Over a decade has passed since Congress passed the Help America Vote Act (HAVA) in 2002, largely in response to the problems experienced in Florida during the 2000 presidential election. Despite the appropriation of more than $3.5 billion, there are still persistent problems with the administration of elections in the United States. Some encompass technical problems involving computerized voting equipment, voter lists, and the software used to manage them, but many others are rooted in more traditional issues of candidates and political parties attempting to advantage themselves at the expense of their opponents. Indeed, contemporary elections provide many opportunities for those participating in election administration to manipulate the system intentionally for partisan advantage or to introduce inconsistencies across jurisdictions as a result of the discretion granted to state and local officials. This study focuses on a key area where partisan manipulation and local variation may be most prominent: provisional ballots.

Provisional ballots are important because they offer voters who otherwise would be denied the right to vote an opportunity to cast a ballot – at least in many cases. The Caltech/MIT Voting Technology Project (2001) provided one of the first arguments in favor of provisional voting and estimated that provisional ballots might have saved 1.5 million votes from being lost in 2000.

On November 4, 2008, my drive to day care with my four-year-old son and two-year-old daughter started off as it generally does. We saw the usual set of things that capture their attention – cars, trucks, buses, and dogs being walked. These sightings sparked the typical set of questions they ask, such as: “Can I buy that car someday?” and “Can I buy a dog when I'm bigger?” But outside a Metro station, while stopped at a red light just before our last turn, we witnessed something we had not seen before. Three young women were chanting and waving Obama/Biden signs. Not surprisingly, this generated a new question: “What are they doing, Daddy?” In answering, I said that it was election day and the women were expressing their support for Barack Obama. Then, as a man and a woman who were walking together approached the three Obama supporters, the man joined in. His partner jumped a few steps into the street and snapped a photo of the group. This activity inspired a fresh line of questioning, first from my son, who asked: “What is the man doing? Why did they take a picture? Is it a party?” Upon hearing the word “party,” my daughter asked hopefully: “Is it my birthday party?” Some confusion from my two-year-old notwithstanding, the excitement on election day was palpable.

If turnout in the United States was ever going to reach new heights, the conventional wisdom was that 2008 was the year for it to happen.

The results presented thus far demonstrate that recent registration reforms have not led to substantial increases in turnout, nor have they amplified the voices of those who are the least likely to vote beyond a blip on the volume meter. In order to address concerns that variations in the effect of election day registration (EDR) across contexts could be the result of poor implementation in the states that adopted it in response to the National Voter Registration Act (NVRA) and/or differences in the strategies political parties have used to mobilize unregistered but eligible citizens, this chapter reports on interviews I conducted with state election officials and state Democratic and Republican party leaders across the EDR states. Systematic data on EDR implementation and party responses, to the best of my knowledge, are simply not available from existing sources. Although the National Election Studies (NES) provides information regarding party mobilization efforts, asking respondents whether or not they were contacted by the political parties, the NES was designed to provide a representative sample of the nation, not individual states. Moreover, the sample size for the EDR states, which have small populations, is insufficient to allow one to draw inferences; in fact, in some study years, individuals in the less populous EDR states are not even sampled. As a result, interviews with both election officials and party leaders are used to provide insight into the implementation of EDR and how the parties' strategies are shaped by its availability.

The essence of Arend Lijphart's (1997) Presidential Address to the American Political Science Association was that who votes matters – for the types of candidates that get elected to office, the policies they enact (see Hill and Leighley 1992), and as an indicator of the health of our democratic society. Lijphart began his address by stating: “Political equality and political participation are both basic democratic ideals” (1997, p. 1). The results offered in Chapters 3 and 4 reveal that relaxed registration laws will not take us as far as reformers might like toward increased participation, the second of these ideals. Motivation, not costs, represent the most significant barrier to higher turnout in the United States. But what have registration laws done to address political equality? Even though the increases in turnout resulting from EDR and motor voter failed to reach the heights imagined by some, previous scholarship has not come to a firm conclusion regarding whether the turnout rates of some types of individuals are more likely to grow than others and, if there is variation, which types of individuals are expected to gain the most ground.

The answer has important implications. If the effects are concentrated among those who tend to be left out of politics, then the conclusion that EDR and motor voter have a limited effect might require revision.

America's status at the bottom of international turnout rankings has shaped the study of voter turnout in the United States. Although it should be quite obvious that a single explanation for low turnout does not exist, researchers have devoted substantial attention to the role of institutions in structuring the rules regarding who registers and votes and the norms relating to whom the parties contact. Scholars and policy experts remain attracted to institutional reforms as solutions because of the collection of concerns that arise out of the decentralized nature of election administration in the United States, coupled with the tools available to make changes. After a review of the nature of these concerns, I highlight the flaws in the existing theoretical approach to the study of the effects of registration laws on voter turnout, discuss the methodological pitfalls that follow, and present a theoretical framework and research design that better reflect the complexities of political behavior in the contemporary United States.

OPPORTUNITY AND VOICE

In the United States, out of the roots of federalism, a sprawling election administration system has sprouted, with roughly 10,000 officials responsible for conducting elections at the state, county, and municipal levels. With the implementation of the National Voter Registration Act of 1993 (NVRA) and Help America Vote Act of 2002 (HAVA), the federal government brought a greater degree of standardization to the electoral process.

In Chapter 3, I demonstrated that the effect of election day registration (EDR) differs across social and political contexts. In the already high turnout states of Minnesota and Wisconsin, EDR led to a small but significant increase in turnout. The effect of EDR in Idaho, New Hampshire, and Wyoming, states that adopted EDR to avoid federal motor voter legislation, fell short of expectations based on previous studies and the mark set by Minnesota and Wisconsin. These conclusions were derived primarily by using the difference in difference method, a method that aims to control for endogeneity by comparing turnout over time in comparable states. The purpose of this chapter is to estimate the effect of motor voter as implemented via state law and via the National Voter Registration Act of 1993 (NVRA). The states with pre-NVRA laws that were deemed suitable for analysis are Michigan, Minnesota, Nevada, and North Carolina.

As discussed in Chapter 1, although some scholars have treated EDR and motor voter as substitutes (Teixeira 1992; Highton and Wolfinger 1998), the policies must be investigated separately. EDR and motor voter policies differ mainly in terms of the costs imposed on voters, but also in terms of the likelihood that interest in voting drives the decision to register, and the ease with which political parties and candidates can contact registrants, all of which lead to differential expectations with respect to the respective policy's effect on turnout.

Previous chapters have shown that the selection of election laws is purposeful, that the overall effects of the two registration reforms projected to have the largest impact on turnout have resulted in much smaller than anticipated (and smaller than previously estimated) increases in turnout, and that the magnitude of the effects varies based on the social and political contexts into which the reforms are placed. That is, lowering the costs of participation does nothing directly, and apparently very little indirectly (I take this up in Chapter 7) to engage those who do not want to vote. Having estimated the overall effects of election day registration (EDR) and motor voter laws, as implemented by the states on their own and via impetus from the federal government, the present chapter provides a bridge to the analysis of how the effects vary across individual characteristics. In this chapter, I explore what can be learned from self-reports of registration behavior when individuals are asked about the timing and method they used to register.

Beginning in 1996, Current Population Survey (CPS) Voter Supplement respondents who voted or were registered but did not vote were asked a series of questions regarding when, relative to the implementation of the National Voter Registration Act (NVRA), and by what means (at a government office, by mail, and so on) they registered.