Equal Vote

Elmendorf on Crawford on Harper

2008-05-06T10:53:00.007-04:00

Chris Elmendorf offers this comment for EL@M, regarding last week's decision from the Supreme Court in Crawford v. Marion County Election Board. It's characteristically thoughtful, provocative, and well worth reading. (Disclosure: He and I co-authored an amicus brief to the Court on the case.)

Particularly interesting is Elmendorf's discussion of the lingering uncertainty over the level of scrutiny to be applied in cases challenging restrictions on participation. As I suggested in my initial post on Crawford last week, one of the most important aspects of the lead opinion is its treatment of the Court's landmark decision in Harper v. Virginia Board of Elections (1966), which struck down a $1.50 poll tax on the ground that it violated equal protection. Elmendorf has this to say about the discussion of Harper in Justice Stevens' lead opinion (joined by the Chief Justice and Justice Kennedy):

Much to my surprise, six Justices unequivocally affirmed that Burdick requires open-ended balancing, rather than a threshold classification of the challenged requirement as "severe" or "not severe." To emphasize this point, Justice Stevens's lead opinion read Harper v. Virginia Bd. of Elections, which held that Virginia may not require would-be voters to pay a $1.50 tax, as standing for the proposition that "[h]owever slight [a] burden on [voting] rights may appear . . . it must be justified by relevant and legitimate state interests sufficiently weighty to justify the limitation." Slip op at 7. (I had treated Harper as the source of a special, per-se rule that express financial conditions on political participation are qualitatively "severe in kind" and thus subject to strict scrutiny, even if not practically burdensome.)

I think Elmendorf is right to say that the lead opinion requires some sort of weighing of interests in all cases involving restrictions on participation, and not just in those which impose "severe" restrictions -- more on this below. But I don't think it's the lead opinion renounces the idea that "severe" restrictions warrant strict scrutiny. Rather, the lead opinion in Crawford understands Harper as a case that did apply something like strict scrutiny. At the start of his discussion of Harper (slip op. at 5). Justice Stevens describes Harper as "[a]pplying a stricter standard" than rational basis. This is consistent with Harper itself, which said that laws burdening fundamental rights like voting must be "closely scrutinized and carefully confined."

Of course, Harper came before the development of the three-tiered approach to levels of scrutiny that we know today. And, of course, Justice Stevens has never been a fan of the Court's three-tiered approach. (See, e.g., his concurring opinion in Craig v. Boren (1976).) Still, I don't think Crawford should be read as discarding the idea that some restrictions on participation are sufficiently severe as to warrant strict scrutiny. Later in the same paragraph, Justice Stevens characterizes Harper as a case involving "invidious" discrimination. Even though there wasn't a showing of intentional discrimination, and even though the tax was "rational," according to Justice Stevens, "it was invidious because it was irrelevant to the voter's qualifications." This suggests that there's still a threshold test, which was met in Harper but not Crawford, for determining whether to apply strict scrutiny.

In the passage Elmendorf quotes, Justice Stevens goes on to say that "[h]owever slight [a] burden on [voting] rights may appear . . . it must be justified by relevant and legitimate state interests sufficiently weighty to justify the limitation" (emphasis mine). But appearances can be deceiving. In Harper, the law was "invidious" despite its relatively benign appearance. After its discussion of Harper, the lead opinion reaffirms the approach in Anderson and Burdick, cases that articulated the severe burden standard, while recognizing that there's no "litmus test for measuring the severity of a burden that a state law imposes ...." (slip op. at 7).

What all this means, I think, is that there's a spectrum of burdens, and that the justification that the state must provide will vary accordingly; but it's not to deny that, at one end of the spectrum, the state will have to show that its law is narrowly tailored to a compelling interest. Most if not all of the justices seem to agree that some voting laws warrant strict scrutiny. Joined by Justices Alito and Thomas, Justice Scalia reads Harper -- properly in my view -- as a case applying "strict scrutin[y]" (although I think he's wrong to characterize the poll tax law as "nondiscriminatory"). And Justice Souter, joined by Justice Ginsburg, acknowledges that burden imposed by Indiana's voter ID law isn't severe, but then go on to explain why that burden shouldn't be ignored altogether.

It thus appears that a majority of the justices agree that even laws that don't impose a severe burden get something more than deferential rational basis review. This is something on which I hadn't really focused before reading Elmendorf's comment. It's a point on which the lead opinion and the Scalia group sharply disagree. Justice Scalia would adhere to a more rigid two-tiered framework, applying a "deferential" standard where a "severe" restriction is lacking (Scalia slip op. at 1-2). By contrast, both the lead opinion and the dissenters seem to think there's got to be some evidence justifying a burden on voting, even one that isn't "severe." See Souter slip op. at 7 ("Although making voters travel farther than what is convenient for most and possible for some does not amount to a 'severe' burden under Burdick, that is no reason to ignore the burden altogether.") ; Breyer slip op. at 1 (looking to whether the law's burden is disproportionate to its benefits).

I thus think it's right to point out, as Elmendorf does, that Crawford softens the two-tiered severe/nonsevere framework that some courts have read Anderson and Burdick to have created. It doesn't reject the idea that "severe" burdens get strict scrutiny, but less now hinges on this question. That's because even laws that don't impose severe burdens on voting should (according to six justices, by my count) get something more than deferential rational basis review.

All this admittedly requires some not inconsiderable between-the-lines reading of the various opinions in Crawford. As Elmendorf puts it, those opinions "do little to clarify how the courts ought to weigh the 'voting related interests' assertedly at stake in constitutional challenges to voting mechanics." Indeed, the case may well leave the law more confused than it found it. The Crawford opinions can, however, be read to support the proposition that the state isn't off the hook just because its law imposes a "nonsevere" burden on voter participation. That's certainly how I'd read them, were I a voting rights lawyer trying to challenge an election administration practice after Crawford.

An Election Day Registration Bill

2008-05-02T09:46:00.006-04:00

A bill has been introduced in the Senate that would require states to allow election day registration (EDR) in federal elections. The bill may be found here and co-sponsor Senator Russ Feingold's statement in support of the bill here.

EDR has been used in nine states some of them since the 1970s. We therefore have quite a bit of evidence on its efficacy. As detailed in our recent report From Registration to Recounts: The Election Ecosystems of Five Midwestern States, it's been very successful in those states, increasing participation without any evidence of an increase in voter fraud -- the main argument that's usually used against EDR. Minnesota and Wisconsin are particularly good examples of the positive effects of EDR, as documented in that report.

Social science research has consistently found an increase in turnout due to EDR, probably in the range of 5-10%, as documented in testimony that I offered to the House Adminstration Committee when it was considering another EDR bill late last year. Senator Feingold's statement in the support of the bill thus aptly cites this as an example of the "moneyball approach to election reform," which I discussed in this post. That's the idea that election reform ought to be based on solid empirical evidence rather than intuition and anecdote -- something that's too often been the case by reformers on both the left and the right.

Another benefit of EDR is that it reduces reliance on provisional ballots and, with it, the likelihood of post-election litigation over whether those ballots should be counted. As I stated in testimony on another EDR bill that was being considered by the House Administration committee late last year:

While the turnout benefits of EDR are widely recognized, another advantage has received virtually no attention: allowing EDR can almost entirely eliminate the need for provisional ballots. That is primarily due to the fact that voters whose registration forms are mishandled need not cast a provisional ballot in EDR states. Instead, they may simply register at the polls. Consider, for example, voters whose names do not appear on registration lists when they appear at the polling place for any of the following reasons:
-a third-party registration group soliciting voter registration inadvertently failed to return the form,
-a public agency that took the voter's registration application, such as a state motor vehicle office, failed to transmit that registration to the appropriate county election office, or
- the county election office made a data-entry error, say in the voter's home address, causing the voter's not to appear on the registration list for his or her polling place.
In each of these circumstances, the voter's name would not appear on the registration list for the proper polling place when he or she shows up to vote on Election Day. In a state without EDR, that voter would be relegated to the provisional voting process. In an EDR state, by contrast, the voter would be permitted to register and vote on Election Day, provided that he or she satisfied state requirements for confirming eligibility.
The data on provisional voting confirms that EDR states are much less reliant on provisional ballots than other states. In the 2004 election, for example, the EDR states of Maine, Wisconsin and Wyoming all had 0.05% or less of their registered voters cast provisional ballots.1 Maine had only 483 provisional ballots cast statewide, while Wisconsin had only 374, and Wyoming just 95.2 In Minnesota, there were zero provisional ballots cast in the 2004 presidential election. By contrast, almost 2% of Ohio's registered voters - a total of 157, 714 people - cast provisional ballots in 2004. Quite clearly, the much larger number of provisional ballots cast increases the likelihood of a close election turning into a disputed election. It is not difficult to imagine the nightmare scenario that would have emerged in Ohio in 2004, had the margin of victory been closer. The two candidates would have wound up arguing over whether provisional ballots should be counted in counties across the state, just as they argued over whether punch card ballots should be counted after Florida's 2000 election.
Greater participation in our democracy is thus only one of the benefits of EDR. Adopting EDR can virtually eliminate the need for provisional ballots and, with it, a potential source of contestation and litigation over close elections.... Put more simply, EDR promotes the value of finality as well as access, and does so without sacrificing electoral integrity.

For these reasons, I hope Congress will give serious consideration to this bill.

Crawford: It Could Have Been Worse

2008-04-29T06:53:00.018-04:00

That's about the best that can be said about yesterday's opinions in Crawford v. Marion County Election Board. Following the precedent set by LULAC v. Perry and Randall v. Sorrell in 2006, the Court issued a splintered decision in an important constitutional election law case. The lack of a majority opinion, and the narrowness of Justice Stevens' lead opinion (joined only by the Chief Justice and Justice Kennedy), will probably limit the harm done by the Court's decision to uphold what is probably the strictest and most exclusionary voter ID law in the country.

One couldn't realistically expect much better from the current Court. In a variety of areas -- most notably abortion and racial integration of public schools -- the Roberts Court has taken a dramatic turn to the right. While nominally adhering to past decisions, the Court has in fact eviscerated precedent, something that Justice Scalia in last year's Wisconsin Right to Life v. FEC referred to as "faux judicial restraint." What almost certainly prevented the Court from doing similar damage in Crawford is the fact that, in something of a surprise, Justice Stevens was part of a majority of justices voting to uphold Indiana's restrictive law, which requires government-issued photo ID in order to vote at a polling place and have that vote counted.

Justice Stevens' lead opinion, to its credit, starts with the precedent most at risk in Crawford: the Court's 1966 opinion in Harper v. Virginia Board of Elections which struck down Virginia's $1.50 poll tax. In that case, the Court applied what Crawford properly characterizes as a "stricter standard" than rational basis to a law that imposing a voting condition that would disproportionately exclude poor people. As I set forth in this recent comment in the University of Pennsylvania Law Review's online supplement, Harper ought to be the starting point for evaluating claims that an election practice disproportionately excludes a particular group from participating in elections.

In Crawford, the lead opinion declines to apply a comparably strict standard, despite the fact that Indiana's law seems comparably designed to exclude poor voters, not to mention those who are elderly, disabled, students, and racial minorities. It's certainly true that the evidence of disparate impact on these and other groups was pretty thin in the case. But the same could be said of Harper, which didn't rest on empirical research that a poll tax would have a disparate impact on poor voters. If anything, Indiana voters' empirical case was stronger.

So what distinguishes Crawford from Harper? The lead opinion's answer seems to be this: "Although the state's justification for the [poll] tax was rational, it was invidious because it was irrelevant to the voter's qualifications." The problem here is that the same can be said for Indiana's law, given the utter lack of evidence showing that the state has any problem whatsoever with voters going to the polls pretending to be someone they're not -- which the lead opinion admits to be the only problem that the state's ID law could possibly remedy. Even if one looks beyond Indiana's borders the evidence of in-person voting fraud is exceedingly scant, as a careful reading of the lead opinion (especially footnotes 11 & 12) makes clear. The evidence of fraud that exists is mostly with absentee ballots, a problem to which a polling place ID requirement is irrelevant, as the lead opinion admits.

The lead opinion goes on to weigh the burdens on the voter against the state's asserted interest, concluding that Indiana's law doesn't impose "substantial burdens." It's here that I have my most serious disagreement with its reasoning. Its analysis focuses on the individual voter, reasoning that, even for poor voters, there are ways of complying with Indiana's law. But this really misses the point of the law -- both from a judicial perspective and almost certainly from the perspective of those who enacted it.

It's probably true that, for most individual voters, the burdens imposed by the law are surmountable. Even a voter who doesn't drive can purchase a birth certificate and then take a bus trip to the BMV to get a photo ID card. But the focus on the individual voter overlooks the systemic impact of such a photo ID requirement. Some voters will surely go through the hoops required to get photo ID. But others won't wait in a line at the BMV to get photo ID, only to wait in another in order to vote. The legislators in Indiana who voted for the law, almost all Republicans, surely know this -- a cynic might say that the disparate burden on likely Democratic voters is exactly why they voted as they did.

In short, the focus on the individual voter misses the likely systemic impact of the law upon certain groups, in this case poor, disabled, minority, elderly, and student voters. The question upon which the Court should have focused is whether Indiana's law would have a skewing effect on the electorate, as Chris Elmendorf and I argued in our amicus brief.

Here again, it's fair to point out that plaintiffs' lawyers put together a pretty weak case. And the Court does point this out, in declining to strike down Indiana's law on its face: "[O]n the basis of the evidence in the record it is not possible to quantify either the magnitude of the burden on this narrow class of voters or the portion of the burden imposed on them that is fully justified." Ironically, though not uncommonly, the weakness of plaintiffs' evidence helps avoid some really bad law.

This leaves the door open to challenges to other election administration laws, where plaintiffs are able to put together a stronger evidentiary case than did the Crawford plaintiffs' lawyers. It's not clear from the lead opinion how much stronger that evidence would have to be to be in order to support a facial challenge -- but from a voting rights perspective, that lack of clarity is almost certainly a blessing. In fact, the Court even leaves the door open to an as-applied challenge to Indiana's law, by someone presenting stronger evidence of a burden.

While Crawford doesn't close the door on facial challenges to election laws, it does continue the Court's inching toward as-applied challenges in these cases, also evident in its recent decision in Washington State Grange. This is a troubling trend, at least when it comes to election administration cases like this one. The reason is that it will often be difficult to know the precise impact of a particular practice on individual voters before an election. It may be difficult for plaintiffs to establish standing before an election to challenge rules pertaining to ID, registration list maintenance, provisional ballots, or polling place operations, since no particular voter can say for sure that she'll be impeded from voting due to that practice.

What this means is that great weight will end up being placed on the back end of the election process, particularly on the post-election mechanisms for resolving disputes. It will put particular pressure on a state's provisional voting system, where a lot of these questions are likely to be worked out. A voter who doesn't have ID, for example, should still be allowed to cast a provisional ballot. It will be left for the parties and candidates, in the event of a close election, to fight things out afterward. We've seen before what can happen when problems are shunted to the back end of the process. Was 2000 really that long ago?

On this point, I am somewhat surprised to find myself in partial agreement with Justice Scalia. Joined by Justices Thomas and Alito, Justice Scalia rejects the focus on "individual impacts" in ascertaining the burden imposed by a law. Here, I'm with him. The main focus should be on the systemic impact of a law.

At this point, Justice Scalia and I would go in opposite directions. He would uphold virtually all laws alleged to systemically exclude certain classes of voters, at least where they don't affect a suspect class. He relegates Harper to his cryptic footnote *, which acknowledges that it "strictly scrutinized" Virginia's poll tax law. Justice Scalia then says that "we have never held that legislatures must calibrate all election laws, even those totally unrelated to money, for their impacts on poor voters ...." I'm not quite sure that I understand what this means. Perhaps Justice Scalia means to sweep Harper into the dustbin of history, or at least confine it to its facts. If he means to suggest that Indiana's law (unlike the poll tax) is really "unrelated to money" -- well, that's a pretty hard one to swallow, even on the relatively meager record in this case.

This brings me to an important lesson for voting rights lawyers who lose in lower courts: Think long and hard before seeking Supreme Court review. That's true even when you've got a poorly reasoned lower court opinion like the one Judge Posner authored for the Seventh Circuit in Crawford (see this post by Bob Bauer). In retrospect, it was a major error in judgment for plaintiffs' lawyers to seek cert in this case. This isn't just Monday-morning quarterbacking. Most of the civil rights community knew that taking this case up to the Supreme Court was a bad idea, given its present composition.

It's also clear that the Supreme Court should not have granted the cert petition in this case. The more prudent course would have been to let the issue percolate in the lower courts and, perhaps even more important, to let the empirical research on the impact of voter ID develop further before taking up the issue. Again, this isn't just Monday-morning quarterbacking. As I said in an article published last year: "Instead of granting certiorari in Crawford, the Court should have awaited a case presenting a more fully developed record that included empirical research on the harms and benefits of voter identification." Perhaps the most conservative justices on the Court (i.e., Justices Scalia, Thomas, and Alito) were hoping for a decision that would give states carte blanche on voter ID laws, but fortunately that's not what this set of opinions does. Instead, Crawford accentuates the lack of coherence in the Court's jurisprudence when it comes to election law.

But, as I say, it could have been worse.

The Problems with All-Mail Elections

2008-03-12T11:40:00.012-04:00

With the Clinton and Obama camps at odds over whether to seat Florida and Michigan delegates, the idea of holding an all-mail election has emerged as a possible solution. The New York Times reports today that Democratic Party officials are "close to completing a draft plan" for a mail-in primary in Florida that would take place in early June. Proponents of all-mail voting often cite Oregon's experience in support of their arguments. If they can do it, the argument goes, why can't we?

Given that Democratic Party rules set clear standards for having delegates recognized, which Florida and Michigan just as clearly failed to abide by, it seems obvious that the delegates selected through those states' prior primaries shouldn't be recognized. At the same time, there are reasons to be very cautious about exporting all-mail elections to these states, especially in a hotly contested and undeniably important race like this one. Here are a few of those reasons:

- Lack of experience. All-mail elections would be new to Florida. It's certainly true that some voters in Florida and other states already vote by mail, in the form of absentee ballots. But having everyone vote by mail is a major change that raises a different set of issues. In Oregon, the transition to all-mail elections was made gradually, over two decades as summarized in this timeline. Trying to implement all-mail voting on an extremely accelerated schedule would invite trouble. This is particularly true for a state like Florida, to put it mildly, doesn't exactly have a trouble-free history of election administration. With so much at stake, this isn't a great time to experiment.

- Security. The likelihood of fraud and other forms of electoral manipulation is frequently exaggerated. But to the extent foul play happens, it's most likely to occur with mail-in ballots. That's partly because the anonymity of the ballot is compromised, allowing people to buy and sell their votes in a way that's not possible with in-precinct voting, as Rick Hasen has pointed out. It's also because lots of things that can happen to a ballot between the time it's goes from election authorities to the voter and back again. Suppose some election insider has a list of "deadwood" on the rolls (i.e., people who've died or moved yet remain on the rolls) and is able to intercept those ballots before they get into the mail? Or suppose someone has a connection at the post office? This isn't to argue that these things often happen -- there's not much evidence of such fraud in Oregon, according to this report by Paul Gronke. But again, Oregon's got a long history of dealing with the problems with mail voting, and not much history of corruption. By contrast, there has been fraud with mail ballots in Florida, specifically in a Miami mayoral election in which absentee ballots were found at the home of a local political boss, as noted by Prof. Gronke (at p. 2).

- Voter mistakes. As we learned in Florida eight years ago, voters make lots of mistakes. Fortunately, the current generation of voting technology can reduce those mistakes, as I've discussed at length in this article. That includes not only electronic touchscreen voting systems, but also paper-based "notice" systems that are used at Florida's precincts. With such "notice" systems, commonly known as precinct-count optical scan, voters run their paper ballots through scanners at each polling place. Those scanners provide voters with notice and the opportunity to erroneous "overvotes" (making more choices than allowed). Such mistakes are more common than you might think, as documented in the media consortium study of ballots in Florida's 2000 election. That study found more than 40 overvotes per 1000 ballots with optical-scan paper ballots. The use of precinct-based notice technology reduced the number of errors to less than 3 per 1000. People voting by mail, of course, don't have access to notice technology and can thus be expected to cast more ballots that won't be counted. And this isn't even taking into consideration the other mistakes than can occur, like sending in the ballot late, failing to include adequate postage, not including adequate identifying information, or not signing in the right place. See this study by Mike Alvarez, Thad Hall, and Betsy Sinclair on the errors that voters make when voting by mail

- Skewing the electorate. To my mind, the most serious risk of all-mail elections is that it will distort participation to the disadvantage of certain demographic groups. Those who are most familiar with voting by mail are likely to have the highest levels of participation; others can be expected to have more trouble and thus lower levels of participation. This includes not only people who have moved or who are homeless, but also those who are illiterate or marginally literate, and therefore may have difficulty following written instructions on mail ballots. At the polling place, such people can of course rely on poll workers' assistance -- not so when they vote by mail.

Empirical research for Oregon provides some support for this concern. It's true that some studies have found a modest increase in overall turnout in Oregon, after many years of experience. But even in Oregon, that increase tends to occur disproportionately among those already most likely to participate, including those who are better educated and more affluent. As one researcher has put it, mail voting can have "perverse consequences" because it tends to "reinforce the demographic compositional bias of the electorate and may even heighten that bias." The end result could be an electorate that's even less representative of the general public than the existing one -- older, richer, and whiter.

Even if one believes that all-mail voting works well in a smaller and relatively homogeneous state like Oregon, there's reason to be very cautious about exporting it to larger, more heterogeneous states. These concerns are especially acute in states such as Florida and Michigan, parts of which are covered by Section 5 of the Voting Rights Act. That means that any change to their election rules -- including an all-mail primary election -- would have to be precleared by the U.S. Department of Justice or the U.S. District Court in Washington, D.C. If the use of all-mail voting would have a retrogressive effect, making racial minorities worse off than they were before, then the change couldn't be made.

There's a reasonable argument that preclearance should be denied, on the ground that an all-mail election will have a negative impact on the participation of minority voters. But even if preclearance is granted, mail voting could still have a disproportionate impact on participation by some groups of voters. And that, of course, would cloud the legitimacy of Florida's election -- and perhaps the selection of our next President. As Yogi Berra (or John Fogerty) might put it, it's like deja vu all over again. If there's going to be a re-vote in Florida, it should be conducted at precincts rather than by mail.

Blackwell Redux?

2008-03-11T10:40:00.004-04:00

The AP has this report on the Ohio Supreme Court's decision rejecting Ohio Secretary of State Jennifer Brunner's request that she be shielded from the deposition in a case regarding a Summit County Board of Elections member whom she refused to reappoint. This follows yesterday's story in the Columbus Dispatch, reporting on allegations that Brunner has retaliated against those who disagree with her by effecting their removal from office. The deputy chair of the Ohio Republican Party, Kevin DeWine, complains that Brunner is "injecting a culture of fear and intimidation" into county boards of elections. See here for more of DeWine's accusations.

This is a particularly significant issue to watch in Ohio, given the allegations of partisanship surrounding Brunner's Republican predecessor, Ken Blackwell. I've previously discussed concerns regarding Brunner's possible role in ousting the Matt Damschroder from his position as Franklin County's elections director. If Brunner is in fact using her power vindictively, to retaliate against local election officials who disagree with her, it can be expected to furthe erode public confidence in the state's administration of elections.

Pay Attention to Provisionals

2008-03-09T14:06:00.007-04:00

That's one piece of advice I'd give to both election officials and candidates this election season. This lesson emerged during the 2004 election, when the large number of provisional ballots cast in Ohio delayed the decision to call the state -- and thus the presidential race -- for President Bush. In Ohio's 2004 election, provisional ballots amounted to 2.8% of those cast, and an even higher percentage of the state's voters cast provisional ballots in 2006.

A large number of provisional ballots can indicate problems in a state's registration system. Also, to the extent a state relies heavily on provisional ballots, it's likely that some voters will be disenfranchised. Moreover, county-to-county discrepancies in the way provisional ballots are verified can alter the result of a close election -- and possibly lead to equal protection concerns.

To illustrate the impact of provisional ballots, I've been trying to find out the number and percentage of provisional ballots cast in Tuesday's primaries. So far, the Ohio Secretary of State's website doesn't appear to have this information. (As I mentioned Wednesday, it's important that this information be released as soon as possible.)

I have learned that a large number of provisional ballots were cast in Franklin County (Columbus area) on Tuesday. The total reported turnout was 299,688, but I'm told that there are approximately 20,000 additional provisional ballots that have yet to be verified or counted. If that's correct, it means that around 6.25% of Franklin County voters cast a provisional ballot. That's a lot.

A large number of provisional ballots could have consequences for the allocation of delegates, as I explained Thursday. Although the statewide result in Ohio's Democratic primary wasn't that close, a relatively small change within a couple of districts ould alter the delegate allocation. Take the 1st Congressional District (Cincinnati area), in which Senator Obama has 66,342 votes to Senator Clinton's 40,112. If Obama were to pick up a little over 500 votes, he'd gain a delegate and she'd lose one. What I don't yet know is how many outstanding provisional ballots there are in the two counties within CD 1 (Hamilton and Butler). If the percentage of provisional ballots is comparable to that in Franklin County, it's quite possible that the delegate allocation could change. The same goes for the counties within CD 17 (Summit, Portage, Trumball and Mahoning), where Senator Clinton could net-gain two delegates with a couple hundred more votes.

Ohio and Texas make a nice contrast, for purposes of demostrating how provisional ballots can throw election results into doubt. According to the latest information available on Green Papers' Texas page, Texas has four districts in which a relatively small shift in vote totals could change the allocation of delegates as between Senators Clinton and Obama. Senator Obama could gain delegates by picking up a votes in state senatorial districts 3, 15 and 19, while Senator Clinton could gain delegates by picking up votes in district 26. But this is less likely in Texas than in Ohio. Texas' Secretary of State reports only 9,744 provisional ballots statewide, meaning that less than 0.08% of the states voters voted provisionally. Given the small number of provisional ballots outstanding, the delegate allocation in Texas is less likely to change (though other factors, such as uncounted absentee ballots or residual votes could still alter vote totals).

It's quite possible that even more voters could be casting provisional ballots in this year's elections, as compared to 2004. That's true for at least a couple of reasons. First, the Help America Vote Act of 2002 (HAVA) required each state to have a statewide registration database in place by 2006. The idea is to make registration systems more accurate, but there probably will be -- and indeed have been -- problems with the new state registration lists that result in some voters' names wrongly being omitted from the rolls. Second, a number of states have imposed stricter ID requirements since 2004, including potential swing states like Ohio, Arizona, and Missouri (the last of whose photo ID requirement was struck down by the state supreme court). This can also be expected to cause more provisional ballots, some of which won't be counted.

As we look forward to the general election, states with large numbers of provisional ballots would be well advised to examine the reasons why. They should also take steps to ensure that provisional ballots cast by eligible voters are counted. It bears emphasis that the process for counting provisional ballots invariably involves some discretion, which means that it gives something for candidates to fight over. This isn't to say that this will happen in the primary but, looking forward to the general election, a large number of provisional ballots provides reason for concern -- both in terms of making sure that all eligible voters have their votes counted, and in terms of reducing the likelihood of post-election disputes over the result.

Uncounted Ballots & Ohio's Delegate Math

2008-03-06T12:29:00.006-05:00

Yeterday, I raised some questions regarding as-yet-uncounted ballots in Ohio -- specifically, the number of provisional, residual, and absentee ballots that aren't included in the official vote totals. In this post, I discuss how these ballots could actually affect the result of yesterday's election, despite the clear margin by which Senator Clinton won the statewide popular vote.

Of Ohio's 141 pledged delegates, 92 are to be assigned based on the vote within congressional districts, as set forth here on the Green Papers site. You can find a map of Ohio CD's here. [Update 3/7/08: Green Papers' Ohio page has now been updated to include the most recent vote totals.] Current unofficial votes totals for each district can be found here on the Ohio Secretary of State's site. CNN's current estimate is that Clinton will net 10 delegates from Ohio.

Based on the latest district-by-district vote totals, I find two congressional districts in which a realistic change from the unofficial results could affect the allocation of delegates.

- CD 1 (Cincinnati area, 4 delegates). Senator Obama has 62.32% of the votes cast for qualifying candidates (i.e., himself and Senator Clinton). If he were to get over 62.50%, he'd get 3 and she would get 1; if not, they split them 2-2. For Obama, picking up a third delegate would require an additional 512 votes.

- CD 17 (7 delegates, northeastern Ohio). Senator Clinton currently has 64.24% of the votes cast for qualifying candidates. If she can pick up enough votes to get to 64.29%, she'd get 5 to his 2; if she stays below this threshold, she gets 4 to his 3. For Clinton to pick up a fifth delegate would require an additional 242 votes.

Each of these scenarios would result in a net change of two delegates, since one candidate would be gaining a delegate and the other losing a delegate. Note that both scenarios assume that the other candidate wouldn't pick up any additional delegates, which isn't realistic; but if the candidate could pick up enough votes relative to his or her opponent, it's possible. What this means is that changes in vote totals might affect delegate allocation in Ohio, particularly when provisional ballots are counted. It's also possible that a recount of paper ballots cast in optical-scan counties could result in additional votes being counted. Could we even see litigation over one or both of these districts, say over the counting of provisional ballots or the recounting of residual votes? I doubt it, since it's probably not worth the resources it would take to litigate such an issue for a two-delegate swing, but in the current environment I suppose you never know.

Interestingly, in the district where the two candidates are running closest -- my own district, CD 15, where Senator Clinton currently has 55,070 votes to Senator Obama's 54,544 -- who "wins" is inconsequential. That's because this district has 4 allotted delegates, which they'll split down the middle.

[Note: If you come across any errors in the above post -- particularly a mathematical error, which is quite possible since my algebra is a bit rusty -- I'd be grateful for your calling them to my attention.]

Lingering Questions in Ohio

2008-03-05T17:09:00.024-05:00

You may have heard the joke about the pre-election prayer of election officials: "Please don't let it be close." Last night, the prayers of Ohio's election officials were answered ... at least for the statewide popular vote. The margin of victory in last night's Democratic presidential primary was sufficiently large that election administration problems -- like the ones I anticipated here and here -- didn't affect the overall outcome. But there are still some unanswered questions. I raise and discuss some of them below and, in a separate post tomorrow, will talk about how these could affect the allocation of delegates.

Senator Clinton's day-after margin of victory in Ohio (228,000+ votes) is much too large to have been significantly affected by election administration problems, however serious. There is a temptation to pronounce an election a "success" when there's no doubt about who won. And we should surely be grateful for the difficult and mostly thankless job that election officials do. At the same time, in this as in any other election, it's important to take a careful look at the evidence before drawing conclusions about how well the election went. That's especially true in Ohio, if we view yesterday's election as a trial run for what's likely to be a competitive and pivotal general election eight months from now.

Taking that perspective, there are some big questions as to which more information is needed, in order to evaluate how well the state's election system is working. Though my focus here is on Ohio, these questions also worth pondering with respect to other states' primaries too.

- How many provisional ballots were cast, statewide and in each county? And how many will be counted?

One of the things that has the potential to result in a post-election fight is the counting of provisional ballots. As I mentioned yesterday, Ohio relies very heavily on provisional ballots -- for voters who move, don't have proper ID, and don't appear on registration lists, among other things. These are not just a big headache for voters and election officials, who have to go through thousands of them. In a close election, we could very likely see disputes over whether and how to count provisionals.

In addition to problems anticipated in yesterday's post, I've heard anecdotes that some voters had to cast a provisional ballot because they were marked as having asked for an absentee ballot. At least some of these voters, it appears, had mistakenly responded to a mailing from the Board of Elections informing them that they could vote absentee. By returning this card -- and presumably being sent an absentee ballot -- the voters had their names marked on the registration list.

I suspect that there we have a large number of provisional ballots in Ohio that have yet to be verified and counted. It will be important to ascertain those numbers on a county-by-county basis, along with the reasons why those voters were required to cast provisional ballots rather than regular ones. And of course, it will be important to track how many of those were counted for each county, as was done in 2004.

- How many residual votes were there, statewide and in each of the counties? And were there more residual votes among those using paper ballots without error notification?

In every election, there are some voters who cast regular ballots that wind up not being counted. The term "residual votes" is used to refer to combined undervotes (a ballot that doesn't register a choice) and overvotes (a ballot that registers more than the allowed number of choices). These are particularly common when voters use voting technology that doesn't give them notice and the opportunity to correct errors. There are both paper-based and electronic systems that have the capacity to provide such notice, but voters in Cuyahoga County (Cleveland area) weren't using such technology. We should therefore expect a larger number of residual votes in Cuyahoga County. In addition, there may be more residual votes among voters in touchscreen counties who asked to vote by paper ballot, as a directive from the Secretary of State allows.

How many residual votes were there, in Cuyahoga and elsewhere? As far as I can tell, the unofficial results on the Ohio Secretary of State's website don't yet include the number of residual votes. Ideally, we'd see this information broken down by undervotes (some of which may be intentional) and overvotes (which are almost never intentional).

The evidence that I can find provides some reason for concern that some voters may not have had their votes counted, at least not yet. For example, according to the Secretary of State's website, voter turnout in Cuyahoga County was 406,450 (41.73%). But adding up this afternoon's totals from Cuyahoga County's website, I get 388,959 votes so far counted for the Republican and Democratic presidential candidates. If my math is right, that leaves a gap between turnout and counted votes of 17,491 votes, or 4.3% of total turnout.

There are of course perfectly legitimate reasons why some ballots don't show up in the vote totals. Some are probably independents who voted an "issues only" ballot with no presidential candidates. Others may be people who intentionally abstained. Still others may be voters who cast provisional ballots that have yet to be verified, and won't be for several days. (I can't tell for sure if voters casting provisional ballots are included in the county-by-county turnout figures, but would assume they are. [Update 3/7/08: I'm told that provisional ballots aren't included in the unofficial turnout numbers, at least for some counties. If that's true statewide, it means that any gaps between turnout and presidential votes counted aren't due to provisional ballots.]) Finally, it's possible that some of this gap uncounted ballots, where the voter intended to make a choice but failed to. More information is needed to draw any conclusions.

- How many people asked for absentee ballots? How many were returned? And of those returned, how many were and weren't counted?

A number of Ohio counties made a concerted effort to encourage voters to vote before election day, either through mail-in absentee ballots or through in-person early voting. Especially for those who requested mail-in absentee ballots, it would be useful to find out how many were returned to county boards of elections. I noted above the confusion among at least some voters, who inadvertently requested an absentee ballot and wound up having to cast a provisional ballots. Also, it would be useful to know how many absentee ballots were disqualified -- for example, because they failed to include appropriate identifying information or because their ballots arrived at the board of elections too late.

The Secretary of State's office should collect and release county-by-county information on provisional ballots, residual votes, and absentee ballots as expeditiously as possible. These aren't just academic points. They're vital assessing future changes, such as the proposed switch to optical-scan voting and the procedures used for voting.

There's also a more pressing reason for answering these questions: Despite the significant statewide margin yesterday, they could affect the allocation of delegates. More on this tomorrow.

[Note: If you come across any errors in the above post -- particularly a mathematical error, which is quite possible -- I'd be grateful for your calling them to my attention.]

Don't Call Ohio Too Soon

2008-03-04T08:32:00.014-05:00

That's my advice to the news media tonight, in the event of a close Democratic primary. As returns start to come in from Ohio this evening, we should keep in mind circumstances that will probably result in more outstanding ballots on Election Night than in other states, and maybe even more than is typical for Ohio. If that's true, a margin that appears insurmountable on paper -- even with all precincts nominally reported -- may actually be smaller than it appears.

Here are the big things that might cause there to be a large number of yet-to-be-counted ballots than usual on election night:

- Provisional Ballots. Ohio heavily relies on provisional ballots, which are used for people who've moved, who don't have required ID, and whose names don't appear on the registration list when they go to vote, among other things. Most of us probably remember the delay in calling the 2004 election, when Bush led Kerry by some 136,000 votes with approximately 158,000 provisional ballots left to be verified and counted. When these ballots were eventually counted, they cut Bush's margin by about 18,000 votes. In November 2006, an even higher percentage of Ohio voters cast provisional ballots, over 3%. In light of Ohio's new ID rules, still not completely familiar to many voters, and potential problems with its statewide registration list, we can expect lots of provisionals today as well. Voters have 10 days after the election to bring in their information, and it will be a while after that before we know how many of the provisionals will be counted and who they're cast for.

- Residual Votes. These are ballots that don't register a valid vote, at least when they're run through automatic tabulators. They include undervotes (a ballot that doesn't register a choice) and overvotes (a ballot that registers more than the allowed number of choices). Both can sometimes result from ambiguous marks with paper-based voting systems, but some of the undervotes may be recoverable through a manual recount. Under Ohio law, a ballot on which a voter circles the candidates name or makes a mark with an instrument that can't be recognized by tabulating equiment should eventually be counted. As I explained in Sunday's post, we can expect a significant number of residual votes in Ohio today, because a large number of voters will be voting with central-count optical scan equipment that doesn't allow voters to check for mistakes before casting their votes. Cuyahoga County will be especially hard hit, but other counties will also be affected given that voters in touchscreen counties are allowed to vote a paper ballot on request. If the race winds up being tight, it will be important to know how many residual votes there are -- especially in the Cleveland area.

- Absentee Ballots. Even with all election-day precincts reporting, it's possible that there will still be some absentee ballots not counted. Ohio has no-excuse absentee voting, and there have been reports of large numbers of absentee ballots this year. Ohio counties were permitted to start tabulating absentee ballots on Saturday, but it's not clear how many have actually been counted across the states 88 counties.

- Lines at the Polls. Under Ohio law, voters are entitled to cast ballots if they're waiting in line at the time polls close, 7:30 pm. We've already seen some long lines before election day for early voting at county boards of elections. Although I don't expect the long lines that we saw in 2004, it's possible some voters won't actually cast their ballots -- much less have them tabulated -- for a while after polls officially close.

There are other things that could prevent us from quickly knowing the outcome in Ohio. Another possibility is delays in transporting or counting optical-scan ballots, especially in Cuyahoga County, which my colleague Ned Foley alluded to here. There could be some administrative problems in Franklin County (Columbus area), given the personnel shake-up on the eve of today's election. And we've already today got word of flooding in one Ohio county (Jefferson) that has closed polling places, which will probably result in more provisional ballots and delays in counting all the votes. See this complaint filed by the Secretary of State.

So what margin can be considered safe? That depends. The media of course have other tools at their disposal besides the official count, most notably exit polls, when they call races. But if those are within or close to the margin of error, there's reason to be careful.

If the official tally is within a few ten-thousand votes -- say less than 20,000 -- it's quite possible that the popular vote result might be affected by yet-to-be-counted provisional, residual, and absentee ballots. On the other hand, if the margin is over 100,000 with all precincts reporting, that's unlikely to change. If it's in between, we'd have to look not only at how many ballots are out there yet to be counted, but also at where those voters are to make a reliable assessment of who "won" the popular vote.

This isn't even taking into consideration the difficulties that may arise in determining how many delegates each candidate got. As in other states, most of the state's pledged delegates are apportioned by congressional district. This means that the candidate who got more popular votes may not be the real "winner" in the all-important race for delegates, something that will be difficult to gauge with precision tonight. Given how many of the state's votes won't be counted tonight, it may take a while to get a clear handle on how many of Ohio's delegates each candidate will get -- and perhaps on who really "won" the election.

Ohio's Primary: What Will Go Wrong?

2008-03-02T18:19:00.009-05:00

With polls showing the Ohio Democratic primary race neck and neck, and with victory in the state likely pivotal to the survival of the Clinton campaign, this question of what might go wrong in Ohio's election is again on many people's minds. Election administration was of course the subject of much discussion here in 2004. As I described in this article, there was litigation in Ohio on a number of subjects including voting machines, provisional ballots, registration rules, voter ID, challenges to eligibility, lines at the polls, recounts, and contests.

Although we've now got a new new Secretary of State -- Democrat Jennifer Brunner, who replaced Republican Ken Blackwell -- election administration remains fraught with controversy in the Buckeye State. In this post, I highlight the current state of play in Ohio, including what we can expect for Tuesday's primary.

Though not quite the lighting rod that her predecessor was, Secretary of State Brunner has endured plenty of criticism, as the Cleveland Plain Dealer's blog notes here. Much of it arose from the Secretary of State's EVEREST report which recommended a statewide move to central-count optical scan. See here for my previous thoughts on this report.

The main problem with central-count optical scan systems is that they don't provide voters with notice and the opportunity to correct errors, therefore resulting in more uncounted ballots than other available technology. That's because central-count optical scan systems don't allow voters to check for "overvotes" (inadvertently marking more choices than is allowed for a particular contest), something that was a problem in Florida counties using this equipment in 2000. In sum, central-count ballots result in more voter mistakes, which in turn result in more ballots not registering a valid vote, commonly known as "residual votes."

Fortunately, it looks like a statewide transition to central-count optical scan ballots won't happen anytime soon. The Ohio legislature has enacted and the Governor now signed a bill (SB 286) that prohibits central tabulation for those counties using optical-scan systems. But that provision has a sort of grandfather clause for boards of election that voted prior to February 1, 2008 to use central-count tabulation at the March 4 primary. ORC 3505.25(B).

That clause allows Ohio's largest county, Cuyahoga (Cleveland area), to go ahead with its decision to use a central-count optical scan system in the coming election. This decision was prompted by problems with its Diebold touchscreen system in prior elections. Last month, a federal district court rejected the ACLU's challenge to Cuyahoga County's decision to use non-notice optical scan equipment in this election. (Disclosure: I'm on the ACLU of Ohio's board and consulted with plaintiffs' counsel on this case.)

The upshot is that we can expect Cuyahoga County to have an unusually high number of residual votes, especially overvotes, in Tuesday's primary. This is especially true given another new provision in the just-signed SB 286, which provides that:

If automatic tabulating equipment detects that more marks were made on an optical scan ballot for a particular office, question, or issue than the number of selections that a voter is allowed by law to make for that office, question, or issue, the voter's ballot shall be invalidated for that office, question, or issue.

If I understand this provision correctly, this means that a ballot with marks in two places on the presidential line won't be counted, even if the voter's choice could be deciphered in a hand recount. Suppose, for example, that a voter marked the oval by Barack Obama's name, then had a change of heart and crossed it out, to then place a mark by Hillary Clinton's name writing in the margin "I wish to vote for Clinton." As I read the new statute, that ballot wouldn't be counted, despite the seeming clarity of the voter's notation. Smudges and stray marks could also result in voters being disenfranchised, as Joe Hall observes here.

Voters using central-count optical scan systems can thus be expected to cast a large number of ballots for which no vote can be counted. How many will do this? My rough estimate, based on past experience, is that we can expect at least 1% of central-count optical scan ballots to have such problems. And these problems won't be confined to Cuyahoga County. As a result of a directive issued by Secretary of State Brunner, and thus far upheld by state courts, voters in counties using touchscreen voting machines are entitled to ask for a paper ballot. Voters using those ballots won't have access to precinct-count technology that would allow them to check for errors such as overvotes. Because most voters are likely to be using touchscreen machines, we can expect a smaller number of residual votes in those counties. Still, there will be enough that it could affect the result in a close election.

Will it be enough to affect the outcome? It's not probable, but certainly possible. That's especially true given that 92 of Ohio's 141 pledged Democratic delegates are selected on district basis. See here for a list of how many delegates each of Ohio's congressional districts has. It's quite possible that we could have enough residual votes to make a difference in one or more districts, even if the statewide race isn't within the margin of error. I've previously noted the possibility on such "small-ball" disputes here, and Cuyahoga County is among the places in which they're most likely, given its use of central-count optical scan ballots and the high number of residual votes that can be expected as a result.

Another County watch is Franklin (Columbus area), in which the Director of the Board of Elections, Matt Damschroder, has been replaced just two days before the primary. The Columbus Dispatch has this story, reporting that the board has replaced Damschroder, a Republican, with Democrat Dennis White. Although Damschroder will reportedly remain on as a consultant through the end of the year, the last-minute decision to replace him cannot help but arouse concern about the administration of elections in the state's largest city.

It appears that Secretary of State Brunner is behind the decision to oust Damschroder. Damschroder has been a leader among county election officials in criticizing some of Brunner's decisions, and was subpoenaed to testify in support of the ACLU's lawsuit challenging Cuyahoga County's decision. If in fact Brunner has moved to remove Damschroder because of his testimony in the ACLU case, as this story suggests, that would be a terrible abuse of the Secretary of State's authority -- one that would give justification to Republican complaints about her, comparable to Democrats' justifiable complaints about Blackwell in 2004.

Still another potential area of trouble in 2008, as in 2004, is Ohio's heavy reliance on provisional ballots. In fact, an even higher of Ohioans cast provisional ballots in 2006 than in 2004. As my colleague Ned Foley and Tova Wang note here, provisional ballots have the potential to create inconsistencies among counties and uncertainty about the result. This risk is heightened by Ohio's confusing voter ID requirement and difficulties with its state registration database, which are probable contributing causes to the large number of provisional ballots in the state.

The bottom line: The more things change, the more they stay the same. Despite all the changes that Ohio has seen since 2004, we can still expect lots of residual votes and lots of provisional ballots in Tuesday's primary. These could very well make a difference in a close race ... which means that we may not know who got the most votes or, more importantly, the most delegates until well after Election Day. We can also expect lots of questions to be raised about the manner in which Ohio's elections are being run, including partisanship on the part of the state's chief election official. That's especially true if the race is tight enough to shine a spotlight on the persistent problems in the administration of the state's elections.

Super Tuesday & Possible Delegate Selection Disputes

2008-02-05T11:38:00.000-05:00

This Super Tuesday is like no other, both in terms of the large number of delegates at stake but also because of the national attention that today's national primary is receiving. From here on in, the focus is likely to shift (as it should) from who "won" each state's popular vote to how many delegates each candidate picked up.

Like many others, I've been trying to understand the parties' complicated delegate selection rules, particularly on the Democratic side, and to anticipate legal issues that might emerge. Here's my attempt to pull together sources I've found on the delegate selection rules and to speculate about the possibilities for post-election legal disputes.

Today's Wall Street Journal has this report on the delegate selection rules of both major parties. While several Republican states have winner-take-all primaries, the Democratic Party allocates most of its delegates proportionally by district, as displayed in this graphic from the New York Times. The Green Papers site has both the Republican and Democratic delegate vote allocations for each state, along with this handy glossary of terms and definitions. You can find the Democrat's Delegation Selection Rules here.

Over at the Atlantic, Marc Ambinder has this post on how the Democrats' district-based allocation creates an incentive to focus on those congressional districts with an odd number of delegates. Those are the ones in which a candidate can net-gain a delegate by getting one more vote than his or her opponent. See also this post on Blue Oregon, which relies in part on stories from Time, the AP, and the SF Chronicle.

As I write this post, it's looking increasingly likely that there won't be a runaway winner in terms of delegates, at least on the Democratic side. This opens up the possibility of post-election disputes, mentioned in a post earlier today by my colleague Steve Huefner and this one from Rick Hasen on Slate.com.

We can divide potential disputes into two categories. The first type might be termed "home run" cases, in which a candidate challenges the party's rules for allocating delegates. We might imagine, for example, a "one person, one vote" challenge to a state Republican party's rules giving disproportionate weight to districts with small numbers of Republican voters. For reasons that Rick explains, it's unlikely that such a challenge would succeed in court, since these types of disputes are generally left to the parties to resolve internally. We might see floor fights, but probably not lawsuits.

The other type of dispute we might envision is a "small-ball" case of the sort that Steve envisions. Rather than challenging the party's rules, the dispute would be over the vote totals in a particular district. Such a district-level dispute is possible in districts where a few more votes would yield an additional delegate. For example, suppose that in a California congressional district with five delegates, Clinton has just a few more votes than Obama; each candidate would get two delegates, and there could be a fight over the third. Or in a district with four delegates, Obama has just over 62.5% of the vote (the dividing line for a 3-1 split instead of a 2-2 split). These would still seem to be cases in which a candidate might ask for a recount, and perhaps even litigate the issue if it's close enough.

Will be see district-level lawsuits over delegates? At this point, I doubt it since I don't think the benefits of such litigation are likely to outweigh the costs. But in a race this close, it can't be ruled out.

Electoral College Symposium

2008-02-01T14:36:00.000-05:00

The Michigan Law Review has published this online symposium on "Recent Proposals for Electoral College Reform." It includes my contribution, entitled "An Unsafe Harbor: Recounts, Contests and the Electoral College," which addresses the interaction between federal laws structuring the Electoral College process and state laws regarding post-proceedings. From the introduction:

Although recent proposals for modifying the Electoral College process have focused mainly on how electoral votes are assigned, another problem with the current system has received less attention: the timetable for resolving post-election disputes over electors. Under 3 U.S.C. 5, the so-called "safe harbor" provision of federal law, a state can be assured of having its chosen slate of electors recognized only if post-election disputes are resolved within thirty-five days of Election Day. As a practical matter, this provision doesn't provide states enough time to complete recount and contest proceedings in the event of a close, contested election....

This Commentary addresses the tension between the federally prescribed Electoral College dates and state procedures for resolving close elections. I first discuss the federal timetable for selecting electors and counting their votes. I then move to a discussion of the difficulties in fitting state post-election proceedings into the federal timetable. Finally, I propose changes to federal law designed to give states more time to resolve post-election disputes.

The symposium also includes commentaries from Thomas Hiltachk, Sam Hirsch, John Mark Hansen, Ethan Leib & Eli Mark, Alexander Belenky, and Daniel Rathburn. All are mercifully concise and well worth reading. A PDF of the entire symposium is available here.

The Crawford Argument

2008-01-08T21:41:00.000-05:00

While most of the country's eyes are on the New Hampshire primary results, the Supreme Court will on Wednesday hear arguments in Crawford v. Marion County, involving the constitutionality of Indiana's law requiring government-issued photo identification for in-person voting. I've recorded this podcast describing the case. For more background, see this New York Times story and this wonderful summary of amicus briefs in the case from Mike Pitts. If you're interested in reading all 39 amicus briefs and those of the parties, see this page on the Brennan Center site.

Conflict in Cuyahoga

2007-12-21T09:46:00.001-05:00

The Cuyahoga County Board of Elections yesterday deadlocked 2-2, along party lines, on whether to purchase new voting technology for the 2008 elections. The Cleveland Plain Dealer has this report. This decision follows Monday's meeting which I blogged here, and last week's report from the office of Secretary of State Jennifer Brunner discussed here. The two Republicans on the board opposed switching equipment, while the two Democrats supported it.

This leaves it up to Secretary of State Brunner to break the tie, and there's not much doubt that she'll support the Democratic board members' vote to replace the county's existing touchscreen system with an optical-scan system. This is a mistake, in my view, especially so close to the election. It appear that Cuyahoga County will move to optical-scan equipment marketed by ES&S, despite the problems with those devices revealed in a study this week from the Colorado secretary of state's office.

What's not entirely clear from the reports so far is whether "notice" technology will be provided to all voters, allowing them to check for overvotes. This is critical, particularly since voters in Ohio's other counties all use systems giving them notice and the opportunity to correct errors. Without notice technology, Cuyahoga County will have a serious legal problem on its hands, given the propensity of non-notice systems to result in large numbers of uncounted votes -- especially in minority and low-income communities.

Also noteworthy is the partisan sniping between Secretary Brunner and Rob Frost, a Republican member of the Cuyahoga Board. Mr. Frost complained (legitimately so, in my view) about Secretary of State Brunner's failure to attend or send a representative to Thursday's meeting. Secretary Brunner fired back by accusing Mr. Frost of "confirm[ing] his critics' worst fears." Is this a harbinger of what's to come in the 2008 election season?

A Hobson's Choice in Cuyahoga County?

2007-12-18T13:01:00.001-05:00

The Cuyahoga County Board of Elections held a hearing yesterday, to consider whether to dump its touchscreen voting system in 2008. The Cleveland Plain Dealer has this report and the Columbus Dispatch this one. This follows Friday's EVEREST report from Secretary of State Brunner's office, recommending that Ohio's existing voting technology be replaced with central-count optical scan equipment.

During yesterday's board of elections meeting, I presented this statement, arguing that against a precipitous shift to new voting technology -- especially to a central-count system that don't provide voters with notice and the opportunity to correct errors -- before the 2008 elections. As I explained in Saturday's post, I don't think Brunner's recommendation that Ohio replace all of its technology by 2008 is supported by the evidence set forth in the EVEREST report.

A more careful review of the EVEREST supporting materials raises further questions about its recommendation for a hasty transition to new technology. Among those materials is this study from SysTest Labs, which contracted with the Secretary of State's office. Among its findings are that "the greatest risks to the voting process and the integrity of elections are not created by voting technology but rather by management practices, operational constraints, inadequate funding and resources, regulatory frameworks as well as less than helpful/useful Vendor documentation" (p. 3, emphasis added). Further down on the same page, it says:

True security is a combination of technology related security techniques and security measures found in thoughtful, well documented policies, procedures, and processes for internal controls that are reflective of both a specific locality and a specific voting system.

I couldn't agree more. EVEREST's recommendation for a statewide transition to new technology is especially puzzling, in light of these findings from the Secretary of State's own consultant. It confirms that there's a wide gulf between EVEREST's findings, which warrant careful attention, and its recommendations, which aren't supported by the evidence and have a seat-of-the-pants feel to them. (For more criticism of EVEREST's recommendations, see this statement from Larry Norden of the Brennan Center. )

That said, there can be no question that Cuyahoga County has especially serious problems with its touchscreen voting system, sold by Diebold (now Premier) Election Systems. As the Columbus Dispatch notes in this editorial, there were exceptionally serious problems in Cleveland's 2006 elections. Not the least of these problems is the toilet-paper-roll-style paper trail that's used to satisfy Ohio's "voter verified paper audit trail" (VVPAT) requirement. As noted here, 20% of these paper records were unreadable in the 2006 election. This is especially troubling given that Ohio law makes the VVPAT the official ballot of record, as I've explained in this comment.

All of this leaves Cuyahoga County in an unenviable position. While I still question whether an expedited transition to new technology is the right thing to do, if the Board of Elections does decide to switch equipment, it would have to start this process quickly. See this timeline that one of the vendors has proposed. And if it does switch equipment, it has to go to a system that provides all voters with notice and the opportunity to correct errors. Given that Cuyahoga County is the biggest county is what's likely to be a -- if not the -- pivotal state in the 2008 election, its actions will and should be watched closely.

EVEREST: Ohio's Voting System Report

2007-12-15T08:27:00.000-05:00

Ohio Secretary of State Jennifer Brunner yesterday released this report of Project EVEREST (Evaluation & Validation of Election-Related Equipment, Standards, & Testing). Today's Columbus Dispatch has this report. The report is billed as a "comprehensive" review of the voting systems used in Ohio, which are also used throughout the United States. Its release provides an opportunity for me to get back into the blogging saddle and consider some of the issues that are likely to arise in Ohio, which is likely to be a pivotal state in this year's presidential election as it was in the last one.

There can be no doubt of the need for thorough analysis of both voting systems and procedures, which EVEREST attempts. Secretary of State Brunner thus deserves credit for taking on this task. There are, however, some problems with the report, particularly with the recommendations it draws. Foremost among them are the elimination of the voting systems used throughout most of Ohio and the replacement of precinct-based voting with vote centers. Some of EVEREST's recommendations are worth experimenting with. But in my view, the most dramatic ones aren't justified by the evidence considered

This is a big report, so I'll focus here on three of the most significant recommendations:

1. Eliminating Precinct-Count Optical Scan and Direct Record Electronic Voting Systems. This is the most headline-grabbing recommendation in the report. Ohio currently uses two types of equipment -- precinct-count optical scan (PCOS) and direct record electronic (DRE) systems. In contrast to the systems that were mainly used in Ohio before 2006, both these systems provide voters with notice and the opportunity to correct errors. But the new technology has, of course, been the subject of security worries. Most of these have surrounded voting technology to date have involved touchscreens and other DRE systems. The EVEREST report also takes on Ohio's existing PCOS systems and finds these lacking as well in terms of security.

All of this is perfectly reasonable. The real question is what to do about it. The report's solution is to eliminate both PCOS systems and replace them with Central-Count Optical Scan (CCOS) systems. These were used in some Ohio counties before 2006. The main problem with them is that they typically don't allow voters notice and the opportunity to correct errors. The result is more uncounted votes than either DRE or PCOS systems, as documented here. It's worth remembering -- though often overlooked -- that the transition to newer technology, along with better procedures, is estimated to have saved one million votes in the 2004 election.

Moving back to an ordinary CCOS system would sacrifice these benefits, and could be expected to result in more lost votes. This could revive the equal protection claims that were made in the Stewart v. Blackwell case, which was dismissed as moot earlier this year. (Disclosure: I served as co-counsel for plaintiffs in this case.)

EVEREST's apparent answer to this (see the parenthetical in Recommendation #2 on p. 77) is to recommend that precinct-based optical scanners be made available to allow voters to check for errors before casting their ballots. Although the report's explanation isn't crystal clear, my understanding is that instead of tabulating votes, these scanners would only allow voters to check for overvotes and undervotes, if they chose to use them. One of the questions, not answered by this report, is whether voters actually would do this or even understand the reason for the scanners. To my knowledge, no such voting method has actually been used elsewhere in the country -- at any rate, I can find no examples provided anywhere in the report showing that this will be effective. Will this modified CCOS system really prevent inadvertent overvotes and undervotes? Wouldn't it be better to test such a system in a real-world election environment, before recommending that the existing system be scrapped in its entirety and replaced with an as-yet untested system?

More fundamentally, it's doubtful that moving to a central-count system would solve the security issues identified. I'll take the report at its word when it says that the lock on a precinct-based optical scanner could be picked (p. 21), allowing someone with the requisite technical know-how to tamper with software and alter results. But someone without any technical know-how could pick a lock on a ballot box to tamper with the results. With a precinct-count system, there's at least a redundant record of the votes cast -- both the ballot and the record of votes stored in the memory card. No such redundancy is present with votes placed in a ballot box, to later be counted at a central location.

The point is that any type of voting system is vulnerable to manipulation, if proper checks aren't in place. In the end, the EVEREST report doesn't make a convincing case for Ohio's scrapping its existing technology -- particularly for 2008, a timetable that would wreak havoc on local election officials and, ultimately, inure to the detriment of voters.

2. Replacing Precincts with "Vote Centers." The other major change recommended by the EVEREST report is to eliminate precinct voting as we know it, and move to "vote centers." What's good about the recommendation is that it would mean expanded early voting (which Ohio somewhat oxymoronically calls "in-person absentee voting"). What's worrisome is that it would mean the closure of existing precincts on Election Day.

EVEREST recommends a pilot program in two or three counties in March 2008. This is a worthwhile idea. The problem is that it recommends the implementation of vote centers, apparently statewide, in November 2008 provided that funding is available. Wouldn't it be better to do a serious analysis of how well vote centers work, before mandating them statewide? There are some advantages to the vote center idea, especially in places where poll worker resources are usually scarce. The downside is that voters will have to travel further from home in order to vote. What if the polling place that used to be off the nearest bus line for an elderly voter is moved to the other side of town? How will this affect voter participation? What sort of public education effort will be needed to inform voters throughout the state of the changes in where they may vote? Will we need additional public transportation to get voters to the polls?

Here again, it would be useful to look to the experience of other states before proposing such a sweeping change. I'm not arguing against experimentation with vote centers, but think it's premature to recommend the elimination of precinct-based voting throughout Ohio, especially on such a rushed schedule.

3. Expanded Mail Voting. Making the process of voting more convenient for voters is a worthy objective. In-person early voting is a good way of doing this. More problematic is expanding mail voting and especially going to all-mail elections of the type that Oregon has. The most often cited problem is that mail voting is more susceptible to manipulation than in-person voting, since one can never be sure who actually voted the ballot -- or whether he or she was paid in order to vote a particular way. Even more serious is the risk of errors with mail voting, such as mismarked ballots, the failure to return them on time, or the failure to sign in the right place. Simply put, there are lots of things that can go wrong when people vote by mail, without the benefit of a poll worker to assist them.

There are also reasons to be concerned about how all-mail elections will affect turnout. In places where it's been tried, there's evidence that it results in modest increases in turnout, particularly in local races where turnout is especially low. But those gains disproportionately occur among those groups already the most inclined to vote. The worry is that moving to all-mail elections in a state more diverse than Oregon might further skew the electorate, making it older, richer and whiter.

EVEREST doesn't recommend moving to all-mail voting for all elections, but it does recommend conducting special elections to be conducted by mail and giving counties room to expans all-mail voting. A better option for increasing voter convenience, as we recommended in our Registration to Recounts report released last week, would be to expand in-person early voting.

A final note: I mentioned at the outset that the report bills itself as being "comprehensive." But in reality, the report is heavy on the technical details on voting technology but light on the real-world consequences of the major changes it proposes. This is consistent with the initial reactions of some of the election officials, who criticized parts of an earlier draft for relying on "pure supposition and bias" and for being "over-hyped" (pp. 44-45). Missing is an examination of the experience with different election administration practices, including thorough analysis of the social science evidence that's been gathered in recent years on the accuracy of voting technology and the practical impact of different administrative practices. In sum, the report is partial rather than comprehensive and, accordingly, its recommendations should be taken with a large grain of salt.

One particularly embarrassing example is the report's discussion of the Automark, a ballot-printing device which it ultimately recommends for the use of people with disabilities. After noting the Automark "could be easily compromised" to mark ballots incorrectly, the report states that "[t]he effects of this attack ... may be minimal, as a voter is able to visually detect any errors on the ballot" before casting it (p. 26). But of course, the very reason for using the Automark is to accommodate people with disabilities, including blindness or other visual impairments. How are those voters supposed to "visually detect" errors? This exemplifies the report's failure to apply a reality check to the issues it discusses.

In the end, the success of Ohio's 2008 election is likely to hinge more on procedures and people than on technology. In the few months that remain between now and the election, Ohio and other states would do better to focus on those issues than to attempt a hasty overhaul of its voting technology.

Pew's Data Collection Proposal

2007-10-22T10:11:00.000-04:00

Among the most serious problems with the administration of American elections is the lack of reliable information. On everything from the number of people registered to the number of votes counted, the information coming from state and local election officials leaves much to be desired. In the Election Assistance Commission's 2004 survey, many states simply did not answer all the questions asked. Even where answers were filled in, many of the responses were inaccurate.

As Thad Hall and I explained in this comment, one way of dealing with this is to provide states with federal funds in exchange for better data on election administration. Better data is not only necessary for researchers to understand our election system. It's also essential to diagnose trouble spots before they emerge into full blown problems. Accordingly, accurate and complete data is a prerequisite to meaningful election reform.

The Pew Charitable Trusts has recently taken a step in the right direction, with a proposed pilot program that's being floated in Congress. Here's how the pilot program would work: A total of $10 million in competitive grants would be provided to five states, for them to improve data collection during the 2008 election season. To compete for a grant, states would have to come up with a plan to improve compliance with EAC data requests. States awarded grants would each get $2 million by mid-2008. Afterwards, the EAC would have to submit a report to Congress making recommendations on how to improve data collection in the future. The long-term goal is to improve and expand the collection of data across the country.

There's no doubt that better data is vital to both research and reform in the area of election administration, and Pew's proposal would move us closer to this goal. Congress should therefore embrace Pew's proposal.