Voter ID Laws and Voter Turnout1 Kyle A. Dropp2

1 Please do not cite without the author’s permission. I would like to thank Jonathan Rodden, Justin Grimmer, Paul Sniderman, Clayton Nall, Jowei Chen, Bobby Gulotty, Arjun Wilkins, Gary Cox and members of the Stanford Methods Workshop for helpful comments. 2 Ph.D. candidate, Department of Political Science, Stanford University, [email protected]

1

Do Voter Identification statutes reduce voter turnout? I demonstrate that the decade-long expansion of Voter ID statutes has demobilized Democratic-leaning individuals including young adults, renters, the poor and African Americans using individual voting records over a series of four elections (2004-2010). I use a difference-in-differences approach to compare changes in turnout between 2004 and 2008 among voter subgroups in Voter ID states with broader statewide turnout and turnout in states with no election law policy change. This research both clarifies mixed findings in the scholarly literature and introduces a new approach for documenting how election law changes impact subgroups.

2

Thirty-one states have enacted and adopted Voter Identification statutes,3 and strict policies requiring a government-issued ID at the polls have become commonplace. Lawmakers in 34 states introduced legislation on the subject and eight adopted new policies in 2011 alone, driven by a U.S. Supreme Court decision upholding an Indiana statute4 and widespread Republican gains in 2010 in state houses.5 The Progressive Era reforms of the early 20th Century and the reapportionment revolution started by Baker v. Carr (1962) transformed the representative-constituent linkage, and the burst of legislative attention on Voter ID may fundamentally alter the composition of the electorate. Overall, has the expansion of Voter ID statutes reduced voter turnout? And, if so, which groups have been disparately impacted? Previous research has yielded puzzling contradictory findings because it assesses earlier, more lenient statutes, aggregate county or state-level studies are not able to detect effects across voter subgroups and survey-based studies contain measurement and sampling error.6 My research examines both lenient and strict policies over the past decade, minimizes sampling error and addresses (and dismisses) many potential threats to validity. In my research, I aggregate tens of millions of individual level voting records over a series of four elections (2004-2010) using a national voter database. I isolate groups with low ID ownership rates such as the working class, renters, African Americans, young adults and Hispanics using demographic information contained in the voter files. Then, I use a difference-in-differences approach to compare the turnout of these voter subgroups before and after a Voter ID law change with turnout statewide and with turnout patterns among voter subgroups in states with no policy change. This paper has two principal findings. First, Voter ID statutes exert a modest but politically meaningful demobilizing effect, especially among the poor, young adults, renters and African Americans. States must demonstrate that ID requirements do not pose an undue burden on voters - the results here indicate that Voter ID laws have modest effects that are substantial enough to influence election outcomes in close races. For example, a one percent reduction among African Americans in Ohio, who cast 95%+ of their ballots for Democrats, corresponds with a loss of more than 5,000 votes for Democrats in a presidential election. My research is the first to demonstrate that Voter ID laws impact the participation of a broad swath of the electorate including renters and the poor. The results are robust to a wide range of data robustness checks.7 3

National Conference of State Legislatures http://www.ncsl.org/documents/legismgt/elect/ Canvass_Apr_2012_No_29.pdf 4 Crawford v. Marion County Election Board, 553 U.S. 181 (2008) 5 National Conference of State Legislatures http://www.ncsl.org/legislatures-elections/ elections/voter-id.aspx 6 For a catalog of studies, see this link http://www.brennancenter.org/content/resource/research_ on_voter_id/ 7 Separate sections address estimation issues related to the non-random assignment of Voter ID policies to

3

Second, Voter ID laws are more likely to reduce turnout in midterm elections. This finding suggests that widespread mobilization and voter outreach efforts during presidential contests can reduce the impact of election laws that impose costs on voters. This study proceeds as follows. First, I describe the widespread adoption of Voter Identification statutes in the past decade and assess the scholarly literature. Then, I discuss hypotheses, research design, data sources and findings. I briefly conclude. Separate sections address data robustness and potential threats to validity.

Voter ID adoption: 2002 to present In 2001, only one in four states required that voters provide an ID at the polls.8 and none of these states turned away voters without a suitable ID. Today, voters in 31 states must show a form of identification at the polls, and there has been a decisive trend toward strict policies asking voters to present a government-issued photo identification.9 Pending court challenges in Mississippi and Wisconsin, 33 states may have Voter ID stautes for the November 2012 election. In a close election, these ballot security measures may influence election outcomes. The widespread adoption of stringent Voter ID statutes has been fueled by four factors: the passage of the Help America Vote Act (HAVA) in 2002, a concerted, coordinated effort among Republican state legislators and governors to enhance ballot security, near unanimous public support for ballot security measures and the perception of pervasive voter fraud among Americans. The Help America Vote Act (HAVA) of 2002 established minimum federal standards for first-time voters, established the Election Assistance Commission (EAC) and replaced outdated voting systems such as punch cards.10 The act passed with overwhelming bipartisan majorities in both the U.S. Senate and U.S. House and was signed into law by President George W. Bush in October 2002. The act provided voters with a range of options to register and verify their identify.11 Citizens who cannot provide an ID or the last four digits states. I also assess the implications of potential violations of the Stable Unit Treatment Value Assumption (SUTVA) such as the possibility that the assignment status of some units affects the potential outcomes of other units, and the variation in treatments across groups caused by heterogeneous Voter ID policies. 8 National Conference of State Legislatures http://www.ncsl.org/documents/legismgt/elect/ Canvass_Apr_2012_No_29.pdf 9 National Conference for State Legislatures http://www.ncsl.org/legislatures-elections/ elections/voter-id.aspx 10 Pub.L. 107-252 http://www.gpo.gov/fdsys/pkg/PLAW-107publ252/html/PLAW-107publ252.htm 11 First-time voters must provide either a Photo ID, Non-Photo ID, their current and valid driver’s license number, a state identification number or the last four digits of their Social Security Number.(Help America Vote Act, Section 303b http://www.fec.gov/hava/law_ext.txt). Individuals who cannot meet these requirements will be assigned a unique voter registration number.

4

of their Social Security number are assigned a unique voter registration number. Americans are very concerned about ballot integrity and voter fraud, and Voter ID laws enjoy widespread public support. Seventy-eight percent of Americans said voters should be required to show an official photo identification on Election Day, including 86 percent of Republicans and 71 percent of Democrats, according to a 2006 Pew poll.12 And fully 48 percent of Americans said that voter fraud – people voting who are not eligible or voters casting multiple ballots – is a major problem, according to a 2012 Washington Post poll.13 Figure 1 on Page 6 displays changes in Voter ID statutes from 2002 to the present. States in white have adopted the HAVA minimum requirement, light blue states require a NonPhoto ID such as a bank statement and blue states require a Photo ID such as a driver’s license. Dark blue states, the most stringent, require voters to present a Photo ID but provide few alternatives for casting a provisional ballot.14 Policies vary dramatically across states. In Indiana, a dark blue state, voters must present an ID issued by Indiana or the United States containing the individuals’ name and photo, along with an expiration date.15 In Washington, a Non-Photo ID state voters may present a valid Photo ID, voter identification card, utility bill, bank statement, paycheck or government document. Further, individuals who cannot provide identification can cast a provisional ballot.16 12

Survey by Pew Research Center for the People and the Press, October 17-22, 2006 based on 2,006 telephone interviews. 13 The poll was conducted July 18-29, 2012, and based on 2,047 telephone interviews. Thirty-three percent say it is a minor problem and 14 percent say it is not a problem. 14 Alvarez et al. (2007) use an 8-point coding scheme rather than a 4-point classification. They assign states to the following Voter ID requirements: state name, signature, matching signature, request ID, require Non-Photo ID, require ID plus signature, request Photo ID, require Photo ID. The four point classification is more appropriate for illustrating the decade-long trend toward more stringent Voter ID policies. 15 If a voter cannot produce this proof of identification, he must return to the election board with proof of identification, or state that he cannot obtain an ID because he is indigent or has a religious objection to being photographed. http://www.ncsl.org/legislatures-elections/elections/voter-id.aspx 16 http://www.ncsl.org/legislatures-elections/elections/voter-id.aspx

5

Figure 1: Voter ID statutes from 2002 to present. White states have no Voter ID statute, light blue states require a Non-Photo ID, blue states require a Photo ID and dark blue state have Strict Photo ID policies with limited provisional ballot options.

6

Figure 1 on Page 6 exhibits three trends: Voter ID policies have spread from the South to much of the country in the past decade, they have become increasingly stringent in recent years and their adoption is driven by partisan factors rather than simple diffusion to neighboring states. In 2002, conservative Southern states including Georgia and Texas were among the only states to have Voter ID laws stricter than the Help America Vote Act minimum requirements. Since then, Voter ID statutes have spread to the Mountain States,the Southwest, and parts of the Midwest. The West Coast and the Northeast are the only regions largely unaffected by the nationwide drive to strengthen Voter ID laws. In 2011, only three states without Voter ID statutes in place – Oregon, Vermont and Wyoming, did not consider legislation to strengthen their Voter ID policy.17 Second, Voter ID statutes has become increasingly stringent in recent years. Many early Voter ID statutes required only a Non-Photo ID and provided ample opportunities for voters to cast a provisional ballot. Eleven states now require voters to present a Photo ID, compared with four in 2004 and none in 2002. Third, there is a strong relationship between GOP control of state legislatures and a state’s Voter ID policy. Twenty-one of the 26 states where Republicans currently control both legislative chambers have enacted Voter ID statutes above the HAVA minimum, and all 11 states with the most stringent Photo ID or Strict Photo ID requirements have unified Republican legislative control. On the other hand, only 4 out of 15 states with current unified Democratic control of state houses have enacted Voter ID statutes.18 Biggers and Hanmer (2011) examine Voter Identification statutes from 1972-2011 and find that partisan factors are more important than diffusion to neighboring states in predicting the passage of Voter ID statutes: “the switch to a Republican governor has a large positive effect on this decision to require identification at the polls” (Biggers and Hanmer, 2011, p. 27). Related, Voter ID policies are supported or opposed along party lines. For example, more than 95% of Republican legislators supported Voter ID laws introduced between 2005 and 2007, compared with just 2% of Democrats.19 20 17

http://www.ncsl.org/legislatures-elections/elections/voter-id-2011-legislation.aspx I obtained data on the party composition of state legislatures from 2000-2011 from the National Conference of State Legislatures and merged it with the four-point classification of Voter ID statutes (1 = HAVA minimum, 2 = Non-Photo ID, 3 = Photo ID, 4 = Strict Photo ID). Overall, the correlation between unified GOP control of state legislatures in 2011 and current Voter ID policy is .57. 19 Brief of Amici Curiae of Historians and Other Scholars in Support of Petitioners. Crawford et al. v. Marion County Election Board et al. Nos. 07-21, 07-25 (U.S., 2007). 20 Similarly, on voter identification provisions, the “average difference between the percent of Republicans and Democrats voting yea on the eleven Senate votes relating to HAVA was ninety-one percentage points” (Lee, 2009) 18

7

Scholarly literature Policymakers and politicians generally believe that Voter ID statutes will influence election outcomes. Republicans have suggested that Voter ID statutes may flip states from Democratic to Republican: “Voter ID, which is gonna allow Governor Romney to win the state of Pennsylvania, done,” said Pennsylvania House Majority Leader Mike Turzai after the passage of a strict statute. Democrats have used these ballot laws to motivate their party’s base: “If you want every American to vote and you think it’s wrong to change voting procedures just to reduce the turnout younger, poorer, minority and disabled voters, you should support Barack Obama” former President Bill Clinton said when he nominated President Barack Obama at the Democratic National Convention in September 2012. And prominent Democrats have compared this legislation to Jim Crow laws: “We are witnessing a concerted effort to place new obstacles in front of minorities, low-income families and young people who seek to exercise their right to vote. A poll tax by another name would smell as vile,” said U.S. Rep. Steny Hoyer (D-Md.), in November 2011. Despite these statements, a growing, methodologically diverse research literature on Voter Identification statutes and voter turnout has yielded puzzling, mixed findings. These mixed findings have resulted from data limitations and the focus on earlier, more lenient statutes. Some research has found that Voter ID laws have a minimal impact on turnout (Ansolabehere, 2009; Lott, 2006; Pastor et al., 2010; Mycoff et al., 2009; Milyo, 2007). These authors find that Voter ID statutes are inconsistently implemented, that most adults have suitable forms of identification and that individuals without IDs are not likely to vote. Ansolabehere (2009) examines Current Population Survey (CPS) data and concludes that “[v]oter ID appears to present no real barrier to access” (Ansolabehere, 2009). Poll workers rarely ask for ID, he finds, and individuals almost never say they did not vote because they lacked an ID. Using largely the same methods and data, other authors have found that Voter ID statutes decrease turnout among subgroups lacking suitable IDs and that more stringent Voter ID laws exert a larger demobilizing effect (Vercellotti and Anderson, 2006; Alvarez et al., 2007; Logan et al., 2007; Alvarez et al., 2011). An overarching theme in the literature, however, is that the available data is not powerful enough to answer this question with confidence.21 Aggregate level studies using counties or 21 There are a series of separate reasons why election administration policies may only impact voters on the margins. Few voters actually would be turned away from casting a ballot solely based on 2008 state-by-state ID requirements, according to two separate studies. One study estimates that .5% of respondents would be prevented from voting (Alvarez et al., 2008) http://www.american.edu/spa/cdem/upload/ VoterIDFinalReport1-9-08.pdf while a 2007 study estimates that one tenth of one percent of voters would

8

states cannot document the impact of state-level interventions on voter subgroups, while individual level survey data includes measurement error from inflated self reports of voter turnout and sampling error caused by small samples. Erikson and Minnite (2009) call into question the use of cross-sectional data from the Current Population Survey. They analyze multiple years of CPS data and find no relationship between Voter ID laws and turnout, contrary to previous authors’ conclusions from analysis of the same data: “[W]e see the existing science regarding voter suppression as incomplete and inconclusive. This is not because of any reason to doubt the suppression effect but rather because the data that have been analyzed do not allow a conclusive test” (Erikson and Minnite, 2009, p. 98). The expansion of Voter ID statutes is a recent phenomenon, and scholars have not had many election cycles to examine their effects: “[S]ince the changes in voter identification requirements have really only started since the passage of HAVA in 2002 and the law we are most interested in – photo identification requirements – was only implemented in 2006, we have only a small amount of information in the available data about how each of voter identification requirements might affect participation” (Alvarez et al., 2011, p. 10). Voter Identification Laws may exert their greatest impact years after implementation, when young adults and first-time voters must obtain an ID to cast their ballot. Previous research has been unable to test this. Further, the trend toward stricter laws in recent years means there have been even fewer data points to study the stringent election policies most likely to adversely affect turnout.

Hypotheses Scholars generally cite three reasons why Voter ID statutes may reduce turnout. First, many Americans do not have suitable or up-to-date forms of identification. Second, some citizens may decide not to cast a ballot because they are confused by the statutes’ rebe unable to vote because of an ID requirement (Ansolabehere, 2007). Second, most individuals without proper IDs are unlikely to cast a ballot, regardless of whether they reside in policy or non-policy states, according to Rick Hasen, an election law expert: “It’s not possible to show, he says, that many people have actually been deterred from voting by these laws. In part, that’s because many of the laws are new, and in part it’s because many of the people who lack an ID card tend not to be interested in voting in the first place” (Firestone, 2012). Third, a majority of states with ID policies allow voters to cast a provisional ballot, after signing an affidavit, for instance, which could eventually be counted. Many of the stringent laws with limited provisional ballot options have been passed only in recent months or are pending. Fourth, studies of Voter ID statutes do not examine the full effect of requiring an ID versus asking for no ID. The Help America Vote Act established minimum identification standards for first-time voters and absentee voters, so any change in Voter ID policy is relative to those federal minimum standards.

9

quirements or concerned about the fairness of the law’s implementation.22 Third, Voter Identification laws may reduce voter impersonation, one type of voter fraud, at polling places.23 My research focuses on differential ID ownership rates. Voter ID statutes inconvenience citizens who lack valid IDs,24 and research indicates that subgroups such as young adults, minorities, the working-class, renters and the elderly are both less likely to own a suitable ID25 and favor the Democratic Party.26 These groups also tend to be less familiar with the electoral system.27 Cumulatively, this leads to the 22

In a Boston Exit Poll, Cobb et al. (2012) find racial differences in the administration of Voter ID statutes: “We find strong evidence that Hispanic and black voters were asked for IDs at higher rates than similarly situated white voters” (?, p. 3). Similarly, Ansolabehere (2009) finds that minorities are more likely to report that they were asked to present an ID: “Both the 2006 and 2008 surveys show considerable racial differences. In the 2006 general election, 47% of white voters reported being asked to show photo identification at the polls, compared with 54% of Hispanics and 55% of African Americans. In the 2008 Super Tuesday primary states, 53% of whites were asked to show photo ID, compared with 58% of Hispanics and a staggering 73% of African Americans” (Ansolabehere, 2009, p. 128). 23 According to the National Conference for State Legislatures, “Little evidence exists that fraud by impersonation at the polls is a common problem” http://www.ncsl.org/documents/legismgt/elect/Canvass_ Apr_2012_No_29.pdf 24 A survey of voters found that hundreds of thousands of eligible voters face challenges in obtaining proper IDs to cast a ballot. According to the Brennan Center for Justice at New York University, nearly 500,000 voters “in 10 states with restrictive voter ID laws live in households without vehicles and reside at least 10 miles from an ID-issuing office open more than two days a week.” http://www.brennancenter.org/content/ resource/study_500000_americans_could_face_significant_challenges_to_obtain_photo_id/ 25 Racial differences: Twenty-five percent of African Americans and 18 percent of adults 65 years and older lack the government-issued photo ID necessary to cast a ballot in stringent Voter ID states, compared with one in 10 adults overall, according to the Brennan Center for Justice. http://www.brennancenter.org/ page/-/d/download_file_39242.pdf. Also see http://www.brennancenter.org/page/-/d/download_ file_39242.pdf Pawasarat (2005) finds that African Americans and Hispanics are less likely to have drivers licenses than whites in Wisconsin: “Less than half (47 percent) of Milwaukee County African American adults and 43 percent of Hispanic adults have a valid drivers license compared to 85 percent of white adults” (Pawasarat, 2005, p. 1). White voters, on the other hand, are more likely to have government-issued IDs (Barreto et al., 2007). Five percent of white registered voters have an up-to-date Driver’s License or State Issued ID Card compared with 10 percent of African Americans, 11 percent of Latinos and 14 percent of Asian Americans, according to a 2008 study (Sanchez et al., 2011). Age, income, rental status differences: Residents who have moved are less likely to own suitable forms of identification: “The population that changes residence frequently is most likely to have a drivers license address that differs from their current residence. This would include lower-income residents who rent and students and young adults living away from home” (Pawasarat, 2005, p. 2). For additional evidence, see (Barreto et al., 2007; Sanchez et al., 2011; Pawasarat, 2005) 26 According to the 2008 National U.S. House Exit poll, 93% of African Americans voted for Democrats, along with 68% of Latinos, 63% of Asians, 63% of 18-29 year-olds. Adults 65 and over split about evenly, 49% Democrat to 48% Republican. Fifty-nine percent of renters identify as Democrats or leaned Democrats, compared with 41% of home owners, according to a Gallup poll conducted January 5-8, 2012. In the same poll, 50% of home owners self-identified as Republican or leaned Republicans, compared with 31% of renters. 27 When there are changes in election administration policies, Americans with higher levels of civic skills, resources or flexibility in work schedule may adapt more readily than those without such skills: “[The] pres-

10

first hypothesis: H1: Varying ID ownership rates across the electorate suggests that Voter ID laws will reduce turnout among Democratic-leaning subgroups such as young adults, renters and minorities. However, campaigns and interest groups react swiftly to election law changes. Voter ID statutes may not decrease turnout when campaigns and interest groups devote additional resources toward voter education and outreach aimed at Americans lacking IDs,28 interest groups synchronize their efforts with traditional allies,29 or organized interests shift their mission to focus on voter outreach and education campaigns rather than persuasion efforts.30 Organized interests have fixed resources, and grassroots mobilization efforts are more intense and comprehensive during presidential elections compared with midterms. Therefore, campaigns will have many more opportunities during presidential election campaigns to educate citizens about Voter ID requirements and assist them in obtaining valid IDs. H2: Voter ID statutes are more likely to reduce turnout during midterm elections than during presidential elections. Overall, I argue that Voter ID statutes impose a cost on citizens without suitable forms of identification. These laws are likely to shift the composition of the electorate in the GOP’s favor when mobilization efforts are not intense, such as during midterm contests or when these Voter ID laws are stringent. ence or absence of resources contributes substantially to individual differences in participation. Resources are, in turn, not equally distributed; some socioeconomic groups are better endowed than others” (Brady et al., 1995, p. 274). 28 During the 2012 presidential election, the Obama campaign has sent teams to educate Americans in Voter ID states: “Field workers for President Obama’s campaign fanned out across the country over the weekend in an effort to confront a barrage of new voter identification laws that strategists say threaten the campaign’s hopes for registering new voters ahead of the November election” (Shear, 2012). Moreover, the AFL-CIO “vowed to mount their biggest voter registration and protection efforts ever to counter these [Voter ID] laws” (Greenhouse, 2012). 29 Election law changes increase interest groups’ coordination efforts with traditional ideological allies: “The federation’s [AFL-CIO] leaders said they would work closely with other groups, including the N.A.A.C.P. and the National Council of La Raza, to maximize voter turnout and provide whatever help is needed to enable elderly, disabled and poor Americans to get voter IDs” (Greenhouse, 2012). 30 During the 2008 presidential campaign, The National Association for the Advancement of Colored People (NAACP) chose to “focus on voter education and outreach ahead of this year’s presidential election in the wake of a U.S. Supreme Court ruling on voter identification laws,” according to a statement (Haines, 2008).

11

Research Design In this section, I outline the difference-in-differences identification strategy (Ashenfelter and Card, 1985)31 and describe the assumptions I am making to estimate treatment effects. Suppose State S implements a Voter Identification law a few months after the 2004 presidential election. We want to understand whether the statute reduces turnout among college students, a group that disproportionately lacks valid forms of identification. Our outcome variable is the percent of eligible adults who cast a ballot, the treatment group is college students in State S and the control group is all other residents in State S. This differencein-differences approach tests whether the policy intervention in State S decreases college students’ turnout relative to statewide turnout. If turnout in State S was 50% for both college students and statewide in Nov. 2004, 50% statewide in Nov. 2008 and 45% for college students in 2008, as displayed in Figure 2 on Page 13, we conclude that Voter ID policies reduced turnout by 5% among college students. This design’s main shortcoming is that factors unrelated to State S ’s new policy may affect the political participation of college students relative to statewide turnout. The Obama campaign’s mobilization efforts in 2008, for example, buoyed turnout among college students nationwide. Therefore, a model focusing solely on policy states could produce the spurious result that Voter ID laws boost turnout among college students. We improve the first approach by using an additional control: one or more non-policy States ∼S, as shown in Figure 3 on Page 13. Then, we can examine whether the change in college student turnout relative to statewide patterns is larger in policy states S than in non-policy states ∼S. This accounts both for factors unrelated to the Voter ID policy that affect turnout among the subgroup nationally, such as Obama mobilization efforts, and features unique to State(s) ∼S that either increase or decrease turnout statewide, such as a competitive gubernatorial race or a weak economic climate. In this hypothetical, turnout among college students rises across policy and non-policy states; however, it rises much more sharply in non-policy states. We conclude that Voter ID laws decrease the turnout of college students. The difference-in-difference-in-differences estimates (DDD) has four components = [(∆¯ yS,G − ∆¯ yS,∼G ] − [(∆¯ y∼S,G − ∆¯ y∼S,∼G ]

δ

where δ is the estimator, ∆¯ y is change in mean turnout before and after an intervention, State(s) are either S (policy) or ∼S (non-policy) and Group(s) are either G (subgroups 31

Difference-in-differences is used widely in economics and scholars have studied Voter ID policy implementations with this research design.

12

Figure 2: Voter turnout in policy states among students (blue) and statewide (red)

Figure 3: Voter turnout in policy states (right) and non-policy states (left) among students (blue) and statewide (red)

such as college students) or ∼ G (statewide).32 The first quantity is the change in turnout for a voter subgroup (i.e., college students) in Policy State(s) S before and after the intervention, while the second sum is the change in turnout statewide in Policy State(s) over the same period. The third quantity is change in the voter subgroup’s turnout in non-policy State(s) ∼ S during the same time frame, whereas the final sum is the statewide change. We find the population analog by taking the expected value of the four quantities. This basic approach can be extended to study different elections, alternate voter subgroups or other state-level election law interventions. Finally, we can also estimate the treatment effect via OLS using a series of indicator variables and interactions shown below y = β0 + β1 S + β2 G + β3 S ∗ G + λ0 T + λ1 T ∗ S + λ2 T ∗ G + λ3 T ∗ S ∗ G + u

(1)

where S (State) is either a policy or non-policy state, G (Group) is either statewide turnout or turnout among the disadvantaged group, T (Time) is pre-intervention or postintervention and λ3 is the coefficient of interest from the triple interaction of State, Group 32

I separately estimate ∆¯ y for turnout changes between Nov. 2004 and Nov. 2008, between Nov. 2006 and Nov. 2010 and between Nov. 2004 and Nov. 2010.

13

and Time. A negative coefficient on λ3 indicates that state Voter Identification laws decrease turnout among a particular subgroup. In summary, I use a difference-in-differences estimator to compare the change in voter turnout before and after a policy change among specific subgroups such as African Americans with statewide voting trends. Furthermore, I analyze individual level voting patterns both in states implementing a new Voter ID policy and in states where election law remained constant over the same period.

Potential threats to validity In a controlled, randomized experiment, the experimenter randomly assigns individuals to conditions and exercises control over the treatment, while subjects are not aware of their treatment assignment. Unfortunately, many questions under inquiry in political science cannot meet the exacting standards of a controlled, randomized experiment. In this section, I address and reject a series of potential threats to validity caused by the observational nature of this study. First, since the individual level observations are grouped within the states in which the Voter ID implementation occurs, we cannot assume that errors are independently and identically distributed. Estimates obtained without clustering observations result in downward biased standard errors because individual observations in clustered data contribute less information to a model than data without clustering.33 Erikson and Minnite (2009), for example, find that Voter ID studies using robust rather than clustered standard errors understate the size of standard errors by a factor of seven, influencing whether findings are significant or not (Erikson and Minnite, 2009, p. 92). Overall, when using state or county-level aggregate data or responses from the Current Population Survey (CPS), the computation of standard errors can determine whether findings are statistically meaningful or not. I use national voter file data, and the standard errors based on a proportion of tens of millions of observations are miniscule. For example, we could multiply the standard error of a proportion of 10 million individuals by a factor of 30 and still obtain a statistically significant finding for a one percentage point treatment effect. The main models in this study estimate standard errors utilizing a cluster bootstrap at the state and county, which causes a modest increase in the size of the standard errors.34 33

Research in social science has shown that not accounting for clustering can bias standard errors (e.g., Carsey and Wright, 1998; Green and Vavreck, 2008) 34 I compute standard errors by using a series of different clusters. First, I aggregate individual level data to the county level. Then I resample observations using a bootstrap at the state and county level. Second, I aggregate individual level data to the ZIP Code level. Then, I resample observations using a bootstrap at the state and ZIP Code level. The bootstrap resamples entire clusters of observations rather than individual

14

Second, in an experiment, there is random assignment and treatment and control groups should be similar across covariates. Since Voter ID laws are more likely to be adopted in conservative states, it is possible that elites are targeting states with voter subgroups likely to be affected by the policies or with historically low levels of voter turnout.35 The evidence suggests that Voter ID laws have been driven mainly by the partisan composition of the state legislature rather than specific characteristics of the state’s electorate. Further, the difference-in-difference approach accounts for variation in turnout rates across states and voter subgroups. Previous studies of Voter ID statutes have used difference-in-differences estimators to address problems associated with non-random assignment (Alvarez et al., 2011).3637 Third, the stable unit treatment value assumption (SUTVA) holds that the assignment status of any unit does not affect the potential outcomes of other units. The clustering of policies at the state level, the broad awareness of Voter ID statutes nationwide,38 and the widespread adoption of these election laws in the past decade calls into question the validity of this assumption.39 It is likely, for instance, that learning of a Voter ID policy in a neighboring state could confuse voters and deter them from voting, even if they live in a state with no Voter ID policy. This actually would bias our estimate in the opposite direction and make it more difficult to conclude that Voter ID statutes decrease turnout among voter subgroups.40 Second, as I discuss in the next section, I study the impact of observations and modestly increases the size of the standard errors. 35 For instance, Voter ID laws may negatively impact turnout among African Americans in Georgia more than African Americans in North Carolina. If legislators pass a law in Georgia but not in North Carolina, then the estimated treatment effect will be biased upward. 36 Alvarez et al. (2011), for example, use a difference-in-differences estimator with Current Population Survey data to estimate the impact of Voter ID statutes: “Finally, identification requirements are not randomly assigned across states. This is a problem if states with historically lower turnout also tend to adopt stricter identification requirements, then we will have trouble isolating whether the low level of turnout is due to the identification requirement or to other factors that lead a given state to have lower turnout rates. The estimation strategy used exploits the temporal and geographic variability in voter identification requirements to sidestep the problem on non-random assignment. This is referred to as a difference-indifferences estimator and our analysis is built on a generalization of this procedure. In particular, we use a multilevel model – also referred to as a random effects model – to assess how voter identification requirements affect participation by registered voters, using data from four years of recent CPS Voter Supplement data (Alvarez et al., 2011, p. 10). 37 We also can address this issue by matching on covariates with Catalist’s 1% sample, which contains detailed information on more than two million voters. With this approach, the sole difference between units is the assignment to treatment. 38 Twenty-one percent of Americans said they heard or read a lot about states putting in place new photo identification requirements for voters, according to a Washington Post poll in 2012. The pool was conducted July 18-29, 2012 and based on 2,047 telephone interviews. Twenty-seven percent said they heard or read some, 15% said not much and 36% said they had heard or read nothing at all. 39 “The [potential outcome] observation on one unit should be unaffected by the particular assignment of treatments to the other units” (Cox 1958). If assignment status influences potential outcomes, then there are several compound treatments, each of which involves a different assignment. 40 It is possible, however, that reports that African Americans are assigned to a Voter ID treatment may

15

relative changes in the stringency of Voter ID statutes, not the absolute level of the policy. It is unlikely that voters residing in Non-Photo ID states who hear of a Photo ID policy implementation in a neighboring state will change their voting behavior.41 The next section discusses the elections and constituencies examined in this study.

Data Assembling a Voter ID Database I assemble a database of Voter ID policies from 2002 to the present using sources such as the National Conference of State Legislatures, LexisNexis and the work of previous scholars. The time series policy database facilitates statistical tests examining the impact of changes in Voter ID policy on voter turnout.42 The National Conference of State Legislatures places ID requirements in four categories: No Voter ID law, Non-photo ID law, Photo ID law and Strict Photo ID law. I follow their methodology.43 Overall, 13 states changed their Voter ID statutes between the 2002 and mobilize African Americans in Mississippi, a non-policy state, leading to a biased estimate of the treatment effect 41 Finally, the second part of SUTVA holds that there is no variation in treatment across groups, that the treatments for all units are comparable. SUTVA posits a fixed potential outcomes for units assigned to either treatment or control. Variation in treatment effects means that there are actually multiple, separate treatments. Voter ID statutes are multi-faceted and diverse, and each state has unique minimum and maximum requirements, making it nearly impossible to calculate a single average treatment effect. These concerns are highlighted by Alvarez et al. (2011). Alvarez et al. (2011) address the variation in treatments: “[T]here are many methodological problems unique to this data, one of which is the ordinality of voter identification requirements. As is apparent from the listing of the types of regimes, it is not the case that a state either requires identification to vote, or does not. States require many different levels of identification from simply stating one’s name to showing a picture identification. This further complicates the question, as we must determine not just one effect but several potentially incremental effects. Second, states may differ in their implementation of similar requirements. While one state may consider a student identification card or discount club membership card to be valid photo identification, another state may only recognize government-issued photo identification cards” (Alvarez et al., 2011, p. 9-10) 42 Assembling a database of Voter ID statutes for each midterm and presidential election since 2002 is not an easy task. States have unique minimum requirements, maximum requirements and policies for handling provisional ballots, multiple courts may render judgment on newly enacted Voter ID policies, and election officials such as the Secretary of State can modify policies in the days leading up to an election. In a report to the Election Assistance Commission, the authors note the complexity of Voter ID policy: “We recognize the difficulties in summarizing each state’s voter ID requirements. The problem is illustrated by the number of footnotes to Table 1 below. The variety of statutory and regulatory details among the states is complex” (Eagleton Institute report to the EAC, p. 20). 43 National Conference on State Legislatures. http://www.ncsl.org/legislatures-elections/ elections/voter-id.aspx The strict photo option refers to states that do not allow voters to cast provisional ballots unless they present a photo ID.

16

2004 election, seven strengthened their laws between 2004 and 2006, two tightened their policies between 2006 and 2008 and three implemented additional Voter ID provisions between November 2008 and November 2010.44 A few states eased their policies during the period. I examine the impact of relative changes in Voter ID policy from one of the four categories to another.45 Other scholars have examined the impact of eight different types of Voter ID laws (Alvarez et al., 2011), and my coding approach can be extended to measure both the impact of a change in a relative and absolute sense.46

Voter File and Census Data My primary elections data source is Catalist, a national voter database containing observations for 180+ million registered voters.47 The query-able database includes detailed voter histories, along with demographic and commercial data appended to each name.48 Overall, the official tallies and demographic variables reported in the Catalist database are highly accurate.49 44

2002-04: AZ, MT, SD, ND, HI, CO, AR, LA, AL, FL, TN, SC, MD; 2004-06: WA, AZ, NM, FL, IN, OH, HI; 2006-08: MI, GA; 2008-10: ID, OK, UT. 45 The majority of states in the sample changed their Voter ID statute by one increment in the four-point classification scheme. Separate models demonstrate that the impact of changing one increment is similar to changing multiple increments. 46 Alvarez et al. (2011) have developed an eight point classification scale based on the strictness of the statute and have categorized each state from 2000 through 2006. Their scheme includes the following categories, ranging from the least intrusive to the most stringent: 1) voter states name, 2) voter signs name, 3) voter signs and signature match, 4) voter is requested to present proof of identification or registration card, 5) voter must present proof of ID or voter registration card, 6) voter must present proof of identification and signature match, 7) voter is requested to present photo id, and 8) voter is required to present photo id (Alvarez et al., 2011). 47 I received access to Catalist through Stanford University’s Academic Subscription. 48 Ansolabehere and Hersh (2010) describe Catalist’s basic data collection process and the rigorous procedures they implement to validate the data: “Several times a year, Catalist purchases the publicly available voter registration files made available by each state or county election office...Catalist then cross-references the registration lists with other public records, such as the National Change of Address (NCOA) database maintained by the Post Office and the Social Security death index. Movers and deceased voters are flagged. Catalist matches the registration files to commercial records from data aggregation firms that compile lists from retailers and direct marketing companies. This allows the firm to correct the records of individuals who may have a typo in their registration record or may have registered with a nickname rather than their legal name” (Ansolabehere and Hersh 2010, p. 5) 49 There may be a slight discrepancy between the official vote tally and the number of votes cast in the database due to voter purges: “A vote tally from a registration file excludes the votes cast by citizens who were purged from the file since the election. For instance, a person who voted in 2006 but was since removed from the rolls would not be included in the county on the registration list but would have an official ballot counted. This presents a minor problem since it only applies to voters who confirmed with the registrar that they moved” (Ansolabehere and Hersh, 2010, p. 15).

17

Previous research on the impact of statewide policy interventions has utilized county or state level data. These data are not well suited for detecting heterogeneous treatment effects and are susceptible to ecological inference concerns. These Catalist aggregations provide precise, accurate outputs of voter turnout with significantly lower sampling error than state-level estimates from the Current Population Survey’s November supplement. I assembled a database from 2004 through 2010 and extracted data across the following background variables: race/ethnicity, age, family income, owner / renter, length of residence and more.50 Catalist Voter Groups. I isolated the number of voters in the following racial groups: African American, Hispanic, White and Nonwhite.51 The racial data in the Catalist database is derived from two sources: self-identified responses in voter files and CPM Ethnics race prediction software. The race variable in the database uses self-identified race in many Southern states, where residents list their racial status when registering to vote. In other states, however, CPM Ethnics assigns a race based on a highly accurate algorithm including the respondent’s first name, middle name, last name, age and characteristics of their Census geography.52 We will have the highest level of confidence when using racial breakdowns based on data coming from the South, where voters self-report their ethnicity. Hersh (2011), for example, finds that the race variable included in the Catalist data file is accurate between 91% and 96% of the time in Southern states.53 States vary in the discrepancy between vote tallies and official results, though the discrepancy is less than 5% in most states: “The 2008 and 2006 vote history discrepancy rates vary considerably by states. In Oregon, North Carolina, Rhode Island, Delaware, and many other states, discrepancies are at a minimum, representing fewer than 5% of all votes. However, in other states like Mississippi, New York, and Texas, the 2008 discrepancy rate is closer to 10%” (Ansolabehere and Hersh, 2010, p. 15). The worst performers are Colorado, Maine, Mississippi and North Dakota. 50 I collected election returns for the 2004 presidential election, the 2006 midterm election, the 2008 presidential election and the 2010 midterm election. Therefore, I can compare the relative impact of Voter ID statutes may exert during presidential contests, midterm elections and multiple cycles after implementation. 51 Separate models account for shifts in statewide population size by using annual Census data from the American Community Survey listing total Voting Age Population and Voting Eligible Population by racial group as a denominator. 52 Given the geographic concentrations of Americans by race, the resulting predictions are highly accurate, though, not without error. Here is a research notes from CPM Ethnics: “In external blind testing against self-reported ethnicity identification, CPM Technologies solutions have shown over 20% more coverage than other established ethnicity appending services...CPM Ethnics software can find over 75% of the African Americans in lists and still maintains an accuracy of over 80%. CPM’s algorithms are based upon modern machine learning techniques and are built using tens of millions of samples with known race.”http:// cpm-technologies.com/cpmEthnics.html 53 “The exact model Catalist uses to predict race is proprietary, but we can check the quality of the prediction using survey responses that have been matched into the Catalist database. The 2009 Cooperative Congressional Election Study (CCES) was matched into Catalist’s database. For the registrants with listed races, 96% of voters’ self-reported races were the same as the publicly listed races. For the registrants whose races were predicted with confidence, 91% had the same self-reported race as predicted by Catalist’s model.

18

I used birth dates listed in voter files to separate voters into the following age cohorts: adults under 25, 25-34, 35-44, 45-54, 55-64, 65-74, 75-84 and 85+.54 I accounted for shifts in statewide population size by age cohort by using annual Census data from the American Community Survey.55 Adults who move more often are more likely to have IDs with outdated addresses. I used household level commercial data to isolate adults who had lived in their current household for the following periods of time: less than one year, one to five years, six to 10 years, 11 to 20 years and more than 20 years. Similarly, I identified individuals who were either renting or who owned their residence. Scholars have hypothesized that Voter ID statutes will place an undue burden on working class and poor Americans. I isolated household incomes in the following ranges using household level commercial data: less than $5,000, $5,001 to $12,50, $12,501 to $20,000, $20,001 to $30,000, $30,001 to $40,000, $40,001 to $60,000, $60,001 to $100,000 and over $100,000.56 Population Denominators. I estimate models with four sets of population denominators. Calculating four separate denominators, each with its own strengths, weaknesses and biases, increases my confidence in the main results. The findings across the four denominators are very similar. The first approach uses the change in total votes cast in a subgroup in the Catalist database over time. This simple, straightforward approach contains limited sampling error because it relies on Catalist aggregations; however, it does not fully account for changes in state subgroup populations between November 2004 and November 2010.57 Though the match between self-reports and the Catalist data is not in perfect agreement, it is sufficiently accurate that each racial group in the Catalist database can be divided in two...” (Hersh, 2011, p. 9-10). 54 Nearly all records contain a birth date. Records with missing age, however, were supplemented using a model including the number of years the individual has been registered to vote, the age of the head of household and the individual’s first name. Importantly, I calculate the age of adults separately for each election. Future analyses can utilize a cohort analysis tracking turnout of the same group of adults over 65, for instance, over time. 55 Ansolabehere and Hersh (2010) find that “1 in 7 records does not have a listed birthdate, and for many voters who do have a listed birthdate, the date entered is inaccurate.” (Ansolabehere and Hersh, 2010, p. 2). The primary inaccuracy, however, is that the voter’s birthdate was entered as the first of the month or as January 1. 56 Both the household income data and the rental status data are largely based on Census block group characteristics. There are more than 200,000 Census block groups, and there is a high level of spatial clustering of income and rental status. Therefore, the data is likely highly accurate and not affected by the fact that the estimates are obtained from data not at the individual level. 57 State subgroup populations change over time, largely due to people changing residence. If, for instance, the number of African Americans in Voter ID states decreases at a higher rate than in non-policy states, we may erroneously conclude that Voter ID laws decrease turnout. For this to affect treatment estimates, though, there have to be systematic differences in population changes across Voter ID and non-policy states.

19

The second approach divides the total votes cast in each subgroup by annual estimates of that group’s Voting Age Population (VAP) and Voting Eligible Population (VEP) from the American Community Survey (ACS). While this approach accounts for changes in subgroup populations over time, it introduces sampling and selection biases. For example, the ACS’s estimates have margins of errors that could swamp a modest treatment effect; further, voting age population estimates inflate the denominator by including groups ineligible to vote such as the large proportion of minorities that are incarcerated or college students who are living in one state but are registered to vote in another.58 The third approach divides the total votes cast in each subgroup by the number of registered voters in Catalist’s national database. This approach estimates the change in the percent of registered voters who cast a ballot, accounts for under coverage in ACS estimates and reduces sampling error. However, across both states and voter subgroups, there may be differential rates of inactive voters, ineligible voters, dead voters or purged voters on registration rolls. The fourth approach divides the total votes cast in a subgroup by the number of individuals in that subgroup in Catalist’s database with a voter history. With this estimation strategy, we reduce the amount of deadwood in voter registration files because we only include citizens who have voted at least once between 2002 and 2010.59

Results and Figures This section isolates groups that scholars have hypothesized will be disparately impacted by Voter ID statutes, such as African Americans, young adults, adults over 65, Hispanics, renters and others. I compare the change in turnout of these subgroups before and after the policy intervention with broader statewide turnout and with turnout among similar groups in control states. I examine the impact of Voter ID statutes separately for presidential and midterm contests. I finally test the long-term impact of legislation by examining changes in turnout among voter subgroups four years after a policy implementation. Difference of means Table 1 displays each component of the difference-in-difference-in-differences estimator.60 First, we estimate the change in turnout among a subgroup, African Americans, before and 58

Again, these selection biases are only problematic from inference if there are systematic differences across Voter ID and non-Voter ID states. Sampling error can be accounted for by repeated simulations. 59 As in the third approach, it is still possible that purging or removal of voters occurs at different rates across states and voter subgroups. 60 Each component is weighted by population.

20

after a policy change. Then, we estimate the change in turnout statewide in policy states. The third and fourth components are group and statewide turnout changes, respectively, in states with no policy implementation. The final estimate of -.005, or half a percentage point, the “DDD estimator,” is the impact of Voter ID statutes among African Americans between 2004 and 2008. Table 1: Change in mean turnout for African Americans and statewide, 2004-08 Policy States Non-policy states 2004 2008 ∆¯ y 2004 2008 ∆¯ y African Americans 0.583 0.675 0.092 0.532 0.629 0.097 Statewide

0.597

0.634

0.037 0.055 -0.005

DDD estimator

0.574

0.611

0.037 0.060

Figures The figures below display treatment effects by subgroup for three separate election comparisons: presidential (2004-08), midterm (2006-10) and midterm (2004-10). The first gauges the impact of Voter ID laws during presidential contests, while the latter two address the impact of Voter ID during midterm elections. In each figure, I aggregate turnout among tens of millions of records in the national voter file by the respective subgroup. Given the size of the data, the confidence intervals are extremely small. The primary model specifications compare turnout between the November 2004 and November 2008 presidential elections. During this period, nine states tightened their ID policies and one relaxed its policies. Overall, I find that Voter ID policies exerted a limited impact on turnout during these presidential elections. The second specification compares turnout between the November 2006 and November 2010 midterm elections. During this period, five states tightened their ID policies and one relaxed its policies. These states include Georgia, Idaho, Michigan, Oklahoma and Utah. Mobilization efforts are less intense during midterm campaigns, and I find that Voter ID laws are more likely to reduce turnout when such mobilization efforts wane.The relative strictness of laws implemented between 2004 and 2008, and those adopted between 2006 and 2010, is similar. Therefore, the differential effects sizes between the two cycles appear to be driven by changes in mobilization rather than increased strictness of election law policies. The third specification, which compares turnout between 2004 and 2010, yields similar results to the second set of models.

21

This section proceeds by analyzing the impact of Voter ID statutes across Americans’ age, income, length of residence, race and home ownership. Age Figure 4 on Page 23 displays the treatment effects across five age cohorts.61 The x-axis displays the percent change in voter turnout for each election and the panels present three separate election models. The treatment effects in the 2004 - 2008 comparison (left panel) are small and suggest that Voter ID statutes have a limited impact on turnout across age cohorts during presidential elections. The youngest and oldest age cohorts are slightly more likely to vote, perhaps due to enhanced mobilization efforts in these Voter ID states. The midterm comparison (center panel) suggests that Voter ID laws reduce turnout among young adults but may lead to a slight increase in turnout among older adults. This accentuates the Republican Party advantage. For example, Voter ID laws reduced turnout among adults under 35 by two percentage points between 2006 and 2010, compared with a two percent increase among adults over 65. This suggests that counter-mobilization efforts inspired by Voter ID statutes may slightly increase turnout among Republican-leaning older adults.62 The right panel looks nearly identical to the center panel, with Voter ID statutes reducing turnout among adults under 35 and adults but generally causing slight increases among older citizens. This figure illustrates the advantages of using granular aggregate data. While Voter ID policies do not appear to affect overall turnout (mean treatment sizes are around 0), there are heterogeneous effects across age cohorts. 61 62

Given the low number of voters under 25, I collapsed the under 25 and 25-34 cohorts. Adults over 65 were the age cohort least likely to support Obama in 2008.

22

Figure 4: Voter ID treatment effects by age cohort and election.

Household income Figure 5 on Page 24 displays the treatment effects across six household income strata ranging from less than $20,000 annually to household income exceeding $100,000. The effects in the 2004 to 2008 comparison suggest that Voter ID laws actually increased turnout among citizens in the lowest income strata. It is possible that during this cycle campaigns targeted such voters in Voter ID states. Voter ID laws do not appear to influence turnout between 2004 and 2008 across other income strata. Among states that changed their policy between 2006 and 2010, Voter ID laws disproportionately demobilize poor and working class voters. Americans in households earning less than $20,000 annually are about three percentage points less likely to cast a ballot in Voter ID states. Voters in the highest income strata exhibit slight turnout increases during this same period. The demobilizing of working class Democratic voters, combined with the mobilization of wealthier Republican voters, heightens the impact of these statutes on the composition of the electorate. The effects are even larger in the model comparing 2004 turnout with 2010 turnout. In 23

states that implemented Voter ID statutes between 2004 and 2008, turnout is reduced among poor voters by approximately five percentage points. Figure 5: Voter ID treatment effects by income strata and election.

Length of Residence Figure 6 on Page 25 displays the impact of Voter ID statutes based on the length of time the current resident has lived in his or her current household. Individuals who have resided in their current household for many years are more likely to own an ID listing their current address. Individuals who have lived in their current residence for less than one year are negatively affected by Voter ID statutes in all three panels, though the effects are most negative among adults who have lived in their household for six to 10 years. Overall, however, the plots (especially the center and right panels) suggest that Voter ID statutes depress turnout among residents who have lived in their current household for a relatively short amount of time. Across the three election comparisons, we never witness declines in turnout among individuals who have lived in their current residence for more than 10 years. 24

Figure 6: Voter ID treatment effects by length of residence and election.

Race Figure 7 on Page 26 displays the impact of Voter ID statutes for African Americans, Hispanics, Whites and all non-whites. The effects are rather small across the three panels. However, the right panel suggests that Voter ID laws cause an approximately 2 percentage point decrease among African Americans while not affecting turnout among white Americans between November 2004 and November 2010. Hispanic voter turnout is not affected in any of the specifications.

25

Figure 7: Voter ID treatment effects by race and election.

Own / rent Finally, I aggregated data based on whether respondents were currently renters or owners of their households. Figure 8 on Page 27 shows that Voter ID statutes do not affect turnout for either group for the 2004 or 2008 presidential elections; however, in the midterm comparison and the 2004-10 comparison, there are significant differences between turnout patterns among renters and owners. The data suggests that Voter ID policies decrease midterm turnout among renters by about seven percentage points and decrease turnout between 2004 and 2010 by 10 percentage points, an even larger treatment effect.

26

Figure 8: Voter ID treatment effects by home ownership and election.

Census Block Group data The previous section analyzes individual level Catalist voter file information. I conduct a separate analysis using aggregate data at the Census block group level. There are more than 200,000 block groups across the country, meaning that the average population for these block groups is approximately 1,500 persons. This secondary analysis presents evidence broadly supportive of the findings in the previous section and introduces a few new measures of interest such as block group level household income. Block Group election returns I obtain Block Group elections returns from the Catalist Voter database. Generally, I divide variables such as income, racial composition and home ownership rates into bins for

27

which there are equivalent U.S. Census Bureau measures. Then, I take the sum of the total votes across the four general election cycles. I download Census block group data using the 2006-2010 American Community Survey’s five-year block group estimates and sum the number of individuals living in block groups with certain characteristics.63 We expect that areas concentration with African Americans, lower income individuals and renters will experience turnout declines subsequent to Voter ID statute implementations. On the other hand, block groups populated with white Americans, the wealthy or home owners should experience little to no impact in their political participation rates. Table 2 on Page 29 displays the change in turnout based on analyses at the block group level.64 The top panel of Table 2 displays voting patterns for individuals based on the household income in their block group. We expect that areas with lower incomes will experience a decline in turnout relative to areas with wealthier residents. Between 2004 and 2008, turnout in block groups in Voter ID states with average household income under $25,000 annually decreased by 1.5 percentage points. In middle class and upper class block groups, the Voter ID intervention either had a minimal effect or a slight positive impact. Voter ID policies reduce turnout among the lowest two income categories for the midterm election comparison and the third comparison as well, while voter turnout among richer households actually increases in policy states. The middle panel of Table 2 displays turnout patterns based on the proportion of African Americans residing in a decile. We expect that areas with mainly African Americans will experience a decline in turnout relative to areas that are homogeneously white. Overall, the data confirm this pattern, though the substantive effects sizes are quite small. Between 2004 and 2008, for example, turnout was unaffected among Americans living in block groups with fewer than 10% African Americans. On the other hand, turnout declined 1.1% in block groups containing 40-50% African Americans and by .5% in block groups with at least half African American residents. The results for the midterm election largely confirm this pattern. 63

For example, I measure the proportion of African Americans living in each block group, divide the measure into deciles and aggregate the total number of individuals living in each decile. These estimations provide a denominator for the Catalist block group election returns discussed in the previous subsection. 64 As a rule, I only display block group categories that contain at least five percent of the American population. This leads me to pool a few results. For example, four percent of Americans live in block groups with fewer than 10% home owners. I sum election returns for all Americans living in block groups with fewer than 30% homeowners. This procedure does not change any substantive takeaways. Similarly, I sum election returns for all Americans who live in block groups with an average household income above $125,000. The Census measure goes up to $250,000, but very few Americans resides in block groups with median household incomes this high. Finally, I sum election returns for Americans living in block group with at least 50% African Americans. Seven in 10 Americans live in a Census block group with fewer than 10% African Americans, meaning that the remaining nine deciles do not have huge populations.

28

Table 2: Block Group Analysis: The percent change in turnout based on Census Block Group characteristics

Household Income at the Block Group Level

2004 vs. 2008 2006 vs. 2010 2004 vs. 2010 % population

< 25k -1.5 -.9 -.3 7

25-50k 0 -2.0 -1.3 39

50-75k .5 -.4 .3 31

75-100k .5 1.4 .2 14

100-125k -.1 1.9 .3 7

125k+ -.3 2.2 3.0 4

% African American at the Block Group Level

2004 vs. 2008 2006 vs. 2010 2004 vs. 2010 % population

2004 vs. 2008 2006 vs. 2010 2004 vs. 2010 % population

0-10 0 .1 .2 70

10-20 .1 .1 -1 11

20-30 -.4 -.2 -.6 6

30-40 -.9 -.5 -.4 3

40-50 -1.1 -.4 .3 2

50-100 -.5 -.8 0 8

% Homeowners at the Block Group Level 0-30 30-40 40-50 50-60 60-70 .2 -.3 -.3 .1 .2 -.1 .2 -.1 -.1 .1 -.7 -.8 -.5 -.9 -.4 11 5 7 9 11

70-80 .2 .3 .3 16

80-90 .2 .2 -.3 21

90-100 -.3 -.5 1.3 20

The bottom panel of Table 2 displays turnout patterns based on the proportion of homeowners at the block group level. We expect that areas populated with renters are more likely to experience decreases in turnout compared with areas with all homeowners. The evidence to support this claim is weaker than in the top two panels of the table. In the main presidential and midterm comparison, the proportion of home owners in your block group does not appear to impact turnout; however, in the 2004-10 comparison, residents residing in block groups with fewer than 50% homeowners experience an approximately 1 percent decline in turnout, while individuals living in block groups with 90%+ home owners experience a 1.3% increase in turnout.

29

Discussion This research suggests that the decade long strategic effort among Republicans shifted the composition of the electorate in the GOP’s favor. I aggregate hundreds of millions of individual voter file records and isolate subgroups that scholars have hypothesized will be adversely impacted by Voter ID statutes. The difference-in-differences approach demonstrates that Voter ID policies can reduce turnout among the young, the poor and among individuals who move frequently. This study presents a template for future scholars to examine heterogeneous treatment effects caused by state level interventions. More research needs to be done on the subject. I qualify my conclusions for the following reasons. First, the results are dependent on a relatively small number of states that modified their election law policies. Second, many of the strictest Voter ID laws have been passed in the last year, and future scholars will need to examine their effects. We will have a more definitive answer to this research question in coming years, after current Voter ID laws have been in effect for a longer duration and after strict Voter ID laws take effect in a series of additional states. Third, while this powerful, data source enables me to aggregate millions of individual level records, it is not without error.

30

References Alvarez, R. M., S. Ansolabehere, A. Berinsky, G. Lenz, C. Stewart III, and T. Hall. “2008 Survey of the Performance of American Elections.” Draft, May 5. Alvarez, R. M., D. Bailey, and J. N. Katz. “The Effect of Voter Identification Laws on Turnout.” Manuscript, California Institute of Technology and Washington University in St. Louis . . “An empirical Bayes approach to estimating ordinal treatment effects.” Political Analysis 19, 1: (2011) 20–31. Ansolabehere, S. “Effects of identification requirements on voting: evidence from the experiences of voters on election day.” PS: Political Science & Politics 42, 01: (2009) 127–130. Ansolabehere, S., and E. Hersh. “The quality of voter registration records: A state-by-state analysis.” Cambridge, Mass.: Department of Government, Harvard University . Ansolabehere, Stephen. “Ballot Bonanza.” Slate . Ashenfelter, O., and D. Card. “Using the longitudinal structure of earnings to estimate the effect of training programs.” The Review of Economics and Statistics 648–660. Barreto, M. A., S. A. Nuno, and G. R. Sanchez. “Voter ID Requirements and the Disenfranchisements of Latino, Black and Asian Voters.” In Annual Meeting of the American Political Science Association, Chicago, Illinois. 2007, volume 30. Biggers, D. R., and M. J. Hanmer. “When Voting Gets Harder: Understanding the Adoption of Voter Identification Laws in the American States.” Presented at the 2011 American Political Science Association Conference . Brady, H. E., S. Verba, and K. L. Schlozman. “Beyond SES: A resource model of political participation.” American Political Science Review 271–294. Cobb, R., D. Greiner, and K. Quinn. “Can Voter ID Laws Be Administered in a RaceNeutral Manner? Evidence from the City of Boston in 2008.” Quarterly Journal of Political Science 7, 1. Erikson, R. S., and L. C. Minnite. “Modeling Problems in the Voter Identification-Voter Turnout Debate.” Election Law Journal 8, 2: (2009) 85–101. Firestone, David. “Will Voter ID Laws Disenfranchise Democratic Voters?” The New York Times . Greenhouse, Steven. “A.F.L.-C.I.O. Takes On Voter ID Laws.” The New York Times .

31

Haines, Errin. “Ga. NAACP focuses on voter outreach after voter ID ruling.” Associated Press . Hersh, E. “The Public Code of Racialized Electioneering.” . Lee, F. E. Beyond Ideology: Politics, Principles, and Partisanship in the US Senate. University of Chicago Press, 2009. Logan, J. R., J. Darrah, and Brown University American Communities Project. The Suppressive Effects of Voter ID Requirements on Naturalization and Political Participation. American Communities Project, Brown University, 2007. Lott, J. “Evidence of voter fraud and the impact that regulations to reduce fraud have on voter participation rates.” . Milyo, Jeffrey. “The Effects of Photographic Identification on Voter Turnout in Indiana: A County-Level Analysis.” Technical report, Institute of Public Policy, University of Missouri, 2007. Mycoff, J. D., M. Wagner, and D. C. Wilson. “The Empirical Effect of Voter-ID Laws: Present or Absent?” PS: Political Science & Politics 42, 1: (2009) 121–126. Pastor, R. A., R. Santos, A. Prevost, and V. Stoilov. “Voting and ID Requirements: A Survey of Registered Voters in Three States.” The American Review of Public Administration 40, 4: (2010) 461–481. Pawasarat, J. “The driver license status of the voting age population in Wisconsin.” Employment and Training Institute, University of Wisconsin-Milwaukee . Sanchez, Gabriel R., Stephen A. Nuno, and Matt A. Barreto. “The Disproportionate Impact of Photo-ID Laws on the Minority Electorate.” Technical report, Latino Decisions, 2011. Shear, Michael D. “Obama Campaign Confronts Voter ID Laws.” The New York Times . Vercellotti, T., and D. Anderson. “Protecting the franchise, or restricting it? The effects of Voter Identification requirements on turnout.” Manuscript, Rutgers University .

32