Registering to Vote

Among the costs of voting, the greatest attention has been paid to the fact that although in many other democracies citizens' names are automatically placed on the voting rolls by the government, most American states require voters to pay this cost themselves by visiting a registration site prior to the election. The most common finding here is that stricter registration rules correlate with lower turnout, but that reforms designed to ease the process do not necessarily increase turnout substantially.

In 1978 Rosenstone and Wolfinger reported that an early closing date for registration had the most powerful effect on turnout. They estimate that if the closing date for registration were eliminated, turnout would increase by 6.1%. Other changes—notably, in the number of hours registration sites stayed open—would increase the impact of registration to 9%. However, an electorate expanded by liberalizing these rules would not look much different from the one they currently observed. Rosenstone and Hansen (Reference Rosenstone and Hansen2003, 206) show that voter turnout in the North dropped 17% between 1896 and 1916 as registration laws were being instituted in the states. But other changes taking place at this time could have affected turnout levels as well. So to gain more purchase on the causal relationships involved, researchers have sought circumstances in which changes in registration requirements could be studied in a longitudinal and quasi-experimental manner.

Most of these over-time analyses support Wolfinger and Rosenstone's findings, though some conclude that the size of the impact was overstated (Mitchell and Wlezien Reference Mitchell and Wlezien1995). One effort has been to explore state laws similar to those of the 1993 National Voter Registration Act (NVRA). Knack (Reference Knack1995) shows that states adopting “active” motor-voter laws, in which driver's license agency employees take the initiative to ask clients if they'd like to register to vote, or use a form combining driver's licensing with voter registration, had almost a 4% turnout increase relative to states with no such program. There was little or no increase in states where agencies simply made registration materials available. Active by-mail registration (in which registrants need not have their forms witnessed or notarized) also had some effect, but allowing prospective voters to register at public agencies had no significant impact on turnout.

Highton and Wolfinger (Reference Highton and Wolfinger1998) find a turnout increase of 4.7% (relative to other states) during the four years after Colorado adopted a registration law that most closely resembled the NVRA; the biggest impact was on young people and recent movers. They also examine Election Day Registration (EDR), which, like active motor-voter laws, eliminates the cost of a second trip, in this case by combining registration and voting.Footnote ¹ They estimate that one-trip voting would increase turnout by 8.7% compared with a 30-day closing date, again especially among the young and the residentially mobile. Other studies of EDR have confirmed increases in the range of 3–9% (Fenster Reference Fenster1994; Knack Reference Knack2001).

Ansolabehere and Konisky (Reference Ansolabehere and Konisky2006) make use of the natural “experiment” that occurred when New York, in 1965, and Ohio, in 1977, imposed registration requirements. By comparing turnout in counties that did not have registration rules until the state imposed them with turnout in counties in the same state where registration had been imposed or voluntarily adopted earlier, they found that requiring registration reduced turnout by 7% in the subsequent election and by 3–5% in the long term. Their over-time analysis shows a smaller effect than does their cross-sectional analysis. But even a drop of 3–5%, projected to the voting-age population, is not trivial.

Powell (Reference Powell1986) concludes that in the 1970s, the need for U.S. citizens to take the initiative to register decreased American voter turnout by 16% compared with other Western democracies. Highton (Reference Highton2004, 511) contends, however, that the liberalization of state registration rules has accomplished as much as it could. He points out that even with no registration requirement, North Dakota's 2004 turnout rate was just 72% of voting-age citizens. Thus, additional costs of voting must be taken into account.

Other Institutional Requirements

Many other institutional costs of voting have been eliminated by court action and legislation. Filer, Kenny, and Morton (Reference Filer, Kenny and Morton1991) show that poll taxes, which literally raised the cost of voting, significantly depressed turnout over time, especially in counties with a higher proportion of poor people. Literacy tests, outlawed by the Voting Rights Act in 1970, were less strongly related to turnout. And Rosenstone and Wolfinger (Reference Rosenstone and Wolfinger1978, 34) did not find residency requirements—how long one must live in a district before voting—to have a sizable effect on turnout, at least by 1972.

Some institutional rules remain. States vary as to whether convicted felons may vote. Manza and Uggen (Reference Manza and Uggen2004, 494–95) estimate that such laws prevented 4.7 million people from voting in 2000. These laws affect voter turnout because of the high rate of incarceration in the U.S., especially among black males. The impact of other types of costs, such as information costs, convenience of the polling place, and state laws giving workers paid or unpaid time off to vote, are discussed in Hershey (Reference Hershey2008).

Costs Imposed by Voter-ID Laws

States have imposed several different types of voter-ID requirements since the passage of the Help America Vote Act (HAVA) in 2002: having to state or sign one's name at the polls, having one's signature matched with the signature on file with the local election board, and/or providing an identification (a driver's license, student or military ID, voter registration card, paycheck, or utility bill), with or without a photo. Arizona requires proof of citizenship in order to vote, and in 2008 the Missouri legislature nearly permitted the requirement that a prospective voter present a passport, original birth certificate, or naturalization papers as proof of citizenship when registering to vote.

To date, Indiana's is the most demanding voter-ID law. Since 2006, all registered voters must provide a document issued by the U.S. government or the State of Indiana with a photo of the individual, with an expiration date, and that “conforms to” the name on the voter-registration record. Anyone without such identification can be given a provisional ballot, which will be counted only if the voter brings the required photo ID to the county election board or the circuit court clerk within 10 days after the election or executes an affidavit of indigence or religious objection to being photographed.

This poses no additional costs to registrants with a current Indiana driver's license, state ID card, passport, or other appropriate ID. For those who lack the required ID, county bureaus of motor vehicles will provide a free ID card for voting purposes to those who bring four of a group of “primary and secondary” documents that could include an authenticated birth certificate, certificate of citizenship, or military ID or passport, but must also include proof of Social Security number and Indiana residency. Several of these documents cost money, time, information, and transportation to locations other than where the photo ID is obtained. The burden rises because most people who need this documentation do so because they lack a driver's license, and public transportation is not available in some counties.

Few studies have examined the impact of voter-ID laws. Alvarez, Bailey, and Katz (Reference Alvarez, Bailey and Katz2008) show that in aggregate data, voter-ID laws had no significant impact on turnout during 2000–2006. Mycoff, Wagner, and Wilson (Reference Mycoff, Wagner and Wilson2007) concur, and Vercellotti and Anderson (Reference Vercellotti and Anderson2006) find only a modest correlation in 2004 between two voter-ID requirements—a signature match and a non-photo ID—and statewide turnout, relative to requiring only that voters state their names at the polls. Berinsky (Reference Berinsky2005) argues that these null findings should be expected; the more politically involved are more likely to learn about voter-ID rules and also more likely to vote.

But researchers differ in their findings at the individual level. Using CPS data, Alvarez, Bailey, and Katz (Reference Alvarez, Bailey and Katz2008) show that stricter rules—the combination of having to present an ID and a signature match, and the photo-ID requirement—did depress the turnout of registered voters relative to the requirement of stating one's name at the polls. Mycoff, Wagner, and Wilson (Reference Mycoff, Wagner and Wilson2007), with a smaller N, did not find an impact. Vercellotti and Anderson, in a cross-sectional study, showed that the signature, non-photo ID, and photo-ID rules were associated with lower voter turnout in 2004, in the range of 3 to 4%.

Lower-Income and Less-Educated Persons

Researchers argue that more restrictive registration rules disproportionately reduce the turnout of low-SES people (Jackson, Brown, and Wright Reference Jackson, Brown and Wright1998, 268; Avery and Peffley Reference Avery and Peffley2005), and that more education and income help overcome the costs of voting (Verba, Schlozman, and Brady Reference Verba, Schlozman and Brady1995). Even when Nagler (Reference Nagler1991) re-estimated the Rosenstone and Wolfinger model with 1978 and 1984 CPS data to correct for a mis-estimation, he still finds that earlier closing dates lowered turnout among the least educated.

Yet making it easier to register doesn't necessarily increase low-SES people's turnout. Highton and Wolfinger (Reference Highton and Wolfinger1998, 84–85) showed that Colorado's active motor-voter law had the greatest effect on those with moderate education levels; those with the least education were unlikely to vote even if costs were reduced. Karp and Banducci (Reference Karp and Banducci2000) found that Oregon's use of voting only by mail increased turnout more among whites, higher-SES individuals, and older people. And Knack and White (Reference Knack and White2000) showed that EDR led to a rise in turnout among young and residentially mobile people but not significantly among lower-SES persons (see also Brians and Grofman Reference Brians and Grofman1999). Labor-union strength seems to increase participation slightly more among high- and moderate-income than low-income people (Leighley and Nagler Reference Leighley and Nagler2007). Those with the lowest SES have multiple barriers to turnout, whereas those with moderate SES levels, who may feel more connected to politics, might be more responsive to reductions in the costs of voting.

Research confirms that stricter voter-ID rules also disproportionately reduce the turnout of the least educated and those with lowest incomes. Vercellotti and Anderson (Reference Vercellotti and Anderson2006) find a stronger relationship between voter-ID requirements and lower turnout among registered voters with less than a high school education. Alvarez, Bailey, and Katz (Reference Alvarez, Bailey and Katz2008, 20) show that the least-educated registrants and those with lower incomes were less likely to vote in states that require a photo ID than in states that require voters only to state or sign their names. And Barreto, Nuño, and Sanchez (Reference Barreto, Nuño and Sanchez2007) report that Indiana registered voters and eligible non-registrants with incomes under $20,000 were much less likely to have the form of ID that the Indiana law requires than were higher-income residents, and less-educated people were somewhat less likely to possess the required photo ID.

Solt (Reference Solt2008, 56–58) shows that the more economic inequality in a nation, the less the turnout of all but the wealthiest quintile; the biggest impact was on the poorest quintile. He concludes (58), “One's political engagement … is shaped not only by how much money one has, but also by how much money everyone else has. Where economic resources are distributed more evenly, power is distributed more equally, and the resulting politics encourage relatively poor citizens to take interest and take part” (see also Filer, Kenny, and Morton Reference Filer, Kenny and Morton1993). A related question involves the impact of economic decline, as opposed to current SES. Rosenstone (Reference Rosenstone1982, 34–41) finds that unemployment and feeling worse off financially were associated with lower turnout during 1948–1974. Radcliff (Reference Radcliff1992) reports that decline in real per capita national income depressed turnout, especially among the economically marginal. This is less likely in nations where the full development of the welfare state cushions the effects of economic decline on the poor.

Presuming that those whose economic situation has deteriorated might be especially inclined to hold incumbents accountable for poor economic conditions, a finding that these individuals are less likely to vote could undermine democratic responsiveness. However, Arceneaux (Reference Arceneaux2003, 70–71) reports that those facing economic adversity and who blame the government for their plight are more likely to vote than are those who do not blame the government. We need to know more about the conditions in which economic decline energizes some individuals or further sensitizes them to the costs of voting.

Black Americans and Others of Minority Races and Ethnicities

The story of the relationships among race, ethnicity, and the costs of voting is less clear. A lot of research suggests that when SES and other resource variables (such as free time and facility in English) are controlled, blacks and Latinos are as likely to vote as whites are (Verba et al. Reference Verba, Schlozman, Brady and Nie1993; Jackson Reference Jackson2003; Cassel Reference Cassel2002). Other studies question this conclusion. Abramson and Claggett (Reference Abramson and Claggett1986) used validated votes to show that black respondents were more likely to over-report voting than whites, which could lead to a mistaken finding of parity in turnout rates. They found, as have others (Leighley and Nagler Reference Leighley and Nagler1992a; Reference Leighley and Nagler1992b), that racial differences in turnout remained even when education and region were controlled.

But this racial effect, at least among blacks, seemed to vary by election year (see Leighley and Nagler Reference Leighley and Nagler1992a). Abramson and Claggett (Reference Abramson and Claggett1986) suggest that the independent effect of race in 1984 may have been triggered by Jesse Jackson's presidential candidacy. The relationship between race or ethnicity and voting costs, then, may be contingent on the stimuli of the election at hand, perhaps by affecting concern about the outcome (Barack Obama's presidential campaign is a case in point) or by prompting group mobilization that reduces the costs of voting for some types of individuals.

Most researchers agree that race and ethnicity can structure the relationship between the costs of voting and turnout, that racial and ethnic groups differ in their access to politically relevant resources, and perhaps even that particular resources and costs may affect different groups in different ways (see Verba et al. Reference Verba, Schlozman, Brady and Nie1993, 458). For instance, blacks and Latinos are much less likely than Anglo-whites to belong to non-political groups, but about as likely to be church members, which could convey political information at low cost. While these skills may be more relevant to political activity beyond voting, they should also make it easier to register to vote and obtain needed documents. And although blacks were more likely than Anglo-whites to report that someone in church had asked them to vote, Latinos were much less likely to say so (Verba et al. Reference Verba, Schlozman, Brady and Nie1993, 484–85). It may be that forces in the black community, such as churches, act to counter the costs of voting to a greater extent than among Latinos. Examining these racial and ethnic differences, however, requires measuring a wide range of variables, from command of English and acculturation to native-born status and ethnic identity, and the literature in this area shows little consensus (see, for instance, Shaw, de la Garza, and Lee Reference Shaw, de la Garza and Lee2000; Cho Reference Cho1999; Jackson Reference Jackson2003).

With respect to voter ID, Alvarez, Bailey, and Katz (Reference Alvarez, Bailey and Katz2008) find no evidence that a photo-ID requirement, or requiring a voter ID plus a signature match, affected turnout among black registrants between 2000 and 2006. Vercellotti and Anderson (Reference Vercellotti and Anderson2006) confirm in the aggregate that voter-ID requirements did not disproportionately reduce turnout among blacks, though they did report a relationship in their individual-level data, and to a greater degree among Latinos and Asian Americans. Further, Barreto, Nuño, and Sanchez (Reference Barreto, Nuño and Sanchez2007) found that black registered voters and eligible non-registrants in Indiana were significantly less likely than whites to have a valid, state-issued ID matching the criteria required by the state's law. It could be that when faced with restrictive requirements, blacks in some areas are even more likely to mobilize to vote than whites are.

Thus, any disproportionate effect of stricter voter-ID rules on blacks may well reflect the fact that blacks tend to be lower in SES (though the fact that the impact is not specifically racial makes it no less real). There are suggestions, however, that there may be real racial differences in the administration of some voter-ID regimes. Alvarez, Atkeson, and Hall (Reference Alvarez, Atkeson and Hall2007, 24–25) find that in New Mexico, poll workers enforced voter-ID rules inconsistently. And a 15-state study (Alvarez et al. Reference Alvarez, Ansolabehere, Berinsky, Hall, Lenz and Stewart2008, 17), reports that a much higher proportion of black voters said they had been asked for an identification in 2007 and in the 2008 Super Tuesday events than white voters did, regardless of the state's voter-ID rules.

In sum, the immense literature on the costs of voting has shown that costs ranging from the registration requirement to strict voter-ID laws do reduce voter turnout to some degree and that the impact seems to fall disproportionately on the least educated and the least wealthy. There is less evidence, however, that reducing the costs of voting necessarily increases the turnout of these groups, probably because advance registration and photo-ID rules are only two of multiple burdens on their likelihood of voting.

We know little about the impact of other costs—the accessibility of polling places (but see Burmila Reference Burmila2009), changes in voting systems, and the interactions among various types of costs. And we need to research the impact of these costs on other large groups: people with disabilities, and racial, ethnic, and language groupings other than black Americans. Researchers should also consider the effects of perceived costs; although Alvarez et al. (Reference Alvarez, Ansolabehere, Berinsky, Hall, Lenz and Stewart2008) find relatively short waits at the polls in most precincts, media stories about long waits could affect people's willingness to vote, even if these stories are inaccurate or unusual.

In particular, research would benefit from classifying costs of voting according to their specific demands on citizens. One important aspect of advance registration, strict voter-ID laws, and laws requiring proof of citizenship, is that those who are unregistered or who lack the specified documents must make a separate trip—on a day other than Election Day, to a place other than the polling place, whose location and open hours will need to be learned, to take a bureaucratic step that does not lead directly to an election outcome. The burden of this extra trip falls most heavily on people with lower education and income, but it would also be likely to affect persons with disabilities and, due to HAVA's requirement of proof of residence for first-time voters, those who have recently moved. Given the importance of turnout to democratic politics, the need for research on the impact of voter-ID laws and those requiring proof of citizenship is compelling. ■