As of 2015, 36 states and the District of Columbia allow any registered citizen to cast a vote before Election Day through the mail or at designated locations (National Conference of State Legislatures 2015). Early voting programs have become a significant feature of American elections. In both 2008 and 2012, more than 30% of the electorate cast ballots before Election Day, including a majority of voters in at least nine states in 2012 (Demos 2014; McDonald 2012). Numerous surveys indicate that voters appreciate the convenience offered by early voting (Kasler 2014; Martens 2014; Warner 2013). Some voters also recognize that voting before Election Day means an early end to the barrage of mailings, phone calls, and canvassing visits directed by parties and candidates (Benac 2010).
States offering early voting do so not only in general elections, but in primary contests as well. While national figures are difficult to obtain, a substantial number of voters cast early ballots during both the 2008 and 2012 presidential nomination cycles. In fact, a majority of primary voters participated early in California and Texas in both 2008 and 2012 (Austin Community College 2010, 6; California Secretary of State 2012; Reynolds 2014). (1) Early voting has become significant enough in presidential primaries that exit pollsters now routinely conduct telephone surveys of early voters to supplement their primary day surveys (Roper Center 2008). (2)
Most scholarly attention devoted to early voting has focused on the degree to which programs increase voter turnout. While many have argued that early voting fails to bring new voters to the polls (Burden et al. 2014; Fitzgerald 2005; Gronke, Galanes-Rosenbaum, and Miller 2007; Larocca and Klemanski 2011; Primo, Jacobsmeier, and Milyo 2007; Wolfinger, Highton, and Mullin 2005), others have reported that programs with abundant sites (Fullmer forthcoming; Losco, Scheele, and Hall 2010; Neeley and Richardson 1996) or sites in convenient areas (Stein and Garcia-Monet 1997) can improve turnout by several percentage points or more.
Research has largely ignored the question of whether early voting alters the information environment in campaigns. Those who vote early may do so before important information becomes available in the final days and weeks of a campaign. Many have speculated that early voting may particularly affect presidential nomination contests, as support for candidates tends to be more unstable and the process often commences before candidates fully launch campaigns in respective states. I speculate that early voting should benefit early front-runners in these contests, as voters may cast early votes for these candidates before fully considering their less-known opponents. Examining exit-poll data from the 2008 Democratic primaries between Senators Hillary Clinton and Barack Obama, I find that Clinton indeed benefitted from early voting in several early primary states.
When voters submit ballots early, they complete their civic duties ahead of schedule. With votes already cast, political campaigns essentially conclude weeks early for these citizens. There is no mechanism by which one can change a submitted vote; early voters can only wait and observe whether election outcomes (reported on Election Night or primary day) match their preferences. Of course, campaigns themselves do not end early. Candidates continue to fund television ads, participate in debates, and grant interviews and news conferences. The news media investigate voting records and past associations. Parties and outside groups remain active as well. Lacking any knowledge of late campaign developments, those who choose to vote early ultimately do so without the same information as Election Day voters.
Late information comes in many forms during political campaigns. The term October Surprise has become synonymous with a news event holding the potential to influence an election's outcome. Examples can be found throughout American political history. In the days leading up to the 1968 presidential election, the Johnson administration (for whom Hubert Humphrey, the Democratic candidate, was vice president) announced a halt of the bombing campaign in North Vietnam; Humphrey quickly gained in the polls, though Republican Richard Nixon ultimately prevailed (Tucker 2011). In late September 2008, as Americans were beginning to vote early in the presidential campaign, the country watched closely as Obama and John McCain navigated the collapse of Lehman Brothers and an ensuing financial panic. Obama was widely credited for his response, even by Republican President George W. Bush (2010). McCain, however, was criticized for seeming erratic and reactionary (Henninger 2008).
More generally, debates have long been considered an important factor in vote selection and a major conveyor of candidate differences in presidential campaigns (Schroeder 2001; Shenkman 2004). In 2012, the four debates did not commence until October 3. By this point, seven states had already begun accepting early ballots through the mail and at designated sites. (3) By the final debate on October 22, the only one devoted to foreign affairs, 26 states had begun early voting (National Association of Secretaries of State 2012).
Late information is also prevalent during presidential nominating campaigns. Modern nomination races feature over 50 state primaries and caucuses held sequentially between January and June of presidential election years. Candidate fields are typically large in the early stages of a contest, though poor performances and insufficient fundraising often cause candidates to withdraw from the race before all state contests have concluded (Aldrich 1980; Bartels 1988; Brams 1978; Matthews 1978; Norrander 1996; Shafer 1988). As a result, early voters may cast votes for candidates who ultimately quit before the state's respective primary day even occurs. Analyzing the 2008 California presidential primary, Meredith and Malhotra (2011) find that precincts with more early voters gave a significantly greater share of their vote to John Edwards, Rudy Giuliani, and Fred Thompson, three candidates who were actively seeking the nomination when the early voting period began (on January 7) but dropped out before the primary date (February 5).
Further, important information can emerge about nomination candidates who remain in the race. In 1992, Bill Clinton was accused of having a 12-year extramarital affair in the weeks before the New Hampshire primary (Sabato 1998). In 2008, Obama was forced to address controversial comments made by his pastor, Jeremiah Wright, during his campaign against Clinton (Ross and El-Buri 2008). More generally, presidential nominations are fluid contests, as candidates are often relatively unknown to large chunks of voters until the final days and weeks before a state's primary or caucus. With the exception of Iowa and New Hampshire, where campaigning begins as much as one year before the contests, most states see high levels of campaign activity only in the week or so before their respective primary or caucus. Candidates can only be in one place at a time, while other campaign resources (i.e., money, staffers) are also limited. Traditionally, winning (or exceeding expectations) in the next primary or caucus produces momentum in the form of positive media coverage and, perhaps most importantly, financial contributions (Abramowitz 1989; Aldrich 1980; Bartels 1985, 1988; Geer 1989; Lichter, Amundson, and Noyes 1988). The ability to win a state contest and slowly generate momentum and resources is seen to benefit poorly funded, low-name-recognition candidates, as they are able to slowly build credible campaigns via retail politics in a single state. However, if a voter has already cast an early ballot, then new information about candidates cannot be incorporated into one's vote choice.
Researchers have found that early voters tend to be more politically astute and partisan (Baretto et al. 2006; Dubin and Kalsow 1996; Patterson and Caldeira 1985; Stein 1998; Stein et al. 2004), leading some to suggest that additional information may not alter their preference. But a deep literature also shows that people tend to be overconfident in decision making (Barber and Odean 2001; Camerer and Lovallo 1999; Svenson 1981). More specifically, voters routinely underestimate the likelihood that their preferences will shift (Meredith and Malhotra 2011) and do become more likely to support a candidate once the media determines that they are viable (Abramowitz 1989; Polsby and Vildavsky 2008).
Many political operatives, journalists, and academics have suggested that early voting produces information asymmetries that may affect primary outcomes. Political analyst Craig Wilson believes early voting can easily produce voter regret, stating, "One downside to early voting is that once you cast that vote, there's still two weeks or a month to go and what happens if something eventful happens with a campaign or a candidate during that period and you change your mind" (quoted in in Domurat 2008). Patt Morrison of the Oakland Tribune agreed, stating "Casting an absentee ballot so far ahead of Election Day is like picking a Super Bowl winner based on who's ahead at halftime. It's like recommending a book you've only halfway read. It's like getting married on the first date" (quoted in in PBS 2008). Paul Gronke, director of the Early Voting Information Center at Reed College, has argued that early voters may miss information that could affect their votes (NPR 2008). Robert Stein, a longtime early voting researcher, offered a similar perspective, suggesting that Clinton did better among early voters in some 2008 primaries because "her polling numbers were higher" when early voting began (quoted inin Hylton-Austin 2008).
I posit that early voting produces information asymmetries that can affect voting behavior in...