caltech/MIT Voting Technology Project, Voting: What Is, What Could Be, estimated that of the 4 to 6 million lost votes in the 2000 presidential election, between 1.5 and 3 million were due to registration mix-ups. Florida was among the states that experienced serious registration problems in the 2000 presidential election, reportedly resulting in many eligible voters being turned away from the polls on the ground that their names did not appear on registration lists.

There are several provisions of HAVA that are designed to deal with the issue of voters being denied access due to registration errors, but the most pertinent are the ones contained in section 303(a), 42 U.S.C. § 15483(a), entitled "Computerized statewide voter registration list requirements." Prior to HAVA, voting lists were maintained at the local level in most states, usually by county registrars or boards of election. Under HAVA, each state's chief election official must implement "a single, uniform, official, centralized, interactive computerized statewide voter registration list defined, maintained, and administered at the State level." This database must contain the name and registration information of every registered voter in the state.

The database must allow "immediate electronic access" to all election officials within the state, including local election officials. Any information that local officials obtained shall be entered "on an expedited basis at the time the information is provided to the local official" (emphasis added). The idea here is that it is essential for the state registration database to reflect current information, including any changes that voters provide to local election officials. State chief election officials must provide "such support as may be required so that local election officials are able to enter information." The implication is that the data entry will actually be done by local election officials, although HAVA does not exclude state-level officials from entering changes or additions into the database.

HAVA also contains requirements as to the maintenance of statewide registration databases. Foremost among those is that states must still adhere to the requirements of the National Voter Registration Act of 1993 ("NVRA" or "Motor-Voter"), in removing voters from the database. Specifically mentioned is Section 8 of the NVRA, which provides that states shall not engage in programs to systemically remove, or "purge," voters from the rolls within 90 days of a federal election. This section also provides that voters' names are not to be removed from the rolls due to a change of residence, unless they either (a) confirm in writing that they have moved, or (b) do not respond to an address-confirmation notice sent by forwardable mail and do not appear to vote in at least two federal general elections after that notice is sent. To illustrate the latter requirement, this means that if a notice is sent in March 2006 and not returned, a voter could not be removed from the rolls, unless and until she does not appear to vote in both the November 2006 and November 2008 general elections.

HAVA requires that list maintenance be performed which ensures that "duplicate names are eliminated." On the other hand, HAVA also requires that there be "[s]afeguards to ensure that eligible voters are not removed in error from the official list of eligible voters." Reconciling these two imperatives – eliminating ineligible voters and duplicate records, without erroneously deleting eligible voters – is sure to be a formidable challenge in implementing HAVA's statewide registration database requirement.

Among the most significant, if somewhat confusing, requirements has to do with the coordination of information on the statewide registration database with other sources. HAVA requires that information be shared between a state's chief election official and its motor vehicle authority – more specifically, they are required to enter an agreement to "match information" from the state registration database with that in the state motor vehicle database, in order to "verify the accuracy" of information provided by the voter. To facilitate such matching, voters who have a state driver's license are now required to provide that number. Those who lack a driver's license must provide the last four digits of their social security number, and those who have neither are to be assigned an identifying number.

State motor vehicle authorities are in turn required to enter into agreements with the Commissioner of Social Security, to facilitate the sharing of information. Congress directed the Commissioner to enter into agreements with states that will allow the verification of specified information, specifically voters' names, dates of birth, social security numbers, and whether they are deceased. This information is supposed to be kept "strictly confidential," and not to be used for purposes other than the matching and verification required by HAVA. Left ambiguous by HAVA is what should happen if the registration information provided by the voter cannot successfully be "matched" with motor vehicle or social security records – a subject I return to below.

States must also coordinate their registration database with other officials records. Specifically, HAVA requires states to coordinate the statewide registration database with other state records, to ensure that ineligible felons and deceased voters are removed.

The original deadline for compliance with the statewide registration database requirement was January 1, 2004. States were allowed to obtain an extension until January 1, 2006, however, by certifying to the Election Assistance Commission ("EAC") that they could not meet this deadline for good cause. The vast majority of states obtained this good-cause extension.

Implementation of HAVA's Requirements

Implementing the statewide registration database requirements has proven to be a challenge for the states. Among the most significant issues that have arisen are: (1) how to allocate responsibilities as between state and local election officials, (2) how to "match" statewide registration database records against other databases, specifically motor vehicle and social security records, and (3) how to remove voters who are truly ineligible, without mistakenly purging eligible voters.

As for the first issue, there have been two contrasting approaches to state-local responsibility that the states have taken in developing their statewide registration databases. A majority of states (38, according to electionline.org) have adopted a "top-down" approach. Under this approach, a unified registration database is maintained by state election officials, with the information needed to update this database provided by local election officials. Seven states, however, opted for a "bottom-up" approach, in which local jurisdictions maintain their own registration lists – as has been the practice in most states before now – with information periodically submitted to the state election authority. In effect, the statewide registration database functions as a compilation of local lists under the "bottom-up" approach. Other states have adopted a hybrid approach that combines elements of both, and one state (North Dakota) is exempt because it does not maintain voter registration lists.

In July 2005, approximately six months before the statewide database requirement took effect, the EAC released its Voluntary Guidance on Implementation of Statewide Voter Registration Lists. The EAC does not have the power to issue binding regulations to implement HAVA's requirements, which is why this guidance is "voluntary." Although the EAC's guidance states that "top-down" databases are "most closely akin" to what HAVA requires, it declined to say that "bottom-up" databases violate HAVA. To the contrary, it concluded that "bottom-up" databases comply with the act "as long as the State database, the data and the data flow are defined, maintained, and administered by the State." In order to make sure that new information is promptly incorporated into the statewide registration database, the EAC recommended that there be synchronization between the local lists and statewide database at least once every 24 hours.

The second issue, the "matching" of statewide registration databases against other official records, has raised concerns that voters may erroneously be deleted from registration rolls due to data entry or other administrative errors. The Brennan Center for Justice highlights this problem in its recently released report, Making the List: Database Matching and Verification Processes for Voter Registration. Based on a survey of the states, the Brennan Center discovered widely divergent practices among the states in this area. It found particularly troubling the expected practice of some states to reject registration applications, if the information could not precisely be matched with social security or motor vehicle records. While HAVA requires that some matching be done to "verify the accuracy" of information provided on accuracy, it does not say what should happen in the event that there is not a match. Especially worrisome is the prospect that registration applications could be rejected, due to trivial discrepancies.

By way of illustration, the Brennan Center notes that typographic errors in data entry (e.g., entering "Pierce" instead of "Peirce") and the use of nicknames (e.g., "Sam Peirce" rather than "Samuel Peirce") could result in the rejection of registration applications. For this reason, the Brennan Center report recommends against requiring an "exact match" to verify registration information, but instead suggests a more flexible "substantial match" standard. Even where a "substantial match" cannot be obtained, it recommends that voters should still be registered. The consequence for voters would be that they would then have to produce some form of documentary identification, if they are among the group of voters to which HAVA's ID requirement applies (first-time voters who registered by mail).

According to the Brennan Center, 24 states will place voters on the rolls even if a match cannot be obtained, with some of those states requiring them to sign an affidavit or produce ID in order to vote. At the other end of the spectrum, 6 states will refuse to register voters entirely if a match cannot be obtained. This stringent standard will make it very likely that some voters will find their names not on the rolls due to data entry errors.

The EAC's guidance speaks to this issue, recommending that states "ensure that properly filed registration applications from eligible voters are not rejected due to a database error or inflexible database coordination or matching rules." It also recommends that states avoid rejecting registration applications as unverifiable until they have "given the individual an opportunity to correct the information at issue and attempted to validate the accuracy of the government information contained in its databases."

The third issue is how states can successfully clear "deadwood" from the rolls without mistakenly purging eligible voters. In the majority of states that disenfranchise some or all ex-felons, this will require care to ensure that voters are not erroneously deleted because they have the same name as a convicted felon – something that occurred in Florida prior to the 2000 election. The same concern applies to voters who have died. The "unique identifier" to be assigned to each voter is supposed to prevent confusion among voters with the same or similar names, but it remains to be seen whether this will function as an adequate check on erroneous purges of eligible voters. Another concern is whether states will in fact adhere to the NVRA's requirements, applicable to voters who change residence, as HAVA requires. As noted above, a voter should not be removed due to an address change, unless she confirms that change in writing, or unless she fails to vote in two consecutive federal general elections after a mailed notice is sent.

States have adopted a variety of approaches in constructing their databases. Most states entered into contracts with vendors to create statewide registration lists. Among the companies that have sought states' business are major voting machine vendors (ES&S, Diebold, and Hart Intercivic) and other companies including IBM, Accenture, and Unisys. In some cases, vendors were unable to produce statewide databases on the timelines required by HAVA.

Among the states that did not comply with the January 1, 2006 deadline for implementation of the statewide registration database was California, which entered into a memorandum of agreement with the U.S. Department of Justice in order to avoid a lawsuit. DOJ subsequently brought this lawsuit against the State of New York, for that state's failure to comply with HAVA's database requirements, as well as its voting equipment requirements. Other states that reportedly missed HAVA's January 1, 2006 database deadline include Colorado, Wyoming and Nevada. Other states, like Indiana, have a registration database in place but have experienced glitches in pre-election testing.

What to Look Out for in 2006

While the missed deadlines are unfortunate, an even more serious problem would be for voters to be mistakenly deleted from registration rolls due to faulty matching or purging practices. Voters in states that have adopted stringent data-matching standards, and have decided not to register voters where a match is not found, are especially at risk.

There are also security and privacy concerns surrounding the implementation of statewide registration databases. That includes not only the people who will have access to the information they contain, but also the ability of malicious insiders to alter registration records. Given the enormous attention that has been given to the security of electronic voting machines, it is somewhat surprising that relatively little attention has been given to the security of voter registration records, where the risks of foul play are at least as serious. For example, an insider with access to state registration records could selectively delete voter records in precincts where voters of one party predominate, thus affecting election results, if proper procedural safeguards are not in place.

Another concern arising from the implementation of statewide registration databases is that it may put additional pressure on provisional ballots, at least in the short term. To the extent that voters are mistakenly deleted or omitted from new voter registration databases, those voters still have the right to cast provisional ballots in federal elections. Whether or not those provisional ballots will actually be counted, however, is another question. That will depend largely on the standards set by the law of that state, and whether or not election officials are able to pinpoint the reason why a voter was not included in the statewide registration database between election day and the deadline for counting provisional ballots. This could obviously be a laborious process but, in close elections, it might spell the difference between victory and defeat.

None of this is to suggest that Congress made a mistake when it decided to require statewide registration databases as a part of HAVA. It is simply too early to make that judgment. So too, it would be a mistake to enact new federal requirements with respect to statewide registration databases, until the results of state's implementation of HAVA-required databases are known and can be carefully studied. Among the questions that ought to be examined are:

  • whether top-down databases function better than bottom-up databases,
  • whether stringent matching standards resulted in the erroneous deletion of eligible voters,
  • whether the statewide registration databases have succeeded in accurately removing "deadwood" from the voting rolls,
  • whether state election officials were successful in coordinating state registration databases with motor vehicle, social security, penal, and death records,
  • whether the implementation of the statewide registration databases resulted in more provisional ballots being cast, and
  • whether provisional voting served as an effective means by which to count the votes of those who were wrongly purged from the voting rolls.

One of the chief benefits of our federalist system of government is that it allows states to operate, in the words of Justice Brandeis, as "laboratories of reform." Seen in this light, there is a useful dimension to the diversity of approaches toward statewide registration databases being taken by the various states. To take advantage of this feature of our democracy, however, it is vital that there be systematic research on how well the reforms of various states are actually functioning in real-world elections. The most important thing that we can do in 2006, then, is to study with care the results of the different approaches that states are taking to statewide registration databases, and evaluate how effectively different methods have worked. Until we have answers to these and other questions surrounding the implementation of state registration databases, it is premature to recommend new changes in federal law.