Acorn v. Bysiewicz (413 F. Supp. 2d 119, 2005)

MEMORANDUM OF DECISION

KRAVITZ, District Judge.

In this action, Plaintiffs — two individual voters, a political party, and various public interest groups — claim that Connecticut’s election law violates their First and Fourteenth Amendment rights to equal protection, to vote, to associate, and to engage in political speech, because Connecticut requires voters who wish to cast a ballot for non-presidential offices to register to vote in advance of election day. Importantly, Plaintiffs do not assert in this case that Connecticut’s pre-election-day registration requirement invidiously discriminates against any group. Nor do Plaintiffs claim that requiring voters to register to vote is itself unconstitutional. Instead, Plaintiffs argue that requiring Connecticut’s voters to register any time before election day is unconstitutional. According to Plaintiffs, the United States Constitution guarantees the citizens of Connecticut the right to register to vote on the same day as they cast their ballots — a concept known among political scientists and election officials as “election-day registration.”

Connecticut is not alone in requiring voters to register to vote in advance of election day; forty-two other states require pre-election-day registration. However, Connecticut’s registration requirement is unique in one important respect. Connecticut currently allows voters to register closer to election day than any other state that requires pre-election-day registration — up to seven days before a general election and one day before a primary. Most other states requiring pre-election-day registration require their citizens to register to vote three to four weeks before a general election. Therefore, if Connecticut’s shortest-in-the-nation pre-election-day registration requirement cannot pass constitutional muster, it is doubtful that any state’s can.

This Court does not write on a clean slate regarding state regulation of the time, place, and manner of exercising the right to vote. In fact, the Supreme Court has explicitly and repeatedly upheld state regulations requiring voters to register to vote in advance of general and primary elections, and has endorsed laws imposing much lengthier deadlines than Connecticut’s. Moreover, contrary to Plaintiffs’ claim in this case, the Supreme Court has expressly held that “a person does not have a federal constitutional right to walk up to a voting place on election day and demand a ballot.” Marston v. Lewis, 410 U.S. 679, 680, 93 S.Ct. 1211, 35 L.Ed.2d 627 (1973). Nevertheless, Plaintiffs insist that this Court can ignore these Supreme Court holdings because (they claim) the record in this case is much more developed than in the cases the Supreme Court considered, and the evidence shows that Connecticut’s pre-election registration requirement is unjustified and unduly burdensome.

Plaintiffs are wrong for two fundamental reasons. First, this Court cannot ignore governing Supreme Court precedent in the manner suggested by Plaintiffs. The Supreme Court’s decisions (and there are several) are directly on point, their rulings are crystal clear, and their holdings remain good law, some having been cited with approval by the Supreme Court as recently as last Term.

Second, even if the Court could wish away these decisions, Plaintiffs have not satisfied the Supreme Court’s standards for evaluating the constitutionality of state restrictions on voting. In recent years, the Supreme Court has clarified that such restrictions are subject to differing levels of scrutiny depending upon the severity of the burden imposed. Restrictions on voting that invidiously discriminate or constitute a severe burden on First and Fourteenth Amendment rights are subject to strict scrutiny. However, when a restriction on voting is reasonable and nondiscriminatory — and that is the case here — a state’s important regulatory interests ordinarily are sufficient to justify the requirement.

Plaintiffs do not claim that Connecticut’s seven-day registration deadline invidiously discriminates against any group, and the Court is convinced that it is not a severe burden on voting. It is the shortest pre-election-day registration deadline in the nation, and it is sufficiently easy to comply with that over two million voters in Connecticut — approximately 75% of the voting-age population of the State — successfully registered to vote in 2000 and 2004. Connecticut law permits citizens to register by mail or in person. Mail-in applications are readily available. In fact, state law requires the Commissioner of Motor Vehicles to include an application for voter registration with each application for an operator’s license or renewal. Those voters wishing to register in person can do so not only at registrar of voters’ offices in each of Connecticut’s 169 towns but also at myriad other locations including libraries, motor vehicle offices, state universities, and state assistance offices. Citizens can also register to vote during the many privately run voter registration drives held in numerous locations throughout the State in the weeks leading up to each election. The registration process itself takes about one minute of a citizen’s time and, once registered, a Connecticut citizen can vote in all subsequent elections without re-registering unless the citizen moves to a different town. Importantly, to register, Connecticut citizens need not decide for whom to vote, or even whether to vote on election day; they need only decide, a mere seven days in advance of a general election, that they wish to preserve the option of voting a week later.

Although it may be true that registering in advance is not as convenient as registering on election day, requiring citizens to take one or two minutes of their time to register to vote seven days before a general election cannot reasonably be characterized as a severe burden on the right to vote. Over thirty years ago,. Justice Stewart, writing for the Supreme Court, upheld the constitutionality of a registration requirement that was far more onerous than Connecticut’s, and his words in that case aptly sum up this case as well. By requiring voters to register in advance of election day, Connecticut “d[oes] not prohibit the petitioners from voting ... or from associating with the political party of their choice. It merely impose[s] a legitimate time limitation on their enrollment, which [unregistered voters] cho[o]se to disregard.” Rosario v. Rockefeller, 410 U.S. 752, 762, 93 S.Ct. 1245, 36 L.Ed.2d 1 (1973).

Thus, because Connecticut’s registration deadline is neither discriminatory nor severe, to withstand constitutional challenge it need only serve important regulatory interests. That standard was more than met here. As the evidence in this case showed, Connecticut’s pre-election registration requirement is amply justified by several important and legitimate state interests, including the State’s interest in minimizing voter fraud (as well as the perception of a vulnerability to fraud) and in avoiding confusion and chaos on election day itself.

It bears particular emphasis that the representatives of the people of Connecticut have not lightly rejected the concept of election-day registration. The General Assembly has long considered the question of registration deadlines and whether to adopt for Connecticut the election-day registration scheme that only six other states have endorsed. Over the years, the state legislature has held numerous and lengthy hearings on this topic, has heard from interested groups, citizens, and public officials about the pros and cons of election-day registration, and has debated its merits in committees and on the floor of the General Assembly. As a result of this deliberative process, the General Assembly has carefully but significantly reduced the length of time between registration and election day. Indeed, Connecticut’s current seven-day pre-election registration deadline for casting a vote in non-presidential general-election races is the most liberal pre-election-day registration deadline in the nation, and the eighth most liberal registration requirement overall, right behind the six states that allow election-day registration and the single state that imposes no voter registration requirement at all. Having carefully considered bills introducing election-day registration every year since 2001, and after consultation with the Secretary of the State, the Attorney General, the State Elections Enforcement Committee, the Registrar of Voters Association of Connecticut, and numerous others, the Connecticut General Assembly has apparently decided that, whatever benefits election-day registration might provide in terms of increasing voter turnout, those benefits are, at least for now, outweighed by the State’s interests in minimizing fraud, crowding, and confusion at the polls. That is a choice, the Court notes once again, that forty-two other state legislatures have also made.

Dissatisfied with the result of this legislative balancing of interests, proponents of election-day registration now turn to an unelected federal judge to achieve what they have been unable to obtain from state legislatures. Plaintiffs ask this Court to re-weigh the competing public interests — in Learned Hand’s famous words, to assume the role of a “third legislative chamber” — and impose election-day registration by fiat upon the citizens of Connecticut. But striking the appropriate balance between promoting smooth and accurate elections, on the one hand, and encouraging voter turnout, on the other, is “quintessentially a legislative judgment with which we judges should not interfere unless strongly convinced that the legislative judgment is grossly awry.” Griffin v. Roupas, 385 F.3d 1128, 1131 (7th Cir.2004). With all due respect to Plaintiffs— who have the best of intentions and have worked very hard to promote a concept about which they are clearly passionate— the judgment of Connecticut’s legislature with respect to election-day registration is not even remotely close to “grossly awry,” and “[t]he Constitution is not so rigid that that determination and others like it may not stand.” Marston, 410 U.S. at 681, 93 S.Ct. 1211.

I.

This case was tried to the Court over the course of several days, with an additional day for extended oral argument. The parties also provided the Court with pre-trial and post-trial briefing, including a lengthy stipulation of facts, proposed findings, and extensive pre-filed testimony. The Court was fortunate to have superb counsel working on this case. Every attorney who worked on this case, from the most experienced to the least, demonstrated time and again the very highest standards of skill, hard work, cooperation, candor, and professionalism. The Court is grateful to counsel and commends them for their superior legal work. It was truly a pleasure working with them on this important matter.

This Memorandum of Decision will serve as the Court’s findings of fact and conclusions of law in accordance with Rule 52 of the Federal Rules of Civil Procedure. Emblematic of the parties’ cooperation in this case, their Statement of Stipulated Facts [doc. #61, Ex. A] (“Stip”) is 67 pages long and consists of 416 separately numbered paragraphs. The Court has not repeated in this Memorandum each of those 416 stipulations of fact but instead has highlighted only those facts that the Court finds particularly noteworthy. The Court refers readers to the full Statement for the entire universe of stipulated facts and will assume familiarity with the facts stipulated by the parties.

The Parties

Plaintiffs in this action are two individuals (Garvin Roos and Susan Sorensen) one political party (the Working Families Party), and six public interest groups (ACORN, Connecticut Citizen Action Group, Connecticut Common Cause, Connecticut Public Interest Research Group, Connecticut Democracy Works, and People for the American Way). Ms. Sorensen and Mr. Roos are citizens who claim they were prevented by Connecticut’s registration deadline from voting for non-presidential offices in the 2004 election. Plaintiff Working Families Party (“WFP”) is a minor political party (as that phrase is defined in Connecticut’s election laws) that seeks to engage individuals disenchanted by the political status quo and therefore less likely to be frequent, and registered, voters. Voter registration is an integral part of WFP’s efforts to promote candidates , and the party’s agenda. Plaintiff public interest groups are various nonpartisan state and national membership organizations and lobbying groups that have a particular interest in the functioning of American democracy, the furtherance of civic participation, and increased voter registration. Connecticut ACORN registered over 5000 voters during the 2004 election cycle. Connecticut Public Interest Research Group is also active in voter registration, with a focus on Connecticut’s students.

Connecticut’s Current Legal Regime for Elections

A. Overvieiv

Like forty-nine other states, Connecticut requires its citizens to register before allowing them to vote in either primary or general elections. However, in accordance with federal legislation, Connecticut law provides an exception to the registration requirement for presidential elections: citizens may vote in presidential races only without first registering to vote. Conn. Gen.Stat. §§ 9-158a-c. Registration occurs at the town level and is overseen by officials known as registrars of voters, of whom each of Connecticut’s 169 towns have two (except Branford, which has four). Thus, although the Secretary of the State is the State’s chief election official, she is not empowered to x-egister voters. Town budgets, and not the State, pay for the wages of registrars of voters as well as poll workers hired for election day.

Any citizen of the State who will have attained the age of eighteen on or before the day of a regular election may apply to register to vote. Registration typically involves completing a simple one-page foxm that requests standard personal information including name, address, age, citizenship, and date of birth, providing some type of identification, and signing, under penalty of perjury, an affirmation that the registrant is eligible to vote. Registration forms, along with instructions on how to complete them and when and where to submit them, are available from multiple state institutions, and can also be obtained from the Secretary of the State’s website or from a registrar of voters office.

Completed registration cards may be returned by mail to a registrar’s office, or in person either to numerous state agencies, or to private groups holding voter registration drives. From the various locations to which voters may return their registration cards, the forms are forwarded to the registrar of voters for the applicant’s hometown. The local registrar of voters for the town where the applicant resides is tasked with ascertaining that the applicant is qualified to vote, and, if so, sending a notice of acceptance to the applicant. The registi’ar of voters also electronically records each qualified registrant’s information in a statewide electronic database that also generates town-by-town voting lists for local use.

Once properly registered, a citizen is eligible to vote in all subsequent elections without the need to re-register unless the citizen moves to another town. Registration is not completed, however, until the registrar of voters confirms that an application is timely and appropriately authenticated. Details about the timing and authentication requirements for proper registration, various exceptions to the usual registration requirements, the consequences of attempting to evade the registration requirements, and the implementation of the statewide electronic database are provided below.

B. Timing of Registration

In order to vote in a primary election, Connecticut voters may register to vote in person until noon on the day before the primary election. Conn. Gen-Stat. § 9-56. However, in order to vote for both presidential and noxx-presidential offices in a general election, voters in Connecticut generally must register to vote at least seven days before the election if registration is done in person, and at least fourteen days before the election if the registration is done by mail. The Court notes that at the time of filing and trial of this lawsuit, the in-person registration deadline was fourteen days before the general election. Conn. Gen.Stat. §§ 9-1, 9-17, 9-23g(d). However, on July 8, 2005, the Governor signed into law Public Act 05-235, Sections 26 and 27 of which extend the period for in-person registration until seven days before the election. See Public Act 05-235 §§ 26, 27, 2005 Conn. Pub. Acts 685-86. Because Public Act 05-235 was enacted while this case was still pending, this Court has assessed the constitutionality of Connecticut’s voter registration law by considering the requirements of Public Act 05-235, and not those of prior law.

Connecticut’s seven-day deadline for in-person registration is the seventh-most liberal voter registration deadline in the nation: only the six states that allow election-day registration (or “EDR”) have shorter deadlines. The relatively short voter registration deadline in Connecticut is the product of two legislative amendments to the election law in the last fifteen years, the Connecticut legislature having reduced the registration deadline from twenty-one days to fourteen days in 199!, and from fourteen days to seven days as recently as July 2005. See Public Act 91-352 § 24, 1991 Conn. Pub. Acts 984; Public Act 05-235 §§ 26, 27, 2005 Conn. Pub. Acts 685-86.

Although the seven-day deadline for in-person registration and the fourteen-day deadline for mail-in registration are the defaults, Connecticut law recognizes a variety of exceptions to these deadlines. Under these exceptions, registration of varying degrees of lateness is permitted for various categories of voters, including: 1. “Matured rights” voters (individuals who become 18 years of age, a U.S. citizen, or a resident of a municipality between the fourteenth day before a general election and the last business day before that election) who may register to vote in person until noon on the last week day before the election, Conn. Gen.Stat. § 9 — 17(b); and 2. Members of the armed forces, including those discharged within the year before an election, who may request in writing to be registered at any time before five p.m. on the last week day before any regular election, Conn. Gen.Stat. § 9-25. In addition, in accordance with federal law, Connecticut allows provisional ballots to be cast by individuals who come to the polls and declare that they are registered and eligible to vote but who cannot be found on or restored to the registration list, or cannot provide required identification. See 42 U.S.C. § 15482 (codifying part of the Help America Vote Act of 2002); Conn. Gen. Stat. § 92321.

For prospective voters not falling within the aforementioned exceptions, failure to register by the seventh day before a general election means that they will not be permitted in that election to vote for non-presidential offices, although untimely but otherwise satisfactory applications will be processed and will serve to register a voter for subsequent elections, see Stip. ¶ 69. However, as required by the Voting Rights Act of 1965, codified as amended at 42 U.S.C. §§ 1973 to 1973bb-1 (2005), Connecticut permits individuals who move to the State after the registration cutoff to vote without registration for president and vice-president through the device of a presidential ballot. See 42 U.S.C.1973aa-1(e); Conn. Gen.Stat. § 9-158b. In addition, Connecticut has extended the option of voting by presidential ballot to all unregistered voters, not only those who miss the registration deadline due to migration. Conn. GemStat. § 9-158b(a). Furthermore, in 1997, the Connecticut legislature decided to make presidential ballots available to unregistered voters on election day itself, see Public Act 97-154 § 6, 1997 Conn. Pub. Acts 347, a system first put into practice during the 2000 presidential election. Thus, in Connecticut, all unregistered voters can cast ballots on election day for president and vice-president. Conn. Gen.Stat. § 9-158c. Presidential ballots are counted in accordance with the rules on absentee ballots. Voting by presidential ballot does not serve to register the voter, although all those voting by presidential ballot are also offered the opportunity to register to vote, and if they do register, they will be permitted to vote for all races in the next election.

C. Authentication of Voter Identity

Identification requirements are an integral part of the State’s effort to prevent voter fraud, and are deployed both at registration and at the polls. However, not all voters are subject to the same identification requirements: the type of identification that a voter must provide varies with the method and time of registration.

Recently enacted federal legislation, the Help America Vote Act of 2002, Pub.L. No. 107-252, 116 Stat. 1666 (codified in scattered sections of 42 U.S.C.) (“HAVA”), imposes special identification requirements on individuals whose first voter registration is submitted by mail. See 42 U.S.C. § 15483(b)(l)-(2). In accordance with HAVA, Connecticut requires first-time mail-in registrants to provide, either with their registration application or at the polls, one of the following forms of identification: a current valid photo-identification; or a current utility bill, bank statement, government check, paycheck, or other government document showing the person’s name and address. Conn. Gen.Stat. § 9-23r. The same identification x-equirements imposed on first-time mail-in registrants also apply to unregistered voters seeking to cast a presidential ballot. Conn. Gen. Stat. § 9-158e.

Voters who register for the second or subsequent time by mail are not required to present any identification with the registration application. Conn. Gexx.Stat. §§ 9-23h, 9-23r. Voters who register in person must present a birth certificate, driver’s license or Social Security card, see id. § 9-20(a), but if that identification is not available, registration will be accepted if the voter proves her identity and eligibility by sworn testimony, or by other proof acceptable to the admitting official, see id. § 9-20(a). However, at the polls, all voters must present a Social Security card, or other form of preprinted identification (like a drivers’ license) displaying the voter’s name, address, and photograph or signature. See id. § 9-261(a)(2)(A). If the voter is unable to present that identification at the polls, she will still be allowed to vote provided that she submits a sworn declaration that she is the voter whose name appears on the official registration list. See id. § 9-261(a)(2)(B). At trial, Mr. Richard Abbate, the President of the Registrar of Voters Association of Connecticut (Registrars’ Association) testified that this provision of Connecticut law— allowing a voter to substitute a declaration of identity for actual identification — is a major concern for registrars with respect to voter fraud.

In addition to verifying identity by requiring voters to px’esent identification or sign sworn statements as to name, age, and eligibility, Connecticut registrars ax-e required to verify the address of a mail-in applicant by sending out an acceptance letter. If the letter is not returned as undeliverable, the presumptively successful delivery is taken as evidence of the voter’s identity. If the acceptance letter is returned undeliverable, the registrar will follow up by phone, and if that does not succeed, the registrar will send an official “confirmation of voter address” letter. If the confirmation of voter address letter is also returned undeliverable, the voter is designated inactive, but may apply for reinstatement even at the polls. See id. §§ 9-23g(c), 9-35. Although this verification process is only required for mail-in applicants, Mr. Abbate testified at trial that the same process is widely followed for in-person registrants, and that he considers it helpful in preventing voter fraud. The Executive Director of the State Elections Enforcement Commission (“Elections Commission”), Mr. Jeffrey Garfield, also testified that voter verification by mail was a useful anti-fraud tool.

State law provides two further methods for maintaining accurate voter lists. Section 9-32 of the Connecticut General Statutes mandates that registrars conduct an annual canvass, whether house-to-house or by mail or by telephone, in order to ascertain changes of address. Id. § 9-32. Finally, Connecticut law requires registrars to make a final list of the registered voters for each town available for public inspection at the town clerk’s office on the second Friday preceding a general election. Id. § 9-38. Candidates for office may also request a copy of the list, and according to the testimony at trial, most do so. In addition, Ms. Rae Tramontano, the Republican Registrar of Voters for New Haven, testified that her office prepares a preliminary version of the registration list in October and displays it in convenience stores, libraries, and any other public place that is willing to accept it. The publication of the eligible voter list enables members of the public to check their own entry on the list for accuracy, as well as to challenge the eligibility of any voter whose name appears on the list, either before election day or even at the polls. The list also allows candidates for office and political parties to focus their efforts to garner support on voters who, by registering, have indicated an intention to vote.

D. Civil and Criminal Penalties for Fraud

In addition to various strategies for verifying identification, Connecticut seeks to deter fraud with criminal penalties. Every voter registration form requires the voter to sign an affirmation that he or she is eligible to vote, and warns: “if you sign this statement even though you know it is untrue, you can be convicted and imprisoned for up to five years and fined up to $5,000.” Fraudulent registration of oneself or others is punishable by imprisonment for up to one year and a fine of up to $500. Id. § 9-357. The same penalty plus disenfranchisement is applicable to someone who falsely votes under the name of another registered voter. Id. § 9-360. Voting for an election in which one is not qualified, or voting more than once in the same election, is punishable by disenfranchisement, imprisonment for one to two years, and a fine of $300-$500. Id. § 9-360. Anyone who falsely testifies before a registrar or election moderator to any material fact concerning the right of any person to be registered or to vote will be disenfranchised and may be imprisoned for up to two years. Id. § 9-358. And intentionally making a false written statement on the application for a presidential or an absentee ballot is a Class D felony, punishable by imprisonment for up to 5 years, and a fine up to $5000. Id. §§ 9-359a, 53a-35, 53a-41(4), 53a-156. Although Public Act 05-235, which became law in July 2005, made some changes to the penalty provisions of the election law, it did not stiffen any of the penalties listed above. See 2005 Conn. Pub. Acts 674-75.

E. The Statemde Computer Database of Registered Voters

HAVA required each state to implement, by January 2004, an “official, centralized, interactive computerized statewide voter registration list defined, maintained, and administered at the State level that contains the name and registration information of every legally registered voter in the State and assigns a unique identifier to each legally registered voter in the State.” 42 U.S.C. § 15483(a)(1)(A). Further, the information in the system must be immediately accessible to every election official in the state, including local officials. Connecticut had a head start on this project, having contracted with IBM to build such a system as early as 1994, and was in full compliance with HAVA by January 2004, as of which date all 169 towns in Connecticut were connected to Connecticut’s statewide Centralized Voter Registration System (CVR database).

The CVR database is a browser-based system that stores all the information gleaned from voter registration forms in a single statewide database, in contrast to the previous system under which registrars maintained independent hard-copy voter lists at the local level. The functions of the CVR database include adding and changing voter registrations, compiling statistics required by federal law, generating official voter lists, conducting an online inquiry of the statewide voter list, and communicating in real time with the database of the Department of Motor Vehicles to verify voters’ driver’s license numbers, or the last four digits of a voter’s Social Security number. When a registrar enters a voter’s registration information, the system checks for exact matches on last name and date of birth, and displays the full name, address, and date of birth of any matching voters, allowing registrars to ask registrants for clarification. However, the system does not provide real-time information on which voters have already cast ballots on election day, and even registration information may not be perfectly up to date since registrars are not required to enter every registration into the system on the day it is received.

HAVA requires that the electronic registration system assign a unique identifier to each legally registered voter in the State. The CVR database actually assigns an identifier to each voter record, so the identifier cannot be used to identify duplicate registrations. Although driver’s license numbers are entered into the system when available, they cannot be used to spot duplicates, because the system does not allow searches by driver’s license number. In fact, out of respect for voters’ rights, the system was designed not to reject apparently duplicative entries. However, Connecticut law authorizes the Secretary of the State to conduct periodic searches of the database to identify voters who are registered in more than one town. Conn. Gen.Stat. § 9-21a. The Secretary performs this task by means of a computer program that searches for matches on last name and date of birth, compiling a list of potential duplicates for further investigation by the registrars of affected towns. The record contains conflicting accounts of how often the Secretary compiles such a list, but the minimum estimate was once a year, in January or June — that is, not during the peak registration period before an election. Registrars receiving a list of potential duplicate registrations from the Secretary of the State are obliged to send a notice regarding the possible duplicate registration to each affected voter. The notice requires the voter to reassert her right to remain on the registered-voter list, or be removed from it. However, a voter may not be removed until the registrars of both towns implicated by a possible duplicate registration have conferred and agreed that the voter is registered in both places.

Terminals for accessing the CVR database are not installed at polling stations, so the system is only accessible to registrars and a few town clerks at the registrar of voters’ offices. Furthermore, the reliability of the CVR system has been called into question as a result of continuous crashes during the final, and very busy, two days of the registration period for the 2004 election, as well as during local elections held in May 2005. On these dates, many registrars were either unable to access the CVR system at all, or found the system unworkably slow. As a result, registrars could not run the usual checks to see whether registrants were already registered elsewhere in the State, and whether the driver’s license and social security information matched that on file with the Department of Motor Vehicles; nor could they enter new registration information, creating a backlog of thousands of registration cards. In addition, Ms. Tramonta-no, the Republican Registrar of Voters for New Haven, testified that the CVR system often slows down at busy times, such as close to cutoff dates, when many towns are accessing the statewide system simultaneously. As a result of the CVR system’s sluggishness, during periods of heavy registration traffic, it takes much longer to perform the standard checks to see whether an individual has registered elsewhere, with the result that each in-person registration may take up to fifteen minutes to perform.

The State and the private contractors responsible for maintaining the CVR database are aware of these performance issues and have been seeking to ameliorate them. The Secretary of the State has indicated that she believes that the problems have been resolved, but local officials remain concerned about the possibility of a recurrence or the development of new difficulties.

Connecticut’s Experience with Voting and Registration 1996-2004

According to statistics for the year 2000, 75% of the voting-age population of Connecticut was registered to vote, compared with a national average of 76%. However, far fewer voters than are registered actually turn out to vote. Indeed, voter turnout in Connecticut has dropped from 76% of voting-age population in 1960, to 58.4% of voting-age population in 2000, with 2004 representing a ten-year high of 58.8% turnout.

According to Plaintiffs’ expert, Professor Jonathan Nagler of New York University, Connecticut’s experience of declining voter turnout parallels a national trend involving a steep decline between 1960 and the early 1970s that established a new, lower, average voter turnout, followed by two decades of fluctuation around that low average, and then an upward bump in the 2004 presidential election. Remarkably, the decline was not arrested nationally by the passage of the Voting Rights Act of 1965, or in Connecticut by the passage of legislation reducing the registration deadline from twenty-four days prior to 1977, to fourteen days between 1991 and 2005. Indeed, data compiled by the Federal Election Commission show that even the states that have introduced EDR have not escaped declining turnout. See Tables of Presidential Voter Registration and Participation Statistics at http://www.fec. gov/votregis/tu rn/ turn.htm. It is therefore clear, as Professor Nagler agreed at trial, that something much more significant and fundamental than time-to-register has been influencing overall voter turnout, and that adopting EDR likely will only reduce the extent of the decline in voter turnout.

As mentioned above, Connecticut permits all unregistered voters to cast a ballot in the presidential races. In 1996, when a seven-day deadline for presidential ballots was in effect, only 1,000 citizens voted by presidential ballot. However, after the legislature made presidential ballots available on election day itself, there was a dramatic increase in the number of voters exercising the presidential ballot option— with more than 30,000 presidential ballots cast in both 2000 and 2004.

Testimony at trial indicated that registrars and town clerks in at least some towns encouraged presidential ballot voters to register at the polls by providing them with registration cards when they cast their ballots. The success of this enterprise appears to have varied from one town to another. Mr. Abbate testified that in Cheshire almost all of the 188 individuals who voted by presidential ballot took advantage of the opportunity to fill out a registration card at the same time. Ms. Hutton, Town Clerk of Middletown, testified that of 530 individuals casting presidential ballots, perhaps 100 completed registration cards. Ms. Hutton speculated that the reason for this might be Middle-town’s large population of students who may know that they will never again vote in the town. Finally, Ms. Tramontano, Republican Registrar of Voters for New Haven, testified that of 1040 unregistered voters who cast presidential ballots, only 240 also took the time to register.

The large increase in voters applying for presidential ballots on Election Day 2000 caused long lines and considerable chaos at town clerks’ offices with a shortage of ballots in some places, but the process ran much more smoothly in 2004. The large volume of presidential ballots cast does not appear to have led to a concomitant increase in electoral fraud. Mr. Garfield of the Elections Commission testified that, after asking town clerks and registrars to look through their records for suspicious activity, the Commission received only one notice of possible fraud in connection with the more than 30,000 presidential ballots cast on Election Day 2000, and that complaint was not substantiated. In 2004, there were claims that up to 500 presidential ballots had been fraudulently cast, and the Elections Commission is still investigating those claims, though Mr. Garfield acknowledged on cross-examination that these cases may be attributable to previously unregistered voters both casting presidential ballots and registering to vote in future elections on the same day.

Indeed, Plaintiffs contend that electoral fraud in general is not a large problem under Connecticut’s current regime and, happily for the voters of Connecticut, Plaintiffs’ position was largely supported by the testimony of the Defendant’s witness, Elections Commission General Counsel Jeffrey Garfield. Mr. Garfield testified that, over the last five years, the State’s Elections Commission has received approximately 1100 complaints of fraud, but only about five substantiated cases per year of voter and registration fraud combined. However, Mr. Garfield further testified that his extensive experience with Connecticut elections leads him to believe that more voter fraud occurs than is reported: “I think we have evidence that voter registration fraud is committed and as I indicated before, there’s a lot more that we don’t see because its difficult to detect and people don’t tend to bring all the cases to us that exist out there.” Mr. Garfield’s view was shared by long-time registrar Ms. Tramontano. The Court finds their testimony credible and accepts the judgment of these witnesses that more voter registration fraud occurs in the State than is ever actually reported or investigated.

A Brief History of Election-Day Registration (EDR)

On August 6, 1965, President Johnson signed into law the Voting Rights Act of 1965, which, among other important reforms, prohibited states from employing durational residency requirements to restrict access to the ballot in presidential races (but not in congressional or local races). See Voting Rights Act of 1965, 42 U.S.C. § 1973aa-1(b). The Voting Rights Act appears to have generated a wave of reform in state election laws during the 1970s. In the period between 1972 and 1974, Maine, Minnesota, and Wisconsin adopted EDR laws as part of a more general process of liberalizing the states’ registration requirements. Then, in 1977, President Jimmy Carter proposed a federal law imposing on states EDR requirements similar to those previously adopted in Minnesota and Wisconsin. However, President Carter’s proposed EDR legislation was rejected by both houses of Congress, and between 1977 and 1993 no other state in the nation adopted EDR.

In 1993, President Clinton signed the National Voter Registration Act (“NVRA”), 42 U.S.C. § 1973gg, which requires, among other things, that states offer voter registration services at numerous state agencies, including those providing public assistance and unemployment compensation, as well as public schools and libraries. Notably, the NVRA did not make election-day registration compulsory for states, but instead provided an exemption from its requirements for those states that voluntarily adopt EDR. Three states — Idaho, New Hampshire, and Wyoming — took advantage of the exemption and passed laws implementing EDR. The Wyoming Secretary of State, who testified via deposition in this case, confirmed that EDR was adopted in Wyoming to avoid the costs of implementing the NVRA. Despite the carrot provided by the NVRA exemption, no other state has adopted EDR.

The NVRA-inspired adoption of election-day registration in Idaho, New Hampshire, and Wyoming raised the number of EDR states to six- — Idaho, Maine, Minnesota, New Hampshire, Wisconsin, and Wyoming (“EDR states”). A number of other states have also considered EDR systems. In 2002, EDR ballot initiatives were defeated in Colorado and California: Oregon experimented with EDR in the 1970s and 1980s, repealed the system in 1985, and is now apparently considering it again. EDR proposals are also currently under consideration in Massachusetts, Montana, New York, North Carolina, and Vermont. In addition, at least four bills providing for national EDR are under consideration by the United States Congress: The Voting Opportunity and Technology Enhancement Act of 2005, S. 17, H.R. 533; The Count Every Vote Act of 2005, S. 450, H.R. 939; The Same Day Voter Registration Act of 2005, H.R. 496; The Voter Outreach and Turnout Expansion Act of 2005, H.R. 3557. To date, however, Congress has chosen not to require states to adopt EDR.

The Evidence Favoring EDR

Supporters of EDR urge its adoption with the purpose of increasing voter turnout, and Plaintiffs presented three highly qualified experts — Professors Donald Green of Yale University, Kenneth Gold-stein of the University of Wisconsin, and Professor Jonathan Nagler of New York University — who testified to the potential of EDR in this regard. Professors Green and Goldstein explained Plaintiffs’ view of the causal relationship between voter turnout and election-day registration. Professor Green testified that requiring registration in advance of elections for non-presidential races is problematic in his view because lower-profile races, and especially state and local races, tend to excite voter interest only very close to election day and usually after most regis: tration deadlines have passed. He explained that most media coverage comes in the two weeks leading up to elections, just as voters themselves are becoming interested in the races, but that, by then, it is too late to register for most unregistered citizens who decide they want to vote. Campaigns for candidates for most offices also typically ramp up their efforts in the week or two before an election, and since that is after most pre-election-day registration deadlines have passed, campaigns tend to focus their efforts on registered voters. Campaign neglect of unregistered voters may also entrench nonvoting because political scientists have found that voting is a habitual activity.

Professor Goldstein presented the results of various studies demonstrating that a high proportion of all campaign advertisements for, and local new stories about, federal races occur in the final week or two weeks before an election. According to Professor Goldstein, in 1998, 35% of all advertisements for federal races occurred in the last two weeks before the election, and 19% occurred in the last week. The pattern for 2000 and 2002 was similar. Professor Goldstein also showed that of all the local news stories covering federal races during the last seven weeks of the campaign, 54% aired in the last two weeks and 36% aired in the last week. In sum, Professors Goldstein and Green explained that EDR encourages higher voter turnout because voters only become interested in political races close to election day, and campaign staff and media organizations respond to this late-blooming voter interest by concentrating the distribution of information about candidates in the last few weeks of the election cycle, thus reinforcing the last-minute interest.

The explanation of the likely quantitative impact of EDR on voter turnout fell to Professor Nagler. Plaintiffs’ argument as to how much EDR might affect voter turnout in Connecticut began with the observation that EDR states have higher voter turnout on average than non-EDR states. In this connection, Professor Nagler cited data showing that in the 2000 presidential election, EDR states enjoyed an average turnout of 65.4% of voting-age population, compared with an average for non-EDR states of 50.5%, and a figure of 58.4% for Connecticut. Based on the experience of the EDR states, Professor Nagler predicted that EDR would increase turnout for presidential elections in Connecticut by 5.2%, which, in the year 2000, would have corresponded to an extra 130,000 voters.

However, Professor Nagler’s supposition appears to the Court to have a number of flaws. As a preliminary matter, the Court notes that despite the superficial allure of statistics comparing EDR and non-EDR states, the extent to which EDR is the driver of higher turnout in EDR states is questionable. To begin with, at the time of adopting EDR, each of the EDR states already had higher turnout than the national average. See Tables of Presidential Voter Registration and Participation Statistic at http://www.fec. gov/votregis/tu rn/ turmhtm. Plaintiffs’ expert from Wisconsin confirmed this fact. Thus, as commonsense would suggest, voting culture is an important factor affecting voter turnout. Moreover, focusing on the averages for EDR and non-EDR states masks a large spread in turnout between the EDR states, and the strong or even superior performance of certain non-EDR states. Thus, Professor Nagler’s report also reveals that turnout in 2000 was only 54.5% in Idaho and 59.7% in Wyoming (both EDR states) compared with 67.3% in Maine and 68.8% in Minnesota (both non-EDR states). Further, Alaska, a non-EDR state, had higher voter turnout than the average EDR state; five non-EDR states (Alaska, Iowa, Montana, Oregon, and Vermont) had higher turnouts than two of the EDR states (Idaho and Wyoming); and fifteen non-EDR states had higher turnouts than Idaho, an EDR state. Similarly, in 1998, New Hampshire had a turnout of only 35.4%, while 60% of Minnesota’s citizens came to the polls, and Alaska and Nebraska had higher turnouts than any of the EDR states except Minnesota. In short, as is often the case with statistics, each side of the EDR debate can find some figures to support their position.

Furthermore, Professor Nagler acknowledged that his 5.2% figure for the magnitude of the impact of adopting EDR in Connecticut might be a significant overestimate for a couple of reasons. First, Professor Nagler conceded his model did not take into account the number of Connecticut citizens already voting by presidential ballot, and that if his model did so, EDR might be expected to boost turnout by only 4.2%. Second, and more seriously, Professor Nagler admitted that his model assumes that Connecticut would implement what is known as one-step EDR. The one-step EDR approach allows voters to register and vote in the same location, usually a polling place, and is employed in five of the six EDR states (Wisconsin, Wyoming, Minnesota, New Hampshire, and Idaho). Maine, however, employs a two-step approach, requiring election-day registrants to register at the office of the registrar, and then travel to a polling place to cast their ballots. This difference in approach is significant because, to date, the majority of proposals for EDR in Connecticut have involved a two-step process like Maine’s, and Plaintiffs have not argued that a two-step EDR process would run afoul of the Constitution. Professor Nagler was unable to predict with confidence the likely impact of a two-step process like that contemplated by Connecticut, but he observed that Oregon had once experimented with such a system, and had experienced only a one or two percent increase in voter turnout. Notably, Oregon has since abandoned its EDR system and now requires pre-election-day registration.

In assessing all three experts’ testimony as to the likely impact of EDR on voter turnout in Connecticut, the Court must take account of one final factor: the decision of the Connecticut legislature in July of this year to reduce its registration cutoff from fourteen to seven days. As noted above, when trial commenced in this case, the deadline for registration was still fourteen days before a general election, and each of the experts based his analysis around that figure. The Court is certain that Plaintiffs are correct that many voters only really begin to pay attention to election campaigns at the last minute. Such is human nature. However, the Court is bound to note that the reports of Professors Goldstein and Green indicate that under a seven-day deadline, voters will have had at least one week of fairly decent media coverage of an election before registration closes, and that Professor Nagler acknowledged that the length of the cutoff period makes a “big difference, [it’s] a key thing that is a component of the model,” and that reducing registration from fourteen to seven days would also increase turnout. Thus, like the implementation of the presidential ballot, Connecticut’s reduction of its registration cutoff will to some extent erode the proportional increase in voter turnout to be gained by implementing EDR. In addition, the Court notes that Plaintiffs’ experts conceded that their conclusions were applicable to voters in general, and were not based on studies devoted to the class of voters at issue in this case — namely, unregistered voters— and that neither they nor anyone else knows much if anything about the habits or motivations of unregistered voters.

Although the Plaintiffs’ expert evidence is susceptible to criticism on the grounds sketched above, on balance the Court agrees with Plaintiffs that it is likely that adopting EDR would have a positive effect on voter turnout in Connecticut. However, on the record before it, the Court believes that the impact on voter turnout would be more modest than Plaintiffs assert — closer to one percent than five percent. That said, as the Court explains below, whether the likely impact is one or five percent is not determinative in this case, since Connecticut’s registration deadline would survive constitutional challenge even if EDR increased voter turnout by five percent.

The Evidence Against EDR

Opponents of EDR raise concerns about administrative burdens, election-day confusion, and the potential for fraud. The evidence presented in this case indicates that the volume of voters who take advantage of same-day registration in EDR states is very large, and much larger than the number of voters that Plaintiffs assert will only vote if they can register on election day. According to Professor Nagler, for the 2004 election, Minnesota reported that 21% of those voting, or 590,467 people, registered on Election Day; in Idaho the proportion was 19%, or 117,622 individuals; in Wyoming it was approximately 16%; and in Wisconsin the proportion was 14.7%. These figures would seem to support the concern expressed by, for example, Mr. Abbate (the President of the Registrars’ Association) that a side-effect of EDR may be to encourage people who would otherwise register in advance of election day to wait and sign up at the polls. To put this concern in perspective, if Connecticut had adopted EDR in 2004 and had experienced a volume of election-day registration similar to Wisconsin, Wyoming, Idaho, and Minnesota, poll workers in Connecticut’s towns would have confronted over 200,000 individuals lining up to register at the polls, even if EDR led to no increase in voter turnout. Chaos reigned in Connecticut in 2000 when only 30,000 unregistered voters showed up to cast presidential ballots, and, although the process ran more smoothly in 2004, two of the three registrars of voters who testified asserted that they likely could not cope with the kind of volume of election-day registrants experienced by EDR states.

As to the impact of EDR on fraud, the record before the Court is rather thin. Plaintiffs submitted various reports by public interest groups that baldly assert that there is no evidence of an increase in voter fraud in EDR states. The paucity of formal findings of fraud in EDR states was reiterated by the first of three fact witnesses presented by Plaintiffs, Professor Lorraine Minnite, who based her views on a compendium of articles and news reports concerning election fraud that she had culled from a selection of electronic databases. In a similar vein was a report by the Connecticut Office of Legislative Research (“OLR Fraud Report”) that was “based on calls to the office of the chief election official [or Attorney General] in each [EDR] state,” and that summarized its findings as follows: “While no state had any formal allegation of fraud, four had anecdotal information peculiar to each state.” Pls.’ Ex. 125 at 1. However, the OLR Fraud Report in fact provided anecdotal information for only three states, and that information was extremely vague— New Hampshire “has had some complaints;” Wyoming reported “allegations] that out-of-state residents are crossing state lines and voting in Wyoming [but] this has not been substantiated”; and Wisconsin’s “elections director has heard of some isolated cases where college students have been placed on deferred prosecution for [voter fraud].” Id. at 2. Joe Meyer, Secretary of State for Wyoming, testified by deposition, but his testimony essentially amounted to an assertion that his thirty-five years of experience as a state official causes him to believe that there is very little voter fraud in his state, and that when fraud does occur, it is adequately detected and reported. However, even Mr. Meyer qualified his testimony by an admission that he may not be aware of every case of voter fraud that arises.

The only really significant information in the record as to the prevalence of fraud in EDR states is restricted to the situation in a single EDR state — Wisconsin. That evidence tends to indicate that the jury is still out on the relationship between EDR and fraud. Mr. Kevin Kennedy, Executive Director of the Wisconsin State Elections Board, testified that in his twenty-six years of service he has never had the sense that voter fraud was a problem in Wisconsin, and indeed has heard about less than ten convictions or prosecutions for election-day registration fraud. He further testified that he believes the face-to-face nature of election-day registration actually decreases fraud compared to mail-in registration. However, Mr. Kennedy acknowledged that in Wisconsin, as elsewhere, voter fraud may go undetected, and that Wisconsin had experienced significant problems in the 2004 election, including the failure to process around 20,000 of the registration cards submitted more than thirteen days prior to the election, and a discrepancy of as many as 7,000 votes between the number of people recorded as voting and the number of votes cast.

Mr. Kennedy offered his personal view that the most likely explanation for the problems in 2004 was administrative error — namely, the failure of poll workers properly to maintain the records on election day. However, Defendant provided the Court with the preliminary report of a joint task force of the Milwaukee Police Department, District Attorney, United States Attorney and the Federal Bureau of Investigation, set up to investigate Wisconsin’s problems in the 2004 election. The Task Force’s investigation focused on some 70,000 election-day registrations and concluded that a large majority of the reported errors were due to poor data entry, supporting Mr. Kennedy’s sense that the real challenge with EDR is proper administration, not necessarily voter fraud. Nonetheless, the Task Force also found evidence sufficient to suggest that there had been more than 100 instances of voter fraud, not including 200 cases of voting by ineligible felons. There has been no final conclusion of the Wisconsin investigation.

In the Court’s view, the evidence in the record as to the EDR states’ experience of with voter fraud illustrates that although elections in the United States are far from riven by fraud, the potential for voter fraud exists, and states are, therefore, right to be concerned about it and to take steps to minimize it.

EDR in Connecticut

Legislation to implement election-day registration has been introduced in the Connecticut legislature every year since 2001, with proposals varying mostly along the dimension of fraud protection. The 2001 bill, sHB 6823, would have introduced a two-step election-day registration system: first, an unregistered voter desiring to vote on election day would visit her town registrar’s office, where she would present a birth certificate, driver’s license, or Social Security card, take an oath that she had not already voted in the election, and receive a notice of acceptance; second, the voter would then proceed to a polling place, where she would present the notice of acceptance and cast her ballots. See Pis.’ Ex. 62, Office of Legislative Research, Bill Analysis, File No. 488 at 10. The 2002 bill, sHB 5700, would have introduced a two-step process similar to that suggested by the 2001 bill, but with the additional requirements that voters supply proof of address, and that registrars send out a confirmation notice, which, if returned undeliverable, would result in the name of the voter being passed on to the Elections Commission for investigation of possible voter fraud. See Pis.’ Ex. 64, Office of Legislative Research, Bill Analysis sHB5700. Defendant, Secretary of the State Susan Bysiewicz, supported the 2002 bill and testified to her sense that EDR “would go a long way toward encouraging young people ... to participate in [the] election process,” Pis.’ Ex. 10, Hearing of the Government Administrations and Elections Committee (March 11, 2002) at 8, and that “for the future ... we can eliminate presidential ballots and move toward a system where we can have protections in place to protect the integrity of the process and at the same time allow people to register to vote on election day,” Pis.’ Ex. 36, Hearing of the Government Administrations and Elections Committee (February 20, 2002) at 22-23.

With bi-partisan support, the' Connecticut legislature in 2003 passed a two-step EDR proposal, sHB6370, Public Act 03-204, which would have allowed a citizen resident in Connecticut to register to vote and cast a ballot on election day if she presented identification displaying her name, address and photograph (or, if no photographic identification was available, submitted to having a picture taken by the registrar), and signed under penalty of perjury a statement affirming her eligibility, the fact that she had not already registered or voted elsewhere, and her understanding of the criminal penalties for making a false statement. As under the 2002 proposal, registrars would send out a confirmation notice, which, if returned undeliverable, would result in the name of the voter being passed on to the Elections Commission for investigation of possible voter fraud. See Pls.’ Ex. 66, Public Act 03-204 §§ 2(2)(A)(6).

The 2003 EDR Act was supported by Secretary Bysiewicz, Attorney General Richard Blumenthal, the Elections Commission, and the Registrars’ Association, all of whom were satisfied with the Act’s strong identification provisions. However, the Act was vetoed by then-Governor John Rowland because at that time the State’s CVR database had not yet been fully implemented, and he was concerned about the potential for voting fraud in the absence of “an accurate, complete, up-to-date and real-time centralized voter registration database.” Pis.’ Ex. 69, Letter from Governor Rowland to Secretary Bysiewicz regardi