Donate
Independent, objective, nonpartisan research

R 114EMR

Authors

R 114EMR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3742 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_114EMR.pdf" ["wpmf_size"]=> string(7) "3962794" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(75732) "www.ppic.org Expanding California’s Electorate Will Recent Reforms Increase Voter Turnout? January 2014 Eric McGhee with research support from Daniel Krimm Supported with funding from the S. D. Bechtel, Jr. Foundation SUMMARY O ver the past 20 years, voter turnout in California has been slipping compared to other states, and this decline may be exacerbating the gap between Californians who vote and the rest of the population. The state has considered or undertaken a variety of reforms to reverse these trends. In this report, we explore three of these reforms: a system of online voter registration, a same-day registration process, and a more relaxed deadline for submitting vote-by-mail ballots. One could argue that all of these reforms have inherent value because no eligible citizen should be prevented from voting for what amount to administrative reasons. But the administrative costs of a reform and the number of people who beneft from it matter as well. We fnd that none of these reforms is likely to produce large gains in turnout but two of the three are likely to cost very little or save money. • California implemented an online voter registration system late in the 2 012 election cycle. It was immediately popular, but it probably did not bring many new voters into the elec - torate or signifcantly change the demographic makeup of new registrants. However, the initial rollout probably saved counties considerable time and money. Online registration might have a greater efect on voter registration in the future, and its cost savings make the new system worthwhile. • California has adopted a new system of same-day registration, which will allow voters to register and cast ballots after the close of the ofcial registration period. This system HILL STREET STU DIO S/BLEN D IMAGE S/COR B I S Expanding California’s Electorate 2 www.ppic.org will probably boost turnout a few percentage points. But a substantial number of people may take advantage of same-day registration, which could result in signifcant costs and complications for county registrars. • Proposed reforms designed to ensure that more vote-by-mail ballots are counted by relaxing the deadline would afect only a tiny fraction of total votes cast but a large share of the ballots that are currently rejected. The cost of counting late ballots is probably minor, and the value of counting legitimate votes is great. The online registration and vote-by-mail reforms are worth pursuing despite the small turnout benefts evident so far. And because it seems to encourage earlier registration among some voters, online registration might mitigate some of the negative efects of same-day reg - istration on county registrars. Same-day registration, by contrast, creates an administrative burden that is quite heavy, given its modest efect on turnout. Instead, it might make sense to switch to a system of automatic registration, which would put every eligible Californian on the registration rolls and eliminate the need for the patchwork measures now in place. In any case, the state will have to do more than remove administrative barriers if it wants to expand the size of its electorate. It will need to do aggressive outreach to communities of potential voters who are underrepresented at the polls and often overlooked in get-out-the- vote drives. For the full report and related resources, please visit our publication page: www.ppic.org/main/publication.asp?i=1083 3 Expanding California’s Electorate www.ppic.org Introduction California has been at the forefront of recent eforts to expand the electorate. Te state has considered or passed a number of reforms in recent years to get more people regis- tered and increase voter turnout. Tese include creating an online system that makes registering to vote as easy as pos- sible, scrapping the registration deadline so that citizens who decide to vote at the last minute are not excluded, and relaxing the deadline for vote-by-mail (VBM) ballots so that fewer go uncounted because they are late. Tere is ample evidence that California has a turnout problem. Among the population of citizens who are eligible to vote, the share that casts a ballot has been declining in California compared to other states. Figure 1 shows that in the 1990s, California’s turnout was consistently higher than in the rest of the country, especially in midterm elec- tions when the presidential contest was not on the ballot. Tis advantage has faded in recent years, and now Califor- nia tends to match the national average or fall behind. 1 Tere is also evidence that California has an “exclusive electorate,” dominated by older, whiter, wealthier Califor - nians (Baldassare 2006). Te decline in turnout matters because it may be widening the rif between this exclusive electorate and those who do not (or cannot) vote. In this report, we evaluate three important voting administration reforms that have been adopted or pro- posed in recent years. First, we look at the state’s promis- ing new online registration system and evaluate its efect on the 2012 election. We then examine the likely efect of the state’s new same-day registration law. And, fnally, we look at ways to improve the state’s VBM system, usually by ensuring that fewer voters miss out on the franchise because their ballots arrive too late. To provide a comprehensive overall picture of the value of each reform, we consider both its efect on turnout and its potential to increase efciency or impose additional burdens or costs on election administration. We conclude by outlining the policy implications of this analysis. Online Registration California’s Internet-only voter registration process was established by Senate Bill (SB) 397, which was signed into law in October 2011 and rolled out on September 19, 2012. Te system allows users to enter voter registration information and click a button to complete the process. It replaces the registration form that had long been available online but that had to be printed, signed, and mailed in. Te new system was instantly popular: more than half of all new registrants in the last month of the 2012 fall registration period used it, exceeding even the most opti- mistic expectations. Moreover, several studies in California and elsewhere have concluded that online registration has brought new voters into the electorate and that these new voters were younger, poorer, and more likely to belong to racial or ethnic minority groups (García Bedolla and Vélez 2013; Romero 2013a, 2013b). There is evidence that California has an “exclusive electorate,” dominated by older, whiter, wealthier Californians. Percentage 20102008200620042002200019981996199419921990 2012 Figure 1. California’s voter turnout is declining more than in other states SOURCE: U.S. Census Current Population Survey, November Supplement, 1990–2012. –4 +2 +4 +6 +8 +10 +12 –2 California vs. average of all other states Average of all other states Expanding California’s Electorate 4 www.ppic.org In purely administrative terms, there is little doubt that online registration is more efcient and vastly cheaper than the old paper system. Tere has not yet been a com - prehensive study of the cost savings in California, but an important study in Arizona found that the cost to process a traditional paper form was about 28 times the cost of an online registration form (Barreto et al. 2010). Moreover, an online system signifcantly reduces the risk of error by com - puterizing the entire registration process, eliminating steps in the middle—such as inputting information from mailed- in forms—that led to errors and logistical problems. As long as the old system is available for those who do not have easy computer access, the state should move aggressively to get as many people as possible to use the new system. Efect on Voter Participation Administrative benefts aside, did online registration change the size or composition of the registered voter population? Tere were several diferences between citizens who registered online and the population that registered the traditional way during the same period. Online regis- trants were somewhat less likely to be Latino (22% versus 25%), slightly more likely to be Asian American (9% versus 8%), and somewhat more likely to end up voting on elec- tion day (84% versus 78%). Tey were also younger, with an average age of 35, compared to age 39 for traditional registrants—and adults in their mid-20s were almost three times more likely than senior citizens to register online (Figure 2). 2 But does this mean that online registration drew more or diferent people into the electorate? Or would the voters who registered online have registered anyway, using the traditional process? 3 Figure 3 shows the number of new or changed registra- tions during the period between the fnal two registration reports—60 and 15 days before the election—in 2004, 2008, and 2012. To account for population growth between election years, the numbers are presented as a share of the total unregistered but eligible population—adults legally able to register who had not yet done so as of 60 days before the election. Te results make clear that the surge in registrations in 2012 was consistent with previous presi- Voting online Given the popularity of online voter registration in California, what are the prospects for casting ballots online? In fact, there are two bills currently under consideration in the Assembly that would permit online voting in specifc circumstances: Assembly Bill (AB) 19 would allow counties to conduct Internet voting pilot programs in local elections, and AB 1360 would allow online voting in homeowners’ association elections. Neither bill had passed out of the Legislature by the time this report went to press. Internet voting raises important security concerns that are difcult to address. Online transactions are not reliably secure, and private frms that rely on such transactions lose billions of dollars annually to fraud (Paget 2009). Since these firms are mostly concerned about proftability, they can balance the costs of preventing fraud against the costs of reimbursing the losses of individual customers. Election administrators, by contrast, must ensure the integrity of the voting process. Privacy is another concern. Both financial transactions and voting require authentication of identity, but voting requires a secret ballot after that initial confrmation. It would be impos- sible to replace a lost vote in the same way as one would replace a lost dollar. Since the barriers to Internet voting tend to be structural rather than technical, successful implementation would probably require the state to adopt standards of security and privacy that match those of Internet commerce more closely, a shift in values that many might resist making. Percentage 95 Figure 2. Younger people have been far more likely than older people to register online SOURCE: Political Data, Inc. 0 20 30 40 50 60 70 10 908580757065605550454035302520100 Age of new registrant 5 Expanding California’s Electorate www.ppic.org dential elections. Registration applications were submitted by 25.9 percent of the potential new voters—no more than a percentage point more than in 2004 (25.5%) and 2008 (2 4 .9 %). 4 Te results are similar if total registration is con- sidered as a share of the eligible population. 5 When the numbers in Figure 3 are broken down by week, the pattern of registrants in 2004 and 2008 is strik- ingly similar: registration is slow at frst, then picks up speed, and there are huge registration numbers in the fnal few days. By contrast, in 2012 registration began to increase much earlier, right afer online registration became available, and then slowed down (Figure 4). In other words, the online registration system may have encouraged voters who were planning to register anyway to do so earlier in the cycle, without bringing many new voters into the electorate. Did online registrants turn out in greater numbers? On the one hand, it seems that online registrants were more likely to vote. But this, too, could be a matter of which people availed themselves of the new system: indeed, the voting rate for registrants who signed up just before online registration became available was virtually identical to the rate just afer. 6 Also, the presidential conventions seem to have played an important role: those who registered afer the conventions were about 10 percentage points more likely to vote than those who registered before, presumably because their interest was piqued by media coverage. It is possible that there was something unusual about the 2012 election that depressed registration rates overall or among key subgoups. Might registration have been even lower without an online registration option? To answer this question, we can compare registration in California and other online registration states to those that have not adopted the reform, and do so across multiple election years, not just 2012. Did the online registration states see a gain afer adopt - ing the reform? We conducted a separate analysis of registra - tion rates over time in all 50 states, paying special attention to changes when states have adopted online registration. Consistent with the analysis so far, registration as a share of the eligible population in other states has not changed on average in response to the adoption of online registration. 7 Efect on the Composition of the Electorate Did the online system increase registration in key sub- groups? Te 2012 registration rates for these groups did not change much: the Latino share of registrants was up about a half a percentage point from 2008, whereas the shares of both young people (ages 18 to 24) and Asian Americans were down about a half a percentage point. 8 But these numbers do not settle the matter; there may have been something unusual about 2012 that would have depressed registration rates for all these groups if online registration had not been available. New registrants as a share of potential (%) Figure 3. Online registration did not produce a signifcant surge in new registrants SOURCES: California Secretary of State, Political Data, Inc. 2004 2012 Online registration available 2008 0 5 10 15 20 25 30 25.5 25.9 24.9 New registrants as a share of potential (%) 1 2 3 4 5 6 0 Weeks until registration deadline Figure 4. Online registration shifted registrations earlier in the cycle without increasing the total number of registrants SOURCE: Political Data, Inc. 0 5 6 7 8 9 10 4 3 2 1 Online registration available in 2012 2004 2008 2012 Expanding California’s Electorate 6 www.ppic.org One way to get at this question is to see if there was a change in the type of person who registered af er online registration went live. Here, we do see signs of some ef ect. Figure 5 shows that the share of both young adult and Asian American registrants began climbing right af er the rollout of online registration. To attribute all of this increase in young people to the new registration system, we would have to believe that without the online option, a smaller share of 18- to 24-year-olds would have registered in 2012 than in any other primary or general election since 2002—despite constituting about the same or a slightly larger share of the pool of eligible voters. 9 T us, it seems likely that—as with registrants overall—young voters who would have regis- tered anyway decided to sign up earlier in the cycle. 10 T e same is true for Asian Americans, who constituted about 5.5 percent of new registrants before online regis- tration and about 8.5 percent af er. 11 It is hard to believe that without the online option, Asian Americans would have remained 5.5 percent of new registrants through the close of registration—not only would this have been a lower share than in any election since 2002, but the share of Asian Americans eligible to vote has grown over the past decade. Nonetheless, online registration probably accounted for some of the 3 percent increase in Asian American registration. 12 By contrast, Latino registration actually fell slightly just af er the online system was introduced. T is does not mean that online registration discouraged Latinos from participating but that the surge in non-Latino registration was not matched by Latinos. And, as Figure 6 shows, the share of Latino registration rebounded by the end of the cycle. 13 Because young people, Latinos, and Asian Americans tend to belong to the Democratic Party, some predicted that online registration would increase the number of Democrats. Did online registration help one or the other major party? Figure 7 shows the share of new voters who registered as Democrats and Republicans in the weeks leading up to election day. T e trend lines for party reg- istration are identical before and af er the online option Share of new registrants (%) Figure 5. Young people and Asian Americans registered at higher rates after the rollout of online registration SOURCE: Political Data, Inc. 0 15 20 25 30 35 40 10 5 2211 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 Adults age 18–24 Asian Americans JuneJulyAugustSeptemberOctober Online registration rollout Latino share of new registrants (%) Figure 6. The online option did not signifcantly improve Latino registration SOURCE: Political Data, Inc. 0 15 20 25 30 35 40 10 5 2211 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 JuneJulyAugustSeptemberOctober Online registration rollout Democrats Republicans Democratic convention Republican convention Party share of new registrants (%) Figure 7. There was no clear sign of a shift in party afliation after online registration became available SOURCE: Political Data, Inc. 0 20 30 50 10 22 40 60 11 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 JuneJulyAugustSeptemberOctober Online registration rollout 7 Expanding California’s Electorate www.ppic.org became available, so there is no clear sign here of any efect. Te real efect appears to have come from the presidential conventions, especially the Democratic ones. Afer the Democratic convention, the share of registrants who chose the Democratic Party was almost 10 percentage points higher—and the Republican Party share 5 percent- age points lower—than in the weeks leading up to the conventions. Again, it is not clear how many of these voters would have registered as they did anyway, but the conven- tions do appear to have afected decisions about timing, if nothing else. A Small Efect on Total Registration In short, online registration seems to have altered the dynamics more than the level of registration. More Cali- fornians registered earlier in the cycle and fewer registered late, resulting in a very small overall increase. Other states that have adopted similar reforms have seen only modest increases in registration rates, or none at all, including among key subgroups. It is far too soon to discount the efect of online reg- istration on turnout. Te system was rolled out quite late in the election cycle, slightly more than a month before the registration deadline. It is possible that as the online option becomes known to more people, and once parties and interest groups have time to experiment with diferent mobilization approaches, there will be more gains in voter participation. Same-Day Registration In September 2012, the governor signed AB 1436, which establishes a system of “conditional” voter registration in California. Tis system allows residents who miss the registration deadline (15 days before the election) to both register and vote on any of the remaining days, includ- ing election day itself. Conditional registrants will have to register and vote at their county registrar’s ofce for their ballots to count. 15 To minimize the potential for fraud, con - ditional registration will not be available until the California Online registration seems to have altered the dynamics more than the level of registration. More Californians registered earlier in the cycle and fewer registered late, resulting in a very small overall increase. Is online registration the right tool for the job? Should we expect an online form to be an efective way to increase registration? Californians who are eligible to vote but have not registered are less educated and more heavily Latino—groups that help form the “digital divide” between those who use the Internet and those who do not (Baldassare et al. 2013). Thus, the very people who could beneft most from an online form might not have easy access to it. It is true that those who are not registered to vote are less likely to use the Internet. According to the most recent PPIC Statewide Survey on Internet use, about 17 percent of unreg- istered adult citizens say that they do not use the Internet, compared to 9 percent of registered adults (Baldassare et al. 2 013). 14 The gap is somewhat larger when framed in terms of convenience and frequency: unregistered citizens are less likely than registered citizens to have broadband access at home (62% versus 77%) and are less likely to go online from home at least once a day (57% versus 67%). And, because young people and Asian Americans are among the most frequent users of the Internet, it is perhaps no coincidence that there are signs of an online registration efect for these groups. Nonetheless, these numbers suggest that most unregis- tered Californians have regular access to the Internet. More- over, mobile Internet devices such as Android phones and iPads are equally common among unregistered and registered citizens: two-thirds of each group have access of this kind. Even among those who are both unregistered and have no Internet access at home, a quarter have mobile access. Overall, most unregistered citizens have regular access to the Internet and could make use of the online registration system. Expanding California’s Electorate 8 www.ppic.org Ansolabehere 2002). Many of these studies also ofen argued for an especially sizable gain in key demographic subgroups such as less-educated adults, young adults, racial and ethnic minorities, and newly arrived residents. Tese initial fndings have been challenged in recent years by research that tests causal links. 17 In these newer studies, the maximum estimated efect falls at the low end of the older range of results, topping out at about 4 percentage points for a state with a 15-day closing date (such as Cali - secretary of state fnishes creating a voter registration database that complies with the Federal Help America Vote Act—the database will not be ready before 2016. 16 Te goal of this reform, as described in the analysis of the bill, is to make it easier to vote and so increase voter turnout overall. Many states have a “same day” registration process similar to conditional registration, but they close of access on or just before election day. Other states take the opposite approach: they have “election day registration” (EDR)—that is, same-day registration on election day— but forbid registration in the days and weeks before. In permitting both same-day registration and EDR, Califor- nia will be adopting one of the most permissive registra- tion systems in the country. Tis signifcant reform generates two important ques- tions. Will conditional registration increase voter turnout in California, as its authors intended? And how many new voters might election administrators need to process close to or on election day? Will It Increase Voter Turnout? To assess same-day registration’s potential efect on turn- out, we focus on EDR states—the ones that currently allow voters to register and vote on election day—and set aside those (such as Ohio) that cut of same-day registration sev- eral days before. Since the excitement of election season is highest on election day itself, allowing same-day registra- tion on that day is likely to have a larger efect. Early studies of EDR suggested a sometimes notable efect on turnout. Te efects ranged between 3 and 6 per- centage points, with extrapolated predictions of up to 9 percentage points for a state like California (Alvarez and Does same-day registration increase the risk of voter fraud? Same-day registration gives election administrators less time to verify voter eligibility. Does that make it harder to prevent ineligible voters from participating or eligible voters from casting multiple ballots? Empirical studies produce very little evidence of voter fraud; there are few intentional attempts, and most of those are prevented by administrative review processes already in place to detect and correct inadvertent errors. Qualifca- tion requirements for registration generally exceed those for voting, even where there are voter ID laws, and this applies to same-day registration the same as any other voter registration process. Verifcation of registration information occurs in due course even for same-day registrants, and intentional violators are subject to criminal prosecution. Comprehensive information on the incidence of voter fraud is not available from a single source. One study of six long-term EDR states (Idaho, Maine, Minnesota, New Hamp- shire, Wisconsin, and Wyoming) found news reports of 10 discrete incidents of voter fraud with merit between January 1999 and February 2005 and fve federal convictions for felon voting between 2002 and 2005. An additional survey of 36 out of 252 local jurisdictions in 2004 and 2005 yielded seven cases that were sent warnings but not prosecuted (Minnite 2007). This is similar to the low levels of voter fraud found in non-EDR states overall. However, proper safeguards that apply to all voter regis- tration should be maintained for EDR to prevent problems in the future. Such a system of checks may be somewhat easier to maintain under the California system, since “conditional” registration will be conducted by experienced employees of the county registrar and not volunteer poll workers. In permitting both same-day registration and EDR, California will be adopting one of the most permissive registration systems in the country. 9 Expanding California’s Electorate www.ppic.org fornia), and resulting in an apparent decline in turnout in some cases (Brians and Grofman 2001; Hanmer 2009; Knee and Green 2011; Keele and Minozzi 2013). Te largest efects are in Minnesota and Wisconsin, which adopted the reform in the 1970s and enjoy the highest level of turnout among states; the efects are much smaller in later adopters. 18 Te analysis in this report suggests results on the lower end of this range (Table 1), with a slight decline in turn- out on average for a state shifing from a 15-day close to EDR. 19 Since there is no clear reason for EDR to discourage people from voting, it seems reasonable to conclude that the reform has done little or nothing to boost turnout in most of the states that adopted it. Te fndings were similar for key subgroups, including the young, those who recently moved, and those without a high school diploma. 20 Are these fndings overly optimistic, given that elec- tion day registrants in most EDR states are not required to go to the county registrar but can register and vote at their polling places? Te California EDR law allows counties to set up satellite ofces for EDR registration, and some care could be taken to place satellite ofces in areas with high numbers of potential new registrants (such as university or community college campuses). But for the average EDR voter, these ofces will likely be farther away than a local polling place. Moreover, many counties may not want to take the trouble to ensure that satellite ofces have the full range of ballots in all languages. In its use of county registrars, California’s new sys- tem closely resembles Montana’s EDR process, which was adopted in 2005. A comparison of turnout before and afer Montana implemented EDR gives results ranging from vir- tually no efect on turnout to a more noticeable increase of about 3.9 percent. Tus, it seems unlikely that this aspect of the Montana reform had much efect on EDR. In fact, Iowa, which adopted EDR about the same time and is similar to Montana in many other ways, saw similar results from EDR despite allowing registration at the polling place. 21 Overall, then, the average efect of election day regis- tration ranges from nothing to about a 4 percentage point increase in voter participation. Although larger efects are certainly possible, one can point to evidence of smaller efects as well. However, there is an important diference between Cali - fornia and other EDR states: even the most diverse of these other states are far more racially and ethnically homoge- neous than California. Tat makes it impossible to evaluate the efect of EDR on racial and ethnic subgroups. We have ofered estimates of EDR’s efect on groups that might over - Table 1. Election day registration has modest efects on turnout Data source U.S. Census Current Population Survey (public opinion poll) Election returns Average efect nationwide –0.4%–0.3% Efect on . . . Those with no high school diploma –1.9 %– Those ages 18–24 +2 . 7 %– Those who moved < 6 months ago +3.4%– Efect in . . . Iowa +1.1%–0.2% Montana +0.5%+3.9 % SOURCE: Author’s calculations. NOTE: With the exception of the estimates for Iowa and Montana, the models include the date for the close of registration, and the predictions are calculated assuming a shift from a 15-day close to an EDR system, as would be true for California. Expanding California’s Electorate 10 www.ppic.org and Montana, no more than 4 percent of voters used EDR, compared to 20 percent or more in some other states. Some of this variation may refect difering reporting procedures and defnitions. 22 Tese EDR voters are not necessarily new registrants: roughly half or more of the election day regis - trants who could be identifed were already on the books but had changed addresses. However, these EDR users still take up administrative time and resources. In states for which we have data from both presidential and midterm elections, use of the system has almost always increased in presidential years, even when considered as a share of total turnout. Tus, EDR may place a propor- tionately larger burden on county registrars in high-profle elections. What do the experiences in other states imply for Cali- fornia? Te smallest share in Table 2 (3% in Iowa in 2008) would have translated to approximately 400,000 EDR forms in California in 2012, ranging from a low of 20 in Alpine County to a high of roughly 97,000 in Los Angeles. Note that even the lower-bound estimate for Los Angeles County is more than three times as large as the estimate in the AB 1436 analysis. If late registrations in California had matched the highest rate in Table 2 (21% in Minnesota in 2004), there would have been 680,000 EDR applications in Los Angeles County alone—upward of 100,000 for each branch ofce of that county’s registrar, assuming that the ofces could be set up to handle such applications. Some mitigating factors might ease the burden. Because EDR’s estimated efect on turnout is modest, the total number of applications will likely be on par with previous election cycles. Tus, EDR will probably shif the work- lap with racial and ethnic minorities in California—such as the less educated, the young, and those who recently moved. But, in a sense, California—with its size and diversity— is breaking new ground in adopting such an aggressive reform, and projections should be tempered accordingly. Will There Be a Late Surge? Even if EDR does not expand the electorate, will more voters decide to register on or close to election day? Te analysis that accompanied AB 1436 suggested that the tide of late registrants could be substantial—as many as 30,000 in Los Angeles County alone. Although these voters could certainly register at any point during the 14 days before the election, the greatest surge is likely to come on election day, creating a potentially signifcant administrative challenge. County registrars, who will have to cope with this surge with limited resources, are understandably concerned about how large it is likely to be. Table 2 shows the share of all voters casting ballots who registered afer the ofcial close of the registration period in each of the EDR states. Many states do not report regularly on their EDR voters, so there is a lot of missing informa - tion, but it is possible to make a few generalizations. First, the use of EDR varies greatly from state to state. In Iowa One analysis of same-day registration suggested that the tide of late registrants could be substantial— as many as 30,000 in Los Angeles County alone. Same-day registration could result in modest increases in turnout but could also place a burden on county registrars. M I K E BLA KE/ REUTER S/COR B I S 11 Expanding California’s Electorate www.ppic.org load of processing these applications to later in the cycle without signifcantly increasing it. Moreover, California’s online registration system may lessen this burden: it seems to have the opposite efect of EDR, encouraging voters to register earlier than they had in previous election cycles. 23 It seems reasonable to assume that EDR registration in California will not match the usage levels of states that allow EDR at local polling places. Moreover, there is no guarantee that the numbers will match even those of Montana, which also requires that EDR voters go to county registrars. But given the size of even the lower esti - mates, it would make sense for registrars to be prepared for a large surge. On balance, the modest potential efect of same-day reg - istration on turnout does not seem worth the cost of process - ing late registrations. Since the system is now law and seems unlikely to be repealed, preparing for the change is likely the best short-term solution. In the conclusion, we touch on longer-term reforms that might alleviate the problem. Improving the Vote-by-Mail Process California adopted “absentee” voting in 1923 so that “any duly registered voter, who, by reason of his [sic] occupation is regularly required to travel about the state” could cast a ballot by mail (California Statutes of 1923, Chapter 283). In 1978, California was the frst state in the nation to permit “no excuse” absentee voting (California Statutes of 1978, Chapter 77), and in 2001 the state created a “permanent absentee” option (California Statutes of 2001, Chapter 922), making it as easy to vote by mail as to vote in person. Since the permanent VBM option was introduced, its popularity has skyrocketed: a majority of ballots are now cast by mail (Figure 8). VBM may be more convenient than voting at a polling place, but it complicates the process. 24 Current law requires that VBM ballots arrive on or before election day. Unless voters decide to drop of ballots at their polling places on election day, they have to rely on the U.S. Postal Service (USPS) to deliver their votes on time. Senator Lou Correa has introduced a bill (SB 29) that would address this issue. Tis reform would consider VBM ballots valid if they are postmarked by election day and arrive by the third day afer the election. Table 2. Election day registration states often have large numbers of late registrants Y ear IowaIdaho Maine Minnesota Montana New Hampshire Wisconsin Wyoming 2000 17 % 11% 2002 12 % 7%7% 2004 20% 2006 12 % 6 % 2008 3% 7 % 4% 15% 16% 2010 9% 2 012 4% 8% 4% 11% SOURCES: Various Secretaries of State, Federal Election Assistance Commission (2008, 2010, 2012). NOTE: Cell entries are the share of all voters in each state and year who registered for the frst time or changed registration after the close of the ofcial registration period, meaning that they took advantage of the election day registration law. VBM may be more convenient than voting at a polling place, but it complicates the process. Expanding California’s Electorate 12 www.ppic.org one of every fve uncounted VBM ballots in most of the counties listed here. Tere is no reason to think that the ballots were fraudulent, which makes their rejection akin to disenfranchisement. And, in a close election, even these small numbers could make a big diference. 25 For 24 of these 31 counties, the CACEO has also provided information about the lateness of the ballots and whether they had valid postmarks. 26 According to those data, ballots later than three days afer the election accounted for about 4 percent of all late ballots across all 24 counties and accounted for more than 15 percent of such ballots in only two of the 24. So a three-day window would probably allow the vast majority of the late ballots to be counted. Some ballots did not have valid postmarks, making it difcult to know whether they were cast on time. Te current version of SB 29 allows those ballots to be counted if the date that accompanies the voter’s signature on the envelope falls on or before election day. Such bal- lots account for 12 percent of all late VBM ballots arriving within the three-day window in these 24 counties, with only two reaching as high as one-third. 27 Are some groups of voters hurt more by late ballots than others? Table 4 presents the results of a regression analysis of how well a variety of voter characteristics corre- lated with submitting a late VBM ballot in the 2012 general Potential Efect Before we can assess the efect of this legislation, we need to determine the scope of the problem. Counties do not consistently gather data on late ballots, but we analyzed data from Political Data, Inc., and the California Associa- tion of Clerks and Elected Ofcials (CACEO) that cover 31 counties for the 2012 election. Tis accounts for two- thirds of all VBM ballots that year. Because these late ballot totals are not always broken down by their day of arrival, they are likely to generate high estimates of SB 29’s potential efect. Table 3 makes clear that late ballots are extremely unusual—they rarely constitute more than a half a percent of all VBM ballots cast. But lateness accounts for at least Percentage of ballots submitted by mail Figure 8. The vote-by-mail option has exploded in popularity SOURCE: California Secretary of State. 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 20120 10 20 30 40 50 60Permanent VBM begins No-excuse absentee begins Table 3. Late ballots are extremely rare but constitute a large share of those that go uncounted Total late ballots statewide 18 ,0 6 4 Share of VBM ballots cast 0.40% Share of uncounted ballots 4 7. 0 0 % SOURCES: Political Data, Inc., CACEO (total late and VBM ballots), California Secretary of State (total uncounted VBM ballots by county). NOTES: Political Data, Inc. provided data for Calaveras, Madera, Mariposa, San Francisco, Sierra, Sonoma, and Tehama counties. CACEO provided data for Amador, Fresno, Humboldt, Inyo, Los Angeles, Mono, Orange, Plumas, Sacramento, San Benito, San Diego, Santa Clara, and Siskiyou counties. Data for the remaining counties were available from both sources. Where the sources did not match, the higher number is reported. Table 4. Some groups are slightly more likely to submit late ballots Likelihood of late ballot New registrant since 2010 No clear diference Voted in 2010 Lower by 0.36% Ag es 18 –24 Higher by 0.86% Democrat Higher by 0.12% Independent Higher by 0.22% Latino No clear diference Asian American/Pacific Islander Higher by 0.13% SOURCE: Political Data, Inc. NOTES: Each number represents the change in the predicted percentage of people who would submit a late ballot for those in each group compared to the rest of the VBM population. Likelihood is calculated from a logit regression of late ballot status on the identifed variables, with fxed efects for counties. The sample was limited to VBM voters who returned a ballot in the 17 counties identifed in Table B11 of Technical Appendix B. “No clear diference” refers to efects of less than 0.10%. 13 Expanding California’s Electorate www.ppic.org analysis suggests that VBM voters in consolidated zip codes were slightly less likely to have a late ballot in 2012, at least in those counties for which we have late ballot information. 29 In sum, late ballots are not a source of concern for overall turnout. But because they constitute a signifcant share of uncounted ballots, SB 29 would signifcantly reduce the number of disenfranchised voters. Moreover, because the number of ballots involved is small, counting them would not impose much additional administrative burden, especially in counties that already have an exten- sive post-election validation process. Tere may be no reason to worry that USPS cost- cutting will result in more late ballots. But the USPS is not election for the 17 counties for which we have data for individual voters. Each number is the separate diference in likelihood that is attributable to that group, independent of all the other characteristics listed in the table. Te larg- est diferences generally have to do with a lack of experi- ence with the act of voting. Voters young enough to have limited voting experience are more likely to submit a ballot late, whereas those who voted in 2010 are less likely. Being a new registrant makes no diference independent of being young, but many of these “new” registrants might have been registered somewhere else before the 2012 election. Relying on the U.S. Postal Service Most VBM voters count on the USPS to get the ballot to elec - tion ofcials on time. Te USPS has a strong delivery record, but some mail does get delayed or lost. If a delivery delay prevents a ballot from arriving on time, then it can be said that the USPS has unintentionally disenfranchised a voter. Tis is not an idle concern. Te use of traditional mail has plummeted even as the popularity of VBM has rapidly increased. Tis has forced the USPS to cut costs; it has laid of employees and closed local post ofces, and it is contemplating ending Saturday delivery. It has also been closing many of its processing and distribution centers (P&DCs). P&DCs are hidden to most postal customers. Tey are not points of contact for the general public but large-scale facilities that sort the mail and direct it to its destination. Tese P&DC closures have been quite signif- cant: 39 percent nationwide were closed between 2006 and the present, including seven in California between 2010 and 2012 alone. Still more closures are planned. Afer a center is closed, the mail it used to process is handled by a nearby existing facility. Because of the greater distance and the potential for slower processing in the newly con- solidated facilities, there is always the chance that voters whose mail is consolidated will see slower delivery times. Te average voter is not likely to know whether his or her P&DC has been closed. Can we attribute any late ballots to the USPS consoli- dation process? To assess this possibility, we factored P&DC consolidation into the regression analysis in Table 4. 28 Tis Late ballots are not a source of concern for overall turnout. But because they constitute a signifcant share of uncounted ballots, counting them would signifcantly reduce the number of disenfranchised voters. The closing and consolidation of USPS processing and distribution centers could afect vote-by-mail ballots. JOH N GRE SS/COR B I S Expanding California’s Electorate 14 www.ppic.org VBM system in the future. But the administrative costs seem small enough—and the efect on the problem of late ballots large enough—to make the reform worth pursuing.Given that the removal of virtually all administrative barriers helps only at the margins, the state should shif its focus to motivating Californians to take advantage of the systems that have been put into place. In short, we should shif from facilitating voters to mobilizing them. 30 A number of options could be pursued. First, online registration should be viewed not as an end in itself but as a frst step toward a diferent system of mobilization. Compared to other forms of communication, the Internet is highly scalable: it costs only slightly more to reach 10,000 people than it does to reach 10. And because the state’s new online registration system is completely Internet-based, it can be integrated seamlessly with any electronic appeal to register and vote. 31 Tis can allow a great deal of experi- mentation with diferent forms of outreach at relatively low cost. Moreover, the secretary of state can take further steps to make the new system easier to use with mobile Internet devices. If online registration can make electronic outreach more efective in getting voters registered, the potential gains could be quite signifcant. Only two (Colorado and Nevada) of the 12 states that have adopted online registra- tion have been in play in a presidential election when the system was in place. A close presidential race is the cir- cumstance most likely to make a mobilization efort strong and efective, so as more states adopt the reform, it may yield greater results. Mobilization eforts need to reach beyond the “low- hanging fruit” of those who are likely to register and vote but have not yet done so. To increase turnout substantially, these eforts will need to target hard-to-reach citizens who probably would not vote at the same rate as those who are currently registered. But the turnout rate among these unregistered citizens is currently zero, so even an extremely low turnout rate would be an improvement. In fact, it would probably make sense to think of both online registration and same-day registration as way sta - tions en route to a system that automatically registers every done with the cuts, and more radical changes could be in store. Relaxing the deadline for VBM ballots seems like a modest way to forestall any adverse efects and make sure that all valid votes are counted. Policy Implications None of the reforms examined in this report is likely to have a large efect on voter turnout. Online registration may have brought a few more people into the system in 2012, but it mostly encouraged voters to register earlier than they might have otherwise. At the same time, it is much cheaper and more accurate than the paper system, so its administrative benefts are potentially substantial. Election day registration is similar in that its efect on voter turnout will probably be no more than a few percentage points (although for California that would still mean hun- dreds of thousands of new voters). But the administrative efect might be large. Certainly, county registrars ought to prepare for a deluge of late registrants under the new sys- tem, since the numbers involved may be larger than even some existing estimates that seem large already. Online registration might ofset this surge somewhat by encourag- ing voters to register earlier in the cycle, but the possibility for this efect in future years is uncertain enough that it would be prudent not to count on it. Finally, eforts to deal with late VBM ballots will afect only a tiny fraction of the ballots cast, and there is no sign that cutbacks at the USPS will create problems for the To increase turnout substantially, mobilization eforts will need to target hard-to-reach citizens who probably would not vote at the same rate as those who are currently registered. 15 Expanding California’s Electorate www.ppic.org eligible California citizen. California does have a “motor voter” system in place so that citizens can register when they get drivers licenses or engage with government in other ways. But this system is not as simple or well pro- moted as it was intended to be. 32 Recent legislation (SB 35) by Senator Alex Padilla that was signed into law would improve many aspects of this system and expand it to more government ofces. But citizens must still actively choose to register. Te system could instead be made “opt out,” mean- ing that voters would automatically be registered based on information provided to various government agencies. To opt out, they would need to remove their names from the registration rolls. 33 Automatic registration would not force residents to divulge any more information to the govern- ment than they had already revealed for other purposes; it would simply use the information already provided to determine voter eligibility. Even if this approach were to register no more citizens than election day registration, it would likely result in a more even distribution of new The turnout rate among unregistered citizens is currently zero, so even an extremely low turnout rate would be an improvement. California should consider reforms that target and mobilize hard-to-reach voters. GET T Y IMAGE S Expanding California’s Electorate 16 www.ppic.org Technical appendices to this report are available on the PPIC website: www.ppic.org/content/pubs/other/114EMR_appendix.pdf registrants throughout the election cycle and save county registrars the challenge of large pre-election surges. What might be the limitations of such a reform? Some voters might not want to register for philosophical or admin - istrative reasons, 34 so it would be important to make the choice to opt out very clear. Tere might also be costs that ofset the administrative gains—for example, the state would need to send election materials to a much larger group of reg - istered voters. Tere would need to be further cost estimates before the system could be considered a viable option. An automatic registration system would not neces - sarily increase turnout much by itself. 35 Election day registration, which in most states is only slightly more burdensome on the average voter, appears to have had only a modest efect in those states. It would have to be coupled with aggressive outreach to bring new voters to the polls. More generally, voter turnout cannot be improved solely by administrative means. Increasing and diversifying voter participation is an ongoing process of motivating more Californians to exercise their right to vote. 17 Expanding California’s Electorate www.ppic.org Notes 1 Tese results are consistent with and extend those found by Ramakrishnan and Baldassare (2004). Te numbers in this fgure and the next were calculated by estimating a separate logit regression for each election year with a dummy for California and then predicting the frst diference estimate for this Califor- nia dummy. Coefcients and model ft for all 12 election years are available from the author on request. 2 All of the data for this section, and for any later analysis that uses voter registration records, were provided by Political Data, Inc. Te frm has a copy of the voter registration fle, to which it has coded race based on registrant surnames (see note 8) and appended a wide range of information about the disposition of each ballot. Te frm is widely considered the best source for data on online registration. 3 Tere are certainly reasons to believe that an online registra- tion system might boost registration rates. For instance, in one study, 29 percent of eligible but unregistered Californians said that they did not know where they could fnd registration forms (Alexander 2004). An online registration system might be easier for these citizens to fnd and use. But this is far from guaranteed: the online system might not help as much as intended, and these unregistered voters might have reasons for not registering other than a lack of convenience. 4 Te secretary of state’s percentages of total change in registra- tion over these same periods are much lower because the secre- tary is obligated to purge old registrants as well as add new ones. Te comparable changes in total ofcial registration are 14.5 percent (2004), 16.2 percent (2008), and 15.1 percent (2012). 5 Total registration as a share of the eligible population was 75.0 percent in 2004, 74.6 percent in 2008, and 76.7 percent in 2012. 6 Tese results are available from the author on request. 7 To obtain these estimates, we used the Current Population Survey data to conduct a logit regression of registration on online registration status, with fxed efects for states and years to control for unchanging diferences across states and uniform changes over time, plus demographic controls to account for unusual changes in a state that had nothing to do with online registration. Te resulting estimates identify how much online registration states changed when they adopted the reform, relative to both the changes in other states in the same year and the online registration states’ own baseline rates from before the reform. We also tried omitting the demographic controls and obtained a similar result. Coefcients and model ft can be found in Technical Appendix B. 8 We should be somewhat cautious when comparing the Latino and Asian American numbers from 2012 to those from earlier elections. Latino and Asian American registrants are identifed in the voter registration fle by surname (which almost certainly leads to some unknown degree of undercounting). Te surname estimates for 2012 come from Political Data, Inc., whereas those from 2002 through 2010 come from the statewide database. Tese sources use slightly diferent methods for identifying sur- names, so diferent numbers could be attributable to the method alone. Tis caveat is not relevant when comparing Latino and Asian American estimates over time within a single election year, or for any analysis involving age. 9 To obtain this estimate, we projected three trend lines from before online registration to the period afer it was available: one using all the pre-online registration data; one using only the last 20 days, which was a period when young people were a declining share of new registrants; and one assuming that the average registration rate just before online registration would remain constant. We then assumed that any diference between this new projected registration rate and the actual one was entirely explained by a loss of young voters (rather than an increase in older ones). Te estimates ranged from a total of 272,000 young registrants lost to 334,000 lost. Tis would have dropped the share of young people in the electorate well below 8 percent for the frst time in any statewide election since 2002. Details of this estimate are available from the author on request. 10 Since registration rates over time for young people in earlier years are not available, it is not possible to confrm this idea d i re c t ly. 11 A regression discontinuity (RD) analysis of these data con- frms a statistically signifcant increase in the Asian American registration rate right afer online registration became available. However, because there was a small surge in Asian American registration immediately before online registration went live, the RD estimate is much smaller than the 3 percent identifed in the text and is sensitive to the size of the bandwidth used in the estimation. We used sofware from Nichols (2012) to implement the Imbens and Kalyanaraman least-squares method to identify the relevant bandwidths. Details of these estimations are avail- able from the author on request. For details of the method, see Imbens and Kalyanaraman (2009). Expanding California’s Electorate 18 www.ppic.org 12 We ran the analysis of the efect of online registration on turn- out in other states for these key subgroups. For Latinos, it did suggest a boost in registration rates of about 2 percentage points under online registration, making California perhaps something of an outlier. However, a similar analysis limited just to young people ages 18 to 24 suggested no efect at all. Tese results are available from the author on request. Te population of Asian Americans in the states that have adopted online registration is generally too small for separate analysis. 13 A regression discontinuity analysis of these data confrms a statistically signifcant drop in the Latino registration rate right afer online registration became available. Te statistical signifcance of this discontinuity is sensitive to the bandwidth employed in the estimation, but its negative sign is not. We used sofware from Nichols (2012) to implement the Imbens and Kalyanaraman least-squares method to identify the relevant bandwidths. Details of these estimations are available from the author on request. For details of the method, see Imbens and Kalyanaraman (2009). 14 Self-reported registration rates in the PPIC samples exceed ofcial rates by about 10 percentage points, because of some unknowable combination of nonresponse bias and exaggeration by the respondents who agreed to participate. Te frst type of bias might exaggerate Internet use among the unregistered if the more Internet-sav vy among them are more likely to participate in the survey. However, the second type of bias will probably understate Internet use among the unregistered, since those most inclined to inaccurately claim that they are registered may also be the sort of educated, engaged population most likely to use the Internet, thus sapping the unregistered category of some of its heaviest Internet users. 15 Although this approach may have been adopted for any num- ber of reasons, it does make it easier to manage the potential for a single voter to cast a conditional ballot at multiple polling places. It also helps minimize the administrative complexity of ensuring that the proper ballot—in terms of either the races and initiatives ofered or the language they are translated into— would be available for any voter who might show up to register. 16 Te Help America Vote Act declares that states must have “a single, uniform, ofcial, centralized, interactive computerized statewide voter registration list” and they “shall enter into an agreement to match information in the database of the statewide voter registration system with information in the database of the motor vehicle authority to the extent required to enable each such ofcial to verify the accuracy of the information provided on applications for voter registration.” 17 Te newer methods try much harder to control for any unmeasured diferences between political entities that have adopted EDR and those that have not, to ensure that EDR itself is actually the cause of any diferences in turnout. Tese range from diference-in-diference models that use the change before and afer the adoption of EDR as a measure of its efect to regres- sion discontinuity designs that compare jurisdictions that just qualify for EDR according to some metric to those that just fail to qualify. 18 Hanmer (2009) argues that the early adopters saw more of an efect because they embraced EDR as part of their established participatory culture. States that adopted EDR grudgingly—such as Idaho and Wyoming, which sought to avoid coverage under the National Voter Registration Act in the early 1990s—had popula - tions that were less amenable to the reforms in general and so less likely to take advantage of them. California likely falls between these two extremes: the state adopted the reform voluntarily but it has not been an especially high-participation state, at least recently. However, see Keele and Minozzi (2013) for evidence of small or even negative efects even in Minnesota and Wisconsin. 19 Our analysis consists of a time-series cross section of voter turnout for all 50 states, with fxed efects for states and years and other key controls. We estimated this model both with actual election returns and with individual-level survey data from the Current Population Survey. Te detailed results of these models are in Technical Appendix B. 20 For California, both Asian Americans/Pacifc Islanders and Latinos are important groups to watch. However, the size of these groups in EDR states is too small for reliable analysis. 21 We conducted a diference-in-diference analysis of the change in Montana, using Idaho, Wyoming, North Dakota, and South Dakota as comparison states and controlling for age, marital status, gender, education, employment status, and mobility as demographic controls. In this model, the estimated efect of EDR was a statistically insignifcant increase in turnout of 1 percent. Coefcients and model ft can be found in Technical Appendix B. 22 Many of these numbers come from the Federal Election Assistance Commission, which conducts a survey of state elec- tion ofcials every two years. Te question wording for this survey refers to “same day registration” and does not ask states to identify what this means. 23 Currently, there are no states with both online registration and election day registration. Tus, it is impossible to say just what the efect of this combination might be. 19 Expanding California’s Electorate www.ppic.org 24 Overall, the evidence is mixed on a positive efect of VBM on turnout, but several studies have found a modest efect in at least certain circumstances. For a good summary, see Gerber, Huber, and Hill (2013). For evidence that VBM may actually reduce turnout in California, see Kousser and Mullin (2007). 25 Te closest congressional or state legislative outcome in the 2012 general election was Assembly District 36, where Demo- crat Steve Fox defeated Republican Ron Smith by just 145 votes. Unfortunately, we do not have late ballot data for that race in particular, so we cannot say whether late ballots could have fipped the outcome. 26 Te 24 counties are Amador, Butte, Colusa, Contra Costa, Fresno, Glenn, Humboldt, Inyo, Lassen, Los Angeles, Merced, Mono, Napa, Orange, Placer, Plumas, Sacramento, San Benito, San Diego, Santa Clara, Shasta, Siskiyou, Sutter, and Ventura. 27 Tere is always the chance that this number could increase under SB 29 if voters waited until they knew the election out- come and then fraudulently declared that their ballot was cast on time. However, since the ballots would have to be certain to arrive without a postmark, the fraud would require impersonat- ing a member of the postal service to deliver the late ballot or ballots without going through the regular mail. Although this is certainly possible in theory, it would seem to be difcult to accomplish without being detected. 28 Tis research design necessarily ofers only an approximate estimate of the causal efect of consolidation, because it is always possible that voters in the consolidated zip codes difer from others in important ways that have not been controlled for in the model. “Consolidation” in this analysis is treated as having one’s P&DC closed and merged with another. We also tried treating as “consolidated” all those facilities that received customers from closed facilities, on the assumption that their workload would increase and reduce the efciency of mail processing. Tere was no diference one way or the other from consolidation in that analysis. 29 Te diference is very small, so its importance should not be overstated, and its direction likely says something about the zip codes that were consolidated rather than the act of consolida- tion itself. Nonetheless, it casts doubt on concerns that continu- ing cuts at the USPS will lead to more late ballots. We also tried a somewhat diferent specifcation that allowed us to use data from all 58 counties. Instead of a late ballot fag, our dependent variable was a fag for whether a registered voter cast a ballot that was counted. We regressed this variable on VBM status, P&DC closure, and an interaction between the two. A nega- tive interaction term would suggest that VBM voters who were also in P&DC zip codes would be especially unlikely to have a counted ballot. We also controlled for the same demographics found in the original model. Te results were quite consistent with the fndings for the late ballot fag, regardless of the defni- tion of “consolidation” that we used. Coefcients and ft for these models can be found in Technical Appendix B. 30 See Berinsky (2005) for a similar point about the relative value of reform versus mobilization. 31 Some research has shown that social media have great poten- tial to mobilize citizens to go to the polls, and it seems relatively easy to incorporate online registration more thoroughly into such eforts (Bond et al. 2012). 32 Project Vote did a study of state agency compliance in Califor- nia and found that it had declined steeply since 1995–1996, when the Federal National Voter Registration Act frst required that such registration be ofered in 1993 (Herman and Hess 2009). 33 Te state with the closest existing approximation of this system is North Dakota, which has no voter registration at all. However, its system predates our data, so we have no robust means of evaluating the efect of its approach. Oregon recently considered adopting exactly the system described here, but the legislation fell one vote short in its state Senate (Holeywell 2013). 34 For instance, university students from out of state might want to maintain residency in their state of origin instead of being registered in California simply because they engage with a gov- ernment agency here. 35 One study that examined the efect of establishing voter registration in Ohio and New York in the 1970s estimated that it reduced turnout by about 3 to 5 percentage points (Ansolabehere and Konisky 2006). Expanding California’s Electorate 20 www.ppic.org 20 References Alexander, Kim. 2004. California Voter Participation Survey: Registering to Vote. Sacramento: California Voter Foundation. Alvarez, Michael, and Stephen Ansolabehere. 2002. California Votes: Te Promise of Election-Day Registration . Available at w w w.vote.caltech.edu/sites/default/fles/california_votes.pdf. Ansolabehere, Stephen, and David M. Konisky. 2006. “Te Introduction of Voter Registration and Its Efect on Turnout.” Political Analysis 14 (1) : 8 3 –10 0 . Baldassare, Mark. 2006. California’s Exclusive Electorate . San Francisco: Public Policy Institute of California. Available at w w w.ppic.org/main/publication.asp?i=705. Baldassare, Mark, Dean Bonner, Sonja Petek, and Jui Shrestha. 2013. PPIC Statewide Survey: Californians and Information Technology . San Francisco: Public Policy Institute of California. Available at w w w.ppic.org/main/publication.asp?i=1064. Barreto, Matt A., Bonnie Glaser, Karin MacDonald, Loren Collingwood, Francisco Pedraza, and Barry Pump. 2010. Online Voter Registration (OLVR) Systems in Arizona and Washington: Evaluating Usage, Public Confdence and Implementation Processes . Washington, DC: Pew Center on the States. Available at w w w.pewstates.org/research/reports/online-voter-registration -85899378 4 69. Berinsky, Adam J. 2005. “Te Perverse Consequences of Elec- toral Reform in the United States.” American Politics Research 33(4): 471– 4 91. Bond, Robert M., Christopher J. Fariss, Jason J. Jones, Adam D. I. Kramer, Cameron Marlow, Jaime E. Settle, and James H. Fowler. 2012. “A 61-Million-Person Experiment in Social Infu- ence and Political Mobilization.” Nature 489(7415): 295–298. Brians, Craig Leonard, and Bernard Grofman. 2001. “Election Day Registration’s Efect on U.S. Voter Turnout.” Social Science Quarterly 82(1): 170 –183. Davern, Michael, Arthur Jones, Jr., James Lepkowski, Gestur Davidson, and Lynn A. Blewett. 2007. “Estimating Regression Standard Errors with Data from the Current Population Survey’s Public Use File.” Inquiry 4 4(2): 211–224 . García Bedolla, Lisa, and Verónica N. Vélez. 2013. Diferences among Latina/o, Asian American, and White Online Registrants in Californi a. Berkeley: University of California, Berkeley, Center for Latino Policy Research. Available at http://escholarship.org /uc/item/44k3s91p. Gerber, Alan S., Gregory A. Huber, and Seth J. Hill. 2013. “Identifying the Efect of All-Mail Elections on Turnout: Stag- gered Reform in the Evergreen State.” Political Science Research and Methods 1(01) : 91–116 . Hanmer, Michael J. 2009. Discount Voting: Voter Registration Reforms and Teir Efects . Cambridge: Cambridge University Press. Herman, Jody, and Doug Hess. 2009. California’s Failure to Comply with the Public Agency Registration Requirements of the NVRA . Research Memo. Washington, DC: Project Vote. Available at w w w.projectvote.org/images/publications/State -SpecificDocuments/California/FINALPVCA NVRA5.7.09.pdf. Holeywell, Ryan. 2013. “Oregon May Be 1st with Automatic Voter Registration.” Politics (blog), Governing, April 5. Honaker, James, Gary King, and James Blackwell. 2009. “Amelia II: A Program for Missing Data.” Available at http://j.mp/k4t8Ej. Imbens, Guido, and Karthik Kalyanaraman. 2009. Optimal Bandwidth Choice for the Regression Discontinuity Estimator . National Bureau of Economic Research, Inc. Available at http://EconPapers.repec.org/RePEc:nbr:nberwo:14726. Keele, Luke, and William Minozzi. 2013. “How Much Is Min- nesota Like Wisconsin? Assumptions and Counterfactuals in Causal Inference with Observational Data.” Political Analysis 21(2): 193 –216. King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data: An Alterna- tive Algorithm for Multiple Imputation.” Te American Political Science Review 9 5 (1) : 4 9 – 6 9. Knee, Matthew R., and Donald P. Green. 2011. “Te Efects of Registration Laws on Voter Turnout: An Updated Assessment.” In Facing the Challenge of Democracy: Explorations in the Analy- sis of Public Opinion and Political Participation , ed. P. M. Snider- man and B. Highton. Princeton, NJ: Princeton University Press, pp. 312–328. Available at http://press.princeton.edu/titles /9602.html. 21 Expanding California’s Electorate www.ppic.org Kousser, Tad, and Megan Mullin. 2007. “Does Voting by Mail Increase Participation? Using Matching to Analyze a Natural Experiment.” Political Analysis 15(4): 428–445. McDonald, Michael. 2012. “Is Minority Voter Registration Really Declining?” Hufngton Post Blog, May 7. Available at w w w.huffingtonpost.com/michael-p-mcdonald/is-minority -voter-registr_b_1497813.html. Minnite, Lorraine Carol. 2007. Election Day Registration: A Study of Voter Fraud Allegations and Findings on Voter Roll Security . Available at www.demos.org/sites/default/files /publications/edr_fraud.pdf. Mitchell, Glenn E. II, and Christopher Wlezien. 1996. Voter Registration and Election Laws in the United States, 1972–1992. Ann Arbor, MI: Inter-university Consortium for Political and Social Research Data set ICPSR01102.v1. Nichols, Austin. 2012. RD: Stata Module for Regression Discontinuity Estimation. Sofware component provided by Boston College Department of Economics. Paget, François. 2009. Financial Fraud and Internet Banking: Treats and Countermeasures . Santa Clara, CA: McAfee, Inc. Available at www.mcafee.com/us/resources/reports/rp-financial- fraud-int-banking.pdf. Ramakrishnan, S. Karthick, and Mark Baldassare. 2004. Te Ties Tat Bind: Changing Demographics and Civic Engagement in California . San Francisco: Public Policy Institute of Califor- nia. Available at w w w.ppic.org/main/publication.asp?i=410. Romero, Mindy. 2013a. Online Voter Registration: Impact on California’s 2012 Election Turnout, by Age and Party Afliation . Davis, CA: Center for Regional Change, University of Califor- nia, Davis. Available at http://regionalchange.ucdavis.edu /ourwork/publications/ccep/ucdavis-ccep-brief-4-online-voter -turnout-impact. Romero, Mindy. 2013b. California’s 2012 Youth Voter Turnout: Disparate Growth and Remaining Challenges . Davis, CA: Center for Regional Change, University of California, Davis. Available at http://regionalchange.ucdavis.edu/ourwork/publications/ccep /ucdavis-ccep-brief-5-youth-voter-turnout. Expanding California’s Electorate 22 www.ppic.org 22 About the Author Eric McGhee is a research fellow at PPIC. His work focuses on elections, political participation, political polarization, legislative behavior, redistrict - ing, and surveys and polling. Before joining PPIC as a research fellow, he was assistant professor of political science at the University of Oregon; he previously worked at PPIC on the PPIC Statewide Survey. He holds a Ph.D. in political science from the University of California, Berkeley. Acknowledgments The author would like to thank a number of people for their help in making this report possible. The report drew on the expertise, and in some cases the data, of a long list of individuals who have committed their careers to understanding and improving California’s elections: Kim Alexander, Steve Boilard, Darren Chesin, Carolyn Crnich, Cathy Darling Allen, Efrain Escobedo, Bonnie Glaser, Kevin Keenan, Jill LaVine, Scott Lay, Dean Logan, Paul Mitchell, Michelle Romero, Don Smeraldi, and Nicole Winger. Special thanks to Paul Mitchell for access to his comprehensive data on voter registration. Kim Alexander also provided invaluable feedback on an earlier draft of the report, as did long-time scholars of elections Benjamin Highton and Karthick Ramakrishnan. The report benefited tremendously from the careful attention of Mary Severance and Lynette Ubois at PPIC. Thanks also to Mark Baldassare for his consistent support of the project. 23 Expanding California’s Electorate www.ppic.org Board of Directors DONNA LUCAS , CHAIRChief Executive Ofcer Lucas Public Affairs MAR K B ALDASSAREPresident and CEO Public Policy Institute of California RUBEN BARRALESPresident and CEO GROW Elect M AR Í A BLANCOVice President, Civic Engagement California Community Foundation B RIGITTE BRENAttorney WA LT E R B. HEWLETTChair, Board of Directors The William and Flora Hewlett Foundation PhIL I SEnBE rgChair Delta Stewardship Council MAS MAS UMOTOAuthor and Farmer STEVEN A. MERKSAMERSenior Partner n ielsen, Merksamer, Parrinello, g ross & Leoni, LLP KIM P OLESEChairman ClearStreet, Inc. THOMAS C. SUT TONRetired Chairman and CEO Pacific Life Insurance Company PPIC is a public charity. It does not take or support positions on any ballot measures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public ofce. PPIC was established in 19 9 4 with an endowment from William r. hewlett. Copyright © 2014 Public Policy Institute of California. All rights reserved. San Francisco, CA Mark Baldassare is President and Chief Executive Officer of PPIC. Donna Lucas is Chair of the Board of Directors. Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source. Research publications refect the views of the authors and do not necessarily refect the views of the staf, officers, or Board of Directors of the Public Policy Institute of California. Library of Congress Cataloging-in-Publication Data are available for this publication. I S B N 978 -1-5 8213 -15 6 - 6 PuBLIC POLIC y In STIT uTE OF C ALIFOrnIA 500 Washington Street, Suite 600 San Francisco, California 94111 Telephone 415.291.4400 Fa x 415 . 2 91. 4 4 01 PPIC SAC rAME nTO C E n TEr Senator Ofce Building 1121 L Street, Suite 801 Sacramento, California 95814 Telephone 916.440.1120 Fax 916.440.1121 Additional resources related to fscal and governance reform are available at www.ppic.org. The Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research." } ["___content":protected]=> string(102) "

R 114EMR

" ["_permalink":protected]=> string(118) "https://www.ppic.org/publication/expanding-californias-electorate-will-recent-reforms-increase-voter-turnout/r_114emr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8897) ["ID"]=> int(8897) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:41:55" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(4333) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 114EMR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_114emr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_114EMR.pdf" ["wpmf_size"]=> string(7) "3962794" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(75732) "www.ppic.org Expanding California’s Electorate Will Recent Reforms Increase Voter Turnout? January 2014 Eric McGhee with research support from Daniel Krimm Supported with funding from the S. D. Bechtel, Jr. Foundation SUMMARY O ver the past 20 years, voter turnout in California has been slipping compared to other states, and this decline may be exacerbating the gap between Californians who vote and the rest of the population. The state has considered or undertaken a variety of reforms to reverse these trends. In this report, we explore three of these reforms: a system of online voter registration, a same-day registration process, and a more relaxed deadline for submitting vote-by-mail ballots. One could argue that all of these reforms have inherent value because no eligible citizen should be prevented from voting for what amount to administrative reasons. But the administrative costs of a reform and the number of people who beneft from it matter as well. We fnd that none of these reforms is likely to produce large gains in turnout but two of the three are likely to cost very little or save money. • California implemented an online voter registration system late in the 2 012 election cycle. It was immediately popular, but it probably did not bring many new voters into the elec - torate or signifcantly change the demographic makeup of new registrants. However, the initial rollout probably saved counties considerable time and money. Online registration might have a greater efect on voter registration in the future, and its cost savings make the new system worthwhile. • California has adopted a new system of same-day registration, which will allow voters to register and cast ballots after the close of the ofcial registration period. This system HILL STREET STU DIO S/BLEN D IMAGE S/COR B I S Expanding California’s Electorate 2 www.ppic.org will probably boost turnout a few percentage points. But a substantial number of people may take advantage of same-day registration, which could result in signifcant costs and complications for county registrars. • Proposed reforms designed to ensure that more vote-by-mail ballots are counted by relaxing the deadline would afect only a tiny fraction of total votes cast but a large share of the ballots that are currently rejected. The cost of counting late ballots is probably minor, and the value of counting legitimate votes is great. The online registration and vote-by-mail reforms are worth pursuing despite the small turnout benefts evident so far. And because it seems to encourage earlier registration among some voters, online registration might mitigate some of the negative efects of same-day reg - istration on county registrars. Same-day registration, by contrast, creates an administrative burden that is quite heavy, given its modest efect on turnout. Instead, it might make sense to switch to a system of automatic registration, which would put every eligible Californian on the registration rolls and eliminate the need for the patchwork measures now in place. In any case, the state will have to do more than remove administrative barriers if it wants to expand the size of its electorate. It will need to do aggressive outreach to communities of potential voters who are underrepresented at the polls and often overlooked in get-out-the- vote drives. For the full report and related resources, please visit our publication page: www.ppic.org/main/publication.asp?i=1083 3 Expanding California’s Electorate www.ppic.org Introduction California has been at the forefront of recent eforts to expand the electorate. Te state has considered or passed a number of reforms in recent years to get more people regis- tered and increase voter turnout. Tese include creating an online system that makes registering to vote as easy as pos- sible, scrapping the registration deadline so that citizens who decide to vote at the last minute are not excluded, and relaxing the deadline for vote-by-mail (VBM) ballots so that fewer go uncounted because they are late. Tere is ample evidence that California has a turnout problem. Among the population of citizens who are eligible to vote, the share that casts a ballot has been declining in California compared to other states. Figure 1 shows that in the 1990s, California’s turnout was consistently higher than in the rest of the country, especially in midterm elec- tions when the presidential contest was not on the ballot. Tis advantage has faded in recent years, and now Califor- nia tends to match the national average or fall behind. 1 Tere is also evidence that California has an “exclusive electorate,” dominated by older, whiter, wealthier Califor - nians (Baldassare 2006). Te decline in turnout matters because it may be widening the rif between this exclusive electorate and those who do not (or cannot) vote. In this report, we evaluate three important voting administration reforms that have been adopted or pro- posed in recent years. First, we look at the state’s promis- ing new online registration system and evaluate its efect on the 2012 election. We then examine the likely efect of the state’s new same-day registration law. And, fnally, we look at ways to improve the state’s VBM system, usually by ensuring that fewer voters miss out on the franchise because their ballots arrive too late. To provide a comprehensive overall picture of the value of each reform, we consider both its efect on turnout and its potential to increase efciency or impose additional burdens or costs on election administration. We conclude by outlining the policy implications of this analysis. Online Registration California’s Internet-only voter registration process was established by Senate Bill (SB) 397, which was signed into law in October 2011 and rolled out on September 19, 2012. Te system allows users to enter voter registration information and click a button to complete the process. It replaces the registration form that had long been available online but that had to be printed, signed, and mailed in. Te new system was instantly popular: more than half of all new registrants in the last month of the 2012 fall registration period used it, exceeding even the most opti- mistic expectations. Moreover, several studies in California and elsewhere have concluded that online registration has brought new voters into the electorate and that these new voters were younger, poorer, and more likely to belong to racial or ethnic minority groups (García Bedolla and Vélez 2013; Romero 2013a, 2013b). There is evidence that California has an “exclusive electorate,” dominated by older, whiter, wealthier Californians. Percentage 20102008200620042002200019981996199419921990 2012 Figure 1. California’s voter turnout is declining more than in other states SOURCE: U.S. Census Current Population Survey, November Supplement, 1990–2012. –4 +2 +4 +6 +8 +10 +12 –2 California vs. average of all other states Average of all other states Expanding California’s Electorate 4 www.ppic.org In purely administrative terms, there is little doubt that online registration is more efcient and vastly cheaper than the old paper system. Tere has not yet been a com - prehensive study of the cost savings in California, but an important study in Arizona found that the cost to process a traditional paper form was about 28 times the cost of an online registration form (Barreto et al. 2010). Moreover, an online system signifcantly reduces the risk of error by com - puterizing the entire registration process, eliminating steps in the middle—such as inputting information from mailed- in forms—that led to errors and logistical problems. As long as the old system is available for those who do not have easy computer access, the state should move aggressively to get as many people as possible to use the new system. Efect on Voter Participation Administrative benefts aside, did online registration change the size or composition of the registered voter population? Tere were several diferences between citizens who registered online and the population that registered the traditional way during the same period. Online regis- trants were somewhat less likely to be Latino (22% versus 25%), slightly more likely to be Asian American (9% versus 8%), and somewhat more likely to end up voting on elec- tion day (84% versus 78%). Tey were also younger, with an average age of 35, compared to age 39 for traditional registrants—and adults in their mid-20s were almost three times more likely than senior citizens to register online (Figure 2). 2 But does this mean that online registration drew more or diferent people into the electorate? Or would the voters who registered online have registered anyway, using the traditional process? 3 Figure 3 shows the number of new or changed registra- tions during the period between the fnal two registration reports—60 and 15 days before the election—in 2004, 2008, and 2012. To account for population growth between election years, the numbers are presented as a share of the total unregistered but eligible population—adults legally able to register who had not yet done so as of 60 days before the election. Te results make clear that the surge in registrations in 2012 was consistent with previous presi- Voting online Given the popularity of online voter registration in California, what are the prospects for casting ballots online? In fact, there are two bills currently under consideration in the Assembly that would permit online voting in specifc circumstances: Assembly Bill (AB) 19 would allow counties to conduct Internet voting pilot programs in local elections, and AB 1360 would allow online voting in homeowners’ association elections. Neither bill had passed out of the Legislature by the time this report went to press. Internet voting raises important security concerns that are difcult to address. Online transactions are not reliably secure, and private frms that rely on such transactions lose billions of dollars annually to fraud (Paget 2009). Since these firms are mostly concerned about proftability, they can balance the costs of preventing fraud against the costs of reimbursing the losses of individual customers. Election administrators, by contrast, must ensure the integrity of the voting process. Privacy is another concern. Both financial transactions and voting require authentication of identity, but voting requires a secret ballot after that initial confrmation. It would be impos- sible to replace a lost vote in the same way as one would replace a lost dollar. Since the barriers to Internet voting tend to be structural rather than technical, successful implementation would probably require the state to adopt standards of security and privacy that match those of Internet commerce more closely, a shift in values that many might resist making. Percentage 95 Figure 2. Younger people have been far more likely than older people to register online SOURCE: Political Data, Inc. 0 20 30 40 50 60 70 10 908580757065605550454035302520100 Age of new registrant 5 Expanding California’s Electorate www.ppic.org dential elections. Registration applications were submitted by 25.9 percent of the potential new voters—no more than a percentage point more than in 2004 (25.5%) and 2008 (2 4 .9 %). 4 Te results are similar if total registration is con- sidered as a share of the eligible population. 5 When the numbers in Figure 3 are broken down by week, the pattern of registrants in 2004 and 2008 is strik- ingly similar: registration is slow at frst, then picks up speed, and there are huge registration numbers in the fnal few days. By contrast, in 2012 registration began to increase much earlier, right afer online registration became available, and then slowed down (Figure 4). In other words, the online registration system may have encouraged voters who were planning to register anyway to do so earlier in the cycle, without bringing many new voters into the electorate. Did online registrants turn out in greater numbers? On the one hand, it seems that online registrants were more likely to vote. But this, too, could be a matter of which people availed themselves of the new system: indeed, the voting rate for registrants who signed up just before online registration became available was virtually identical to the rate just afer. 6 Also, the presidential conventions seem to have played an important role: those who registered afer the conventions were about 10 percentage points more likely to vote than those who registered before, presumably because their interest was piqued by media coverage. It is possible that there was something unusual about the 2012 election that depressed registration rates overall or among key subgoups. Might registration have been even lower without an online registration option? To answer this question, we can compare registration in California and other online registration states to those that have not adopted the reform, and do so across multiple election years, not just 2012. Did the online registration states see a gain afer adopt - ing the reform? We conducted a separate analysis of registra - tion rates over time in all 50 states, paying special attention to changes when states have adopted online registration. Consistent with the analysis so far, registration as a share of the eligible population in other states has not changed on average in response to the adoption of online registration. 7 Efect on the Composition of the Electorate Did the online system increase registration in key sub- groups? Te 2012 registration rates for these groups did not change much: the Latino share of registrants was up about a half a percentage point from 2008, whereas the shares of both young people (ages 18 to 24) and Asian Americans were down about a half a percentage point. 8 But these numbers do not settle the matter; there may have been something unusual about 2012 that would have depressed registration rates for all these groups if online registration had not been available. New registrants as a share of potential (%) Figure 3. Online registration did not produce a signifcant surge in new registrants SOURCES: California Secretary of State, Political Data, Inc. 2004 2012 Online registration available 2008 0 5 10 15 20 25 30 25.5 25.9 24.9 New registrants as a share of potential (%) 1 2 3 4 5 6 0 Weeks until registration deadline Figure 4. Online registration shifted registrations earlier in the cycle without increasing the total number of registrants SOURCE: Political Data, Inc. 0 5 6 7 8 9 10 4 3 2 1 Online registration available in 2012 2004 2008 2012 Expanding California’s Electorate 6 www.ppic.org One way to get at this question is to see if there was a change in the type of person who registered af er online registration went live. Here, we do see signs of some ef ect. Figure 5 shows that the share of both young adult and Asian American registrants began climbing right af er the rollout of online registration. To attribute all of this increase in young people to the new registration system, we would have to believe that without the online option, a smaller share of 18- to 24-year-olds would have registered in 2012 than in any other primary or general election since 2002—despite constituting about the same or a slightly larger share of the pool of eligible voters. 9 T us, it seems likely that—as with registrants overall—young voters who would have regis- tered anyway decided to sign up earlier in the cycle. 10 T e same is true for Asian Americans, who constituted about 5.5 percent of new registrants before online regis- tration and about 8.5 percent af er. 11 It is hard to believe that without the online option, Asian Americans would have remained 5.5 percent of new registrants through the close of registration—not only would this have been a lower share than in any election since 2002, but the share of Asian Americans eligible to vote has grown over the past decade. Nonetheless, online registration probably accounted for some of the 3 percent increase in Asian American registration. 12 By contrast, Latino registration actually fell slightly just af er the online system was introduced. T is does not mean that online registration discouraged Latinos from participating but that the surge in non-Latino registration was not matched by Latinos. And, as Figure 6 shows, the share of Latino registration rebounded by the end of the cycle. 13 Because young people, Latinos, and Asian Americans tend to belong to the Democratic Party, some predicted that online registration would increase the number of Democrats. Did online registration help one or the other major party? Figure 7 shows the share of new voters who registered as Democrats and Republicans in the weeks leading up to election day. T e trend lines for party reg- istration are identical before and af er the online option Share of new registrants (%) Figure 5. Young people and Asian Americans registered at higher rates after the rollout of online registration SOURCE: Political Data, Inc. 0 15 20 25 30 35 40 10 5 2211 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 Adults age 18–24 Asian Americans JuneJulyAugustSeptemberOctober Online registration rollout Latino share of new registrants (%) Figure 6. The online option did not signifcantly improve Latino registration SOURCE: Political Data, Inc. 0 15 20 25 30 35 40 10 5 2211 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 JuneJulyAugustSeptemberOctober Online registration rollout Democrats Republicans Democratic convention Republican convention Party share of new registrants (%) Figure 7. There was no clear sign of a shift in party afliation after online registration became available SOURCE: Political Data, Inc. 0 20 30 50 10 22 40 60 11 18 25 2 9 16 23 30 6 13 20 27 3 10 17 24 1 8 15 JuneJulyAugustSeptemberOctober Online registration rollout 7 Expanding California’s Electorate www.ppic.org became available, so there is no clear sign here of any efect. Te real efect appears to have come from the presidential conventions, especially the Democratic ones. Afer the Democratic convention, the share of registrants who chose the Democratic Party was almost 10 percentage points higher—and the Republican Party share 5 percent- age points lower—than in the weeks leading up to the conventions. Again, it is not clear how many of these voters would have registered as they did anyway, but the conven- tions do appear to have afected decisions about timing, if nothing else. A Small Efect on Total Registration In short, online registration seems to have altered the dynamics more than the level of registration. More Cali- fornians registered earlier in the cycle and fewer registered late, resulting in a very small overall increase. Other states that have adopted similar reforms have seen only modest increases in registration rates, or none at all, including among key subgroups. It is far too soon to discount the efect of online reg- istration on turnout. Te system was rolled out quite late in the election cycle, slightly more than a month before the registration deadline. It is possible that as the online option becomes known to more people, and once parties and interest groups have time to experiment with diferent mobilization approaches, there will be more gains in voter participation. Same-Day Registration In September 2012, the governor signed AB 1436, which establishes a system of “conditional” voter registration in California. Tis system allows residents who miss the registration deadline (15 days before the election) to both register and vote on any of the remaining days, includ- ing election day itself. Conditional registrants will have to register and vote at their county registrar’s ofce for their ballots to count. 15 To minimize the potential for fraud, con - ditional registration will not be available until the California Online registration seems to have altered the dynamics more than the level of registration. More Californians registered earlier in the cycle and fewer registered late, resulting in a very small overall increase. Is online registration the right tool for the job? Should we expect an online form to be an efective way to increase registration? Californians who are eligible to vote but have not registered are less educated and more heavily Latino—groups that help form the “digital divide” between those who use the Internet and those who do not (Baldassare et al. 2013). Thus, the very people who could beneft most from an online form might not have easy access to it. It is true that those who are not registered to vote are less likely to use the Internet. According to the most recent PPIC Statewide Survey on Internet use, about 17 percent of unreg- istered adult citizens say that they do not use the Internet, compared to 9 percent of registered adults (Baldassare et al. 2 013). 14 The gap is somewhat larger when framed in terms of convenience and frequency: unregistered citizens are less likely than registered citizens to have broadband access at home (62% versus 77%) and are less likely to go online from home at least once a day (57% versus 67%). And, because young people and Asian Americans are among the most frequent users of the Internet, it is perhaps no coincidence that there are signs of an online registration efect for these groups. Nonetheless, these numbers suggest that most unregis- tered Californians have regular access to the Internet. More- over, mobile Internet devices such as Android phones and iPads are equally common among unregistered and registered citizens: two-thirds of each group have access of this kind. Even among those who are both unregistered and have no Internet access at home, a quarter have mobile access. Overall, most unregistered citizens have regular access to the Internet and could make use of the online registration system. Expanding California’s Electorate 8 www.ppic.org Ansolabehere 2002). Many of these studies also ofen argued for an especially sizable gain in key demographic subgroups such as less-educated adults, young adults, racial and ethnic minorities, and newly arrived residents. Tese initial fndings have been challenged in recent years by research that tests causal links. 17 In these newer studies, the maximum estimated efect falls at the low end of the older range of results, topping out at about 4 percentage points for a state with a 15-day closing date (such as Cali - secretary of state fnishes creating a voter registration database that complies with the Federal Help America Vote Act—the database will not be ready before 2016. 16 Te goal of this reform, as described in the analysis of the bill, is to make it easier to vote and so increase voter turnout overall. Many states have a “same day” registration process similar to conditional registration, but they close of access on or just before election day. Other states take the opposite approach: they have “election day registration” (EDR)—that is, same-day registration on election day— but forbid registration in the days and weeks before. In permitting both same-day registration and EDR, Califor- nia will be adopting one of the most permissive registra- tion systems in the country. Tis signifcant reform generates two important ques- tions. Will conditional registration increase voter turnout in California, as its authors intended? And how many new voters might election administrators need to process close to or on election day? Will It Increase Voter Turnout? To assess same-day registration’s potential efect on turn- out, we focus on EDR states—the ones that currently allow voters to register and vote on election day—and set aside those (such as Ohio) that cut of same-day registration sev- eral days before. Since the excitement of election season is highest on election day itself, allowing same-day registra- tion on that day is likely to have a larger efect. Early studies of EDR suggested a sometimes notable efect on turnout. Te efects ranged between 3 and 6 per- centage points, with extrapolated predictions of up to 9 percentage points for a state like California (Alvarez and Does same-day registration increase the risk of voter fraud? Same-day registration gives election administrators less time to verify voter eligibility. Does that make it harder to prevent ineligible voters from participating or eligible voters from casting multiple ballots? Empirical studies produce very little evidence of voter fraud; there are few intentional attempts, and most of those are prevented by administrative review processes already in place to detect and correct inadvertent errors. Qualifca- tion requirements for registration generally exceed those for voting, even where there are voter ID laws, and this applies to same-day registration the same as any other voter registration process. Verifcation of registration information occurs in due course even for same-day registrants, and intentional violators are subject to criminal prosecution. Comprehensive information on the incidence of voter fraud is not available from a single source. One study of six long-term EDR states (Idaho, Maine, Minnesota, New Hamp- shire, Wisconsin, and Wyoming) found news reports of 10 discrete incidents of voter fraud with merit between January 1999 and February 2005 and fve federal convictions for felon voting between 2002 and 2005. An additional survey of 36 out of 252 local jurisdictions in 2004 and 2005 yielded seven cases that were sent warnings but not prosecuted (Minnite 2007). This is similar to the low levels of voter fraud found in non-EDR states overall. However, proper safeguards that apply to all voter regis- tration should be maintained for EDR to prevent problems in the future. Such a system of checks may be somewhat easier to maintain under the California system, since “conditional” registration will be conducted by experienced employees of the county registrar and not volunteer poll workers. In permitting both same-day registration and EDR, California will be adopting one of the most permissive registration systems in the country. 9 Expanding California’s Electorate www.ppic.org fornia), and resulting in an apparent decline in turnout in some cases (Brians and Grofman 2001; Hanmer 2009; Knee and Green 2011; Keele and Minozzi 2013). Te largest efects are in Minnesota and Wisconsin, which adopted the reform in the 1970s and enjoy the highest level of turnout among states; the efects are much smaller in later adopters. 18 Te analysis in this report suggests results on the lower end of this range (Table 1), with a slight decline in turn- out on average for a state shifing from a 15-day close to EDR. 19 Since there is no clear reason for EDR to discourage people from voting, it seems reasonable to conclude that the reform has done little or nothing to boost turnout in most of the states that adopted it. Te fndings were similar for key subgroups, including the young, those who recently moved, and those without a high school diploma. 20 Are these fndings overly optimistic, given that elec- tion day registrants in most EDR states are not required to go to the county registrar but can register and vote at their polling places? Te California EDR law allows counties to set up satellite ofces for EDR registration, and some care could be taken to place satellite ofces in areas with high numbers of potential new registrants (such as university or community college campuses). But for the average EDR voter, these ofces will likely be farther away than a local polling place. Moreover, many counties may not want to take the trouble to ensure that satellite ofces have the full range of ballots in all languages. In its use of county registrars, California’s new sys- tem closely resembles Montana’s EDR process, which was adopted in 2005. A comparison of turnout before and afer Montana implemented EDR gives results ranging from vir- tually no efect on turnout to a more noticeable increase of about 3.9 percent. Tus, it seems unlikely that this aspect of the Montana reform had much efect on EDR. In fact, Iowa, which adopted EDR about the same time and is similar to Montana in many other ways, saw similar results from EDR despite allowing registration at the polling place. 21 Overall, then, the average efect of election day regis- tration ranges from nothing to about a 4 percentage point increase in voter participation. Although larger efects are certainly possible, one can point to evidence of smaller efects as well. However, there is an important diference between Cali - fornia and other EDR states: even the most diverse of these other states are far more racially and ethnically homoge- neous than California. Tat makes it impossible to evaluate the efect of EDR on racial and ethnic subgroups. We have ofered estimates of EDR’s efect on groups that might over - Table 1. Election day registration has modest efects on turnout Data source U.S. Census Current Population Survey (public opinion poll) Election returns Average efect nationwide –0.4%–0.3% Efect on . . . Those with no high school diploma –1.9 %– Those ages 18–24 +2 . 7 %– Those who moved < 6 months ago +3.4%– Efect in . . . Iowa +1.1%–0.2% Montana +0.5%+3.9 % SOURCE: Author’s calculations. NOTE: With the exception of the estimates for Iowa and Montana, the models include the date for the close of registration, and the predictions are calculated assuming a shift from a 15-day close to an EDR system, as would be true for California. Expanding California’s Electorate 10 www.ppic.org and Montana, no more than 4 percent of voters used EDR, compared to 20 percent or more in some other states. Some of this variation may refect difering reporting procedures and defnitions. 22 Tese EDR voters are not necessarily new registrants: roughly half or more of the election day regis - trants who could be identifed were already on the books but had changed addresses. However, these EDR users still take up administrative time and resources. In states for which we have data from both presidential and midterm elections, use of the system has almost always increased in presidential years, even when considered as a share of total turnout. Tus, EDR may place a propor- tionately larger burden on county registrars in high-profle elections. What do the experiences in other states imply for Cali- fornia? Te smallest share in Table 2 (3% in Iowa in 2008) would have translated to approximately 400,000 EDR forms in California in 2012, ranging from a low of 20 in Alpine County to a high of roughly 97,000 in Los Angeles. Note that even the lower-bound estimate for Los Angeles County is more than three times as large as the estimate in the AB 1436 analysis. If late registrations in California had matched the highest rate in Table 2 (21% in Minnesota in 2004), there would have been 680,000 EDR applications in Los Angeles County alone—upward of 100,000 for each branch ofce of that county’s registrar, assuming that the ofces could be set up to handle such applications. Some mitigating factors might ease the burden. Because EDR’s estimated efect on turnout is modest, the total number of applications will likely be on par with previous election cycles. Tus, EDR will probably shif the work- lap with racial and ethnic minorities in California—such as the less educated, the young, and those who recently moved. But, in a sense, California—with its size and diversity— is breaking new ground in adopting such an aggressive reform, and projections should be tempered accordingly. Will There Be a Late Surge? Even if EDR does not expand the electorate, will more voters decide to register on or close to election day? Te analysis that accompanied AB 1436 suggested that the tide of late registrants could be substantial—as many as 30,000 in Los Angeles County alone. Although these voters could certainly register at any point during the 14 days before the election, the greatest surge is likely to come on election day, creating a potentially signifcant administrative challenge. County registrars, who will have to cope with this surge with limited resources, are understandably concerned about how large it is likely to be. Table 2 shows the share of all voters casting ballots who registered afer the ofcial close of the registration period in each of the EDR states. Many states do not report regularly on their EDR voters, so there is a lot of missing informa - tion, but it is possible to make a few generalizations. First, the use of EDR varies greatly from state to state. In Iowa One analysis of same-day registration suggested that the tide of late registrants could be substantial— as many as 30,000 in Los Angeles County alone. Same-day registration could result in modest increases in turnout but could also place a burden on county registrars. M I K E BLA KE/ REUTER S/COR B I S 11 Expanding California’s Electorate www.ppic.org load of processing these applications to later in the cycle without signifcantly increasing it. Moreover, California’s online registration system may lessen this burden: it seems to have the opposite efect of EDR, encouraging voters to register earlier than they had in previous election cycles. 23 It seems reasonable to assume that EDR registration in California will not match the usage levels of states that allow EDR at local polling places. Moreover, there is no guarantee that the numbers will match even those of Montana, which also requires that EDR voters go to county registrars. But given the size of even the lower esti - mates, it would make sense for registrars to be prepared for a large surge. On balance, the modest potential efect of same-day reg - istration on turnout does not seem worth the cost of process - ing late registrations. Since the system is now law and seems unlikely to be repealed, preparing for the change is likely the best short-term solution. In the conclusion, we touch on longer-term reforms that might alleviate the problem. Improving the Vote-by-Mail Process California adopted “absentee” voting in 1923 so that “any duly registered voter, who, by reason of his [sic] occupation is regularly required to travel about the state” could cast a ballot by mail (California Statutes of 1923, Chapter 283). In 1978, California was the frst state in the nation to permit “no excuse” absentee voting (California Statutes of 1978, Chapter 77), and in 2001 the state created a “permanent absentee” option (California Statutes of 2001, Chapter 922), making it as easy to vote by mail as to vote in person. Since the permanent VBM option was introduced, its popularity has skyrocketed: a majority of ballots are now cast by mail (Figure 8). VBM may be more convenient than voting at a polling place, but it complicates the process. 24 Current law requires that VBM ballots arrive on or before election day. Unless voters decide to drop of ballots at their polling places on election day, they have to rely on the U.S. Postal Service (USPS) to deliver their votes on time. Senator Lou Correa has introduced a bill (SB 29) that would address this issue. Tis reform would consider VBM ballots valid if they are postmarked by election day and arrive by the third day afer the election. Table 2. Election day registration states often have large numbers of late registrants Y ear IowaIdaho Maine Minnesota Montana New Hampshire Wisconsin Wyoming 2000 17 % 11% 2002 12 % 7%7% 2004 20% 2006 12 % 6 % 2008 3% 7 % 4% 15% 16% 2010 9% 2 012 4% 8% 4% 11% SOURCES: Various Secretaries of State, Federal Election Assistance Commission (2008, 2010, 2012). NOTE: Cell entries are the share of all voters in each state and year who registered for the frst time or changed registration after the close of the ofcial registration period, meaning that they took advantage of the election day registration law. VBM may be more convenient than voting at a polling place, but it complicates the process. Expanding California’s Electorate 12 www.ppic.org one of every fve uncounted VBM ballots in most of the counties listed here. Tere is no reason to think that the ballots were fraudulent, which makes their rejection akin to disenfranchisement. And, in a close election, even these small numbers could make a big diference. 25 For 24 of these 31 counties, the CACEO has also provided information about the lateness of the ballots and whether they had valid postmarks. 26 According to those data, ballots later than three days afer the election accounted for about 4 percent of all late ballots across all 24 counties and accounted for more than 15 percent of such ballots in only two of the 24. So a three-day window would probably allow the vast majority of the late ballots to be counted. Some ballots did not have valid postmarks, making it difcult to know whether they were cast on time. Te current version of SB 29 allows those ballots to be counted if the date that accompanies the voter’s signature on the envelope falls on or before election day. Such bal- lots account for 12 percent of all late VBM ballots arriving within the three-day window in these 24 counties, with only two reaching as high as one-third. 27 Are some groups of voters hurt more by late ballots than others? Table 4 presents the results of a regression analysis of how well a variety of voter characteristics corre- lated with submitting a late VBM ballot in the 2012 general Potential Efect Before we can assess the efect of this legislation, we need to determine the scope of the problem. Counties do not consistently gather data on late ballots, but we analyzed data from Political Data, Inc., and the California Associa- tion of Clerks and Elected Ofcials (CACEO) that cover 31 counties for the 2012 election. Tis accounts for two- thirds of all VBM ballots that year. Because these late ballot totals are not always broken down by their day of arrival, they are likely to generate high estimates of SB 29’s potential efect. Table 3 makes clear that late ballots are extremely unusual—they rarely constitute more than a half a percent of all VBM ballots cast. But lateness accounts for at least Percentage of ballots submitted by mail Figure 8. The vote-by-mail option has exploded in popularity SOURCE: California Secretary of State. 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 20120 10 20 30 40 50 60Permanent VBM begins No-excuse absentee begins Table 3. Late ballots are extremely rare but constitute a large share of those that go uncounted Total late ballots statewide 18 ,0 6 4 Share of VBM ballots cast 0.40% Share of uncounted ballots 4 7. 0 0 % SOURCES: Political Data, Inc., CACEO (total late and VBM ballots), California Secretary of State (total uncounted VBM ballots by county). NOTES: Political Data, Inc. provided data for Calaveras, Madera, Mariposa, San Francisco, Sierra, Sonoma, and Tehama counties. CACEO provided data for Amador, Fresno, Humboldt, Inyo, Los Angeles, Mono, Orange, Plumas, Sacramento, San Benito, San Diego, Santa Clara, and Siskiyou counties. Data for the remaining counties were available from both sources. Where the sources did not match, the higher number is reported. Table 4. Some groups are slightly more likely to submit late ballots Likelihood of late ballot New registrant since 2010 No clear diference Voted in 2010 Lower by 0.36% Ag es 18 –24 Higher by 0.86% Democrat Higher by 0.12% Independent Higher by 0.22% Latino No clear diference Asian American/Pacific Islander Higher by 0.13% SOURCE: Political Data, Inc. NOTES: Each number represents the change in the predicted percentage of people who would submit a late ballot for those in each group compared to the rest of the VBM population. Likelihood is calculated from a logit regression of late ballot status on the identifed variables, with fxed efects for counties. The sample was limited to VBM voters who returned a ballot in the 17 counties identifed in Table B11 of Technical Appendix B. “No clear diference” refers to efects of less than 0.10%. 13 Expanding California’s Electorate www.ppic.org analysis suggests that VBM voters in consolidated zip codes were slightly less likely to have a late ballot in 2012, at least in those counties for which we have late ballot information. 29 In sum, late ballots are not a source of concern for overall turnout. But because they constitute a signifcant share of uncounted ballots, SB 29 would signifcantly reduce the number of disenfranchised voters. Moreover, because the number of ballots involved is small, counting them would not impose much additional administrative burden, especially in counties that already have an exten- sive post-election validation process. Tere may be no reason to worry that USPS cost- cutting will result in more late ballots. But the USPS is not election for the 17 counties for which we have data for individual voters. Each number is the separate diference in likelihood that is attributable to that group, independent of all the other characteristics listed in the table. Te larg- est diferences generally have to do with a lack of experi- ence with the act of voting. Voters young enough to have limited voting experience are more likely to submit a ballot late, whereas those who voted in 2010 are less likely. Being a new registrant makes no diference independent of being young, but many of these “new” registrants might have been registered somewhere else before the 2012 election. Relying on the U.S. Postal Service Most VBM voters count on the USPS to get the ballot to elec - tion ofcials on time. Te USPS has a strong delivery record, but some mail does get delayed or lost. If a delivery delay prevents a ballot from arriving on time, then it can be said that the USPS has unintentionally disenfranchised a voter. Tis is not an idle concern. Te use of traditional mail has plummeted even as the popularity of VBM has rapidly increased. Tis has forced the USPS to cut costs; it has laid of employees and closed local post ofces, and it is contemplating ending Saturday delivery. It has also been closing many of its processing and distribution centers (P&DCs). P&DCs are hidden to most postal customers. Tey are not points of contact for the general public but large-scale facilities that sort the mail and direct it to its destination. Tese P&DC closures have been quite signif- cant: 39 percent nationwide were closed between 2006 and the present, including seven in California between 2010 and 2012 alone. Still more closures are planned. Afer a center is closed, the mail it used to process is handled by a nearby existing facility. Because of the greater distance and the potential for slower processing in the newly con- solidated facilities, there is always the chance that voters whose mail is consolidated will see slower delivery times. Te average voter is not likely to know whether his or her P&DC has been closed. Can we attribute any late ballots to the USPS consoli- dation process? To assess this possibility, we factored P&DC consolidation into the regression analysis in Table 4. 28 Tis Late ballots are not a source of concern for overall turnout. But because they constitute a signifcant share of uncounted ballots, counting them would signifcantly reduce the number of disenfranchised voters. The closing and consolidation of USPS processing and distribution centers could afect vote-by-mail ballots. JOH N GRE SS/COR B I S Expanding California’s Electorate 14 www.ppic.org VBM system in the future. But the administrative costs seem small enough—and the efect on the problem of late ballots large enough—to make the reform worth pursuing.Given that the removal of virtually all administrative barriers helps only at the margins, the state should shif its focus to motivating Californians to take advantage of the systems that have been put into place. In short, we should shif from facilitating voters to mobilizing them. 30 A number of options could be pursued. First, online registration should be viewed not as an end in itself but as a frst step toward a diferent system of mobilization. Compared to other forms of communication, the Internet is highly scalable: it costs only slightly more to reach 10,000 people than it does to reach 10. And because the state’s new online registration system is completely Internet-based, it can be integrated seamlessly with any electronic appeal to register and vote. 31 Tis can allow a great deal of experi- mentation with diferent forms of outreach at relatively low cost. Moreover, the secretary of state can take further steps to make the new system easier to use with mobile Internet devices. If online registration can make electronic outreach more efective in getting voters registered, the potential gains could be quite signifcant. Only two (Colorado and Nevada) of the 12 states that have adopted online registra- tion have been in play in a presidential election when the system was in place. A close presidential race is the cir- cumstance most likely to make a mobilization efort strong and efective, so as more states adopt the reform, it may yield greater results. Mobilization eforts need to reach beyond the “low- hanging fruit” of those who are likely to register and vote but have not yet done so. To increase turnout substantially, these eforts will need to target hard-to-reach citizens who probably would not vote at the same rate as those who are currently registered. But the turnout rate among these unregistered citizens is currently zero, so even an extremely low turnout rate would be an improvement. In fact, it would probably make sense to think of both online registration and same-day registration as way sta - tions en route to a system that automatically registers every done with the cuts, and more radical changes could be in store. Relaxing the deadline for VBM ballots seems like a modest way to forestall any adverse efects and make sure that all valid votes are counted. Policy Implications None of the reforms examined in this report is likely to have a large efect on voter turnout. Online registration may have brought a few more people into the system in 2012, but it mostly encouraged voters to register earlier than they might have otherwise. At the same time, it is much cheaper and more accurate than the paper system, so its administrative benefts are potentially substantial. Election day registration is similar in that its efect on voter turnout will probably be no more than a few percentage points (although for California that would still mean hun- dreds of thousands of new voters). But the administrative efect might be large. Certainly, county registrars ought to prepare for a deluge of late registrants under the new sys- tem, since the numbers involved may be larger than even some existing estimates that seem large already. Online registration might ofset this surge somewhat by encourag- ing voters to register earlier in the cycle, but the possibility for this efect in future years is uncertain enough that it would be prudent not to count on it. Finally, eforts to deal with late VBM ballots will afect only a tiny fraction of the ballots cast, and there is no sign that cutbacks at the USPS will create problems for the To increase turnout substantially, mobilization eforts will need to target hard-to-reach citizens who probably would not vote at the same rate as those who are currently registered. 15 Expanding California’s Electorate www.ppic.org eligible California citizen. California does have a “motor voter” system in place so that citizens can register when they get drivers licenses or engage with government in other ways. But this system is not as simple or well pro- moted as it was intended to be. 32 Recent legislation (SB 35) by Senator Alex Padilla that was signed into law would improve many aspects of this system and expand it to more government ofces. But citizens must still actively choose to register. Te system could instead be made “opt out,” mean- ing that voters would automatically be registered based on information provided to various government agencies. To opt out, they would need to remove their names from the registration rolls. 33 Automatic registration would not force residents to divulge any more information to the govern- ment than they had already revealed for other purposes; it would simply use the information already provided to determine voter eligibility. Even if this approach were to register no more citizens than election day registration, it would likely result in a more even distribution of new The turnout rate among unregistered citizens is currently zero, so even an extremely low turnout rate would be an improvement. California should consider reforms that target and mobilize hard-to-reach voters. GET T Y IMAGE S Expanding California’s Electorate 16 www.ppic.org Technical appendices to this report are available on the PPIC website: www.ppic.org/content/pubs/other/114EMR_appendix.pdf registrants throughout the election cycle and save county registrars the challenge of large pre-election surges. What might be the limitations of such a reform? Some voters might not want to register for philosophical or admin - istrative reasons, 34 so it would be important to make the choice to opt out very clear. Tere might also be costs that ofset the administrative gains—for example, the state would need to send election materials to a much larger group of reg - istered voters. Tere would need to be further cost estimates before the system could be considered a viable option. An automatic registration system would not neces - sarily increase turnout much by itself. 35 Election day registration, which in most states is only slightly more burdensome on the average voter, appears to have had only a modest efect in those states. It would have to be coupled with aggressive outreach to bring new voters to the polls. More generally, voter turnout cannot be improved solely by administrative means. Increasing and diversifying voter participation is an ongoing process of motivating more Californians to exercise their right to vote. 17 Expanding California’s Electorate www.ppic.org Notes 1 Tese results are consistent with and extend those found by Ramakrishnan and Baldassare (2004). Te numbers in this fgure and the next were calculated by estimating a separate logit regression for each election year with a dummy for California and then predicting the frst diference estimate for this Califor- nia dummy. Coefcients and model ft for all 12 election years are available from the author on request. 2 All of the data for this section, and for any later analysis that uses voter registration records, were provided by Political Data, Inc. Te frm has a copy of the voter registration fle, to which it has coded race based on registrant surnames (see note 8) and appended a wide range of information about the disposition of each ballot. Te frm is widely considered the best source for data on online registration. 3 Tere are certainly reasons to believe that an online registra- tion system might boost registration rates. For instance, in one study, 29 percent of eligible but unregistered Californians said that they did not know where they could fnd registration forms (Alexander 2004). An online registration system might be easier for these citizens to fnd and use. But this is far from guaranteed: the online system might not help as much as intended, and these unregistered voters might have reasons for not registering other than a lack of convenience. 4 Te secretary of state’s percentages of total change in registra- tion over these same periods are much lower because the secre- tary is obligated to purge old registrants as well as add new ones. Te comparable changes in total ofcial registration are 14.5 percent (2004), 16.2 percent (2008), and 15.1 percent (2012). 5 Total registration as a share of the eligible population was 75.0 percent in 2004, 74.6 percent in 2008, and 76.7 percent in 2012. 6 Tese results are available from the author on request. 7 To obtain these estimates, we used the Current Population Survey data to conduct a logit regression of registration on online registration status, with fxed efects for states and years to control for unchanging diferences across states and uniform changes over time, plus demographic controls to account for unusual changes in a state that had nothing to do with online registration. Te resulting estimates identify how much online registration states changed when they adopted the reform, relative to both the changes in other states in the same year and the online registration states’ own baseline rates from before the reform. We also tried omitting the demographic controls and obtained a similar result. Coefcients and model ft can be found in Technical Appendix B. 8 We should be somewhat cautious when comparing the Latino and Asian American numbers from 2012 to those from earlier elections. Latino and Asian American registrants are identifed in the voter registration fle by surname (which almost certainly leads to some unknown degree of undercounting). Te surname estimates for 2012 come from Political Data, Inc., whereas those from 2002 through 2010 come from the statewide database. Tese sources use slightly diferent methods for identifying sur- names, so diferent numbers could be attributable to the method alone. Tis caveat is not relevant when comparing Latino and Asian American estimates over time within a single election year, or for any analysis involving age. 9 To obtain this estimate, we projected three trend lines from before online registration to the period afer it was available: one using all the pre-online registration data; one using only the last 20 days, which was a period when young people were a declining share of new registrants; and one assuming that the average registration rate just before online registration would remain constant. We then assumed that any diference between this new projected registration rate and the actual one was entirely explained by a loss of young voters (rather than an increase in older ones). Te estimates ranged from a total of 272,000 young registrants lost to 334,000 lost. Tis would have dropped the share of young people in the electorate well below 8 percent for the frst time in any statewide election since 2002. Details of this estimate are available from the author on request. 10 Since registration rates over time for young people in earlier years are not available, it is not possible to confrm this idea d i re c t ly. 11 A regression discontinuity (RD) analysis of these data con- frms a statistically signifcant increase in the Asian American registration rate right afer online registration became available. However, because there was a small surge in Asian American registration immediately before online registration went live, the RD estimate is much smaller than the 3 percent identifed in the text and is sensitive to the size of the bandwidth used in the estimation. We used sofware from Nichols (2012) to implement the Imbens and Kalyanaraman least-squares method to identify the relevant bandwidths. Details of these estimations are avail- able from the author on request. For details of the method, see Imbens and Kalyanaraman (2009). Expanding California’s Electorate 18 www.ppic.org 12 We ran the analysis of the efect of online registration on turn- out in other states for these key subgroups. For Latinos, it did suggest a boost in registration rates of about 2 percentage points under online registration, making California perhaps something of an outlier. However, a similar analysis limited just to young people ages 18 to 24 suggested no efect at all. Tese results are available from the author on request. Te population of Asian Americans in the states that have adopted online registration is generally too small for separate analysis. 13 A regression discontinuity analysis of these data confrms a statistically signifcant drop in the Latino registration rate right afer online registration became available. Te statistical signifcance of this discontinuity is sensitive to the bandwidth employed in the estimation, but its negative sign is not. We used sofware from Nichols (2012) to implement the Imbens and Kalyanaraman least-squares method to identify the relevant bandwidths. Details of these estimations are available from the author on request. For details of the method, see Imbens and Kalyanaraman (2009). 14 Self-reported registration rates in the PPIC samples exceed ofcial rates by about 10 percentage points, because of some unknowable combination of nonresponse bias and exaggeration by the respondents who agreed to participate. Te frst type of bias might exaggerate Internet use among the unregistered if the more Internet-sav vy among them are more likely to participate in the survey. However, the second type of bias will probably understate Internet use among the unregistered, since those most inclined to inaccurately claim that they are registered may also be the sort of educated, engaged population most likely to use the Internet, thus sapping the unregistered category of some of its heaviest Internet users. 15 Although this approach may have been adopted for any num- ber of reasons, it does make it easier to manage the potential for a single voter to cast a conditional ballot at multiple polling places. It also helps minimize the administrative complexity of ensuring that the proper ballot—in terms of either the races and initiatives ofered or the language they are translated into— would be available for any voter who might show up to register. 16 Te Help America Vote Act declares that states must have “a single, uniform, ofcial, centralized, interactive computerized statewide voter registration list” and they “shall enter into an agreement to match information in the database of the statewide voter registration system with information in the database of the motor vehicle authority to the extent required to enable each such ofcial to verify the accuracy of the information provided on applications for voter registration.” 17 Te newer methods try much harder to control for any unmeasured diferences between political entities that have adopted EDR and those that have not, to ensure that EDR itself is actually the cause of any diferences in turnout. Tese range from diference-in-diference models that use the change before and afer the adoption of EDR as a measure of its efect to regres- sion discontinuity designs that compare jurisdictions that just qualify for EDR according to some metric to those that just fail to qualify. 18 Hanmer (2009) argues that the early adopters saw more of an efect because they embraced EDR as part of their established participatory culture. States that adopted EDR grudgingly—such as Idaho and Wyoming, which sought to avoid coverage under the National Voter Registration Act in the early 1990s—had popula - tions that were less amenable to the reforms in general and so less likely to take advantage of them. California likely falls between these two extremes: the state adopted the reform voluntarily but it has not been an especially high-participation state, at least recently. However, see Keele and Minozzi (2013) for evidence of small or even negative efects even in Minnesota and Wisconsin. 19 Our analysis consists of a time-series cross section of voter turnout for all 50 states, with fxed efects for states and years and other key controls. We estimated this model both with actual election returns and with individual-level survey data from the Current Population Survey. Te detailed results of these models are in Technical Appendix B. 20 For California, both Asian Americans/Pacifc Islanders and Latinos are important groups to watch. However, the size of these groups in EDR states is too small for reliable analysis. 21 We conducted a diference-in-diference analysis of the change in Montana, using Idaho, Wyoming, North Dakota, and South Dakota as comparison states and controlling for age, marital status, gender, education, employment status, and mobility as demographic controls. In this model, the estimated efect of EDR was a statistically insignifcant increase in turnout of 1 percent. Coefcients and model ft can be found in Technical Appendix B. 22 Many of these numbers come from the Federal Election Assistance Commission, which conducts a survey of state elec- tion ofcials every two years. Te question wording for this survey refers to “same day registration” and does not ask states to identify what this means. 23 Currently, there are no states with both online registration and election day registration. Tus, it is impossible to say just what the efect of this combination might be. 19 Expanding California’s Electorate www.ppic.org 24 Overall, the evidence is mixed on a positive efect of VBM on turnout, but several studies have found a modest efect in at least certain circumstances. For a good summary, see Gerber, Huber, and Hill (2013). For evidence that VBM may actually reduce turnout in California, see Kousser and Mullin (2007). 25 Te closest congressional or state legislative outcome in the 2012 general election was Assembly District 36, where Demo- crat Steve Fox defeated Republican Ron Smith by just 145 votes. Unfortunately, we do not have late ballot data for that race in particular, so we cannot say whether late ballots could have fipped the outcome. 26 Te 24 counties are Amador, Butte, Colusa, Contra Costa, Fresno, Glenn, Humboldt, Inyo, Lassen, Los Angeles, Merced, Mono, Napa, Orange, Placer, Plumas, Sacramento, San Benito, San Diego, Santa Clara, Shasta, Siskiyou, Sutter, and Ventura. 27 Tere is always the chance that this number could increase under SB 29 if voters waited until they knew the election out- come and then fraudulently declared that their ballot was cast on time. However, since the ballots would have to be certain to arrive without a postmark, the fraud would require impersonat- ing a member of the postal service to deliver the late ballot or ballots without going through the regular mail. Although this is certainly possible in theory, it would seem to be difcult to accomplish without being detected. 28 Tis research design necessarily ofers only an approximate estimate of the causal efect of consolidation, because it is always possible that voters in the consolidated zip codes difer from others in important ways that have not been controlled for in the model. “Consolidation” in this analysis is treated as having one’s P&DC closed and merged with another. We also tried treating as “consolidated” all those facilities that received customers from closed facilities, on the assumption that their workload would increase and reduce the efciency of mail processing. Tere was no diference one way or the other from consolidation in that analysis. 29 Te diference is very small, so its importance should not be overstated, and its direction likely says something about the zip codes that were consolidated rather than the act of consolida- tion itself. Nonetheless, it casts doubt on concerns that continu- ing cuts at the USPS will lead to more late ballots. We also tried a somewhat diferent specifcation that allowed us to use data from all 58 counties. Instead of a late ballot fag, our dependent variable was a fag for whether a registered voter cast a ballot that was counted. We regressed this variable on VBM status, P&DC closure, and an interaction between the two. A nega- tive interaction term would suggest that VBM voters who were also in P&DC zip codes would be especially unlikely to have a counted ballot. We also controlled for the same demographics found in the original model. Te results were quite consistent with the fndings for the late ballot fag, regardless of the defni- tion of “consolidation” that we used. Coefcients and ft for these models can be found in Technical Appendix B. 30 See Berinsky (2005) for a similar point about the relative value of reform versus mobilization. 31 Some research has shown that social media have great poten- tial to mobilize citizens to go to the polls, and it seems relatively easy to incorporate online registration more thoroughly into such eforts (Bond et al. 2012). 32 Project Vote did a study of state agency compliance in Califor- nia and found that it had declined steeply since 1995–1996, when the Federal National Voter Registration Act frst required that such registration be ofered in 1993 (Herman and Hess 2009). 33 Te state with the closest existing approximation of this system is North Dakota, which has no voter registration at all. However, its system predates our data, so we have no robust means of evaluating the efect of its approach. Oregon recently considered adopting exactly the system described here, but the legislation fell one vote short in its state Senate (Holeywell 2013). 34 For instance, university students from out of state might want to maintain residency in their state of origin instead of being registered in California simply because they engage with a gov- ernment agency here. 35 One study that examined the efect of establishing voter registration in Ohio and New York in the 1970s estimated that it reduced turnout by about 3 to 5 percentage points (Ansolabehere and Konisky 2006). Expanding California’s Electorate 20 www.ppic.org 20 References Alexander, Kim. 2004. California Voter Participation Survey: Registering to Vote. Sacramento: California Voter Foundation. Alvarez, Michael, and Stephen Ansolabehere. 2002. California Votes: Te Promise of Election-Day Registration . Available at w w w.vote.caltech.edu/sites/default/fles/california_votes.pdf. Ansolabehere, Stephen, and David M. Konisky. 2006. “Te Introduction of Voter Registration and Its Efect on Turnout.” Political Analysis 14 (1) : 8 3 –10 0 . Baldassare, Mark. 2006. California’s Exclusive Electorate . San Francisco: Public Policy Institute of California. Available at w w w.ppic.org/main/publication.asp?i=705. Baldassare, Mark, Dean Bonner, Sonja Petek, and Jui Shrestha. 2013. PPIC Statewide Survey: Californians and Information Technology . San Francisco: Public Policy Institute of California. Available at w w w.ppic.org/main/publication.asp?i=1064. Barreto, Matt A., Bonnie Glaser, Karin MacDonald, Loren Collingwood, Francisco Pedraza, and Barry Pump. 2010. Online Voter Registration (OLVR) Systems in Arizona and Washington: Evaluating Usage, Public Confdence and Implementation Processes . Washington, DC: Pew Center on the States. Available at w w w.pewstates.org/research/reports/online-voter-registration -85899378 4 69. Berinsky, Adam J. 2005. “Te Perverse Consequences of Elec- toral Reform in the United States.” American Politics Research 33(4): 471– 4 91. Bond, Robert M., Christopher J. Fariss, Jason J. Jones, Adam D. I. Kramer, Cameron Marlow, Jaime E. Settle, and James H. Fowler. 2012. “A 61-Million-Person Experiment in Social Infu- ence and Political Mobilization.” Nature 489(7415): 295–298. Brians, Craig Leonard, and Bernard Grofman. 2001. “Election Day Registration’s Efect on U.S. Voter Turnout.” Social Science Quarterly 82(1): 170 –183. Davern, Michael, Arthur Jones, Jr., James Lepkowski, Gestur Davidson, and Lynn A. Blewett. 2007. “Estimating Regression Standard Errors with Data from the Current Population Survey’s Public Use File.” Inquiry 4 4(2): 211–224 . García Bedolla, Lisa, and Verónica N. Vélez. 2013. Diferences among Latina/o, Asian American, and White Online Registrants in Californi a. Berkeley: University of California, Berkeley, Center for Latino Policy Research. Available at http://escholarship.org /uc/item/44k3s91p. Gerber, Alan S., Gregory A. Huber, and Seth J. Hill. 2013. “Identifying the Efect of All-Mail Elections on Turnout: Stag- gered Reform in the Evergreen State.” Political Science Research and Methods 1(01) : 91–116 . Hanmer, Michael J. 2009. Discount Voting: Voter Registration Reforms and Teir Efects . Cambridge: Cambridge University Press. Herman, Jody, and Doug Hess. 2009. California’s Failure to Comply with the Public Agency Registration Requirements of the NVRA . Research Memo. Washington, DC: Project Vote. Available at w w w.projectvote.org/images/publications/State -SpecificDocuments/California/FINALPVCA NVRA5.7.09.pdf. Holeywell, Ryan. 2013. “Oregon May Be 1st with Automatic Voter Registration.” Politics (blog), Governing, April 5. Honaker, James, Gary King, and James Blackwell. 2009. “Amelia II: A Program for Missing Data.” Available at http://j.mp/k4t8Ej. Imbens, Guido, and Karthik Kalyanaraman. 2009. Optimal Bandwidth Choice for the Regression Discontinuity Estimator . National Bureau of Economic Research, Inc. Available at http://EconPapers.repec.org/RePEc:nbr:nberwo:14726. Keele, Luke, and William Minozzi. 2013. “How Much Is Min- nesota Like Wisconsin? Assumptions and Counterfactuals in Causal Inference with Observational Data.” Political Analysis 21(2): 193 –216. King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data: An Alterna- tive Algorithm for Multiple Imputation.” Te American Political Science Review 9 5 (1) : 4 9 – 6 9. Knee, Matthew R., and Donald P. Green. 2011. “Te Efects of Registration Laws on Voter Turnout: An Updated Assessment.” In Facing the Challenge of Democracy: Explorations in the Analy- sis of Public Opinion and Political Participation , ed. P. M. Snider- man and B. Highton. Princeton, NJ: Princeton University Press, pp. 312–328. Available at http://press.princeton.edu/titles /9602.html. 21 Expanding California’s Electorate www.ppic.org Kousser, Tad, and Megan Mullin. 2007. “Does Voting by Mail Increase Participation? Using Matching to Analyze a Natural Experiment.” Political Analysis 15(4): 428–445. McDonald, Michael. 2012. “Is Minority Voter Registration Really Declining?” Hufngton Post Blog, May 7. Available at w w w.huffingtonpost.com/michael-p-mcdonald/is-minority -voter-registr_b_1497813.html. Minnite, Lorraine Carol. 2007. Election Day Registration: A Study of Voter Fraud Allegations and Findings on Voter Roll Security . Available at www.demos.org/sites/default/files /publications/edr_fraud.pdf. Mitchell, Glenn E. II, and Christopher Wlezien. 1996. Voter Registration and Election Laws in the United States, 1972–1992. Ann Arbor, MI: Inter-university Consortium for Political and Social Research Data set ICPSR01102.v1. Nichols, Austin. 2012. RD: Stata Module for Regression Discontinuity Estimation. Sofware component provided by Boston College Department of Economics. Paget, François. 2009. Financial Fraud and Internet Banking: Treats and Countermeasures . Santa Clara, CA: McAfee, Inc. Available at www.mcafee.com/us/resources/reports/rp-financial- fraud-int-banking.pdf. Ramakrishnan, S. Karthick, and Mark Baldassare. 2004. Te Ties Tat Bind: Changing Demographics and Civic Engagement in California . San Francisco: Public Policy Institute of Califor- nia. Available at w w w.ppic.org/main/publication.asp?i=410. Romero, Mindy. 2013a. Online Voter Registration: Impact on California’s 2012 Election Turnout, by Age and Party Afliation . Davis, CA: Center for Regional Change, University of Califor- nia, Davis. Available at http://regionalchange.ucdavis.edu /ourwork/publications/ccep/ucdavis-ccep-brief-4-online-voter -turnout-impact. Romero, Mindy. 2013b. California’s 2012 Youth Voter Turnout: Disparate Growth and Remaining Challenges . Davis, CA: Center for Regional Change, University of California, Davis. Available at http://regionalchange.ucdavis.edu/ourwork/publications/ccep /ucdavis-ccep-brief-5-youth-voter-turnout. Expanding California’s Electorate 22 www.ppic.org 22 About the Author Eric McGhee is a research fellow at PPIC. His work focuses on elections, political participation, political polarization, legislative behavior, redistrict - ing, and surveys and polling. Before joining PPIC as a research fellow, he was assistant professor of political science at the University of Oregon; he previously worked at PPIC on the PPIC Statewide Survey. He holds a Ph.D. in political science from the University of California, Berkeley. Acknowledgments The author would like to thank a number of people for their help in making this report possible. The report drew on the expertise, and in some cases the data, of a long list of individuals who have committed their careers to understanding and improving California’s elections: Kim Alexander, Steve Boilard, Darren Chesin, Carolyn Crnich, Cathy Darling Allen, Efrain Escobedo, Bonnie Glaser, Kevin Keenan, Jill LaVine, Scott Lay, Dean Logan, Paul Mitchell, Michelle Romero, Don Smeraldi, and Nicole Winger. Special thanks to Paul Mitchell for access to his comprehensive data on voter registration. Kim Alexander also provided invaluable feedback on an earlier draft of the report, as did long-time scholars of elections Benjamin Highton and Karthick Ramakrishnan. The report benefited tremendously from the careful attention of Mary Severance and Lynette Ubois at PPIC. Thanks also to Mark Baldassare for his consistent support of the project. 23 Expanding California’s Electorate www.ppic.org Board of Directors DONNA LUCAS , CHAIRChief Executive Ofcer Lucas Public Affairs MAR K B ALDASSAREPresident and CEO Public Policy Institute of California RUBEN BARRALESPresident and CEO GROW Elect M AR Í A BLANCOVice President, Civic Engagement California Community Foundation B RIGITTE BRENAttorney WA LT E R B. HEWLETTChair, Board of Directors The William and Flora Hewlett Foundation PhIL I SEnBE rgChair Delta Stewardship Council MAS MAS UMOTOAuthor and Farmer STEVEN A. MERKSAMERSenior Partner n ielsen, Merksamer, Parrinello, g ross & Leoni, LLP KIM P OLESEChairman ClearStreet, Inc. THOMAS C. SUT TONRetired Chairman and CEO Pacific Life Insurance Company PPIC is a public charity. It does not take or support positions on any ballot measures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public ofce. PPIC was established in 19 9 4 with an endowment from William r. hewlett. Copyright © 2014 Public Policy Institute of California. All rights reserved. San Francisco, CA Mark Baldassare is President and Chief Executive Officer of PPIC. Donna Lucas is Chair of the Board of Directors. Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source. Research publications refect the views of the authors and do not necessarily refect the views of the staf, officers, or Board of Directors of the Public Policy Institute of California. Library of Congress Cataloging-in-Publication Data are available for this publication. I S B N 978 -1-5 8213 -15 6 - 6 PuBLIC POLIC y In STIT uTE OF C ALIFOrnIA 500 Washington Street, Suite 600 San Francisco, California 94111 Telephone 415.291.4400 Fa x 415 . 2 91. 4 4 01 PPIC SAC rAME nTO C E n TEr Senator Ofce Building 1121 L Street, Suite 801 Sacramento, California 95814 Telephone 916.440.1120 Fax 916.440.1121 Additional resources related to fscal and governance reform are available at www.ppic.org. The Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research." ["post_date_gmt"]=> string(19) "2017-05-20 09:41:55" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_114emr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:41:55" ["post_modified_gmt"]=> string(19) "2017-05-20 09:41:55" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_114EMR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }