Souls to the Polls: Early Voting in Florida in the Shadow of House Bill 1355* Michael C. Herron+ Daniel A. Smith? May 22, 2012 Abstract Over the past 30 years an increasing number of American states have made it more convenient for voters to cast early ballots. Despite the rapid diffusion of what is known as early in-person voting and praise for this practice by voting rights advocates and election administrators alike, a new Florida law in 2011 truncated the state's early voting period from a total of 14 days to eight, eliminated early voting on the Sunday immediately preceding Election Day, and reduced the total number of hours that early polling stations were required to be open. We assess the effects that these changes might have on Florida voting by analyzing early voting patterns from the 2008 General Election in this state. By merging a Florida voter file with county-level records of approximately 2.6 million early voters, we are able not only to identify which types of voters cast early ballots in the run-up to the 2008 General Election, but also to determine the precise days of the two-week early voting period in which various voter types cast their ballots. We find that Democratic, African American, Hispanic, younger, and first-time voters were disproportionately likely to vote early in 2008 and in particular on weekends, including the final Sunday of early voting. We expect these types of voters to be disproportionately affected by the recent changes to Florida's voting laws that altered the practice of early voting across the state. An earlier version of this article was presented at the 12th Annual State Politics and Policy Conference, Rice University, February, 2012. In addition, the substance of this article was presented on January 27, 2012, in Tampa Bay, Florida, at a hearing held by the United States Senate Committee on the Judiciary, Subcommittee on the Constitution, Civil Rights and Human Rights. + Professor of Government, Dartmouth College. 6108 Silsby Hall, Hanover, NH 03755-3547 (michael.herron@dartmouth.edu). Herron was Visiting Professor of Applied Methods, Hertie School of Governance, Berlin, Germany, when he completed work on this article, and he thanks the Hertie School for its support. ? Professor of Political Science, 234 Anderson Hall, University of Florida, Gainesville, FL 32611-7325 (dasmith@ufl.edu). * 1 Introduction Until quite recently the expansion of convenience voting across the United States seemed to be moving inexorably forward. Today, all but 15 states provide an extended window prior to Election Day during which electors may vote in person or mail in an absentee ballot, no excuses needed. According to estimates by the Associated Press, more than 44 million early in-person (hereinafter, EIP) and absentee ballots were tallied before the official voting day of the 2008 General Election, accounting for approximately 34 percent of total votes cast (Gronke and Tokaji, 2011). Nowhere in 2008 was early voting more utilized, and perhaps more scrutinized, than in Florida.1 Of the roughly 8.3 million Floridians who cast ballots in the 2008 General Election, over half did so prior to Election Day. More than 1.87 million Floridians voted by "no-excuse" absentee ballots, but even more-- some 2.62 million voters--cast EIP ballots at hundreds of non-precinct voting sites that dotted Florida's 67 counties.2 Media reports surrounding the 2008 election indicated that African Americans seemed more inclined to cast EIP ballots, as many were driven--sometimes literally--with "getting your souls to the polls"3 mobilization efforts sponsored by religious organizations, nonpartisan advocacy groups, and the Democratic Party.4 Despite considerable praise for the expansion of early voting in Florida by voting rights advocates and See, for example, "Early voting suggests 2008 may see record turnout, expert says," CNN, October 21, 2008, available at http://articles.cnn.com/2008-10-21/politics/early.voting_1_ early-voting-early-ballots-count-ballots?_s=PM:POLITICS (last accessed April 11, 2012); Kendric Ward, Absentee Ballots Front-Load Florida Elections, Sunshine State News, October 6, 2008, available at http://www.sunshinestatenews.com/story/absentee-ballots-front-load-florida-elections (last accessed April 11, 2012); Kate Phillips, Florida Extends Early Voting Hours, New York Times, October 28, 2008, available at http://thecaucus.blogs.nytimes.com/2008/10/28/florida-extends-early-voting-hours (last accessed April 11, 2012). 2 In Florida, absentee ballots can be turned in early or on Election Day. We define "early in-person" voting as casting a nonabsentee ballot in-person, prior to Election Day, at an early voting center. 3 The National Association for the Advancement of Colored People is credited with creating the slogan in 2000, "Get all souls to the polls." During the final days of the presidential campaign, Democrat Al Gore exhorted to crowds in Florida, "[G]et your souls to the polls" on Election Day, as early voting had yet to be adopted. See Rupert Cornwell, "With 48 hours to go, the candidates try to close the deal with America," The Independent, November 6, 2000, available at http://www.independent.co.uk/news/world/americas/ with-48-hours-to-go-the-candidates-try-to-close-the-deal-with-america-623223.html (last accessed April 11, 2012). 4 See, for example, The Florida Senate (2010). The report states on p. 3, "Thus, early voting has increased in popularity among electors since it was first introduced in 2004." On Democratic efforts see, Nathan Crabbe, "City ready for Michelle Obama's visit," Gatorsports.com, October 21, 2008, available at http://www.gatorsports.com/article/20081021/NEWS/ 810220978?template=printpicart (last accessed April 11, 2012). 1 2 local election administrators,5 in 2012 citizens of the Sunshine State have fewer opportunities to cast early ballots. In May, 2011, the Florida state legislature passed House Bill 1355 (hereinafter, HB 1355), an omnibus election reform bill, which Governor Rick Scott signed into law.6 Among its many provisions designed to restrict EIP voting, HB 1355 reduced from 14 to eight the total number of days county Supervisors of Elections could offer early voting; altogether eliminated EIP voting on the Sunday immediately preceding Election Day; and, gave election supervisors the discretion to reduce, from 96 hours to 48, the number of hours EIP voting stations are required to be open.7 Changes in voting laws and procedures almost invariably affect some types of voters more than others, and we seek here to understand how the restrictions to EIP voting in Florida wrought by HB 1355 might have differential effects across the span of Florida voters, particularly with respect to the upcoming 2012 General Election. We gain leverage on this issue by studying the most recent General Election in Florida, that which took place in 2008. Specifically, we pose and then offer answers to the following questions: who voted EIP in the 2008 General Election in Florida; on what days in particular before this election did Florida early voters tend to cast their early ballots; and, in 2008 was there variance in preferred days of early voting across the partisan and socio-demographic groups that together make up the Florida electorate?8 Our partisan breakdown of voters turns on a standard categorization (voters are either Democratic, Republican, third party, or have no party affiliation) and our demographic groups of interest are defined based on race/ethnicity, age, voter registration date, and participation history in prior elections. Although the scholarly literature on convenience voting--and EIP voting is a special case of this type of voting--has grown considerably in recent years, scholars know relatively little about who votes EIP in elections as prominent as presidential contests and even less about the exact days on which citizens present See, for example, Anthony Man, "Early voting locations slashed, Sunday hours eliminated," South Florida Sun-Sentinel, August 3, 2010, available at http://weblogs.sun-sentinel.com/news/politics/broward/blog/2010/08/ early_voting_locations_slashed.html (last accessed April 13, 2012). 6 HB 1355 amended the Florida Election Code (chapters 97-106, Florida Statutes) and became law (Chapter 2011-40, Laws of Florida) on May 19, 2011. The legislative history of HB 1355 is detailed on http://www.flsenate.gov/Session/ Bill/2011/1355 (last accessed April 13, 2012). 7 Early voting under HB 1355 is to commence on a Saturday, ten days prior to an Election Day, and it must end on a Saturday, three days prior to Election Day. See Levitt (2011). 8 For the motivation of our study we thank Justin Levitt, who on May 23, 2011, on Rick Hasen's Election Law Blog offered some thoughts about who might be most affected by the elimination of EIP voting in Florida on the Sunday prior to Election Day. Levitt noted that, "The racial breakdown of Florida's early voters isn't publicly available-but it is tracked, and could be analyzed," and that he would "be surprised if African-American citizens weren't a substantial portion of the Floridians choosing to vote in their Sunday best." Levitt's comments, titled "A Devil in the Details of Florida's Early Voting Law," are available at http://electionlawblog.org/archives/019579.html (last accessed April 13, 2012). 5 3 themselves at early voting centers. The dynamics of early voting--when precisely early votes are cast across a given early voting period--has not been a major part of the convenience voting literature. This is unfortunate insofar as the contemporary debate in Florida about EIP voting--most notably, the debate about HB 1355--to a large extent turns on the question of when, precisely, EIP votes should be allowed. Our research on EIP voting in Florida in the 2008 General Election provides insight on early voting dynamics. By merging a Florida state-level voter file, comprised of more than 12.3 million registered voters, with 67 early voting files for the 2008 General Election, one per each of Florida's counties, we are able to assess and study the race/ethnicity, party registration, age, county registration, and vote history of almost every EIP voter in Florida in 2008. Although our individual-level data does not allow us to directly assess the personal motivations of those who chose to vote EIP in the 2008 General Election, that we are able to specify which day of the week an early voter showed up to the polls in this election allows us to gauge in anticipation of the 2012 General Election the differential effects of HB 1355 across various Florida voter groups. Although the richness of our data on EIP voting in 2008 allows us to discuss early voting dynamics in a way that is new to the literature on convenience voting, we cannot at this point assess the effect on overall election turnout in Florida of restrictions to EIP voting in the state. In particular, the matter of whether HB 1355's recent changes to EIP voting rules in Florida will affect actual 2012 General Election turnout-- particularly among those citizens who voted early in the 2008 General Election--is a matter that must wait until after November 2012. We expect to focus on this issue in future research. Before turning to our empirical investigation of Florida early voting in the 2008 General Election, we first provide a brief background on the passage in 2011 of Florida's HB 1355, which was ultimately the trigger for this study. This is followed by a review of the literature on EIP voting in the American states and then by our statistical results. 4 Early In-Person Voting in Florida In the aftermath of the 2000 General Election, the Florida legislature in 2004 passed legislation that, among other things, required local Supervisors of Elections to offer EIP voting.9 This bipartisan legislation (Senate Bill 2566, which passed the Florida House by a 100-12 margin and the Florida Senate 30-8), was signed into law by then-Governor Jeb Bush and took effect on July 1, 2004.10 The 2004 EIP legislation, which technically speaking allowed absentee ballots to be returned up to 15 days before Election Day, was viewed by Republican and Democratic lawmakers alike as a practical way to reduce Election Day mishaps at the polls.11 Under the 2004 statute, an elector could cast a vote as many as 15 days before Election Day at an early voting site designated by his or her county Supervisor of Elections, with early voting ending on the second day prior to a scheduled election. Although the law required EIP voting centers to be open a total of 96 hours during the aforementioned 15 day stretch--including eight hours each week day and a total of eight hours over both weekends--election supervisors retained some flexibility in the setting of early voting hours.12 In 2005, the Florida legislature amended state law so as to eliminate EIP voting on the Monday proceeding a Tuesday Election Day (Levitt, 2012). Given the history of broad consensus in support of convenience voting in Florida, as evidenced by the bipartisan coalitions in the Florida House and Senate that supported EIP efforts in the past, the decision in 2011 by the Florida legislature to curtail EIP voting transpired with what one might call " head-spinning speed."13 Moreover, the rhetoric supporting the recent reduction in EIP voting was particularly pointed. For details on the 2000 election, its aftermath, and what led to the many reform efforts in Florida and beyond, see Richard Posner, Breaking the Deadlock: The 2000 Election, the Constitution, and the Courts, Princeton University Press, Princeton, NJ, 2001, and Martin Merzer, The Miami Herald Report: Democracy Held Hostage, St. Martin's Press, New York, 2001. 10 The legislative history of Senate Bill 2566 is available at http://archive.flsenate.gov/session/index. cfm?Mode=Bills&SubMenu=1&BI_Mode=ViewBillInfo&BillNum=2566&Year=2004&Chamber=Senate# BillText (last accessed April 13, 2012). The Florida state legislature actually passed a bill (Senate Bill 1402) in 1998 that opened the door to early voting, and the legislative history of this bill is available at http://archive.flsenate. gov/session/index.cfm?Mode=Bills&SubMenu=1&BI_Mode=ViewBillInfo&BillNum=1402&Year= 1998&Chamber=Senate#BillText (last accessed April 13, 2012). 11 Section 101.657, F.S. In 1998, the Florida legislature passed a bill allowing voters to personally deliver their absentee ballots to the office of the Supervisor of Elections. See Levitt (2012) and Monica Scott, "Voters Cast Early Ballots," Orlando Sentinel, August 29, 2004, available at http://articles.orlandosentinel.com/2004-08-29/news/0408290050_ 1_early-voting-early-ballots-tavares (last accessed April 13, 2012). 12 Under the 2004 legislation, EIP voting hours could fall any time between the hours of 7am and 7pm, but election supervisors had the option of opening centers on Saturday and/or Sunday. As such, between 2004 and 2011 the application of EIP remained uneven across Florida's 67 counties, with some election supervisors maximizing early voting days and hours (96 hours spread over all 15 days, including both Sundays), and others opting for the minimal requirements (96 hours but no Sunday voting). See The Florida Senate (2010). 13 Aaron Deslatte and Kathleen Haughney, "Legislature passes broad overhaul of 9 5 During the floor debate on HB 1355, for example, Republican state senator Michael S. Bennett (representing Florida's 21st District) asserted that voting was a privilege and that the state should not make voting too easy. Speaking in support of HB 1355, Bennett asked:14 Do you read the stories about the people in Africa? The people in the desert, who literally walk two and three hundred miles so they can have the opportunity to do what we do, and we want to make it more convenient? How much more convenient do you want to make it? Do we want to go to their house? Take the polling booth with us? This is a hard-fought privilege. This is something people die for. You want to make it convenient? The guy who died to give you that right, it was not convenient. Why would we make it any easier? I want 'em to fight for it. I want 'em to know what it's like. I want them to go down there, and have to walk across town to go over and vote. In May 2011, the Florida legislature passed HB 1355, an omnibus elections bill that was the product of an 11th hour, strike-all amendment,15 and Governor Rick Scott signed the bill into law 13 days after receiving it.16 In addition to putting restrictions on voter registration drives, the casting of provisional ballots, and several other voting and elections issues, HB 1355 as earlier reviewed reduced from 14 to eight the total number of days that county Supervisors of Elections were permitted to offer early voting. HB 1355 also completely eliminated early voting on the Sunday immediately preceding Election Day--which in 2008 was offered to voters in ten of Florida's 67 counties. Moreover, HB 1355 gave county election supervisors the discretion to reduce from 96 to 48 the total required number of hours early voting polling stations are required to be open.17 Despite the clear EIP voting reductions embedded in HB 1355, former Florida Secretary of State Kurt Browning stated that the number of total hours of early voting hours required under HB 1355 would remain elections law," Orlando Sentinel, May 5, 2011, available at http://articles. orlandosentinel.com/2011-05-05/news/os-elections-bill-passes-20110505_1_ early-voting-league-of-women-voters-statewide-voter-database (last accessed April 13, 2012). 14 Politifact Florida, "Says people in Africa 'literally walk two and three hundred miles in order to vote," Politifact Florida, May 6, 2011, available at http://www.politifact.com/florida/statements/2011/may/06/mike-bennett/ think-we-have-it-tough-africa-people-walk-300-mile (last accessed April 13, 2012). Similar language to Bennett's was echoed by Republican legislators in Maine, Minnesota, New Hampshire, and Wisconsin. See Ed Kilgore, The Truth About Voter Suppression, Salon, available at http://www.salon.com/2011/09/30/votesuppresion (last accessed April 13, 2012). 15 For details on the bill's legislative history see http://www.myfloridahouse.gov/sections/Bills/ billsdetail.aspx?BillId=46543 (last accessed April 13, 2012). 16 HB 1355 amended the Florida Election Code (chapters 97-106, Florida Statutes) and became law (Chapter 2011-40, Laws of Florida) on May 19, 2011. 17 Early voting under HB 1355 is to commence on a Saturday, ten days prior to an Election Day, and it must end on a Saturday, three days prior to Election Day. See Levitt (2011). 6 the same. "The new law makes early voting more accessible now than ever before," Secretary Browning averred, "by expanding the number of hours that election supervisors can open early voting sites--from just eight hours before, to 12 hours a day." Secretary Browning went on to claim that the reduction in the number of early voting days was designed "[t]o combat voter fraud,"18 as well as "[to] reduce the burden on poll workers and provide needed flexibility to local election supervisors." Summarizing the legislation in a St. Petersburg Times op-ed article, Secretary Browning reiterated that, "Early voting remains at 96 hours, with greater flexibility for counties."19 In fact, the aggregate number of early voting hours that county Supervisors of Elections were required to remain open under HB 1355 was cut in half, from 96 hours in 2008, to 48 hours. Well before HB 1355 was signed into law Democrats widely denounced the bill, referring to it as a "voter suppression" bill, which among other provisions, would deter minorities from voting.20 According to its critics, the Republican-sponsored legislation--at the time of HB 1355's passing both chambers in the 18 If preventing voter fraud were the true intent of HB 1355, then we suspect that in this new law there likely would have been attention placed on cracking down on what appears to be the relatively not-so-uncommon practice of fraudulent absentee ballots in Florida. Absentee ballot fraud is not limited to mayoral races in Florida's metropolitan areas, which are well documented. See, for example, Dahleen Glanton, "Differences Cited in 98 Miami Vote Case," Chicago Tribune, November 13, 2000, available at http://articles.chicagotribune.com/2000-11-13/news/ 0011130119_1_absentee-ballots-miami-mayoral-election-new-election (last accessed April 13, 2012). In November 2011, for example, several people in rural Madison County, including a candidate on the local ballot for school board, were arrested by the Florida Department of Law Enforcement and charged with obtaining absentee ballots for other people without the voters' knowledge or consent. The school board candidate and her accomplices then provided an alternate address for the ballots to be mailed to the registered voters by the county Supervisor of Elections and allegedly then retrieved the ballots from the fraudulent locations, brought the ballots to the voters-sometimes with the ballots already filled out-and had the voters sign the absentee ballot signature envelopes. See "Madison County officials arrested for voter fraud," WTXL, November 1, 2011, available at http://www.wtxl.com/content/localnews/ story/Madison-County-officials-arrested-for-voter-fraud/-Nq7HTFCCU-kRpw39U_2tA.cspx (last accessed April 13, 2012). Rather than directly addressing absentee ballot fraud, HB 1355 actually eliminated a provision that existed in 2010 (when the above fraud occurred) that may make it more difficult to prosecute voting fraud. In particular, prior to the passage of HB 1355 Supervisors of Elections were required to send absentee ballots to voters registered addresses unless said voters were absent from the county, hospitalized, or temporarily unable to occupy residences. Now, under HB 1355, instead of being required with the forgoing exceptions to send an absentee ballot "[b]y nonforwardable, return-if-undeliverable mail to the elector's current mailing address on file with the supervisor," Supervisors of Elections may mail an absentee ballot "to any other address the elector specifies in the request." See Chapter 101.62 "Request for absentee ballots," available at http://statutes.laws.com/florida/TitleIX/chapter101/101_62 (last accessed April 13, 2012). We note that in Florida there is a notable partisan divide on absentee voting. Registered Republicans in Florida are likely to use this form of convenience voting much more than their Democratic counterparts, and in the 2008 General election Republicans had a 10.8 percent lead over Democrats voting absentee ballots through Election Day. Forty-seven percent of those who voted absentee were Republicans and 36 percent were Democrats. See The Florida Senate (2010). 19 Kurt Browning, "Florida's early voting remains at 96 hours," Tampa Bay Times, May 23, 2011, available at http: //www.tampabay.com/opinion/columns/floridas-early-voting-remains-at-96-hours/1170931 (last accessed April 13, 2012). 20 For example, see https://secure.aclu.org/site/Advocacy?cmd=display&page=UserAction&id= 3505 (last accessed April 13, 2012). 7 Florida state legislature had Republican majorities and the Florida governor was also a Republican--was a thinly veiled effort to suppress minority voter turnout under the cover of reducing fraudulent voting. Florida Senate Minority Leader Nan Rich was blunt in her criticism of the legislation, saying, "I have to wonder if the only problem that can be found with our electoral process is that some people don't like the outcome of the last presidential election."21 The Florida League of Women Voters also decried the decision by the state legislature to curtail early voting, accusing the state legislature of "reverting to Jim Crow-like tactics."22 Notwithstanding the criticisms offered by Democratic lawmakers and their political allies as well as a flurry of lawsuits in federal courts,23 62 county Supervisors of Elections in concert with the Florida Secretary of State have implemented the voting and election changes under HB 1355, including the reduction of EIP voting. However, in accordance with the Voting Rights Act of 1965 (hereinafter, VRA), five of Florida's 67 counties--Collier, Hardee, Hendry, Hillsborough, and Monroe--must receive federal preclearance from the United States Department of Justice before making any changes to electoral administration or voting rights in their jurisdictions.24 As such, Florida's January 31, 2012, Presidential Preference Primary operated under conflicting sets of electoral administration rules and voting rights. Howard Simon, the Executive Director 21 David G. Savage, "Election laws tightening in GOP-run states," Los Angeles Times, October 30, 2011, available at http:// www.latimes.com/news/nationworld/nation/la-na-vote-florida-20111031,0,1962738.story (last accessed April 13, 2012). 22 League of Women Voters, Press Release, JIM CROW TACTICS RETURN TO FLORIDA: VOTER SUPPRESSION LAW MOVING QUICKLY DESPITE PROTESTS, April 15, 2011, available at http://leagueorangecounty.typepad.com/blog/2011/04/ jim-crow-tactics-return-to-florida-voter-suppression-law-moving-quickly-despite-protests. html (last accessed April 13, 2012). Former President Bill Clinton also compared HB 1355 to Jim Crow. See Gabriella Schwarz, Bill Clinton compares new voting laws to Jim Crow, CNN, July 6, 2011, available at http: //politicalticker.blogs.cnn.com/2011/07/06/bill-clinton-on-todays-jim-crow (last accessed April 13, 2012). 23 For example, State of Florida v. United States of America, and Eric H. Holder, Jr., United States District Court for the District of Columbia, 2011, Case 1:11-cv-01428-CKK-MG-ESH; League of Women Voters of Florida, et al. v. Kurt S. Browning, et al., United States District Court, Northern District of Florida, 2011, Civil No. 4:11-cv-00628-RH-WCS; The League of Women Voters of Florida, et al. v. Rick Scott, et al., United States District Court, Southern District of Florida, Case No. 4:11-cv-10006-KMM, 2011. 24 At the time of this writing, the status of three provisions of HB 1355 being challenged for the five Section 5 counties remains uncertain. On October 18, 2011, a federal judge in Miami refused to hear a lawsuit against the implementation of HB 1355 in all of Florida's counties. Then, on October 28, 2011, the United States District Court of the District of Columbia turned down Governor Scott's request for an expedited review of the four major changes to the state's electoral code being reviewed by the United States Department of Justice in accordance with the VRA. See Dara Kam, "In push for its new election law, Florida challenges U.S. Voting Rights Act," Palm Beach Post, October 11, 2011, available at http: //www.postonpolitics.com/2011/10/federal-court-tosses-elections-lawsuit (last accessed April 13, 2012); Mary Ellen Klas, "Federal judges reject Browning's request to expedite ruling on voting law case," St. Petersburg Times, October 28, 2011, available at http://www.tampabay.com/blogs/the-buzz-florida-politics/content/ federal-judges-reject-brownings-request-expedite-ruling-voting-law-case (last accessed April 13, 2012). 8 of the American Civil Liberties Union of Florida, which unsuccessfully sued in federal court to prevent the implementation of HB 1355, flagged the conundrum here, saying, "Florida is in position yet again to turn our elections into a mockery by conducting an important, nationally significant election under two different sets of election rules depending on where you live."25 For many observers the passage of HB 1355 smacks of partisan politics: Republican lawmakers, according to such a narrative, convinced that the expansion of convenience voting laws contributed to the victory of Barack Obama in 2008, were determined to not let such a thing happen again in 2012. Marge Baker, executive vice president of People for the American Way, observed that,26 So-called anti-fraud laws are almost always thinly veiled attempts to prevent large segments of the population from making it to the ballot box... low-income voters, college students, people of color, the elderly. The people behind these laws know that there is no voter fraud epidemic. They just want to make it as difficult as possible for certain types of people to vote. Similarly, according to a recent study by the Brennan Center for Justice at New York University School of Law (Weiser and Norden, 2011) the restrictive voting laws in aggregate "could make it significantly harder for more than five million eligible voters to cast ballots in 2012" with the heaviest burdens falling "most heavily on young, minority, and low-income voters, as well as on voters with disabilities" (p. 1). Supporters of HB 1355 thought otherwise, not surprisingly. Florida House Majority Leader Carlos Lopez-Cantera said the following upon passage in April 2011, of the bill: "In a representative democracy, it is imperative that we continue to improve upon our elections process and optimize our citizens' opportunity to make their voices heard. I commend [Dennis] Baxley[, sponsor of HB 1355,] for producing a piece of legislation that will not only protect citizens' voting rights but also increase voter access."27 The Florida legislature's successful effort in 2011 to truncate the number of early voting days and required hours did not occur in isolation. Paralleling Florida's effort to reduce EIP voting, in June 2011, the Mary Ellen Klas, "Federal judges reject Browning's request to expedite ruling on voting law case." Quoted in Kilgore, "The Truth About Voter Suppression." Some of the legislation appears to have designed to eviscerate federal protections covered by Section 5 of the VRA. In defending their statutes, several states have gone further, expressly asking federal courts to strike down Section 5 of the VRA. One key case is Shelby County, Alabama v. Holder, for which associated documents are available at http://moritzlaw.osu.edu/electionlaw/litigation/ShelbyCountyAlabamav.Holder.php (last accessed April 17, 2012). 27 The press release with this quote is available at http://dennisbaxley.com/2011/04/ house-votes-to-improve-elections-process-protect-voter-rights (last accessed April 13, 2012). 26 25 9 Ohio state legislature enacted House Bill 194 which reduced by half the number of early voting days and eliminated early voting on Saturdays and Sundays.28 In Georgia, the legislature passed House Bill 92, which shortens the early voting period in Georgia and among other things permits counties not to have early voting on the Saturday preceding a General Election in the absence of a federal contest on the ballot.29 Beyond reductions in EIP, there have been as of late concerted efforts in many states to tighten voting laws and electoral codes.30 Who Votes Early In-Person? According to a recent report issued by the Pew Charitable Trusts, EIP voting continues to increase across the United States. Drawing on data from the Census Bureau's Current Population Survey, 8.4 percent of those who reported voting in 2010 said they did so casting an in-person early ballot, up from just 5.8 percent in the 2006 midterm election. Four states--Georgia, Nevada, New Mexico, and North Carolina--reported that their EIP voting rates have increased since 2006 by more than ten percentage points.31 Given this surge of interest in EIP voting, it is natural to inquire as to, among other things, who takes advantage of EIP voting laws and whether these laws lead to increased overall turnout (i.e., some voters who vote EIP would not have voted at all had EIP not been permitted) or simply redistributed turnout (i.e., early voters would have voted on Election Day in the absence of an early voting option). The literature on EIP and The text of Ohio House Bill 194 is available at http://www.legislature.state.oh.us/BillText129/ 129_HB_194_EN_N.html (last accessed April 13, 2012). On the referendum aimed at this bill, see Mary Wisniewski and Jo Ingles, "Referendum effort to challenge Ohio absentee voting limits," Reuters, November 22, 2011, available at http://www.reuters.com/article/2011/11/22/us-ohio-idUSTRE7AL2JU20111122 (last accessed April 13, 2012) and Jim Siegel, "Dem group's poll shows wide margin favor HB194 repeal," The Columbus Dispatch, February 21, 2012, available at http://www.dispatchpolitics.com/content/blogs/the-daily-briefing/2012/02/ 2-21-12-194-poll.html (last accessed April 13, 2012). 29 For details on Georgia House Bill 92, see the Georgia State Assembly's legislative history of this bill, available at http://www1.legis.ga.gov/legis/2011_12/sum/hb92.htm (last accessed April 13, 2012), and Jeremy Redmon, "Early voting window smaller," The Atlanta Journal-Constitution, February 25, 2012, available at http://www.ajc.com/ news/georgia-politics-elections/early-voting-window-smaller-1362031.html (last accessed April 13, 2012). 30 According to the National Conference of State Legislatures, legislation curtailing voter registration drives, reducing early voting days, requiring stricter voter ID requirements, and banning voting by ex-felons was introduced in 38 state legislatures and adopted in a dozen states. See National Conference of State Legislatures, "Absentee and Early Voting," 2011, available at http://www.ncsl.org/default.aspx?tabid=16604. See also, Michael Cooper, "New State Rules Raising Hurdles at Voting Booth," New York Times, October 2, 2011, available at http://www.nytimes.com/2011/10/03/us/ new-state-laws-are-limiting-access-for-voters.html (last accessed April 13, 2012). 31 Pew Charitable Trusts, "The Rise of Non-Precinct Place Voting," Election Data Dispatches, November 1, 2011, available at http://www.pewcenteronthestates.org/initiatives_detail.aspx?initiativeID=85899362969# Nov1 (last accessed April 13, 2012). 28 10 convenience voting addresses both of these questions, although often with research that involves elections that are not extremely prominent. In addition, entrants in the EIP literature often rely on aggregate turnout data from a single county or exit poll/post-election survey data. According to Gronke (2008), the conventional wisdom on early voting is that, "[It] does encourage turnout among regular voters for low-intensity contests, but it does not help solve the participation puzzle for new voters or those outside the system for reasons of disinterest, language, disability, or other burdens" (p. 450).32 Similarly, Berinsky (2005) argues that convenience voting reforms, including EIP voting, appear to make it easier for citizens who are already politically interested and engaged in the electoral process to cast their ballots: "Those lacking political interest remain nonvoters" despite EIP reforms insofar as "political engagement currently follows, rather than crosses, demographic divisions in the electorate" (p. 484). Citizens taking advantage of EIP voting tend to be likely voters who merely utilize early voting laws to move ahead the timing of when they cast their ballots. These citizens, according to this logic, were already motivated to vote and for them EIP voting just makes such a task that much more convenient. In a pioneering study of EIP voting, Stein (1998) found that strong partisans were more likely than other voters to utilize this procedure. Analyzing 1994 EIP voting data from Harris County, Texas, Stein found that strong partisans--especially Democrats--were more likely to cast EIP ballots than other registered voters. Nonetheless, Stein found little socio-demographic variance between those who reported casting an early ballot and those who reported voting on Election Day. Reasoning that if partisans tend to be more likely than nonpartisans to have already made up their minds on how to cast their ballots, Stein argued that early voting simply allowed them to express their civic (and partisan) duty sooner. Similarly, in his individuallevel study of voters who submitted mail-in ballots early to the supervisor of elections of Multnomah County, Oregon, in three elections between 2002 and 2003, Gronke (2008) found that partisans were significantly more likely to mail in their ballots early as were those living in zip-code areas with longer commute times (according to the 2000 United States Census). Other early voting research has drawn on self-reported survey responses to estimate the likelihood of a voter casting an early vote. In their study of EIP voters in Knox County, Tennessee, in the 1996 presidential election, Neeley and Richardson (2001) found few differences between respondents who, in a telephone 32 See Stein and Vonnahme (2011) for a recent review of the literature. 11 survey, said they voted early versus those who voted on Election Day, concluding that their results "provide no support for [a] mobiliz[ing] effect" (p. 389) of early voting and instead argue that early voting makes it easier for voters who would have voted regardless to vote. Similarly, in a study using exit poll data of EIP and Election Day voters in the 2004 General Election in Miami-Dade County, Florida, that was conducted between October 22 and November 2, 2004 (Election Day), Gronke and Galanes-Rosenbaum (2008) found that there were few differences between early voters and traditional in-precinct voters along strong partisan lines. They did find that Hispanics were more likely to utilize EIP voting, speculating that it was likely due to the mobilizing efforts by the Republican Party and Mel Martinez's United States Senate campaign. Relatedly, Gronke and Toffey (2008), who rely on early voters' self-reports in several national polls, reported some demographic differences among early and traditional Election Day voters at the descriptive level--namely, that early voters tend to be older, better educated, and more politically engaged--but these relationships largely disappeared in multivariate models. Of course, partisanship alone may not be enough to drive individuals to the polls to cast an early vote. As witnessed in Florida during the 2008 General Election, political parties, candidates, and other voting rights activists often intervened with voters, mobilizing citizens to the polls.33 Indeed, Stein, Leighley and Owens (2005) argue that, without the mobilizing efforts of candidates and political parties, early voting turnout in Texas's 2000 gubernatorial election would likely have been marginal. The motivation to vote EIP may not be "self-actuating" in that "[s]imply put, in-person early voting has been used by those who otherwise would have been most likely to vote on Election Day" (p. 11). Regardless of one's interpretation of the scholarly literature on turnout and convenience voting, it is fair to say that the literature says nothing about timing in EIP voting. Rather, extant studies essentially treat all early votes identically, whether they are cast on a weekend or a weekday. The literature is also silent as to the matter of the optimal length of time that early voting should last, whether weekends should be included, and so forth. See, for example, Michael Vasquez, et al., "Voters in critical Florida casting ballots as early voting begins," Miami Herald, October 20, 2008, available at http://www.mcclatchydc.com/2008/10/20/54452/ voters-in-critical-florida-casting.html (last accessed April 13, 2012) and Matthew Bigg, "Black vote could help swing key states to Obama," Reuters, October 23, 2008, available at http://in.reuters.com/article/2008/10/ 23/idINIndia-36109820081023 (last accessed April 23, 2012). 33 12 Research Design and Expectations Unlike previous studies of EIP in the American states, our dataset allows us to pinpoint precisely not only which Florida voters--identified by race/ethnicity, age, party affiliation, registration date, and prior voting history--were more likely to cast early in-person ballots in the 2008 General Election--but also to identify on which days they cast early ballots. In order to identify these patterns, we followed a two-step procedure. First, we assembled from public sources the complete set of 2008 General Election early voting files, one file for each of Florida's 67 counties.34 Each such early voting file contains a list of voters who voted early, and to the best of our knowledge these lists are comprehensive. Each early voting file consists of records, one record per early voter, and across the combined 67 early voting files there are 2,642,724 records of individuals who voted early in the 2008 General Election. Each record among the approximately 2.6 million total early voting records specifies the date on which a particular early voter voted, for example, October 23, 2008 (a Thursday) or perhaps November 2, 2008 (a Sunday). These dates are key to the analysis that follows. Besides dates of early voting, each record among the 67 early voting files is associated with a voter identification number. The records in the 67 early voting files do not specify voter demographics, e.g., voter race, voter date of registration, voter age, voter gender, and so forth. They also do not specify voter party registration or voter history, e.g., whether a voter cast a vote in the 2004 General Election. Second, we took a state voter file, created in March, 2010, and merged this file with the aforementioned 67 early voting files.35 The reason we took this step is because the Florida voter file contains demographic data on voters as well as voter histories. In our Florida voter file there are a total of 12,387,165 records, each record is associated with one registered voter, and each record has a voter identification number. This latter number allows us to associate records in our Florida voter file with records in the aforementioned early voting files. Given the presence of voter identification numbers in the 67 early voting files and in our overall Florida voter file, we successfully merged 2,592,061 early voting records from the 67 early voting files into the 34 The Florida Division of Elections maintains its repository of early voting files at https://doe.dos.state.fl.us/ fvrscountyballotreports/FVRSAvailableFiles.aspx. 35 The Florida voter file we use was created on March 15, 2010, from the VAN/VoteBuilder database maintained by the Florida Democratic Party. 13 main Florida voting file. This means that approximately 98 percent of voters whose identification numbers appear in the early voting files were successfully merged. We suspect that the 2 percent of early voters who were not so merged reflect people who moved out of Florida or died between the date of the 2008 General Election and the date that our voter file was created.36 We discovered a number of voters whose counties of residence in our overall Florida voting file did not match counties of residence based on the 67 early voting files. For these voters we assumed that the county of residence in the early files is correct. When our calculations required us to know the county in which a voter lived as of November 2008, we therefore used the latter. We suspect that differences in counties of residence between early voting files and the overall voting file reflects moving within Florida. As for our expectations about patterns in early voting, given the fact that anecdotal and media reports in vicinity of November 2008, suggested that Democrats, African Americans, younger, and first-time voters were more likely to cast EIP ballots in the 2008 General Election, we are somewhat dubious of much of the scholarly conventional wisdom regarding convenience voting as it pertains to the experience in Florida. First, we do not expect EIP voting to be uniformly distributed among partisans; rather, we expect to find Democrats disproportionately utilizing EIP voting compared to Republicans. Indeed, we agree with Stein, Leighley and Owens (2005) that the motivation to vote EIP is likely not "self-actuating." As such, due to the mobilization efforts of the Obama campaign and its allies, it is likely that the daily pattern of EIP voting over the two week period prior to Election Day will reveal differences among Democrats and Republicans as well as between partisans and nonpartisans. Based on contemporaneous reports from 2008, there is good reason to expect that Democrats voted disproportionately more on the weekends prior to Election Day 2008 and that nonpartisans who voted EIP were more likely to do so in the waning days of the early voting period.37 Second, we expect there to be considerable socio-demographic differences among EIP voters in Florida over the two week EIP voting period in the 2008 General Election. Not only do we expect early voters not This small, unmatched population is excluded from the analyses that follow. In the process of working with the combined Florida voter file and the 67 early voting files, we also discovered a small number of data errors. Among the 67 early voting files, for example, there are four records from Palm Beach County with dates of early voting from 2005. Obviously, such dates are wrong, and we ignored early voting records with clearly erroneous early voting dates. We also discovered records of individuals who, according to early voting files, voted early in the 2008 General Election but, according to the Florida voter file, did not vote early. For these records we assume that the early voting files are correct. 37 See, for example, Michael Luo and Ron Nixon, "More Democrats Casting Early Ballots, Data Show," New York Times, October 21, 2008, available at http://www.nytimes.com/2008/10/22/us/politics/22early.html (last accessed April 13, 2012). 36 14 to be reflective of the electorate as a whole, we have good reason to expect that not all socio-demographic groups will vote EIP consistently over the two week period. Rather, we anticipate finding that certain socio-demographic groups utilize EIP voting on different days. Specifically, we expect racial and ethnic minorities--especially African Americans--to vote early more often on the weekends, especially Sundays, given what is known about Get-Out-The-Vote efforts by the Obama campaign and its surrogate. In addition, and in keeping with much of the scholarly literature on EIP voting, we also expect older voters, those who have been registered for more than a year, and regular voters to cast EIP ballots earlier during the two week window. Finally, we expect younger, newly registered, and first-time voters to cast EIP ballots later during the two week period as they may have been activated to vote in the waning days of the 2008 early voting time period. African American voters in particular deserve special mention here because of the putative connections between early voting and religious services. Evelyn Garcia, as of July, 2011, the president of the Democratic Haitian-American Caucus of Florida, conveyed the conventional wisdom as to why African Americans are thought to have high early voting rates on Sundays when she stated, "We go to church on Sunday, and then we go together and early-vote. . . People try to help each other because transportation was a problem and knowing where to vote was a problem with some people who were new in the community."38 Findings We begin by comparing overall EIP rates from the 2008 General Election, first broken down by partisanship and then by racial and ethnic group. After explaining the patterns that we observe, we then disaggregate our results by day of early voting and explore how partisanship, race/ethnicity, and other important variables are associated with daily trends in EIP voting.39 Quoted in Frank Cerabino, "Early voting change might reduce black participation," The Palm Beach Post, July 6, 2011, available at http://www.palmbeachpost.com/news/ cerabino-early-voting-change-might-reduce-black-participation-1587638.html (last accessed April 13, 2012). See also Steven Gray, "Is Obama Doing Enough to Get Out the Black Vote?" Time, Wednesday, October 15, 2008, available at http://www.time.com/time/nation/article/0,8599,1850263,00.html (last accessed April 13, 2012). 39 The results that follow in this section are based on the previously described Florida voter file and the 67 county early voting files. Our turnout numbers in general do not match certified totals because we were only able to match approximately 98 percent of early voters with corresponding records in our Florida voter file. In addition, and as already discussed, we discovered errors in a very small number of early voting records, and this rendered this set of records unusable. 38 15 Comparing Non-Early and Early Electorates Third Party, 193,997 NPA, 934,520 Democrat, 2,157,743 Republican, 2,385,067 (a) Non-Early Electorate Third Party, 81,281 NPA, 375,986 Democrat, 1,346,018 Republican, 788,772 (b) Early Electorate Figure 1: Partisan Comparison of Non-Early and Early Electorates, 2008 General Election Figure 1 describes the partisanship of non-early and early voters, and a comparison of the two panels in the figure makes it clear that Democrats were more likely than Republicans to cast EIP votes in 2008. In particular, nearly twice as many registered Democrats, almost 1.35 million, voted EIP in 2008 compared to almost 800,000 registered Republicans. In the 2008 General Election, Democrats comprised nearly 42 percent of the registered voters in Florida, but they accounted for nearly 52 percent of the EIP voters. Repub- 16 licans, on the other hand, comprised approximately 35 percent of voters who were registered for the 2008 election, but they accounted for only approximately 30 percent of EIP voters. A sizable number of No Party Affiliation (NPA) voters cast EIP ballots--around 375,000 NPA voters--accounting for approximately 14.5 percent of all EIP ballots. In 2008, NPA registered voters accounted for around 19 percent of all registered voters in the state. Finally, approximately 3.1 percent of all EIP ballots were cast by voters registered with various third parties, and this percentage is nearly identical to the percentage of voters registered with third parties in the state in 2008. Other, 215,682 African American, 518,922 Asian, 71,339 Hispanic, 634,609 White, 4,262,190 (a) Non-Early Electorate Other, 104,878 African American, 575,699 Asian, 34,414 White, 1,581,594 Hispanic, 294,858 (b) Early Electorate Figure 2: Racial and Ethnic Comparison of Non-Early and Early Electorates, 2008 General Election 17 Turning now to the issue of whether racial and ethnic minorities--most notably African Americans-- were more likely to vote EIP than whites, we see in Figure 2 that, not only did African Americans cast more EIP ballots than they cast on Election Day, but also that African Americans accounted for a much greater proportion of the early voting electorate than they did on Election Day, Tuesday, November 4, 2008.40 Perhaps due to the early voting mobilization efforts by the Obama campaign and their allies which overtly targeted African Americans,41 Blacks ended up casting 22 percent of the total EIP votes in the 2008 General Election even though they comprised approximately 13 percent of the state's total registered electorate. Daily Trends in Early Voting As we have already emphasized, HB 1355 has implications for the days on which EIP ballots can be cast. In light of this we now disaggregate in a temporal way the above two figures and in so doing consider the dynamics of the EIP process. We focus in particular on when early voters actually cast ballots in the two week period before the 2008 General Election. Figure 3 plots over time when EIP ballots were cast, and several patterns in this figure are notable. As is evident by the solid dots in it, there were more 2008 General Election early voters in the second week of early voting than in the first. In addition, in the early voting period there were what one might call weekend effects. Namely, in both the first and second weeks of early voting, the number of early voters was smaller on Saturday and Sunday than on the preceding weekdays. The Friday to Saturday dropoff was relatively larger in the first week than in the second. However, the Saturday to Sunday dropoff was greater in the second week of early voting. The drop shown in Figure 3 of the number of early voters on the two Sundays of early voting could be thought of as visually misleading because only ten counties in Florida offered early voting on Sunday during the run-up to the 2008 General Election.42 To this end, Figure 4 is analogous to Figure 3 but is based only The pie charts presenting in Figure 1 are based on county-level early voting files. Approximately 98 percent of all early voters from the 2008 General election are represented in Figure 1b, and we do not know the racial and ethnic identities of the remaining approximately 2 percent, coded "Other." 41 Luo and Nixon, "More Democrats Casting Early Ballots, Data Show." During a campaign stop at a barbershop in Ft. Lauderdale, Florida, then-Senator Obama intoned, "Whoever comes and sits in that chair, tell them to early-vote. No excuses." 42 These counties were Bradford, Broward, Dixie, Duval, Jackson, Miami-Dade, Palm Beach, Pinellas, Sarasota, and Seminole. We determined the counties that offered Sunday voting based on their official filings to the Florida Secretary of State. See, Florida Secretary of State, Division of Elections, "Early Voting Sites by County: General Election - November 4, 2008," available at http://election.dos.state.fl.us/pdf/CountyEarlyVotSitesGen08.pdf (last accessed April 13, 2012). 40 18 Figure 3: Counts of Early Voters, 2008 General Election 300000 q q q 250000 q Total Early Voters 200000 q q q 150000 q q q q q 100000 q 50000 q 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date on the counties that had early voting on Sundays. The patterns in this figure are similar to those seen for all 67 Florida counties except for the fact that, among the counties that offered early voting on Sunday, early voting totals on Saturday are not as different from Sunday totals as they are across all of Florida. We now take our aggregate results from Figures 3 and 4 and break them down by voter type, and our first slicing of the data in this way produces Figure 5, in which we plot EIP voting by party registration. This figure describes for each partisan group of EIP voters the composition of each group that voted on a given day. "Composition" means here, and in corresponding figures that appear later, the fraction of early voters on a day that is of a particular partisanship. By construction, the four partisan fractions on a given day sum to one (i.e., to 100 percent). Recall that these numbers are based on 98 percent of all voters who voted early in the 2008 General Election. Because the Democratic points in Figure 5 (solid squares) are above the other types of points in the figure, we know that, on all early voting days, the largest voter group in terms of partisanship consisted of Democrats. Moreover, weekdays notwithstanding it is also clear from Figure 5 that, in relative terms, registered Democrats were more likely to cast EIP ballots on the first and second weekends of early voting and 19 Figure 4: Counts of Early Voters, Counties with Sunday Voting, 2008 General Election 120000 q q q Total Early Voters 100000 q q 80000 q q q 60000 q q q q 40000 q q 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date especially on Sundays. In other words, the general pro-Democratic bias within the early voting electorate was accentuated on Sundays. Early-voting Republicans, on the other hand, were relatively less likely to cast early ballots on the weekends, especially on the two Sundays of the early voting period. Although the fraction of early ballots cast by those registered with a third party remained constant over the two week period, there is some evidence in Figure 5 that NPA individuals who voted early cast a disproportionate share of their ballots in the waning days of the early voting period. Note in particular the general upward slope of the NPA points in the figure. Figure 6 is similar to the aforementioned figure except that it describes the composition of the early voting electorate by race/ethnicity (as opposed to partisanship) as it varied from October 20, 2008, through November 2, 2008.43 One can see from the black squares in Figure 6 that, on all early voting days, whites made up the largest racial and ethnic group of early voters. African Americans were the second largest group, again on all early voting days, followed by Hispanics and then Asians. 43 We examine the breakdown of race/ethnicity along four categories, White, Hispanic, Asian, and African American. In our race/ethnicity analyses, i.e., Figure 6, we ignore voters who opted not to report race/ethnicity. 20 Figure 5: Partisan Composition of Early Voters, 2008 General Election 1.0 q Composition Early Voters 0.4 0.6 0.8 Democrat (1,345,635) Republican (788,660) Third Party (81,265) No Party (375,883) q q q q q q q q q q q q q q 0.0 0.2 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date Nonetheless, despite the general dominance of white voters in the early voting electorate, the extent to which whites comprised the largest racial or ethnic group varied by day, mostly obviously on Sundays. In particular, the greatest drop-offs in white early voting rates occurred on Sundays and on these days the fraction of African American early voters jumped. There is a slight decline in the composition of white early voters on Saturdays, but Figure 6 makes it clear that the Saturday effect is not nearly as pronounced as the Sunday effect. Another perspective on racial and ethnic trends across early voting days can be seen in Figure 7. This figure describes, for each of the four groups racial and ethnic already mentioned, the fraction of each group that voted on a given day. For each racial and ethnic group in Figure 7 the sum of the early voting fractions across the 14 days of early voting is one (i.e., 100 percent). In Figure 7 we see a notable view of racial and ethnic early voting trends. In particular, white early voters tend to vote in the first half of the early voting period, not including Sundays. This latter point is clearly evident in Figure 7 in the fact that, on both Sundays of early voting, the solid black squares are the lowest of the four symbols. Across all early voting days, the two days that featured the lowest white 21 Figure 6: Racial and Ethnic Composition of Early Voters, 2008 General Election 1.0 q Composition Early Voters 0.4 0.6 0.8 White (1,581,594) Hispanic (294,858) Asian (34,414) African American (575,699) q 0.2 q q q q q q q q q q q q q 0.0 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date participation rates, relatively speaking, were both Sundays. In contrast, on the first Sunday of early voting, the racial and ethnic group with the highest relative participation rate was African American voters. And on the last Sunday, the group with the highest relative participation rate was Hispanic voters, followed by African American voters.44 Finally, on the two Saturdays of early voting, the racial and ethnic group with the highest participation rate was Asian voters. We cannot explain definitely with our voting data why the members of a particular racial or ethnic group might have a preference for early voting on a given day of the week. Nonetheless, the literature we have reviewed has described African American early voter mobilization efforts and in particular efforts associated with churches. These efforts would be expected to produce Sunday spikes in black early voting, and in fact According to the official early voting reports filed by all 67 Florida counties to the Florida Secretary of State, the ten counties that reported voting on a Sunday in the 2008 General Election early voting period were Bradford, Broward, Dixie, Duval, Jackson, Miami-Dade, Palm Beach, Pinellas, Sarasota, and Seminole. Some of the ten counties that filed early voting reports for either or both of the Sundays differ from what the Florida Secretary of State reported on its website. See Florida Secretary of State, Division of Elections, "Early Voting Sites by County: General Election - November 4, 2008." In a separate analysis, which is not pictured here, we analyze early voters from the ten so-called Sunday counties. In these counties, white early voters cast ballots disproportionately less often on Sundays compared to Hispanic and African American early voters. This pattern holds as well on the Saturday before the 2008 General Election and was true for African American albeit not Hispanic early voters on the middle Saturday of the 2008 early voting period. 44 22 Figure 7: Racial and Ethnic Trends of Early Voters, 2008 General Election 0.25 q Fraction Early Voted 0.15 0.20 White (1,581,594) Hispanic (294,858) Asian (34,414) African American (575,699) q q q 0.10 q q q q 0.05 q q q q q q q 0.00 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date this is what we see in Figures 6 and 7. As we already noted, only ten counties in Florida offered early voting on Sundays. We now show plots analogous to the above Figures 6 and 7 but restricted to the ten so-called Sunday counties. In particular, Figure 8 shows trends in the composition of the early voting electorate in this group of counties, and we see here a familiar drop in weekend white early voting. The drop is not as pronounced as in the earlier Figure 6. Similarly, Figure 9 displays early voting trends among voter types as broken down by racial and ethnic group. Notice here that the same white voter pattern is apparent as was seen in Figure 7, i.e., white voters disproportionately vote less often on weekends and in particular on Sundays. In Figure 9 one sees as well that Asian early voters voted disproportionately on Saturdays and that Sundays tended to feature disproportionate numbers of Hispanic and African American voters. Having established that there was daily variance across the 2008 early voting period in the rates at which partisan and racial and ethnic groups voted, we now turn to the issue of age. To consider whether there is a relationship between age and date of early voting, we partition all early voters into three classes: those 22 years old and younger as of November 4, 2008, those 22 to 65, and then those 65 years and older. Our 23 Figure 8: Racial and Ethnic Composition of Early Voters, Counties with Sunday Voting, 2008 General Election 1.0 q Composition Early Voters 0.4 0.6 0.8 White (498,040) Hispanic (193,973) Asian (15,276) African American (316,965) 0.2 q q q q q q q q q q q q q q 0.0 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date age-based results are summarized in Figure 10, and this figure shows that the age-based composition of the early voting electorate varied dramatically across the 14 days of early voting in 2008. Namely, it is always true that the largest group of voters is the middle group, and this is hardly surprising given the age ranges considered here: the middle age group spans over forty years in contrast to the young group, which spans only five. Even so, Figure 10 has one clear implication: compared to mid-week voting patterns, older early voters vote relatively infrequently on weekends.45 If we look at the fraction of each age group as it varies across the 14 days of early voting--see Figure 11--we see that older voters tend to vote in the early part of the early voting period, very infrequently on Sundays, and to some extent on Saturdays. Young voters, in contrast, tend to vote toward the end of the early voting period and disproportionately often on weekends. Why would older voters prefer, ceteris paribus, to vote in the first half of the earlier voting period as op45 Our data are not broken down by retirement status. This is unfortunate as one interpretation of Figure 10 is that it reflects the fact that older Florida voters are disproportionately retired and thus have the time in the middle of the week to vote early. Regardless of the causal explanation for Figure 10, it is clear that old voters disproportionately prefer to vote on weekdays rather than weekends. 24 Figure 9: Racial and Ethnic Trends of Early Voters, Counties with Early Voting, 2008 General Election 0.25 q Fraction Early Voted 0.15 0.20 White (498,040) Hispanic (193,973) Asian (15,276) African American (316,965) q q q q q 0.10 q q 0.05 q q q q q q q 0.00 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date posed to its second half? Our data do not offer an answer to this question, although one could speculate that younger voters have less free time during the traditional work-day than their older counterparts. Regardless, there are clear age-related trends in Figures 10 and 11, and this by itself, regardless of the explanation for why older and younger voters have different early voting preferences, implies that changes to Florida's early voting laws will have differential effects on voters as defined by their age classes. Related to age is date of registration, and registration date is politically notable because of the possibility that new registrants may vote during particular time periods during an early voting window. We thus divide our set of early voters into two groups: those whose year of registration was prior to 2008 and those whose year of registration was 2008. It is apparent from Figure 12 that early voters registered prior to 2008 were relatively more likely to cast a ballot during the first week of early voting. And, we see as well that early voters who registered in 2008 were relatively more likely to vote in the final week of early voting, especially on the final four days, including the final Sunday before Election Day. Finally, with respect to General Election vote histories of voters who voted early in the 2008 General Election, it is clear from Figure 13 that first-time, early voters tended to wait until the end of the early voting 25 Figure 10: Age Composition of Early Voters, 2008 General Election 1.0 q 22 and Under (122,362) Between 23 and 65 (1,874,741) 65 and Over (594,340) q q q q q q q 0.8 Composition Early Voters q q q 0.6 q q q q 0.0 0.2 0.4 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date period to cast their votes, compared to non-first-time, early voters. Note that Figure 13 includes only those early voters who were registered as of October 6, 2000. The number of "Prior Votes (see the legend for Figure 13) is the number of General Elections in which a voter voted, starting with 2000 and ending with 2006. 26 Figure 11: Age Trends of Early Voters, 2008 General Election 0.25 q 22 and Under (122,362) Between 23 and 65 (1,874,741) 65 and Over (594,340) Fraction Early Voted 0.15 0.20 q q 0.10 q q q q q q 0.05 q q q q q q 0.00 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date Figure 12: Date of Registration Trends of Early Voters, 2008 General Election 0.25 q Reg. in 2008 (259,872) Reg. prior to 2008 (2,331,571) Fraction Early Voted 0.15 0.20 0.10 q q q q q q q q q q q q 0.05 q q 0.00 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date 27 Figure 13: Voting History Trends of Early Voters, 2008 General Election 0.25 q 0.20 q Prior Votes 0 (40,986) Prior Votes 1 (75,713) Prior Votes 2 (141,149) Prior Votes 3 (269,353) Prior Votes 4 (807,995) Fraction Early Voted 0.15 q q q 0.10 q q q q q q q q q q q q q q q q q q q q q 0.05 q q q q 0.00 10/20 M 10/21 T 10/22 W 10/23 Th 10/24 F 10/25 Sat 10/26 Sun 10/27 M 10/28 T 10/29 W 10/30 Th 10/31 F 11/01 Sat 11/02 Sun Date 28 Conclusion The research described here was motivated by the recent passage in Florida of a piece of legislation, House Bill 1355, that restricts the early voting period in this state from what existed during the 2008 and 2010 General Elections. We sought to understand whether the new Florida law might have disproportionate effects on some Florida voters, and, based on our analysis of voting patterns from the 2008 General Election, we find that it will.46 In particular, we find clear differences between the partisanship of 2008 General Election EIP voters and the partisanship of corresponding non-early voters. We also show that the racial and ethnic breakdown of early voters in 2008 was quite different than the racial and ethnic breakdown of non-early voters. As such, it follows that changes to Florida's early voting laws will affect certain partisan groups of voters more than others and certain racial and ethnic groups as well. The evidence we have adduced suggests in particular that Democrats and African American voters will be disproportionately affected by new Florida legislation that changes early voting in Florida because, simply put, these two groups tend to vote early more than other partisan and racial and ethnic groups, respectively. We have described how Florida's recently passed HB 1355 reduced the number of early voting days in the state, cut in half the required number of early voting hours, and eliminated early voting on the final Sunday before Election Day. If the 2008 General Election is a guide as to how early voters would have sorted themselves in 2012 had HB 1355 not existed, then certainly the changes wrought by this piece of legislation will have (and perhaps have had, as the 2012 Presidential Primary is over as of the writing of this article) differential effects on Florida's electorate. Even though, for instance, African Americans comprised only 13 percent of total voters and 22 percent of early voters in Florida in the 2008 General Election, they accounted for 31 percent of early voters on the final Sunday of early voting. While older and more regular voters enjoyed the convenience of voting early in the first week of early voting, younger, first-time, and newly registered voters disproportionately chose to vote toward the latter half of the early voting period and often on the final Sunday of early voting. We note as well that these Florida-wide conclusions are also evident in the five Florida counties subject to preclearance under Section 5 of the Voting Rights Act. The 2012 Florida Presidential Preference Primary took place on January 31, i.e., after HB 1355 became law. This article is being written in late spring, 2012, and hence effects of HB 1355 may have already been felt. 46 29 It is important to acknowledge several caveats pertaining to our findings. First, our study analyzes EIP voting in a single election, albeit a historical one, in just one state. It could be that the 2008 General Election was anomalous and that patterns in EIP voting in Florida are not representative of what one would observe in other states.47 This caveat can be easily resolved if scholars were to replicate the type of Florida analysis we have presented here using other states as laboratories. Second, given the recency of the expansion of Florida early voting opportunities and then the very proximate contraction, we cannot assess at this time whether EIP voting is habitual, with voters becoming accustomed to voting early and perhaps on a particular day, from one election to another. In other words, we are unable to determine using our data whether early voters in one election necessarily remain early voters in subsequent elections. Third, although our data are at the individual-level, they do not allow us to determine the motivations of those who voted EIP or why they did so on a particular day. For the most part, we leave that issue to other scholars. Fourth, we note that our analysis speaks to the issue of early voting timing in the run-up to the 2008 General Election but not to the question of turnout and in particular the matter of whether voters who have voted early in past elections like the previous presidential contest will abstain from voting in elections governed by HB 1355. Suppose, that is, a resident of Florida wants to vote early on the Sunday before an election but cannot because of this new law. Will this individual vote early on a different day, vote on election day, or simply not turn out to vote at all? The literature on voting in American elections is silent on this question as are we at this time. The effect of changes in convenience voting on voter turnout is not something that is known, and it remains an important matter for future research. We conclude with the suggestion that changes to convenience voting laws, including but not limited to the truncation of EIP voting in Florida, may have considerable effects in future elections. As Richard Hasen notes, "These laws will have an effect on the margin on who votes. And in a state like Florida, a small difference matters. It could easily decide the outcome."48 Whether or not one believes that the Florida legislature's effort to restrict EIP voting in anticipation of the 2012 General Elections parallels "methods See Gronke, Hicks and Toffey (2009) for comments on generalizing from the 2008 General Election. Quoted in David G. Savage, "Election laws tightening in GOP-run states," Los Angeles Times, October 30, 2011, available at http://www.latimes.com/news/nationworld/nation/la-na-vote-florida-20111031,0, 1962738.story (last accessed April 13, 2012). 48 47 30 pioneered by the white supremacists from another era that achieved the similar results," as Risa Goluboff and Dahlia Lithwick contend,49 it very well could negatively impact turnout among Democratic, minority, younger, occasional, and first-time voters in the Sunshine State. References Berinsky, Adam J. 2005. "The Perverse Consequences of Electoral Reform in the United States." American Politics Research 33(4):471-491. Gronke, Paul. 2008. "Early Voting Reforms and American Elections." William & Mary Law Review 17(2):423-452. Gronke, Paul and Daniel Krantz Toffey. 2008. "The Psychological and Institutional Determinants of Early Voting." Journal of Social Issues 64(3):503-24. Gronke, Paul and Daniel P. Tokaji. 2011. "The Party Line." Election Law Journal 10(2):71-72. Gronke, Paul and Eva Galanes-Rosenbaum. 2008. The Growth of Early and Non-Precinct Place Balloting: When, Why, and Prospects for the Future. In America Votes! A Guide to Election Law and Voting Rights, ed. Benjamin E. Griffith. Chicago, IL: American Bar Association. Gronke, Paul, James Hicks and Daniel Krantz Toffey. 2009. "N=1? The Anomalous 2008 Election and Lessons for Reform." Paper presented at the Annual Meeting of the American Political Science Association, Toronto, Canada. URL: http://papers.ssrn.com/sol3/papers.cfm?abstract id=1451900 Levitt, Justin. 2011. "New State Voting Laws: Barriers to the Ballot?" Testimony of Professor Justin Levitt, Loyola Law School, September 8, Los Angeles, before the United States Senate Committee on the Judiciary Subcommittee on the Constitution, Civil Rights and Human Rights. URL: http://www.judiciary.senate.gov/pdf/11-9-8LevittTestimony.pdf Levitt, Justin. 2012. "Election Reform: The Pursuit of Unwarranted Electoral Regulation." Election Law Journal 11(1):97-117. Neeley, Grant W. and Lilliard E. Richardson. 2001. "Who is Early Voting? An Individual Level Examination." The Social Science Journal 38(3):381-392. Stein, Robert M. 1998. "Early Voting." The Public Opinion Quarterly 62(1):57-69. Stein, Robert M. and Greg Vonnahme. 2011. "Voting at Non-Precinct Polling Places: A Review and Research Agenda." Election Law Journal 10(3):307-311. Stein, Robert M., Jan Leighley and Christopher Owens. 2005. "Who Votes, Who Doesn't, Why and, What Can Be Done?" A Report to the Federal Commission on Electoral Reform. URL: http://www1.american.edu/ia/cfer/0630test/stein.pdf Risa Goluboff and Dahlia Lithwick, "A Fraudulent Case: The ugly parallels between Jim Crow and modern votesuppression laws," Slate, October 20, 2011, available at http://www.slate.com/articles/news_and_politics/ jurisprudence/2011/10/voter_id_laws_their_proponents_should_have_to_answer_for_the_ ugl.html (last accessed April 13, 2012). 49 31 The Florida Senate. 2010. "The Effect of Early Voting on Voter Turnout in Florida Elections; 2010 Update." Interim Report 2011-118, Committee on Ethics and Elections. URL: http://www.flsenate.gov/Committees/InterimReports/2011/2011-118ee.pdf Weiser, Wendy R. and Lawrence Norden. 2011. "Voting Law Changes in 2012." Report published by the Brennen Center for Justice. URL: http://brennan.3cdn.net/92635ddafbc09e8d88 i3m6bjdeh.pdf 32