Skip Navigation

Testimony Before Congress Regarding the EAC

Testimony before Congressional Subcommittee on Financial Services and General Government regarding the U.S. Election Assistance Commission (the “EAC”). Includes oversight recommendations.

Published: February 27, 2008

Down­load testi­mony [pdf]

Testi­mony of
Deputy Director, Demo­cracy Program
Center for Justice at NYU School of Law
Before the
Subcom­mit­tee on Finan­cial Services and General Govern­ment of the
House Appro­pri­ations Commit­tee

  Febru­ary 27, 2008


Wendy Wesier On behalf of the Bren­nan Center for Justice at NYU School of Law, I thank the Subcom­mit­tee on Finan­cial Services and General Govern­ment for hold­ing this hear­ing and for provid­ing me the oppor­tun­ity to discuss the U.S. Elec­tion Assist­ance Commis­sion (the “EAC”) and elec­tion admin­is­tra­tion issues.

My name is Wendy Weiser, and I direct the Bren­nan Center’s work on voting rights and elec­tions. The Bren­nan Center is a nonpar­tisan think tank and legal advocacy organ­iz­a­tion that focuses on issues of demo­cracy and justice.  Among other things, we seek to ensure fair and accur­ate voting proced­ures and systems and to promote policies that maxim­ize citizen enfran­chise­ment and parti­cip­a­tion in elec­tions.  We have done extens­ive work on a range of issues relat­ing to elec­tion admin­is­tra­tion, includ­ing work to remove unne­ces­sary barri­ers to voter regis­tra­tion; to make voting machines more secure, reli­able, usable, and access­ible; and to expand access to the fran­chise.  Our work on these topics has included the public­a­tion of stud­ies and reports; assist­ance to federal and state admin­is­trat­ive and legis­lat­ive bodies with respons­ib­il­ity over elec­tions; and, when neces­sary, parti­cip­a­tion in litig­a­tion to compel states to comply with their oblig­a­tions under federal law and the Consti­tu­tion.  We have also monitored the work of the EAC on a vari­ety of elec­tion admin­is­tra­tion topics.

My testi­mony today will address how Congress can help ensure that the EAC is effect­ive this year in assist­ing the states to admin­is­ter smooth and fair elec­tions.  I will focus in partic­u­lar on the areas in which we believe this Subcom­mit­tee can provide help­ful over­sight of the EAC.  Although we believe that such over­sight is needed, we also recom­mend that this Commit­tee support the EAC because its work is crit­ical to our coun­try’s elec­tion admin­is­tra­tion.

I.          Congres­sional Over­sight is Needed to Ensure that the EAC Oper­ates in a Trans­par­ent Manner and Follows Regu­lar Proced­ures

First and fore­most, over­sight is needed to ensure that the EAC func­tions pursu­ant to regu­lar proced­ures in a trans­par­ent manner that facil­it­ates public account­ab­il­ity.  Over the past few years, the EAC has come under criti­cism for its fail­ure to adopt and follow regu­lar proced­ures and for the lack of trans­par­ency of its oper­a­tions and decision-making processes.  These fail­ures have, unfor­tu­nately, under­mined public confid­ence in the agency.  Although the EAC has taken steps to increase the trans­par­ency of its oper­a­tions this year-espe­cially by publi­ciz­ing more inform­a­tion about the agency’s oper­a­tions on a regu­lar basis-more can and should be done to foster public account­ab­il­ity and to regu­lar­ize and stand­ard­ize the agency’s oper­a­tions.

Perhaps the most vivid illus­tra­tion[1] of this prob­lem is the way in which the EAC handled two research projects it commis­sioned: a report on voter iden­ti­fic­a­tion and a report on voting fraud and voter intim­id­a­tion, both of which were completed by mid-2006.  Both reports were commis­sioned pursu­ant to the agency’s respons­ib­il­ity to conduct and make avail­able to the public stud­ies regard­ing elec­tion admin­is­tra­tion issues.[2]  Nonethe­less, the EAC refused to release those reports until mid-2007, after it came under intens­ive scru­tiny and pres­sure from members of this Subcom­mit­tee and others. 

The Bren­nan Center consist­ently denounced the EAC’s decision to suppress or delay the release of those reports.[3]  There is simply no legit­im­ate reason for a public agency to with­hold research on matters of public concern, partic­u­larly when the research was commis­sioned with taxpayer dollars from estab­lished experts in the field pursu­ant to a stat­utory mandate that the agency make stud­ies on elec­tion admin­is­tra­tion issues avail­able to public.  The duty to disclose is espe­cially strong because the EAC’s primary purpose is to serve as “a national clear­ing­house and resource” on elec­tion admin­is­tra­tion inform­a­tion and proced­ures.[4]  Worse, the EAC with­held these reports at a crit­ical time when the issues addressed in the reports were the subjects of debate and decisions at all levels of federal and state govern­ment.  These debates and decisions took place without the bene­fit of the reports’ find­ings on the signi­fic­ant impact of voter ID laws in redu­cing voter turnout and the low preval­ence of the kind of voter fraud targeted by ID laws.

The decisions to with­hold these reports from the public, as well as the subsequent decisions to release these reports, were not made in a public vote by the Commis­sion­ers, nor were they publicly justi­fied or explained in any way.  There were no agency policies or proced­ures in place to guide the Commis­sion’s actions with respect to the research.  Indeed, the Commis­sion did not even have published proced­ures for hand­ling requests for public inform­a­tion, as required by the Free­dom of Inform­a­tion Act.  To the best of our know­ledge, no system­atic and regu­lar proced­ures have been put in place since that time.

In April 2007, the Bren­nan Center urged the EAC to adopt a number of recom­mend­a­tions designed to increase the trans­par­ency of the agency’s research oper­a­tions, to bring the EAC’s prac­tices in line with gener­ally accep­ted research proto­cols and meth­ods, to promote greater confid­ence in the EAC’s processes, and to enhance its public legit­im­acy.  I attach a copy of those recom­mend­a­tions as an exhibit to my testi­mony.[5]  To the best of our know­ledge, these recom­mend­a­tions have not yet been adop­ted by the Commis­sion.  We respect­fully request that this Commit­tee consider these recom­mend­a­tions in its over­sight of the EAC.

Also in April 2007, the EAC asked its Inspector General to review its internal proced­ures, includ­ing its proced­ures for award­ing and managing research contracts.  We welcome any recom­mend­a­tions by the Inspector General designed to achieve the above-refer­enced goals.  Given how crit­ical the work of the Elec­tion Assist­ance Commis­sion is to the effect­ive admin­is­tra­tion of elec­tions, it is essen­tial that its proced­ures and processes be reformed and regu­lar­ized as soon as possible.

II.        Congres­sional Over­sight is Needed to Ensure That the EAC Does Not Make Recom­mend­a­tions That Could Risk Disen­fran­chising Eligible Voters This Year

The Help Amer­ica Vote Act of 2002 (“HAVA”) grants the EAC limited author­ity over voter regis­tra­tion in three signi­fic­ant respects.  First, HAVA charges the EAC with adopt­ing volun­tary guid­ance to assist states in meet­ing the require­ments of the stat­ute’s provi­sions regard­ing statewide voter regis­tra­tion data­bases, among other things.[6]  The EAC exer­cised this author­ity in July 2005, provid­ing the states with some useful non-bind­ing guid­ance in imple­ment­ing the new compu­ter­ized voter regis­tra­tion data­bases mandated by HAVA.[7]  Second, HAVA confers upon the EAC the author­ity previ­ously held by the Federal Elec­tion Commis­sion under the National Voter Regis­tra­tion Act of 1993 (the “NVRA” or the “Motor Voter” Act), includ­ing limited rule-making author­ity over the federal mail-in voter regis­tra­tion form.[8]  The EAC has acted under this author­ity on two occa­sions,[9] but it has not revised or under­taken any rule-making actions concern­ing the federal mail-in voter regis­tra­tion form since 2006.  And third, HAVA charges the EAC with conduct­ing and making avail­able to the public stud­ies regard­ing elec­tion admin­is­tra­tion issues, includ­ing the issue of voter regis­tra­tion data­bases and main­tain­ing a secure and accur­ate voter list.[10]  Pursu­ant to this author­ity and possibly also its guid­ance author­ity, the EAC has recently sponsored a three-year study, begin­ning in 2007, on voter regis­tra­tion data­bases, with two prin­cipal goals: (i) to advise the states on how to better imple­ment their statewide voter regis­tra­tion data­bases, and (ii) to study and make recom­mend­a­tions concern­ing inter­op­er­ab­il­ity among differ­ent states’ voter regis­tra­tion data­bases.[11]  An interim report focus­ing on the second goal is expec­ted on April 1, 2008.[12]

For the reas­ons stated below, we believe that the EAC should not exer­cise its author­ity to make or recom­mend any further changes to the voter regis­tra­tion process this year.  Instead, the EAC should focus on assist­ing the states in improv­ing their voter regis­tra­tion data­bases.

A.        The EAC Should Not Recom­mend Changes to the Way in Which States Use Voter Regis­tra­tion Data­bases in 2008

Although the EAC can and should play an import­ant role in assist­ing the states to improve their statewide voter regis­tra­tion data­bases, we believe that the EAC should not recom­mend any changes that could affect how voters are registered and who is added to or removed from the voter rolls in 2008.  In partic­u­lar, the EAC should not adopt any recom­mend­a­tions concern­ing cross-state data match­ing this year.  We are concerned that the EAC may be prepar­ing to issue recom­mend­a­tions concern­ing the use of voter regis­tra­tion data­bases this year, and we hope that this Commit­tee will urge it to refrain from doing so. 

The nation’s exper­i­ence to date shows that changes in the way statewide voter regis­tra­tion data­bases are used often have unin­ten­ded consequences or glitches that could disen­fran­chise eligible voters.  There is simply not enough time in 2008 to prop­erly vet such changes, to identify and correct any prob­lems, and to make sure that there are adequate voter protec­tions in place.  In partic­u­lar, there is insuf­fi­cient time in 2008 to under­take any new program of cross-state data match­ing that could result in changes to voters’ regis­tra­tion status without risk­ing wide­spread prob­lems that would affect the Novem­ber elec­tions. 

The risk of unwar­ran­ted disen­fran­chise­ment is espe­cially high in connec­tion with efforts to elec­tron­ic­ally match records in a voter regis­tra­tion data­base against records in other govern­ment data­bases.  This risk arises primar­ily in two contexts: (1) when match­ing is used as a barrier to getting on the voter rolls, and (2) when match­ing is used to create lists with which to purge the voter rolls of persons who have become ineligible to vote or of duplic­ate entries.  In both contexts, data­base match­ing efforts have yiel­ded signi­fic­ant errors that threatened to disen­fran­chise eligible voters. 

First, data­base match­ing has created prob­lems when states have attemp­ted to match the inform­a­tion provided by applic­ants for voter regis­tra­tion against inform­a­tion main­tained in other govern­ment data­bases, includ­ing data­bases main­tained by motor vehicle author­it­ies and the Social Secur­ity Admin­is­tra­tion.  Although such match attempts can be useful tools when used for the purposes inten­ded by Congress in HAVA,[13] they are unre­li­able when used to determ­ine whether applic­ants are ineligible to vote.  Nonethe­less, over the past few years, a small but signi­fic­ant minor­ity of states have inap­pro­pri­ately used data­base match­ing as a barrier to getting on the voter rolls.[14]  These states have refused to place eligible citizens on the rolls unless elec­tion offi­cials could find a match between the voter’s regis­tra­tion inform­a­tion and inform­a­tion in other govern­ment data­bases.  Although HAVA does provide for such data­base match­ing efforts,[15] the stat­ute does not permit states to condi­tion access to the voter rolls on success­ful matches.  The only two federal courts to have considered the issue to date have held that “no match, no vote” proced­ures viol­ate HAVA and the Voting Rights Act.[16]

Regard­less of whether “no match, no vote” proced­ures are lawful, they risk block­ing large numbers of eligible voters from the voter rolls for no good reason.  As the Bren­nan Center has cata­logued,[17] the fail­ure to find a computer match between a voter regis­tra­tion record and another govern­ment record usually does not indic­ate any prob­lem with an applic­ant’s eligib­il­ity or her voter regis­tra­tion inform­a­tion.  Rather, match fail­ures are far more frequently the result of typos, the switch from maiden names to married names, and a vari­ety of other common data­base incon­sist­en­cies.  When New York City conduc­ted an audit of its 2005 effort to match the records in its voter data­base against those in its depart­ment of motor vehicles data­base, it found that almost all of the match fail­ures resul­ted from typos.  When states use match­ing as a condi­tion to getting on the voter rolls, a substan­tial number of eligible voters are kept off the rolls because of computer and human errors.

The number of eligible voters who could be affected by “no match, no vote” policies is alarm­ing.  In several juris­dic­tions, match fail­ure rates have been as high as 20–30%.  Accord­ing to a Febru­ary 2007 report by the Social Secur­ity Admin­is­tra­tion, 46.2%-almost half-of all voter regis­tra­tion records submit­ted to the Admin­is­tra­tion for veri­fic­a­tion failed to match its records.  Moreover, match fail­ures are more common for members of minor­ity groups, whose names may be less famil­iar to elec­tion offi­cials, as well as for women, who are more likely to have changed their surnames.  The Bren­nan Center found signi­fic­ant racial dispar­it­ies in match fail­ure rates in connec­tion with lawsuits chal­len­ging a “no match, no vote” rules in Flor­ida and Wash­ing­ton State.[18]

Second, data­base match­ing has also created prob­lems in the purge context.  Although new statewide voter regis­tra­tion data­bases, when fully developed, can facil­it­ate better main­ten­ance of the rolls, without adequate protec­tions for voters, purge prac­tices raise signi­fic­ant concerns.  Compu­ter­ized voter rolls enable states to purge voters with the push of a button.  Most states are now able to develop lists of voters to be purged from the rolls by elec­tron­ic­ally match­ing names on voter rolls against govern­ment data­bases of persons ineligible to vote.  Unfor­tu­nately, the match­ing processes used are inac­cur­ate and may result in many eligible voters being purged from the voter rolls.  Since states rarely provide effect­ive notice of a purge, voters whose names have been removed from the rolls usually do not learn of the prob­lem until they show up at the polls on Elec­tion Day and are denied a regu­lar ballot.  The secrecy of the process makes it easier for elec­tion offi­cials to manip­u­late purges to target certain groups of citizens.

The most notori­ous examples of flawed purges occurred in Flor­ida in 2000 and 2004.  In 2000, thou­sands of legal voters were purged from Flor­id­a’s voter rolls simply because their names shared 80% of the char­ac­ters of the names on a list of people with felony convic­tions.  In 2004, the Bren­nan Center uncovered evid­ence of another erro­neous purge list in Flor­ida, contain­ing 47,000 “suspec­ted felons.”  The flawed process used to gener­ate the list iden­ti­fied only 61 voters with Hispanic surnames, notwith­stand­ing Flor­id­a’s sizable Hispanic popu­la­tion.[19]  To compound the prob­lem, the purge list over-repres­en­ted African Amer­ic­ans[20] and mistakenly included thou­sands who had had their voting rights restored under Flor­ida law.  Although the Flor­ida flawed purge lists were widely publi­cized, similar errors across the coun­try typic­ally escape public scru­tiny.  Other purges that have come to public light have been found to ensnare many eligible voters.[21]

Because the data­base match­ing tech­niques used by many states in the voter regis­tra­tion context have been inac­cur­ate; because there are insuf­fi­cient safe­guards in place to guard against unwar­ran­ted disen­fran­chise­ment due to data­base and match­ing errors; and because there is insuf­fi­cient time to improve these processes and provide the neces­sary protec­tions for voters this year, the EAC should not recom­mend that states engage in addi­tional forms of data­base match­ing this year to main­tain their voter rolls.

B.        The EAC Should Not Make Recom­mend­a­tions For Inter-State Data­base Match­ing in 2008

In partic­u­lar, the EAC should not recom­mend that states under­take new efforts to match voter regis­tra­tion data across state lines to identify voters who have allegedly moved or who are other­wise ineligible to vote in a juris­dic­tion.[22]  The prob­lems asso­ci­ated with data­base match­ing voters are compoun­ded as states attempt to increase the inter­op­er­ab­il­ity of their systems for the purposes of prun­ing their voter rolls based on inform­a­tion gleaned from other states.  In several instances in which purges based on cross-state data match­ing have come to public light,[23] the purge attempts were found to have been based on erro­neous assump­tions, such as the assump­tion that two records show­ing the same name and date of birth refer to the same person.  (To the contrary, in a suffi­ciently large group such as a state, two differ­ent indi­vidu­als will share the same name and date of birth surpris­ingly often.[24])

Given the novelty of the prac­tice, the current tech­no­lo­gical limit­a­tions of state voter regis­tra­tion data­bases, and the lack of suffi­cient voter protec­tions in place, any effort to promote purges based on cross-state data­base match­ing presents too great a risk of voter disen­fran­chise­ment this year.  Indeed, the EAC should not make any recom­mend­a­tions on data­base match­ing without ample oppor­tun­ity for study and public comment and a demon­stra­tion that the recom­mend­a­tions include adequate protec­tions to prevent eligible voters from being inap­pro­pri­ately disen­fran­chised.  (At the very least, HAVA requires the agency to provide notice of any such recom­mend­a­tions and to allow ninety days for public comment.[25])

C.        The EAC Should Assist the States to Improve Their Voter Regis­tra­tion Data­bases

The risk of voter disen­fran­chise­ment as a result of purges based on data­base match­ing programs is unlikely to be abated until the states improve the qual­ity of their voter regis­tra­tion data­bases.  Accord­ingly, the Bren­nan Center recom­mends that EAC focus its efforts this year on help­ing states improve their voter regis­tra­tion data­bases, many of which do not func­tion adequately, instead of on making recom­mend­a­tions concern­ing data­base inter­op­er­ab­il­ity.

D.        The EAC Should Not Recom­mend Changes to the Federal Mail-In Voter Regis­tra­tion Form in 2008 Other Than Tech­nical Correc­tions

Congress provided for a federal mail-in voter regis­tra­tion form in the NVRA to guard against a range of state barri­ers to voter regis­tra­tion and to ensure that all eligible Amer­ic­ans can register to vote using a single form.  Most changes to this federal form require a formal vote by the EAC Commis­sion­ers after a formal rule-making process.  At this point, there is insuf­fi­cient time in 2008 to imple­ment any changes to the federal mail-in voter regis­tra­tion form, other than tech­nical correc­tions like address updates, without risk­ing confu­sion and error that could prevent eligible voters from being registered.  The EAC should, however, imme­di­ately update the addresses of the state elec­tion offices on the instruc­tions to the federal form to ensure that eligible regis­trants are not disen­fran­chised due to lost voter regis­tra­tion forms sent to old addresses.

III.             Congres­sional Over­sight Is Needed to Ensure That the EAC Adequately Fulfills Its Neces­sary Func­tions, Espe­cially Relat­ing to Test­ing and Certi­fic­a­tion of Voting Machines

HAVA has placed the EAC in charge of many crucial federal elec­tion admin­is­tra­tion tasks.  Among its many respons­ib­il­it­ies, it is charged with acting as the lead federal agency for accred­it­ing voting system test­ing labor­at­or­ies and certi­fy­ing voting systems, and alloc­at­ing elec­tion-related federal fund­ing to the states.[26]  Both of these func­tions are essen­tial.  The EAC’s perform­ance with respect to the latter can and should be improved.

For the last several years, the test­ing and certi­fic­a­tion process conduc­ted by the EAC for voting machines has been flawed in at least three ways: (1) vendors choose and pay the labor­at­or­ies that eval­u­ate their systems, (2) the process lacks trans­par­ency, and (3) the formal report­ing process is too limited.

Vendors Choose and Pay the Labor­at­or­ies that Revalu­ate their Systems.  Under the current process, vendors choose and pay for the labor­at­or­ies that eval­u­ate their systems.[27]  This process creates an appear­ance of conflict of interest for the test­ing labs.  Worse still, it creates perverse incent­ives for the test­ing labor­at­or­ies to certify machines to ensure that vendors choose them in the future.  The test­ing labor­at­or­ies them­selves have done little to build public confid­ence in their inde­pend­ence from voting machine vendors.  In a fairly well-publi­cized writ­ten submis­sion to the EAC, a test­ing labor­at­ory recently stated that it “view[ed] the rela­tion­ship between an inde­pend­ent test­ing labor­at­ory and it’s [sic] clients as similar to that between lawyer and client or between doctor and client.”[28]

Accord­ingly, the Bren­nan Center recom­mends taking action to ensure that voting system test­ing labor­at­or­ies are and appear to be inde­pend­ent of vendors.  First, we should end the process whereby the voting system test­ing labor­at­or­ies are chosen and directly paid by the vendors whose machines they eval­u­ate.  Second, the peri­odic eval­u­ations of test­ing labor­at­or­ies conduc­ted by the National Volun­tary Labor­at­ory Accred­it­a­tion Program should be made public promptly, regard­less of whether the labor­at­or­y’s accred­it­a­tion is gran­ted, denied or revoked.

Process Lacks Trans­par­ency.  The EAC’s fail­ure to timely publish a damning Assess­ment Report of CIBER, Inc. after it was completed in July 2006 provides a text­book case of how a lack of trans­par­ency can severely shake public confid­ence.  The report concluded, among other things, that CIBER had not shown the resources to provide a reli­able product.[29]  As a result of the Assess­ment Report, the EAC determ­ined it could not accredit CIBER under the interim accred­it­a­tion process.[30]  However, it did not publi­cize this decision, release the Assess­ment Report, or notify the State of New York, which was using CIBER to test its voting systems at the time.  Only after the New York Times repor­ted that CIBER had been barred from certi­fy­ing elec­tion equip­ment and weeks of public pres­sure follow­ing that news article, did the EAC finally release the Assess­ment Report and other docu­ments related to its decision.[31]

New York’s exper­i­ence with CIBER is also an excel­lent illus­tra­tion of the import­ance of trans­par­ency in the voting machine certi­fic­a­tion process, and in partic­u­lar the need to ensure that all test plans, Tech­nical Data Pack­ages and test reports are made public.  Concur­rent with its hiring of CIBER to conduct its certi­fic­a­tion test­ing, New York also hired NYSTEC, a private, not-for-profit engin­eer­ing company to conduct an inde­pend­ent review of CIBER’s test plan.  NYSTEC’s review showed that the test plan lacked several secur­ity and func­tional test­ing require­ments under state law and the federal Volun­tary Voting System Guidelines of 2005 (to which CIBER had agreed to test). Among the items miss­ing from the test plan were: a require­ment that voting systems did not include any device poten­tially capable of extern­ally trans­mit­ting or receiv­ing data via the inter­net, radio waves or other wire­less means; a require­ment that voting system soft­ware not contain any viruses or other devices that could cause the system to cease func­tion­ing prop­erly at a future time; a require­ment for voting systems to provide a means by which the ballot defin­i­tion code could be posit­ively veri­fied to ensure that it corres­pon­ded to the format of the ballot face and elec­tion config­ur­a­tion; and test meth­ods or proced­ures for the major­ity of the state’s voting system require­ments.[32]  These prob­lems were only discovered because CIBER’s test plans were subject to inde­pend­ent scru­tiny.

Short of mandat­ing that juris­dic­tions hire inde­pend­ent review­ers for all certi­fic­a­tions of voting machines, it is imper­at­ive that the EAC publish docu­ments neces­sary for the public to ascer­tain the value of a test­ing labor­at­or­y’s certi­fic­a­tion.  This means not only publish­ing all test­ing labor­at­ory test plans for a partic­u­lar machine, but also the tech­nical data pack­ages submit­ted by the vendor to the test­ing labor­at­ory, and the labor­at­or­y’s reports that assess the machines.  The EAC’s Voting System Test­ing and Certi­fic­a­tion Program Manual now requires the public­a­tion of test­ing labor­at­ory reports and test plans. It does not, however require the public­a­tion of all Tech­nical Data Pack­ages provided by the vendors for the reports; this omis­sion will make it more diffi­cult for the public and inde­pend­ent experts to judge the conclu­sions made in the labor­at­ory reports.[33]  This is a glar­ing gap in the EAC’s report­ing require­ments and should be changed.

Accord­ingly, the Bren­nan Center recom­mends that the EAC make the voting machine certi­fic­a­tion process more trans­par­ent.  If the public is to regain its trust in this process, it is crit­ical that the EAC publish: (1) all test plans submit­ted by the test­ing labor­at­or­ies; (2) the vendor’s Tech­nical Data Pack­ages, which the vendor submits to the EAC to provide the specif­ics of a voting system; as well as (3) the test report that a test­ing labor­at­ory submits to the EAC after it has tested that voting system.[34]

Formal Report­ing Process is Too Limited.  Under the new Voting System Test­ing and Certi­fic­a­tion Program Manual, the EAC will accept reports from “[s]tate or local elec­tion offi­cials who have exper­i­enced voting system anom­alies in their juris­dic­tion.”[35] This is an import­ant step.  Unfor­tu­nately, indi­vidual voters and tech­nical experts perform­ing usab­il­ity, access­ib­il­ity and secur­ity tests on voting machines appear to be excluded from filing such reports with the EAC.[36]  This is prob­lem­atic for two reas­ons.  First, the EAC has no method in place to protect the anonym­ity of elec­tion offi­cials filing reports.  Many elec­tion integ­rity and secur­ity experts have argued that an elec­tion offi­cial “might be reluct­ant to report an irreg­u­lar­ity in a system he was respons­ible for admin­is­ter­ing,” both because he may have also been respons­ible for purchas­ing that system and because he would prob­ably need to continue to rely on tech­nical assist­ance from the vendor.[37]  Second, voters and tech­nical experts using these machines would be an excel­lent source of inform­a­tion about prob­lems with these machines; in many instances, they will be in a far better posi­tion than elec­tion offi­cials to know how the machines actu­ally perform when used.

Accord­ingly, the Bren­nan Center recom­mends that the report­ing process be opened to include report­ing inform­a­tion from voters and tech­nical experts who have used the voting machines to amend voting system stand­ards, where neces­sary.[38]

IV.       The EAC Should Provide Assist­ance to State and Local Elec­tion Offi­cials to Avoid Basic Ballot Design Mistakes

The EAC can make a signi­fic­ant differ­ence this year if it provides assist­ance to state and local elec­tion offi­cials to prevent basic ballot design mistakes.  The EAC recently produced a 266-page report on effect­ive ballot design.  It has been repeatedly demon­strated that ballot design is very import­ant to smooth elec­tion admin­is­tra­tion.  A poorly designed ballot in Sara­sota County is thought to be respons­ible for 13% of voters fail­ing to cast their vote for the House of Repres­ent­at­ives, even though they voted for the Senate.[39]  The “butter­fly ballot” in Palm Beach was blamed for many of the 19,000 over-votes in that county which had to be discarded.[40]  Even more recently, in the 2008 Cali­for­nia Pres­id­en­tial Primary, as many as 49,500 votes may have been lost in Los Angeles County because voters mis-marked their ballots as a result of a confus­ing ballot design.[41]

Notwith­stand­ing the import­ance of the inform­a­tion in the report, the dense report on ballot design produced by the EAC may not, stand­ing alone, be suffi­cient to commu­nic­ate the inform­a­tion in the report to relev­ant decision-makers and actors.  The EAC has exper­i­ence using multiple and compli­ment­ary meth­ods of public educa­tion.  For example, in July of 2007, the EAC hosted a series of brief­ings with respect to improv­ing the qual­ity and recruit­ment of poll work­ers and also published a manual on poll worker recruit­ment, train­ing, and reten­tion.  By way of another example, the EAC published a gloss­ary of common elec­tion terms in Span­ish and provided a mech­an­ism for trans­lat­ing its website.

The EAC should simil­arly under­take further efforts to ensure that ballot design mistakes are avoided-such as provid­ing train­ings and work­shops to those offi­cials respons­ible for design­ing ballots, provid­ing an easy to use “check­list” for ballot design, and publish­ing best prac­tices.  There is still enough time to dissem­in­ate this inform­a­tion this year to ensure that it helps avoid ballot design errors this Novem­ber.

At this time, however, the EAC’s oper­at­ing budget and staff­ing levels are insuf­fi­cient for its many tasks.  Accord­ingly, we urge Congress to provide the EAC with adequate resources to under­take this import­ant public educa­tion effort.

V.        Congress Should Provide Adequate Resources to the EAC and Fully Fund HAVA

As noted above, HAVA has placed the EAC in charge of many of the most import­ant federal elec­tion admin­is­tra­tion tasks.  In addi­tion to its respons­ib­il­it­ies over the accred­it­a­tion of the Voting System Test­ing Labor­at­or­ies and certi­fic­a­tion of voting systems, the EAC is also charged with distrib­ut­ing HAVA fund­ing to the states, acting as a clear­ing­house of inform­a­tion on the exper­i­ences of State and local govern­ments in imple­ment­ing the guidelines and in oper­at­ing voting systems in general, conduct­ing stud­ies and carry­ing out other activ­it­ies to promote the effect­ive elec­tion admin­is­tra­tion of Federal elec­tions, and carry­ing out admin­is­trat­ive duties under the NVRA, includ­ing devel­op­ing and main­tain­ing a mail voter regis­tra­tion applic­a­tion form for elec­tions for federal office.

In contrast to its signi­fic­ant respons­ib­il­it­ies, the EAC receives very little support.  In 2006, it had an oper­at­ing budget of just $15 million and employed less than 30 people.  Even the best oper­at­ing proced­ures and prac­tices will have little effect if the EAC does not have the resources and staff to carry them out.  Accord­ingly, the Bren­nan Center recom­mends that Congress provide the EAC with suffi­cient resources to carry out its respons­ib­il­it­ies effect­ively.

In addi­tion, the Bren­nan Center recom­mends that Congress fully fund HAVA. The scope and speed with which the nation has trans­formed the way it registers, votes, and counts votes is unpre­ced­en­ted.  Such a trans­form­a­tion requires adequate fund­ing to be success­ful.  The fund­ing thus far has proven insuf­fi­cient, and the chal­lenges facing elec­tion admin­is­trat­ors are not abat­ing.  Full fund­ing is neces­sary to ensure that states prop­erly imple­ment statewide voter regis­tra­tion data­bases, keep up with the spiral­ing costs of purchas­ing and main­tain­ing voting equip­ment, and ensure proper poll worker train­ing and voter educa­tion during a period of signi­fic­ant change.  Suffi­cient resources are needed to fulfill HAVA’s prom­ise.

Thank you very much.


[1] Another illus­tra­tion, discussed in Part III, infra, is the EAC’s fail­ure to timely publish a report criti­ciz­ing CIBER, Inc., a voting system test­ing labor­at­ory.

[2] See 42 U.S.C. § 15381.

[3] The Bren­nan Center reques­ted both reports in Octo­ber 2006 under the Free­dom of Inform­a­tion Act (FOIA), but the EAC denied our request and our subsequent appeal, arguing that the reports were priv­ileged.  See http://bren­ (appeal letter).  All docu­ments relat­ing to this dispute are avail­able at http://www.bren­nan­cen­­la­tion_testi­mony/category/eac_over­sight.

[4] 42 U.S.C. § 15322.

[5] The Bren­nan Center’s recom­mend­a­tions are also avail­able here: http://www.bren­nan­cen­­nan_center_calls_for_greater_trans­par­ency_and_account­ab­il­ity_of_us_elec/

[6] 42 U.S.C. § 15501(a).

[7] See­tion/docs/statewide_regis­tra­tion_guidelines_072605.pdf/attach­ment_down­load/file.  The Bren­nan Center provided extens­ive comment­ary on this guid­ance.  See Testi­mony of Wendy R. Weiser before the U.S. Elec­tion Assist­ance Commis­sion Concern­ing Statewide Voter Regis­tra­tion Data­bases, April 26, 2005, avail­able at http://www.bren­nan­cen­­load_file_9238.pdf; Comments by Voting Rights, Civil Rights, and Civic Organ­iz­a­tions on the EAC’s Draft Volun­tary Guid­ance on the Imple­ment­a­tion of Voter Regis­tra­tion Lists, May 25, 2005, avail­able at http://www.bren­nan­cen­­load_file_9233.pdf.  The Bren­nan Center’s chief criti­cism of the guid­ance the EAC issued is that it did not go far enough to ensure that states protect voters’ rights as they use their compu­ter­ized voter lists.  Specific­ally, the guid­ance does not address the prob­lem, discussed infra, of certain states refus­ing to register new voters, contrary to HAVA, solely because elec­tion offi­cials cannot find a compu­ter­ized match for their voter regis­tra­tion inform­a­tion.

[8] See 42 U.S.C. § 1973gg-7(a).

[9]  On both occa­sions, the EAC refused requests-by Flor­ida in July 2005 and Arizona in March 2006-to add to the condi­tions set by Congress for accept­ing the federal mail-in voter regis­tra­tion form.  See Ray Martinez III, Posi­tion State­ment, July 10, 2006, avail­able at­pos­i­tion­state­mentaz.pdf/attach­ment_down­load/file (discuss­ing both actions).

[10] 42 U.S.C. § 15381(a) & (b)(3).

[11] See Testi­mony of Dr. Herbert Lin, Chief Scient­ist, Computer Science and Tele­com­mu­nic­a­tions Board, National Research Coun­cil, before the Elec­tion Assist­ance Commis­sion, Novem­ber 13, 2007, avail­able at­ings-and-hear­ings-11–13–07-meet­ing-herb-lin.pdf/attach­ment_down­load/file (testi­mony by study director). 

[12] Id.

[13] Under HAVA, states must attempt to match the inform­a­tion supplied by applic­ants for voter regis­tra­tion against inform­a­tion in the data­bases main­tained by motor vehicle depart­ments and the Social Secur­ity Admin­is­tra­tion. 42 U.S.C. § 15483(a)(5)(B).  The purpose of this match­ing proced­ure is to enable states to asso­ci­ate a unique identi­fy­ing number-a driver’s license number, a partial Social Secur­ity number, or a newly assigned number-with each eligible voter, to keep better track of voters moving within the state, and to ensure that a John Smith in one county would not be confused with a John Smith else­where.  Thanks to an amend­ment offered by Senator Wyden, this match­ing proced­ure serves a second purpose as well-to exempt first-time voters who register by mail from HAVA’s iden­ti­fic­a­tion require­ments if state offi­cials are able to “match” the inform­a­tion on their regis­tra­tion records with other state records.  42 U.S.C. § 15483(b)(3).

[14] See Justin Levitt, Wendy Weiser & Ana Munoz, Making the List: Data­base Match­ing and Veri­fic­a­tion Processes for Voter Regis­tra­tion, March 24, 2006, avail­able at http://www.bren­nan­cen­­base_match­ing_and_veri­fic­a­tion_processes_for_voter_regi/ (cata­loguing state data­base prac­tices based on research in late 2005 and early 2006).  Currently, we are aware of only three states that still refuse to register voters based on the results of unsuc­cess­ful data­base match­ing efforts, though it is diffi­cult to be sure since states rarely codify such prac­tices.

[15] See supra, note 8.

[16] See Wash­ing­ton Ass’n of Churches v. Reed, 492 F. Supp. 2d 1264 (W.D. Wa. 2006); Flor­ida State Confer­ence of the NAACP Braches v. Brown­ing, No. 07-CV-402-SPM/WCS (N.D. Fla. Dec. 18, 2007) (order grant­ing motion for prelim­in­ary injunc­tion), avail­able at http://www.bren­nan­cen­­cracy/Flor­id­aNAACPvBrown­ing%20-%2012–18–07%20Order%20Grant­ing%20PI.pdf.  While the Wash­ing­ton decision has been made final, the Flor­ida decision is currently on appeal.

[17] See, e.g., Making the List, supra, note 14.

[18] In Flor­ida, we found that Hispanic Amer­ic­ans, who comprised 15% of the voter applic­ant popu­la­tion, consti­tuted 34% of those on the failed match list; simil­arly, African-Amer­ic­ans, who comprised 13% of the voter applic­ant pool, consti­tuted 22% of those on the failed match list.  In contrast, whites, who comprised 66% of the applic­ant pool, accoun­ted for only 29% of the un-matched applic­a­tions. See Plaintiffs’ Supple­mental Evid­en­tiary Submis­sion in Support of Motion for Prelim­in­ary Injunc­tion, Flor­ida State Confer­ence of NAACP Branches v. Brown­ing, No. 07 CV 402 (N.D. Fla. Nov. 14, 2007).  In Wash­ing­ton State, we found that Asian Amer­ic­ans were vastly over-repres­en­ted in the un-matched popu­la­tion.

[19] Specific­ally, 0.1% of the people on the list were Hispanic, even though in Flor­ida, 12% of the disen­fran­chised popu­la­tion, or one in eight citizens, is Hispanic.  The contractor that compiled that list did so by match­ing names on the voter list against records main­tained by the state depart­ment of correc­tions. For a match to be found, the contractor required matches in a vari­ety of fields, includ­ing a field for race.  The prob­lem was that one data­base had a category for Hispan­ics and the other did not.  The result was a list that system­at­ic­ally excluded Hispan­ics.

[20] African-Amer­ic­ans comprised 46% of the purge list but only 30% of the disen­fran­chised popu­la­tion.

[21] For example, the Bren­nan Center worked with polit­ical science professor Michael McDon­ald to analyze a New Jersey purge list prepared by a polit­ical party using “match­ing” tech­niques.  We found that the list was compiled using a number of faulty assump­tions and thus that it would have harmed eligible voters if used as the basis for a purge.  See http://bren­­b­d­db2c45834e0_pom6bx3bk.pdf.

[22] It bears noting that noth­ing in HAVA or any other federal stat­ute requires states to attempt to compare their voter regis­tra­tion data­base records with those of other states.

[23] For example, in 2006, the Secret­ary of State of Kentucky attemp­ted to purge the state’s rolls based on a flawed attempt to identify voters who had moved from Kentucky to neigh­bor­ing South Caro­lina and Tennessee.  The state Attor­ney General learned of the purge and brought a success­ful lawsuit to reverse it on the grounds that the Secret­ary of State did not follow the voter protec­tion proced­ures outlined in the state law analog to the NVRA.  The lawsuit uncovered the fact that eligible voters who had not, in fact, moved out of the state of Kentucky were caught up in the purge.  (Neither South Caro­lina nor Tennessee relied on the results of the flawed match exer­cise.)

[24] Stat­ist­ics teaches that in a group of 23 people, it is more likely than not that two will share the same birth­day; in a group of 180, two will prob­ably share the same birth date, includ­ing day, month, and year.  Partic­u­larly for voters with common names in a pool as large as a state, there will be several differ­ent indi­vidu­als who share the same basic inform­a­tion.  See Michael P. McDon­ald & Justin Levitt, Seeing Double Voting: An Exten­sion of the Birth­day Prob­lem, 11 (July 1, 2007) (unpub­lished manu­script, submit­ted to the 2007 Confer­ence on Empir­ical Legal Stud­ies), avail­able at­ery.cfm/SSRN_ID997888_code698321.pdf?abstractid=997888&mirid=1.

[25] 42 U.S.C. § 15501.

[26] 42 U.S.C. § 15322 (2003).

[27] See ACCUR­ATE, Public Comment on the Manual for Voting System Test­ing & Certi­fic­a­tion Program Submit­ted to the United States Elec­tion Assist­ance Commis­sion (Oct. 31, 2006), joined by the Bren­nan Center, avail­able at http://accur­­ATE_VSTCP_comment.pdf (here­in­after “ACCUR­ATE Comment on VSTCP”); U.S. Elec­tion Assist­ance Commis­sion Public Meet­ing and Hear­ing, Pasadena, CA (July 28, 2005) (Testi­mony of David L. Dill, Professor of Computer Science, Stan­ford Univer­sity and Founder of Veri­fied Voting Found­a­tion and Veri­fied­Vot­ avail­able at (here­in­after “Testi­mony of David Dill”).; Voting Machines: Will the New Stand­ards and Guidelines Help Prevent Future Prob­lems?: Joint Hear­ing Before the H. Comm. on H. Admin. and the Comm. on Science, 109th Cong. 66–71 (2006) (Writ­ten State­ment of David Wagner, Professor of Computer Science, Univer­sity of Cali­for­nia-Berke­ley), avail­able at http://www.votet­rust­ (here­in­after “Testi­mony of David Wagner”).

[28] U.S. Elec­tion Assist­ance Commis­sion Public Meet­ing, Wash­ing­ton, D.C. (Oct. 26, 2006) (Writ­ten State­ment of Frank Padilla, Test Super­visor, Wyle Labor­at­or­ies, Inc.) avail­able at­tems%20Brief­ing%20-%20Frank%20Pa­dilla%2010–18–06%20Fi­nal.pdf.

[29] U.S. Elec­tion Assist­ance Commis­sion, Assess­ment Report: CIBER & Wyle (conduc­ted July 17–22, 2006) avail­able at­ment%20(July%202006).pdf.

[30] Chris­topher Drew, Citing Prob­lems, U.S. Bars Lab From Test­ing Elec­tronic Voting, N.Y. TIMES (Jan. 4, 2007) avail­able at­ted/article?res=F50811F63C540C778CD­DA80894D­F404482.

[31] These docu­ments are avail­able at:

[32] Howard Stan­islevic, Voting System Certi­fic­a­tion: Who’s Mind­ing the Store?, VoteT­rust­USA (Jan. 9, 2007) avail­able at http://votet­rust­

[33] Aaron Burstein & Joseph Lorenzo Hall, Unlike Ballots, EAC Should­n’t Be Secret­ive, Roll Call (Jan. 22, 2007) avail­able at–1.html.

[34]ACCUR­ATE Comment on VSTCP, supra note 27.

[35] U.S. Elec­tion Assist­ance Commis­sion, Test­ing and Certi­fic­a­tion Program Manual section 8.7.2 (draft, Sept. 28, 2006) avail­able at­tem%20Test­ing%20and%20Cer­ti­fic­a­tion%20Pro­gram%20Manual%20FR%20DRAFT%20(Sept%2028).pdf.

[36] ACCUR­ATE Comment on VSTCP, supra note 27, at 8

[37] Id. at 9.

[38] Id. at 8.

[39] See Edit­or­ial, Fixes for Accur­ate Vote Counts, Chris­tian Science Monitor, Feb. 27, 2007, at 8.

[40] See Jeff Kunerth, Scott Maxwell, & Maya Bell, Vote Never Had Chance; Ballots and Laws Were Confus­ing, Poll Work­ers Weren’t Well-Trained and Voters Were Care­less, Orlando Sentinel, Dec. 17, 2000, at A1.

[41] Richard C. Paddock, Some Inde­pend­ent Votes Won’t Count in L.A. County, L.A. Times, Feb. 12, 2008, at A1.