Abstract
In mid-2011, the Florida state legislature passed House Bill 1355 (HB 1355) and in so doing placed new regulations on community organizations that historically have helped eligible Floridians register to vote. Among the legal changes promulgated by this bill were new regulations on the operations of groups like the League of Women Voters and a new oath, warning of prison time and fines, that voter registration agents were required to sign. Such changes raised the implicit costs that eligible Florida citizens faced when registering to vote, and we show that voter registrations across the state in the second half of 2011 dropped precipitously compared with registrations in the second half of 2007. This pattern is evident among registrants in general, among registrants age 20 and younger, and among individuals who registered as Democrats. Outside of HB 1355, we know of no credible explanations for these results. Our findings thus show how restrictions on the way that third-party organizations register voters can have tangible effects on actual registrations. Given that registration prior to an election is a civic necessity in Florida and in many other states, such restrictions have the potential to affect electoral outcomes as well.
Keywords
Introduction
In mid-2011, the Florida state legislature passed House Bill 1355 (HB 1355), a controversial and multifaceted election-reform bill that placed new regulations on community organizations that historically have helped eligible Floridians register to vote. These new regulations restricted the environment in which such organizations were permitted to operate and led to concerns that they might diminish the number of voter registrations across Florida. Many observers also saw in the new regulations an attempt to manipulate the partisan and racial composition of the Florida electorate. In August 2012, a federal judge permanently enjoined the facets of HB 1355 that were involved with new voter registration regulations, but by that time the regulations had been in force for over a year.
The proximity of HB 1355’s passage to the 2012 General Election motivates our focus on the consequences of this legislation for voter registration in Florida. Although our empirical analysis consists solely of an analysis of HB 1355, the framework of this study is a general one and points toward the exploration of the effect of a formal law on an important form of civic behavior, in our case registering to vote. Registration is a necessity in Florida for those seeking to vote, and thus any law that affects registration is conceivably one that influences election outcomes as well.
Generally speaking, there is no doubt that laws and administrative rules affecting voter registration can profoundly shape the electorate. The contrasting histories of the pre– and post–Civil Rights South are ample evidence of this (e.g., Keyssar 2000). Here, however, we are not exploring whether in the aftermath of HB 1355 obvious race-based barriers to registration affected the composition of the registered voter pool in Florida. Rather, the contemporary political debate in the United States about registration and voting is more subtle and turns on issues like voter identification and the availability of early voting. Rules governing these issues remain in flux in many states, and race remains a dominant theme in the overall picture on voting even in the absence of obviously race-based rules.
As to why we study voter registration in Florida, the answer is a relatively simple one: citizens must in this state register prior to voting, and thus scholars interested in the factors that shape election outcomes must focus at least in part on what shapes the pool of registered voters. There is a literature on the relationship between voter registration laws and subsequent behaviors—see Tokaji (2008) for a review—and within it scholars have studied the effects of election day registration on turnout, identifying consequences that run from modest (Brians and Grofman 2001) to more significant (Fenster 1994), and the effect of the National Voter Registration Act (e.g., Highton and Wolfinger 1998). The literature on registration has very little to say to the question of which state laws have facilitated or hampered community-based voter registration efforts, and this is unfortunate given the large and prominent organizations, like the League of Women Voters and the National Association for the Advancement of Colored People (NAACP), that conduct mass voter registration drives.
We contribute to the literature on registration both substantively—with a focus on Florida and in particular HB 1355—and methodologically—by isolating the effects of HB 1355 with daily voter registration counts. In contrast to our approach, many studies of voter registration rely on survey data; this may not be ideal because surveys on voter registration can be unreliable (Ansolabehere and Hersh 2012; Bernstein, Chadha, and Montjoy 2003; Converse and Niemi 1971; Wolfinger and Rosenstone 1980). The state of Florida maintains records that specify the day on which every registered Floridian actually registered, and our analysis takes advantage of this. In particular, we study the effects of HB 1355 on voter registration in Florida by comparing daily voter registration counts in 2007 with those in 2011. Our focus on these two years, which are identically placed in the four-year General Election cycle, reflects a recognition that a study of HB 1355 would be incomplete were we simply to ascertain whether, say, Florida registration numbers fell in the latter half of 2011 compared with the first half of that year.
We take advantage as well of the fact that the legal changes promulgated by HB 1355 did not immediately apply to all 67 Florida counties. As a consequence of the preclearance provision of the Voting Rights Act (VRA), five of Florida’s counties were initially exempt from this law, and these counties provide a control group for our research, albeit not one that was randomly assigned. The presence of Section 5 areas within Florida allows us to compare 2011 registration counts with those from 2007 for the 62 Florida counties where HB 1355 took effect and then to carry out a comparable exercise using the five preclearance counties where it did not. The latter five counties are not a perfect control group, but the existence of an imperfect control is certainly fortuitous compared with the situation we would have faced had HB 1355 uniformly affected all 67 of Florida’s counties as soon as it passed the Florida state legislature.
Briefly, we find that Florida voter registrations in the end of 2011 were sharply lower compared with registrations late in 2007, and we find no such 2011 versus 2007 difference when comparing the first few months of these years. The post–HB 1355 registration drop we find is evident across Florida in the aggregate, among registrants age 20 and younger, in the 62 non–Section 5 counties in Florida, and among the number of individuals who registered as Democrats. Although our analysis cannot directly speak to whether lower registration rates under HB1355 resulted from altered civic group strategies or the rational decisions of potential new registrants, it is nonetheless clear that this law produced differential effects across subgroups of Florida voters.
We now provide a short overview of the legal restrictions on voter registrations that followed from the passage and implementation of HB 1355. We then offer a theoretical way to think about the act of registering to vote, present our research design, and turn to results. The concluding section contains thoughts on the implications of our results for voter registration across the United States and comments on how the research design here can be instructive to scholars attempting to isolate the causal effects of changes in state election rules.
HB 1355 and Limits on Third-Party Voter Registration Organizations
On May 19, 2011, the Republican-controlled state legislature in Florida passed HB 1355, an omnibus election-reform bill. 1 Among its many provisions, which were publicly motivated by allegations of voter fraud, HB 1355 reduced from a maximum of 14 to just eight, the number of days county Supervisors of Elections could offer early voting; eliminated early voting on the Sunday immediately preceding election day; gave county Supervisors of Elections the discretion of offering between 48 and 96 hours of early voting over the reduced eight days as opposed to requiring them to provide 96 hours; required voters seeking to update their addresses at the polls to cast provisional ballots; and placed restrictions on individuals and community organizations wishing to engage in voter registration efforts (Herron and Smith 2012).
Upon becoming law in Florida, HB 1355 placed several new restrictions on what are commonly known as “Third-Party Voter Registration Organizations,” or 3PVROs. 2 A 3PVRO is “any person, entity, or organization that solicits (for collection) or collects any voter registration application,” although this definition allows for some exceptions like official state agencies that play a role in the voter registration process. Specifically, HB 1355 states that registration agents of 3PVROs must preregister, sign an oath warning of prison time and fines, and submit background information to the Florida Division of Elections. Prior to HB 1355, preregistration of 3PVROs was not enforced and no such oath was required by Florida law. Moreover, HB 1355 specifies that all 3PVROs in Florida are required to communicate to the Florida State Division of Elections within ten days any changes concerning their registration agents, file monthly reports with the Division accounting for all registration forms provided to and received from their registration agents, and ensure that they assign identification numbers to all voter registration forms in possession of their registration agents. Finally, once a registration agent of a 3PVRO receives a completed voter registration application, under HB 1355, he or she has 48 hours to deliver it to the Florida Division of Elections or the appropriate county Supervisor of Elections; these Supervisors, one per county for each of Florida’s 67 counties, are elected, constitutional officers of the state of Florida.
Despite objections raised by various nonprofit groups, Florida Republican Governor Rick Scott signed HB 1355 when it was presented to him, and substantial portions of the legislation went into effect immediately, on May 19, 2011. However, existing 3PVROs had a window of 90 days, to August 17, 2011, to comply with some of the law’s new regulations. Moreover, because HB 1355 had not, on May 19, 2011, received preclearance by the U.S. Department of Justice under Section 5 of the VRA, the legislation’s new 3PVRO regulations did not apply to five of Florida’s 67 counties. These counties—Collier, Hardee, Hendry, Hillsborough, and Monroe—fall under federal oversight, and thus the initial application of HB 1355 was uneven across Florida. 3PVROs operating in Florida after HB 1355 was signed were required accordingly to act under two separate sets of laws.
Following the passage of HB 1355 and concerned that they would not be able to meet the law’s new regulatory standards, the League of Women Voters of Florida, the National Council of La Raza, Rock the Vote, and several other organizations completely ceased their voter registration activities in Florida. As of May 2011, it was unclear to these organizations whether they would “be subject to the law’s requirements” if they limited their activities to the five Section 5 counties in Florida or whether they would be permitted to operate in these counties as they had before HB 1355 was implemented. 3
On May 31, 2012, approximately one year after HB 1355 had gone into effect in Florida, U.S. District Judge Robert L. Hinkle in the case of League of Women Voters of FL v. Browning granted a preliminary injunction blocking the enforcement of what appeared to be the most burdensome provisions placed on 3PVROs. Judge Hinkle enjoined the enforcement of the 48-hour turnaround deadline for collected voter registration forms and blocked the oath registration agents were required to sign before registering voters, calling the 48-hour deadline “harsh and impractical” (p. 2) and writing that the sworn oath “could have no purpose other than to discourage voluntary participation in legitimate, indeed constitutionally protected, activities” (p. 18). 4 In his published opinion, the judge ruled that the Florida law based on HB 1355 made voter registration drives “a risky business,” and he ruled that the plaintiffs would likely succeed on their claims that the voter registration provisions in HB 1355 violate First Amendment rights protected under the U.S. Constitution and the National Voting Registration Act of 1993.
A week after the Hinkle ruling, the League of Women Voters, Rock the Vote, Florida PIRG (Public Interest Research Group), and various other organizations resumed their statewide efforts to register voters. Judge Hinkle’s preliminary injunction against HB 1355’s voter registration provisions became permanent in August 2012, and the key features of this bill that many feared would diminish voter registrations across Florida were thrown out. Nonetheless, neither Judge Hinkle’s preliminary nor his permanent injunction assessed empirically whether the “harsh and impractical” regulations promulgated by HB 1355 actually did cause lower voter registrations in Florida. This is the subject of our upcoming analysis.
The Returns and Costs of Voter Registration
How might a change in the laws that regulate community groups that engage in voter registration drives affect actual voter registrations? From a theoretical perspective, registering to vote involves a rational calculus as does the decision to vote, the antecedent of casting a ballot. As Downs (1957, 260) noted, “Every rational man decides to vote just as he makes all other decisions: if the returns outweigh the costs, he votes; if not, he abstains.” We do not believe that the act of registering to vote should be thought of any differently. Because the most egregious barriers to registration and voting—racial and gender prohibitions, taxpayer or landholder requirements, poll taxes, literacy tests, and so forth—no longer are embedded in the American electoral landscape, the rational approach to registration is more amenable than it was, say, 50 years ago.
In considering the necessarily prior act of voter registration from a Downsian perspective, we argue that the motivation undergirding an individual’s decision to register to vote is likely determined by some combination of potential returns offset by potential costs. The plausible individual benefits or returns of registering to vote are roughly constant for eligible individuals holding one’s state constant. In contrast, costs or institutional barriers that confront these individuals when they try to register to vote can be quite variable.
The potential returns of registering to vote include a delayed instrumental benefit—being eligible to cast a decisive vote in a future election—combined with an expressive benefit—being a responsible citizen. With respect to the former, when an individual considers whether to register, the associated instrumental benefit is very abstract, even more abstract than the potential instrumental return of casting a decisive vote in an election. And beyond abstraction, there is the fact that the instrumental value of registration is often delayed: with few exceptions, any gratification an individual gleans from registering to vote can be delayed by months and possibly by years. Any expressive benefits derived from registering to vote, too, are likely tertiary. Unlike in the aftermath of casting ballots, citizens usually do not receive Red, White, and Blue “I Registered” stickers after having filled out voter registration forms.
With respect to such costs, the institutional barriers to registering to vote can be considerable, and there is no reason to think that these costs are roughly uniform across individuals. In contrast to the gradual expansion and federal standardization of voting rights, in the latter half of the twentieth century state legislatures and election administrators continue to wield considerable discretion over regulating voter registration. As such, barriers to registration can vary greatly depending on where a citizen lives, and in part this is because registration laws within states—and even across local jurisdictions—are uneven and occasionally capricious (Burden et al. 2011; Leighley and Nagler 1992; Rosenstone and Wolfinger 1978). Within Florida, the 67 independently elected Supervisors of Elections do not offer an identical number of hours of operations, and they do not allocate proportionally (in terms of county population) similar numbers of staff members to the task of voter registration outreach. 5
Because of the rather abstract benefits combined with the considerable costs of registering to vote, it should come as little surprise that many American citizens who are eligible to register do not. According to a recent report commissioned by the PEW Charitable Trusts, an estimated 51 million citizens in 2012 were unregistered, roughly 24% of the eligible population. And according to the U.S. Bureau of the Census, there are racial and ethnic differences among those eligible citizens who have registered to vote; in 2008, for example, eligible voters residing in households earning more than $100,000 per year were much more likely to be registered as those living in impoverished households, say, those earning less than $30,000 per year. 6 Unlike other western industrial countries, for example Canada as discussed in Rosenberg and Chen (2009), American state governments have no affirmative role to register citizens as qualified voters. Florida state law nonetheless requires that “every qualifying educational institution” give its enrolled students “the opportunity to register to vote or to update a voter registration record on each campus at least once a year,” and these efforts are usually assisted by county Supervisors of Elections. 7
Despite the fact that there is a dearth of scholarship devoted to the voter registration efforts of community groups in the American states, 3PVROs serve an important function registering eligible citizens to vote. For example, the nonpartisan League of Women Voters of Florida has registered voters in the state for seven decades, and Rock the Vote claims to have registered 90,000 citizens to vote in Florida, aged 18 to 29, in 2008 alone. 8 In Leon County, home of the state capital, Tallahassee, as well as Florida State University and Florida A&M University, in 2008 roughly 27,000 new voters were registered by third-party registration efforts, amounting to 15% of the approximately 175,000 county residents on the rolls, according to Leon County Supervisor of Elections, Ion Sancho. Four years earlier according to Sancho, 3PVROs working in the county signed up 22,000 new voters, or 13% of those who were registered to vote. 9
Notwithstanding the fact that HB 1355 was in force in 62 of Florida’s 67 counties, 556 organizations and individuals registered as 3PVROs between July 1, 2011, and May 10, 2012. Only a handful of 3PVROs were actively engaged in registering new voters during this period, and of the 71,492 new voter registration applications received by the Florida Division of Elections nearly a year after HB 1355 went into effect, almost 90% of them were collected by just ten 3PVROs and their registered agents. 10 A cursory analysis of comparable months (July through December) in 2007 and 2011 similarly reveals a steep drop in the number of new registrations. Most notably, 25,000 fewer voter registration forms were submitted by 3PVROs in 2011 compared with 2007. 11
Research Design
We have provided background on HB 1355 and the voter registration regulations that came out of this legislation, and we elucidated a simple cost-and-benefit approach that one can use to think about the act of registration. We now turn to our attempt to ascertain the impact of HB 1355’s regulations that until mid-2012 governed in Florida the environment of community groups who register voters, among other things. We surmise that these regulations raised the cost of registration to vote and thus may have diminished voter registrations across Florida. Whether this latter implication is true empirically is the subject of our research design and the results follow from it.
Daily Registration Data
We take a Florida voter file that was valid as of April, 2012, and from this file we calculate the number of registrants on every day of 2011. We acquired the April 2012 voter file and all voter files mentioned in this article via a public records request to the Florida Secretary of State, Division of Elections. Because Florida voter files contain information on voter race and ethnicity, party registration, age, and gender, we can disaggregate our daily registration counts into counts of registered Democrats, registered Republicans, registered voters in Broward County, registered Democrats in Broward County, and so forth. We perform a similar exercise using a Florida voter file that was valid as of April, 2008, and from this file we assemble daily voter registration counts for each day in 2007.
As we noted earlier, we rely here on comparisons between Florida voter registrations from 2007 with those from 2011. The year 2007 is an appropriate comparison year for 2011 because both of these years fall one year before a year with a regularly scheduled presidential election. In addition, the contemporary Florida Voter Registration System—called the FVRS—went into effect starting in 2006. Because the FVRS was new as of 2006, comparing voter registration patterns in Florida in 2011 with, say, those that occurred in 2003 (another year that precedes a presidential election) would not be ideal.
After we have retrieved both 2011 and 2007 daily registration series (for all registrants, for young registrants, for Democratic registrants, and so forth), we subtract them. We always subtract 2007 counts from 2011 counts, and this exercise produces a series of what we call “2011–2007 daily registration differences.” We use these differences as opposed to absolute counts to draw conclusions about the effect of HB 1355 on voter registration in Florida.
Smoothing
Both 2007 and 2011 contained 365 days, and this is fortuitous insofar as it facilitates our comparing registrations from these two years. Nonetheless, the days in 2007 and in 2011 do not line up perfectly. For example, January 1, 2007, was a Monday, and January 1, 2011, was a Saturday. It would obviously not make sense to compare voter registrations on a day in 2007 that occurred during a workweek and a day in 2011 that was on a weekend. Many other holidays are similar to New Year’s Day in this way, for example Independence Day and Thanksgiving Day. The point here is that ordering the 365 days in 2011 and 2007 and then calculating registration differences will not necessarily provide for useful comparisons.
On account of this problem, we turn to smoothing. To be precise, after we calculate differences (2011 registration counts minus 2007 counts), we apply a loess smoother to the results. Smoothing calculations were done in the R computing environment (R Development Core Team 2011), and we used a first order smoother with a value of α = 0.6. We emphasize that we smooth differences in counts and not raw counts themselves.
The primary reason that we smooth our registration differences is because of the way that the 365 days in 2011 line up with the 365 days of 2007. However, smoothing is also useful in light of the cycles and occasionally severe fluctuations that we observe in daily registration counts. Not surprisingly, registration counts tend to be higher on weekdays and lower on weekends, and this induces a registration count cycle with a seven-day period.
Although weekend registration counts are relatively low—often fewer than 10 per Saturday and Sunday—in 2007 and 2011, they were in general not zero. A Floridian’s official date of voter registration is governed by the following: “[T]he date a signed voter registration application is postmarked or hand delivered to [a] county Supervisor of Elections office will be [the] registration date.” 12 Post offices are in general not open on Sundays and neither are Supervisor of Election offices. However, Supervisors of Elections sometimes venture into their communities on weekends to register voters. Such behavior could lead to Sunday registrations. Still, we suspect that the nonzero registration counts we observe on Sundays embody some looseness in the way that official registration dates are recorded by the Florida Department of State (FDOS).
A Floridian who completed his or her voter registration form on a Saturday (assuming it was before post office closing time) could in principle have a true Saturday registration date. However, prior to the implementation of HB 1355, the lag between the date an eligible registrant completed his or her paperwork and his or her formal registration date could have been up to ten days; HB 1355 reduced this time period to two days. We suspect that lags of this type are what lead to reported registrations on (federal) holidays.
Beyond weekly cycles, another source of registration count fluctuations is the periodic occurrence of voter registration drives; such drives can generate large spikes in registrations that are intense yet very brief in duration. During the spring of 2012, for example, when many civic groups were no longer engaged in voter registration efforts, the Miami-Dade Public School system registered itself as a 3PVRO with the Florida Division of Elections. On a single day, April 4, 2012, 60 district schools “registered more than 10,000 high school students.” 13 A sense of the magnitude of this number can be gleaned from noting that the mean number in 2011 of daily registrations was approximately 1,327 (s ≈ 967) and the median, 1,652.
The Section 5 Counties in Florida
We have already noted that the implementation of HB 1355 upon passage was not uniform across Florida; this is because the regulations that followed from this legislation were not enforced in Florida’s five counties covered by Section 5 of the VRA. Consequently, the costs of registering to vote for Floridians living in Section 5 counties were arguably less than those of eligible Floridians residing elsewhere. The presence in Florida of a set of Section 5 counties provides a natural experiment for us, but of course, this experiment is only meaningful to the extent that there was variance in 3PVRO behavior between the five Section 5 counties and their 62 counterparts.
The public record suggests that many if not the majority of 3PVROs either stopped their voter registrations when HB 1355 became the law or radically curtailed their efforts. Thus, to ensure that some 3PVROs were active in Florida’s Section 5 counties, we contacted several of them. From personal correspondence, we learned that some 3PVROs did register voters in Florida’s Section 5 counties; for example, Mi Familia Vota Education Fund continued registering voters in Hillsborough as did the Florida Consumer Action Network. Still, the natural experiment comparing Section 5 versus non–Section 5 is imperfect because many 3PVROs in the summer of 2011 suspended their registration drives throughout Florida irrespective of the fact that HB 1355 was not in force in Florida’s Section 5 counties. This will make it difficult to detect differences between these two sets of counties; therefore, to the extent that we do find differences, our results are conservative.
Potential Caveat: Presidential Primaries
One limitation of our research design is the fact that the treatment we are studying—the implementation of HB 1355—was not randomly assigned, and neither are the five Section 5 counties in Florida. Simply put, we cannot control for all exogenous events that might have affected Florida voter registrations between 2007 and 2011, and one such exogenous event that theoretically could cause problematic inferences is the fact that there was no Democratic Presidential Preference Primary in January 2012. There was such a primary in 2008.
One might conjecture that the 2008 Democratic Presidential Preference Primary led to a surge in Democratic registration in late 2007. Such a surge might make the lack of a surge in 2011 appear to be a consequence of HB 1335 when in reality it was a reflection of the lack of a 2012 Democratic Presidential Preference Primary.
Fortuitously for our research design, the 2008 Florida Presidential Preference Primary was not officially sanctioned by the Democratic National Committee, no delegates were awarded based on its results, and the Democratic candidates for president did not officially campaign in Florida prior to it. Moreover, and particularly relevant for our objectives here, Barack Obama’s field operation did not begin registering voters in Florida in earnest until May 2012. 14 This means that a possible Florida registration surge reflecting the historic 2008 General Election should not be visible in our registration counts, which end on the last day of 2007. Of course, one could argue that in 2008 there was not a “real” Democratic Presidential Preference Primary because officially this primary was effectively disowned by the Democratic National Committee. Nonetheless, we wish to be methodologically conservative, and hence we treat the 2008 Democratic Presidential Preference Primary as real.
Perhaps counterintuitively, in light of the fact that there was no Democratic Presidential Primary in January 2012, the Florida Democratic Party’s voter registration efforts in the fall of 2011 far exceeded its efforts in 2007. Beginning in May 2011, President Obama’s field campaign, Organizing for America, was already setting up shop in Florida to conduct voter registration drives in conjunction with the state party. According to a vice chair of the Florida Democratic Party, the party’s voter registration effort in 2011 was “massive” and “night and day” from its efforts in 2007. In the summer and fall of 2011, the Democratic party trained “thousands of volunteers” in three-day sessions so as to comply with the 3PVRO requirements wrought by HB 1355. 15 This assists our research design because it means that a relative drop in voter registrations in 2011 cannot be attributed to the fact that the Florida Democratic Party was making only a lackluster effort at registration in the latter months of 2011.
If there were a registration surge in 2007 that might artificially make it appear that registrations in 2011 were relatively few in number—this could lead us wrongly to blame HB 1355 for depressing registrations—we lessen the chance of this surge causing problematic inferences by focusing almost all of our attention on voters who were no older than 7,300 days (20 years) when they registered to vote. We do this because, had an older voter registered during a 2007 surge in registrations, she would not need to register again in 2011. As noted, we have no reason to believe that such surges existed insofar as the Obama campaign began its registration push in 2008 as opposed to 2007. Still, as we exclude older voters from our analysis, this potential problem is not an issue for us. Florida law allows 16-year-olds to (pre)register to vote, and with a cutoff of 20 years, we can be sure that any registered voter younger than 20 years in 2011 could not have been legally registered to vote in 2007.
Summary of Research Design
After we sort our 2007 and 2011 registration counts, take differences (based on all registrants, broken down by party, broken down by racial group, and so forth), and then smooth the differences, we assess the differences by constructing a series of plots. Most of the plots involve young registrants only for the reasons noted above. To complement our plots, we also fit a series of linear regressions to our daily differences. To keep the presentation simple, though, we do not report complete regression results in the article (all results are available from the authors). Each regression contains a constant, an indicator variable for days before May 19, and an indicator for days after August 17. These dates correspond to the initial signing of HB 1355 and the day on which existing 3PVROs in Florida had to comply fully with its directives.
We expect to see registration differences through May 19 that are roughly zero, thus connoting that registration counts in 2011 and 2007 were roughly similar through May 19. If HB 1355 depressed voter registrations, however, then we expect to see registration differences drop either after May 19 or after August 17. Recall that our differences are calculated based on 2011 counts minus 2007 counts, and hence, a falling difference connotes a drop in 2011 registrations compared with 2007 registrations.
Findings
We now turn to results, and to provide context for our analysis we begin with a simple summary in Table 1 of 2007 and 2011 registration counts. This table reports counts of all voter registrations (see columns labeled “All”) as well as counts of registrations of voters who were no older than 20 years on the day of registration (see columns labeled “Under 20”). Note as well that Table 1’s results are broken down by date, specifically pre- and post–July 1. We use this date for convenience (it is halfway between January 1 and December 31) because it is approximately halfway through the 90-day period that started on May 1, 2011, and because some provisions of HB 1355 went into effect on July 1, 2011. Ideally, of course, the date of HB 1355’s implementation would have cleanly divided 2011 into a pre- and a post-period. However, given the staggered way that HB 1355 went into effect, the reality is not so neat. Our upcoming figures are not based on July 1, but here we invoke this date for simplicity.
Summary of 2007 and 2011 Voter Registrations.
Note: Columns labeled “Under 20” contain counts of registrations who were no more than 20 years of age upon registration.
Table 1 shows that total Florida voter registrations were lower in 2011 than in 2007. The decrease in registration was approximately 14% among all voters and approximately 8.0% among those voters less than or equal to 20 years on the day of registration. What is particularly notable about Table 1, though, is the extent to which the 2011 to 2007 registration drops differ depending on whether they calculated before or after July 1. For example, the pre–July 1 drop among total registrants in Florida was approximately 11% and the post–July 1 drop, approximately 16%. In other words, there was a greater percentage decrease in registrations from the second half of 2011 to the second half of 2007 compared with the first half of 2011 to the first half of 2007. Among younger registrants, the pre–July 1 registration drop was not a drop at all insofar as there were more young voters registered pre–July 1 in 2011 than in 2010. After July 1, however, the young registrant drop was almost 17%. If the full force of HB 1355 were not felt until approximately August 19, 2011, then the registration drops described here reflect registrations under both pre- and post–HB 1355 conditions, and hence the post–July 1 registration drops are conservative.
The calculations that follow from Table 1 are consistent with the conjecture that the implementation of HB 1355 led to decreased voter registrations in Florida. The data in the table are rough, however, and we now disaggregate them to further our ability to compare voter registration trends from 2007 with those of 2011.
2011–2007 Daily Registration Changes across Florida
Figure 1 contains two panels that describe daily 2011–2007 registration differences; the differences in Figure 1a reflect all voter registrations executed in 2007 and in 2011, and the differences in Figure 1b include registrants no older than 20. Both Figures 1a and 1b display smoothed differences with observed differences in the background in gray.
2011–2007 daily registration differences.
The observed registration differences in Figure 1 exhibit significant temporal variance, and this accords well with reasons described earlier. One can see in Figure 1a that some differences are relatively large, around 2,000, for example. A number of these differences reflect 2011 to 2007 comparisons that involve one weekday and one weekend (where there were very few registrations). A similar statement applies to differences that hover around -2,000. There are also differences in Figure 1 that are much closer to zero; these involve for the most part comparisons between equivalent types of days in 2011 and 2007. The sort of variability in both panels of Figure 1 is a visual motivation for our plotting of smoothed differences.
Figure 1a highlights a sharp drop in 2011–2007 registration differences several weeks after August 17. Most important, the steepness of the drop is clearly unlike the trend in registration differences that existed prior to May 19. There is if anything a positive trend in such differences prior to this date—not a drop. Figure 1a is thus highly suggestive that something dramatic occurred in Florida in mid- to late-2011 that affected 2011 voter registrations in the state. Whatever the cause, it began to manifest itself around September 2011.
Notwithstanding this point, Figure 1b is based solely on relatively young registrants yet bears similarity to Figure 1a: both figures show 2011–2007 registration differences that started dropping shortly after August 2011. Because Figure 1b uses young voters only, one cannot argue that its downward trend in 2011–2007 registration differences reflects a surge in voter registration in late 2007 that could not have been repeated in 2011 because the voters registered in the surge were already registered by the time 2011 occurred.
Because Figure 1 plots registration changes, it contains no information about overall registration levels in 2007 and in 2011. That is, one cannot tell from the figure whether 2007 registrations increased throughout the year while 2011 registrations were flat, whether 2007 registrations were flat while 2011 registrations dropped as the year progressed; and so forth. It turns out that 2007 registrations were relatively flat from January through the fall of the year with two exceptions: a temporary upward bump around April and a surge that started in October and lasted through December. Registrations in 2011, in contrast, were relatively flat throughout the year excepting a temporary April bump. What this means is that the differences plotted in Figure 1 reflect a 2011 registration series that failed to keep up with a 2007 series that had a notable increase starting in the last quarter of the year.
Registrant age notwithstanding, one might be concerned about the extent to which 3PVROs in Florida acted strategically in the period surrounding the passage of HB 1355. During this time frame, these organizations could have changed their behaviors in ways that are problematic for our analysis, but we do not believe that they did so. In particular, there is neither evidence in the public record nor from our conversations with 3PVROs that these groups stepped up their voter registration drives in May and June 2011, prior to HB 1355’s regulations going into effect. Alternatively, there is no indication that Florida 3PVROs lowered their efforts in this period, anticipating that any additional costs incurred expanding their operations might be for naught. We have also found no evidence that existing 3PVROs combined immediately prior to or after HB 1355’s passage or that organizations scrambled to register as new 3PVROs during HB 1355’s legislative hearings or around the time when Governor Scott signed this bill in May 2011. Indeed, of the 190 3PVROs that complied with HB 1355 by registering with the FDOS during the 2011 calendar year (out of the 739 3PVROs that registered between March 30, 2009, and the 2012 General Election), only 14 registered during the months of April and May 2011. According to state records, these 14 3PVROs subsequently registered a total of 19 voters prior to the 2012 presidential election. 16
Comparing Section 5 and Non–Section 5 Counties
We have noted already that there was variance across Florida in the extent that HB 1355 was implemented. Specifically, the presence in this state of five Section 5 counties provides us with a natural experiment that we can use to assess the effects of HB 1355. There were almost certainly spillover effects into these counties from the rest of Florida because some civic groups opted in the aftermath of HB 1355’s passage to suspend voter registration activities in Florida overall. Nonetheless, patterns in 2011–2007 registration differences from both Section 5 and non–Section 5 counties are shown in Figure 2; this figure is based on registrants 20 years or younger.
2011–2007 registration differences by Section 5 status, registrants no older than 20.
Figure 2a shows that 2011–2007 registration differences among Section 5 counties in Florida did not vary excessively across the entire time period studied. While there was a downward trend in registration differences after August 17, the trend was very similar to that which existed before May 19. Between May 19 and August 17, there was a temporary shift up in registration differences, meaning that in this window, there were more registrations in 2011 than in 2007. In contrast, Figure 2b shows the familiar and sharp downward trend in 2011–2007 registration differences. We saw this trend earlier (see Figure 1b) when we discussed overall trends in registration differences among Floridians no greater than 20 years of age.
The vertical axes in Figure 2a and 2b are different because there are only five Section 5 counties in contrast to 62 non–Section 5 counties. Therefore, one cannot directly compare the slopes of the lines in these two figures: that there are larger drops in the non–Section 5 counties could reflect the fact that more people live there. We can control for this problem with a simple regression analysis, and thus we estimated linear regressions of daily registration differences on time indicator variables, one indicator for the period between May 19 and August 17 and another for the period after August 17 (the omitted category is the period from January 1 through May 18). In the case of Section 5 counties, the estimated slope for the first time indicator is positive and statistically significant (t = 2.73) and the estimated slope for the second time indicator is negative and statistically insignificant (t = 0.993). All t values are based on heteroskedastic-consistent standard errors. In contrast, the non–Section 5 slope estimate for the May 19 through August 17 indicator is negative and insignificant (t = 0.868), and the slope estimate for the post–August 17 indicator is significant at the 0.06 level (t = 1.95). Given expectations for the direction of this slope, a one-tailed test is arguably appropriate, in which this latter estimate is significant at the 0.03 level.
These regressions in conjunction with the visual patterns in the two panels of Figure 2 highlight a difference in voter registration patterns. In the Section 5 counties, there is no significant evidence of a relative drop-off in registrations after HB 1355 was passed; if anything, there was a slight bump in registrations in 2011 relative to 2007. This bump is hard to rationalize, but it certainly is not consistent with the registration patterns observed in Florida’s non–Section 5 counties. In the latter, we see that a significant drop in registration occurred after HB 1355 was fully in force.
Our comparison of Section 5 versus non–Section 5 counties is methodologically useful in the sense that it flows from a natural experiment. More important, though, this comparison speaks to the value of the VRA. In Shelby County v. Holder, the U.S. Supreme Court is poised to rule on a challenge to Section 5 (oral arguments took place on February 27, 2013). The two panels in Figure 2 suggest that Section 5 of the VRA still has purchase 40 years beyond its initial passage. While racial breakdowns of the results in Figure 2b are not particularly useful (there are not enough minority registrations in the Section 5 counties to say much of anything regarding trends in registrations over time), one could still argue based on this figure that the VRA is working as it should be, protecting the registrants in areas of the United States known historically with barriers to electoral participation.
Partisanship and Registration
Figure 3 breaks our results down by party, and we consider here only registrants 20 years or younger. We focus now on registrant party affiliation because it is important to determine whether the results we have described thus far reflect exclusively registrants of one political party. It is fair to say that most of the objections to HB 1355 were associated publicly with supporters of the Democratic Party—to the extent that they had any partisan affiliation at all. Based on the rhetoric of such supporters, one might expect to see particularly negative trends in 2011–2007 registration changes among Democratic registrants.
2011–2007 registration differences by party, registrants no older than 20.
Democrats and Republicans are the two largest partisan groups in Florida, and the third largest group consists of so-called No Party Affiliation (NPA) registrants. As of the end of March 2012, NPA registrants made up approximately 20% of all Florida registered voters; Democrats constituted 40% of this pool and Republicans, 36%. Our analysis of HB 1355 and partisanship thus considers registration trends in these three groups.
Figure 3 plots 2011–2007 registration differences by party, and visually speaking all party groups in the figure suffered from 2011–2007 registration drops after the implementation of HB 1355. Note, however, that the slopes in the three party-specific panels are not directly comparable on account of size differences across partisan groups in Florida.
Figure 3a highlights a distinct drop in 2011–2007 registration differences among Democrats. The drop is stark, and as we have seen before, it starts a few weeks after August 17. We fit a linear regression to the registration differences in Figure 3a, and the slope estimate for the post–August 17 indicator variable is negative and easily significant (t = −2.28). In contrast, the post–August 17 slope estimate from the Republican Figure 3b is not significant (t = −1.29), and the same is roughly true of the corresponding NPA estimate from Figure 3c (t = −1.65). This estimate would be considered significant were we to use one-sided tests at the 0.05 level, but such a standard is not very demanding.
Figure 3a’s drop in Democratic registration could in theory be rationalized by arguing that it reflects the lack of a 2012 Democratic Presidential Preference Primary. However, as we noted previously, the Florida Democratic Party had a much stronger registration effort in 2011 compared with 2007, and because of this our results on the Democratic gap in Figure 3 are conservative. That Democratic registrations were down despite a relatively strong registration push from the Florida Democratic Party means that corresponding estimates of the putative effect of HB 1355 are biased down or lower than they would have been in the absence of a strong party registration effort.
Figure 3d contains another perspective on the relationship between partisanship and registration. In particular, this figure plots the Democratic–Republican difference in 2011–2007 registration differences. In other words, each point in this plot is a difference from Figure 3b subtracted from a difference from Figure 3a. When this difference-in-difference series decreases—as it does in Figure 3d—it implies that the Democratic registration difference is large compared with the associated Republican registration difference. As registration differences have been defined throughout as 2011 registrations minus 2007 registrations, a decreasing difference-in-difference in Figure 3d connotes a decreasing Democratic advantage in registrations in 2011 compared with 2007.
Indeed, we see from Figure 3d that by August 2011, Republicans started to gain (compared with 2007) in terms of young voter registrations. Moreover, the difference-in-difference series in Figure 3d is negative by the end of the year pictured. And recall that the Florida Democratic Party worked much harder in 2011 to register voters than it did in 2007. Thus, after HB 1355 was implemented, it appears that Democratic registrations among those 20 years and younger dropped at a much faster rate than corresponding Republican registrations.
Our findings about the relatively large number of young, Republican registrants in December 2011 are not an artifact of the smoothing processes behind Figure 3. To see this, consider Table 2’s breakdown of monthly registration figures by party. According to this breakdown, there were 3,498 Republican registrations (age no greater than 20 years) in December 2011 and 3,208 Democratic registrations in this same month. Indeed, across 2007 and 2011, December 2011 was the only month in which Republican registrations outweighed Democratic ones.
Monthly Registrations and Partisanship, Registrants No Older than 20.
Note: Differences in the table are Democratic registrations minus Republican registrations. The Difference-in-Difference column reflects 2011 differences minus 2007 differences.
For both 2007 and 2011, the columns in Table 2 labeled “Difference” contain the number of Democratic registrations minus the number of Republican registrations. The “Difference-in-Difference” column is then the difference (2011 minus 2007) of these partisan differences. Starting in May, the Democratic minus Republican difference was smaller in 2011 than in 2007. This means that the advantage Democrats held in daily registrations between May and December was larger in 2007 than in 2011. And, the partisan difference-in-difference across 2007 and 2011 decreased in the second half of the year, meaning that the Democratic advantage in registrations was greater before May than after. This result is identical to that seen earlier in Figure 3 except that there is no smoothing in Table 2.
What is particularly notable about Table 2’s results is the fact that the Democratic advantage in registrations (2011 vs. 2007) started to slip in May. We have mentioned before that HB 1355 was signed on May 19, 2011, and went immediately into effect. While existing 3PVROs had 90 days in which to comply with the law’s regulation requirements, new 3PVROs had to comply with them right away upon registration with the state. The staggered implementation of HB 1355 helps rationalize the partisan differences in Table 2. Indeed, the numbers in Table 2 suggest that HB 1355 started to have an effect on the partisan composition of Florida’s electorate very soon after it was signed. This conclusion would hold even if one were hesitant to ascribe excessive importance to the May difference in Table 2 insofar as the May 2007 registration count was so high. May 19, that is, is closer to June 1 than it is to May 1, and the fact that all of the partisan registration differences in Table 2 are negative starting in May is notable.
The partisan effects in Figure 3 and Table 2 are arguably more meaningful than the overall decreases in registrations apparent in Figure 1. In this latter figure, we showed that absolute counts of registrations were down in 2011 compared with 2007, albeit not seriously until after HB 1355 was fully in force. While drops in overall registrations are not necessarily normatively pleasing, if such drops affected all Florida voter types equally, one could in theory argue that said drops would have no effect on election outcomes. Such an argument would not be compelling in light of our partisan findings. Elections are won based on vote margins between candidates, and because of this it would be very difficult to argue that the partisan composition of the Florida electorate is not important. HB 1355, by this logic, was not politically neutral.
Race and Ethnicity in Registrations
We now turn to the racial and ethnic side of HB 1355 and assess whether this law had disparate effects on the different racial and ethnic groups in Florida. To this end, Figure 4 describes the racial and ethnic dimensions of 2011–2007 registration differences in the same way that we earlier studied partisan groups in Florida. African American, Hispanic, and white registrants constituted 94% of the Florida electorate as of the 2012 General Election. 17
2011–2007 registration differences by race, registrants no older than 20.
Visually speaking, one takes from Figure 4 the idea that the three main racial groups in Florida all suffered from drops in voter registrations in late 2011 as compared with late 2007. Nonetheless, when we fit regressions to the registration differences in the three panels of Figure 4, out of the six estimate coefficients (two time indicator variables per regression), the only statistically significant estimate is that for the post–August 17 variable in the African American regression. This is evidence that HB 1355’s effects varied by racial/ethnic group and were particularly pronounced for a group that in Florida is closely tied to the Democratic Party. This raises the specter of what might befall nonregistered African Americans residing in Florida if Section 5 of the VRA is curtailed or eliminated.
Two Possible Confounds: Growth and Regional Trends
We end our presentation of results by considering two possible confounds that may affect our conclusions. The first confound reflects the fact that our research design holds Florida constant between 2007 and 2011. That is to say, we have thus far ignored migration in and out of the state, and this is in theory problematic. It could be argued that one explanation for our claims about HB 1355 is that Florida’s population shrunk between 2007 and 2011. If there were relatively few eligible registrants in Florida in 2011, it would be no surprise to see relatively small registration counts for that year, HB 1355 notwithstanding.
According the U.S. Census Bureau, the population of Florida in 2007 was estimated at 18,251,243 and in 2011, 19,057,542. Florida’s growth between 2007 and 2011 implies that a shrinking state is unlikely a confound for us. Notwithstanding this point, we generated a plot akin to Figure 1 that contains 2011–2007 daily registration differences but restricted to the 54 Florida counties that experienced positive population growth between 2007 and 2011 (plot available from the authors). By focusing on these counties in particular, we are biasing ourselves against finding low 2011–2007 registration differences. 18 Despite this fact, the patterns we observed earlier when analyzing all Florida counties exist as well in the counties whose populations grew between 2007 and 2011. From this, we conclude that decreasing populations in selected Florida locales cannot explain observed drops in 2011–2007 registration differences.
Another possible confound for our results is the presence of regional trends that affected Florida yet were independent of HB 1355. In the spirit of Campbell and Ross (1968), we therefore compare daily registration trends in Louisiana, a southern state physically proximate to Florida, between 2007 and 2011. In particular, we acquired 10,000 random draws of Louisiana registrants less than 34 years of age (this particular limit reflects a data purchasing system used by the state of Louisiana), and from these draws, we calculated daily 2007 registration counts for voters aged 18 to 21 and daily 2011 registration counts for voters aged 22 to 25. We then calculated the difference between the 2011 counts and the 2007 counts, and we plot the difference series in Figure 5.
2011–2007 registration differences in Louisiana, registrants age 18 to 21 for 2007 and 22 to 25 for 2011.
We observe in Figure 5 nothing that would make us believe that our Florida findings reflect general trends that affected the American South (or an area greater than the South). To be precise, trends in Louisiana registration differences do not mimic Florida trends. Figure 5 shows that 2011–2007 registration differences in Louisiana increased in the latter half of the pictured year, perhaps after August 1 or thereabouts. We saw no such increase in Florida.
Our consideration of Louisiana registration trends is particularly important because of the nature of our earlier Section 5 versus non–Section 5 comparison in Florida. As we pointed out in the course of that comparison, there was spillover between Florida’s 62 non–Section 5 counties and the five covered ones; this weakens inferences from the Section 5 natural experiment although we argued that spillover makes our results conservative. Regardless, we can be reasonably sure that there was no such spillover in Louisiana unless one were to argue that Florida residents disproportionately moved to Louisiana during the period surrounding HB 1355; this would be hard to sustain given the surge in registrations that occurred in Florida after HB 1355 was enjoined in mid-2012.
Conclusion
This article assesses the extent to which a set of Florida state regulations placed in mid-2011 on third-party community groups that traditionally conduct voter registration drives affected the number of voter registrations in the state. These regulations were implemented because of the passage of HB 1355, a controversial piece of election-related legislation whose implementation many feared would depress voter registrations. With a Downsian perspective as background, we have compared daily voter registration patterns in Florida for two comparable pre–General Election years. Our analysis concluded that voter registrations in the latter part of 2011 were significantly lower than registrations in the latter part of 2007. Furthermore, we identified a strong partisan aspect of the 2011–2007 registration drop, namely, that Democratic registrations were down much more than corresponding Republican registrations.
As of late August 2012, HB 1355’s voter registration restrictions were no longer in force; by that time, a federal judge had found them to be in violation of the National Voting Registration Act. These restrictions were nonetheless on the books in Florida for almost a year and, notably, during the run-up to the 2012 General Election. How meaningful were they? Not until January 2012 did the FDOS include a category of 3PVRO registrations in its monthly registration reports, and thus we cannot know the precise number of 3PVRO registrations executed before and after HB 1355. Nonetheless, in October 2012, 3PVROs in Florida registered 46,673 voters, approximately 25% of the 189,736 voters registered in the state during this month. 19 If a voter registration method responsible for, roughly speaking, a quarter of registered Floridians has a partisan bias, then arguably the entire Florida registration system has a partisan bias. Given historical vote margins in Florida in important elections, it would be difficult to argue that the numbers of voters registered by 3PVROs is anything but important.
While our findings should be accompanied by the usual caveats regarding single-state case studies, our brief analysis of Louisiana assures us that Florida’s voter registration experiences do not reflect broader trends in the American South. Moreover, Florida’s experiences with HB 1355 and this law’s focus on 3PVROs are not entirely unique. In the same year that HB 1355 was signed into law, bills were introduced in California, New York, North Carolina, and Texas to prohibit payments to individuals engaged in voter registration efforts; a bill in Texas sought to levy fines on 3PVROs that were tardy in turning in voter registration applications, and a bill in Mississippi tried to force 3PVROs to register with the state. These pieces of legislation died in session, but no one should be surprised if other states experiment as Florida did with efforts to restrict the actions of 3PVROs. The possible repercussions of restrictions on 3PVROs in states across the country are conceivably very significant because fewer than half of all registered voters reported in 2008 that they registered at county or government registration offices or when obtaining driver’s licenses. Indeed, 3PVROs account for a large share of voter registration applications across the entire United States. 20
In our presentation of results, we described an analysis that compared Florida’s five Section 5 counties with its 62 non–Section 5 counties. And, in the course of the said analysis, we offered evidence that post–HB 1355 registration drops in the former were muted compared with those in the latter. This implies, simply, that through 2011, Section 5 was continuing to provide meaningful restrictions on possibly discriminatory voting laws. The importance of this point cannot be understated in light of the contemporary legal debate about the continued relevance of the VRA and Section 5 in particular. As for this writing, the status of the VRA is uncertain as the Supreme Court has not yet ruled in Shelby County v. Holder. Independent of the court’s eventual position in this case, the results shown here are consistent with the argument that a lack of overtly discriminatory voting practices does not imply that the VRA and the preclearance provisions in Section 5 are obsolete. Indeed, our evidence points to the impact of preclearance even when a voting law in question—here, HB 1355—lacks obviously racially discriminatory aspects.
Arguments about the relevance of Section 5 of the VRA are occurring in the United States concomitant with vigorous debates about voting procedures and the exigencies of voter identification. Ideally, we believe the promulgation of new voting laws across the United States will lead to a surge in research on the consequences of these laws. The methodological approach we have implemented here can be thought of as one way for scholars to study the consequences of new state laws. We have emphasized here the importance of temporal comparisons (e.g., pre–HB 1355 vs. post–HB 1355) within state comparisons (e.g., Section 5 counties vs. non–Section 5 counties), as well as across state comparisons (e.g., Florida vs. Louisiana). Different research venues will confront different methodological challenges of course, but we believe that the types of comparisons illustrated here should be part of the continuing research agenda on the effects of changing voting laws on various forms of electoral participation.
Relatedly, it would be beneficial if future research on the impact of voting laws were to focus on the mechanisms at work when registrations drop (or increase). The research design employed here is ultimately agnostic over whether the effects we have identified are entirely supply-based (meaning, the registration drops we have found reflect a lack of registration opportunities) or possibly somewhat demand-based (meaning that potential registrants in Florida changed their behaviors post–HB 1355 and shunned previously available opportunities to register). Although we cannot pin down the precise mechanism or mechanisms by which HB 1355 drove down registration, identifying such mechanisms should be a goal of future research.
Finally, our results raise concerns about excessive partisanship in the way that elections are run in the United States. Minnite (2012, 90) observed that, “[S]tate election codes can be read as records of past partisan conflict reflecting the efforts of the victors to impose or redesign rules that advantage their side, weaken their opponents, and institutionalize their power,” and this seems very apropos given the history of HB 1355. This piece of legislation passed the Florida state legislature in a very partisan way; the margin in the Florida House was 77–38 (complete party-line vote) and in the Florida Senate, 25–13 (party-line except for two Republicans who voted nay). On top of this, we have shown that this law advantaged Republican registrations. The combination of these facts should draw attention to the question of whether Florida is best-served by having a partisan body dictate the laws that control access to voting. It is probably neither impossible nor appropriate to divorce completely partisanship from election administration (e.g., Hasen 2012), but one might honestly question whether a state legislature is best-suited to controlling the minutiae of registration details in the way that the Florida legislature did with HB 1355.
Footnotes
Acknowledgements
The authors thank Diana Kasdan for helpful comments on an earlier draft as well as Pam Carpeuter, Alachua County Supervisor at Electious, and her staff for insights on voter registration procedures in Florida.
Authors’ Note
This article was initially prepared for presentation at the 2012 Annual Meeting of the American Political Science Association, New Orleans, Louisiana, but not presented on account of Hurricane Isaac.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.