Equal Vote
12/01/2004
"Equal Vote" Has Moved
This blog has moved to the Election Law @ Moritz site and can be found here: http://moritzlaw.osu.edu/blogs/tokaji/. Please reset your bookmarks! Archives up to November 30, 2004 will remain here, but can be accessed through the new location.
11/30/2004
"Equal Vote" Will Be Moving
Starting on December 1, the "Equal Vote" blog will be moving to a new location on the Election Law @ Moritz website, http://moritzlaw.osu.edu/blogs/tokaji/. Although I regret the longer address, the new website will provide the advantage of integrating this blog with the other election law resources available at the Moritz site. See you there.
My Findlaw Column
Findlaw.com has published this column on problems that this election revealed, which could come back to haunt us if they're not fixed. I discuss the lack of standards for counting provisional ballots, the long lines at some polling places, and the continuing use of unreliable voting equipment. In a closer election -- say a margin in Ohio that was half of what it was -- any or all of these things could've been determinative.
Conference on 2004 Election Process
The Century Foundation, Common Cause, and the Leadership Conference on Civil Rights are jointly sponsoring an event entitled "Voting in 2004: A Report to the Nation on America's Election Process." It will take place in D.C. on December 7, from 8:30 am to 4:30 pm, and will include testimony from experts on what went wrong in this year's election. The press and public are invited. Details are available here.
11/29/2004
Nevada's DRE+CPR Experiment
For several months, some critics of current electronic voting technology have called for laws mandating a contemporaneous paper record (or "CPR") of the electronic ballot, also known as the "voter verified paper audit trail." The idea is that a printer would be attached to every direct record electronic ("DRE") unit. In the most commonly suggested model, a printed replica of the electronic ballot would print out behind a glass or screen, so that the voter could see but not touch it.
Although some have misleadingly referred to the printed replica as a "receipt," the voter actually wouldn't take any printed record of the vote with him or her. This is important with the CPR/VVPAT model, in order to protect the anonymity of the ballot and to prevent vote buying schemes.
In this year's election, Nevada became the first state to experiment with an electronic voting system capable of generating a CPR on a statewide basis. See here for my previous discussion of Nevada's experiment. Previously, the County of Sacramento and a handful of jurisdictions in Connecticut have experimented with a CPR. Many advocates are election officials are watching Nevada carefully, to assess whether the CPR presents a workable and effective solution.
So far, there's been relatively little analysis of Nevada's experience with the CPR in this election. One of the big questions has been whether voters will actually check the paper record. And even if they do, there's concern that it may slow down the voting process -- a problem that became particularly acute in this election with the long lines in some places. Another isssue is what to do in the event of printer jams. Finally, there are questions about whether the CPR will actually do much to increase the integrity of electronic voting, given that a full-scale manual recount of paper ballots is very rare. I discussed these issues in a letter that I sent to the Election Assistance Commission in May, posted here. I've also discussed this debate at great length in my working paper "The Paperless Chase: Electronic Voting and Democratic Values," which is on SSRN and can be found by clicking on the link in the right margin of this blog.
While I think it would be imprudent to mandate a CPR at this point, Nevada's experience in this election warrants careful examination. The first piece of data that I've seen is a survey by Lombardo Consulting Group, a brief description of which can be found here. This survey reportedly shows that 81% of voters would prefer an actual receipt that they could take with them, to the CPR that's retained by election officials. The survey also reportedly found that only 31% of voters actually checked over the whole paper replica of the electronic ballot. This is part of my reason for hesitancy in using the term "voter verified paper audit trail," since it's not at all clear that voters will actually verify their choices -- and if they don't, it's utility as an audit trail is limited at best.
The problem with having an actual receipt that the voter would keep, as noted above, is preserving the secrecy of voter choices. If the receipt actually listed the voters' choices, it would be a recipe for vote selling and other forms of fraud. A company called VoteHere is marketing a technology (the VHTi) that can be incorporated into existing DREs. The basic idea is provide voters with a receipt while maintaining privacy. It's an interesting idea, worthy of further examination. But it's premature to leap to the conclusion that the VHTi, the CPR, or any other particular device is the magic bullet for the legitimate security concerns that have been raised about electronic voting. This is one area in which the Election Assistance Commission will have some very important work to do in coming months.
New Lawsuit on Provisional Voting in Ohio
The People for the American Way Foundation and individual voters have filed a state court action against Ohio Secretary of State Ken Blackwell and the Cuyahoga County Board of Elections. The complaint, which can be found here, challenges the way in which provisional voting was implemented in Cuyahoga County -- the state's largest county, which includes Cleveland.
The complaint alleges that election officials and poll workers were confused about the circumstances in which a voter should be given a provisional ballot rather than a regular ballot, causing them to give inconsistent and sometimes erroneous instructions to voters. It also challenges the state's failure to set "clear, uniform, and legally valid standards" for handling provisional ballots, a claim that sounds very much like the equal protection claim made in the Schering v. Blackwell complaint pending in U.S. District Court.
A provisional ballot is cast, where the voter's name doesn't appear on the rolls when he or she appears at the polling place. The provisional ballot should be counted, if the voter is determined eligible.
The complaint alleges that election officials violated state law, by relying on electronic registration records rather than the registration cards filled out by voters, in determining whether a provisional voter was really eligible. Because the poll books were prepared from these databases, the complaint argues, simply consulting the database will be of little help in ascertaining whether the voter was really registered to vote.
The complaint also alleges a violation of the Voting Rights Act, which prohibits the denial of the right to vote due to an "error or omission on any record or paper relating to any application, registration, or other act requsite to voting, if such error or omission is not material in determining whether such individual is eligible under State law to vote in such election." The argument is that the rejection of provisional ballots for trivial errors -- such as the absence of a countersignature from a poll worker -- violates the Voting Rights Act. Their complaint also includes allegations that voters' rights under the Help America Vote Act, Equal Protection Clause, and Due Process Clause have been denied.
In terms of relief, the complaint asks that election officials be barred from rejecting provisional ballots, until they've checked the paper registration records. It also asks that registration forms not be rejected where a signature is lacking on the provisional voting envelope, so long as the voter signed elsewhere. It also asks that voters be provided with written notice, if the county plans to reject their provisional ballots.
New Critique of Hout Paper on Electronic Voting
B.D. McCullough of Drexel University and Florenz Plassmann of SUNY Binghamton have issued a stinging rebuttal of the paper by Michael Hout and other researchers at Berkeley, which purported to find that the use of electronic added 130,000 - 260,000 "excess votes" for Bush. The McCullough/Plassmann critique can be found here and the original paper by Hout and his colleagues here.
McCullogh & Plassmann assert that the Hout study "is entirely without merit and its results are meaningless." They find that Hout and his colleagues made a basic error in their model, in attempting to quantify the number of "excess votes" for Bush. This can be shown, McCullogh and Plassmann explain, by relying on the same data and model to reach the opposite conclusion -- namely that electronic voting favored Kerry.
They also note Hout and company's failure to include other variables that might have caused higher support for Bush in certain counties that used electronic voting, a problem I've previously noted here and here, citing other criticisms of the Hout study that have appeared in the blogosphere. For example, McCullogh and Plassmann note, the large turnout for Bush might be due to evangelical Christians who turned out in larger numbers this time around, or to Jewish voters who switched their votes to Bush.
In sum, the authors find that:
- because the Hout paper's concept of "excess votes" is meaningless, it provides no evidence of irregularities in counties using electronic voting,
- the paper wasn't really modelling increases in support for Bush from 2000 to 2004, but only support of Bush in 2004,
- there's no credible basis for asserting that Bush received "excess votes," such as the 72,000 Hout and company allegedly found in Broward, and
- while the Hout paper claims 99.9% confidence that its results weren't the product of chance, McCullogh and Plassmann are "99.999% sure" that the Hout results don't show any problem with electronic voting.
On these bases, they conclude that the Hout working paper "has been discredited."
My take: When the Hout working paper first came out, I urged skepticism. The most recent analyses, including McCullough & Plassmann's as well as the ones noted here suggest that such skepticism was justified. While I'm interested in seeing how Hout & his associates respond to the criticisms of their work, it seems pretty clear at this point that their data don't support the conclusion that electronic voting caused an increase in the number of votes for Bush.
11/28/2004
Counting Provisional Ballots in Ohio
In Ohio, approximately 78% of the 155,000 provisional ballots cast have been validated and will therefore be counted, according to this report. In Hamilton County (Cincinnati area), about 80% of provisional ballots were counted. A lower percentage were validated in Ohio's largest county, Cuyahoga, where Cleveland is located. According to this report, only 16,373 of Cuyahoga County's 24,472 provisional ballots (68%) were actually counted. Of those not counted, about 70% were disqualifed because the county could find no registration record on file. As voting rights activist Norm Robbins states, it's difficult to imagine that so many people would take the trouble to go out and vote -- many of them braving lengthy lines -- if they really weren't registered. The L.A. Times reports here that People for the American Way has filed a federal lawsuit challenging the manner in which Cuyahoga County has determined whether provisional ballots should count.
11/23/2004
Preliminary Relief Denied in Ohio Recount Case
U.S. District Judge James Carr has denied a preliminary injunction and TRO, in a case seeking to compel Ohio election officials to complete a recount by the December 7 "safe harbor" date. His order can be found here on Rick Hasen's blog, and the moving papers in the case (Rios v. Blackwell) are here on the Election Law @ Moritz website.
The plaintiffs in the case include the Libertarian and Green Party candidates for President. They claim that the failure to conduct a recount by the "safe harbor" date -- discussed in the post immediately below -- violates their constitutional and statutory rights. Judge Carr doesn't reach the merits of their claims, instead resting on the fact that there's no possibility of the plaintiffs being harmed, given that they can't possibly prevail in the election.
My take: I think Judge Carr was correct to deny preliminary relief. He was also right to avoid the merits, at least at this stage, given the tricky issues that it raises. But this leaves a serious problem unresolved.
As I mentioned yesterday, Ohio's recount and contest scheme makes it difficult if not impossible for the state to meet the December 7 "safe harbor" date, by which contests over presidential electors must be resolved in the state is to ensure its chosen electors will be the ones counted in Congress. In fact, Ohio's scheme makes it difficult for the recount and contest process to be completed by the date on which the Electoral College meets (December 13). Perhaps some of the blame lies with Secretary Blackwell for not proceeding more quickly. But the biggest problem, it seems to me, lies in Ohio's statutory scheme.
It seems clear that the statute needs to be fixed. Whether the courts have a role to play in the process is less clear.
As an initial matter, it's hard to see how the Plaintiffs can plausibly claim a violation of any right under Art. I, Sec. 2 of the Constitution, providing that electors shall be appointed in such manner as the Legislature may direct. The only real argument is that Blackwell should declare the election by December 3, the date on which the Secretary of State is by statute supposed to send notice to the chosen electors. (He's said that he may not do it before December 6.) But this doesn't really solve the problem, since there still wouldn't be enough time for a recount and contest. Nor can Plaintiffs plausibly claim that the federal safe harbor law (3 U.S.C. 5) has been violated. It's up to states to determine whether they wish to take advantage of this safe harbor, and I think it's doubtful that this statute confers any privately enforceable rights.
That leaves Plaintiffs' claims under the First and Fourteenth Amendment. Maybe there's a due process argument, arising from the fact that the State of Ohio has prescribed a recount and contest procedure that can't realistically be completed in time for it to be meaningful. While this claim isn't very clearly fleshed out in Plaintiffs' moving papers, and it might be declared moot after the electoral college meetts, it would be nice to see this issue litigated now -- rather than in the context of a truly disputed election four, eight, or twelve years hence. Capable of repetition yet evading review?
11/22/2004
Ohio: It Ain't Over 'Til It's Over
Taking Yogi Berra's words to heart, Ohio voting rights advocates have announced their plan to proceed with a contest of the election. This is in addition to the recount planned by the Green and Libertarian candidates, now reportedly to be joined by the Democratic Party as well. And to top things off, a new federal lawsuit was filed today that reportedly seeks to make sure that the recount is completed before the electoral college meets on December 13. Thus, although the result in Ohio is clear (news flash: Bush won), it's not quite as clear when the election will officially be over. Nor is it entirely clear -- at least to me -- how the process will actually work. But I'll try to sort out what's most likely.
The Contest
The Ohio News Network has this report and the Cleveland Plain-Dealer this one on the plan to file a contest of the election results. Activists reportedly plan to cite exit polls, provisional voting glitches, and "hundreds of voter horror stories" in support of their petition. Under Ohio law, election contests are to be heard by the Chief Justice of the Ohio Supreme Court. Ohio R.C. 3515.08. (See here for the Election Law @ Moritz description by my colleague Chris Fairman.)
A contest requires a petition, signed by at least 25 voters and filed within 15 days of the date the election results are “ascertained and announced by the proper authority” -- or in the event of a recount, within 10 days after the recount results have been ascertained and announced. R.C. 3515.09. A hearing must then be scheduled between 15 and 30 days of the petition.
As of last report, the lawyers planning the contest couldn't say when they'd file their petition.
The Recount
But wait ... it gets even more complicated when we get to the recount provisions of Ohio law. As noted here, the Green and Libertarian candidates have announced their intention to seek a recount. Earlier this evening, Keith Olbermann reported here that the Democrats plan to join the recount, and the A.P. later issued this report confirming it. (Update: K.O. is now reporting here that the Dems now say they plan to "participate in" the recount, rather than "join" it ... whatever that means.)
The recount request must be filed within five days of when the election results are officially declared. Ohio R.C. 3515.02 ("the[recount] application shall be filed within five days after the day upon which the secretary of state declares the results of such election") . That hasn't happened yet, and may not for several days, as described below.
Apparently concerned that the electoral college would meet before the recount is complete, the Greens and Libertarians (which K.O. glibly labels the "Glibs") filed a federal lawsuit today in Toledo. That lawsuit reportedly seeks to compel the recount in sufficient time for it to be completed by December 13, the day the electoral college meets in each state's capital. See here for a press release announcing the lawsuit.
Declaring the Results and the "Safe Harbor"
Under Ohio law, as I read it, county boards of election may only "declare" election results after the they've completed their canvass, which was to begin between November 13 and 17. Ohio R.C. 3505.33. According to the Dayton Daily News, the counties have until December 1 to complete their count and the Secretary of State has until December 6 to declare the statewide results.
Ohio law also requires that, for elections in which persons are elected to office, "certificates of election" be issued after the time for requesting a recount has expired and any recounts are completed. Ohio R.C. 3505.38. (I'm not completely sure whether this last provision applies to the presidential election, but I think it does.) In any event, ten days prior to the meeting of the electoral college, the Secretary of State is supposed to send a letter to "qualified electors"notifying them of the time and place of the meeting of the state's presidential electors. Ohio R.C. 3505.39.
This year, Secretary of State Blackwell may not actually certify the election until December 6 according to the A.P. It's that event that starts the five-day clock for requesting a recount ticking.
By federal law, the "safe harbor" date is one day later, December 7. That's the date by which disputes regarding electors must be resolved, in order to make sure that those electors are the ones whose votes are actually counted by Congress. The "safe harbor" date falls six days before the date on which the electoral college meets, which this year is December 13. (See here for Peter Shane's description of how the electoral college works.)
Confused Yet?
Well you're not the only one. Here's the problem: It doesn't seem like there's enough time for the recount and contest process to be completed by the time Ohio's electors are required to meet in Columbus, and certainly not by the "safe harbor" date.
First of all, it's not clear when Ohio's results will officially be "declared" thus starting the five-day clock for requesting a recount. The election calendar posted on the Secretary of State's website doesn't provide a date. According to this Knight Ridder report, the Secretary of State's office is saying that a recount might not even be able to begin before December 11, two days before the electoral college meets -- and obviously too late for a recount to take place before then.
But let's suppose the best of all worlds, in which Blackwell were to declare the results tomorrow. Under Ohio law, there would be five days from that declaration for the candidates to ask for a recount. Say that a recount is requested on November 26 (four days from tomorrow). And let's further assume, quite optimistically, that it takes one week to complete the recount. That would mean the recount ends on December 3. There would still be 10 days from that date to file a contest. Thus, a contest petition could be filed as late as December 13, and a hearing on the contest can't take place until at least 15 days after that. Of course, December 13 is the day the electoral college is to meet and is six days after the "safe harbor" date.
If that's all correct, it seems to me that Ohio can't possibly comply with the December 7 safe harbor deadline under it's existing statutes. Let's take a look at the safe harbor statute (3 U.S.C. 5):
If any State shall have provided, by laws enacted prior to the day fixed for the appointment of the electors, for its final determination of any controversy or contest concerning the appointment of all or any of the electors of such State, by judicial or other methods or procedures, and such determination shall have been made at least six days before the time fixed for the meeting of the electors, such determination made pursuant to such law so existing on said day, and made at least six days prior to said time of meeting of the electors, shall be conclusive, and shall govern in the counting of the electoral votes as provided in the Constitution, and as hereinafter regulated, so far as the ascertainment of the electors appointed by such State is concerned.Well, unless I'm wrong (and I may well be), Ohio hasn't provided by law for a "final determination of any controversy or contest" that will have been made by December 7. As I understand it, Congress isn't required to accept the state's slate of electors, if this provision isn't satisfied.
Of course, all of this is a moot point since both houses of Congress are controlled by the Republicans. In fact, given that Ohio's Governor and Secretary of State are both Republicans, it would be enough for one of the houses of Congress to be controlled by Republicans. As noted above, the counting of electoral votes takes place before a joint session of Congress on January 6. Under 3 U.S.C. 15 "the votes of the electors whose appointment shall have been certified by the executive of the State, under the seal thereof, shall be counted," unless there are competing slates of electors and "the two Houses, acting separately, shall concurrently decide such votes not to be the lawful votes of the legally appointed electors of such State." In Ohio, the electors certified by the Governor and Secretary of State will undoubtedly be Bush's -- regardless of whether the December 7 safe harbor date is met. That's true even if there isn't time to comply with Ohio's contest and recount procedures by December 13.
The Constitutional Crisis That Won't Happen
All this could mean a real constitutional crisis if (1) the chief executive of Ohio weren't a Republican and at least one of the houses of Congress were controlled by the Democrats, or (2) both houses of Congress were controlled by the Democrats. But neither of those things are the case. So what's most likely to happen is that:
- Secretary of State Blackwell will certify the election results on or around December6 and, with that certification, Blackwell and Taft will issue a "certificate of ascertainment" identifying the electors by name, along with the number of votes that each candidates' electors received (see here for Ohio's 2000 certificate of ascertainment) ,
- there won't be enough time to complete either a recount or a contest before December 13, the date that Ohio's electors meet and vote; so technically, Ohio won't have complied with the federal safe harbor law, since there won't be a "final determination" of either the recount or the contest provided for by Ohio law; but it won't matter because,
- the certificate of vote , signed by the Republican electors and Blackwell, will be transmitted to Congress (See here for Ohio's 2000 certificate of vote), and
- both houses of Congress will recognize Ohio's certified electors, despite the state's failure to comply with 3 U.S.C. 5 -- and there aren't likely to be any objections to Ohio's designated electors, given that an objection requires a signature of both a Senator and a member of the House.
The bottom line: Bush still wins. This is more confusing than a Yogi Berra sentence, but it appears that Ohio doesn't have recount and contest laws that allow for a final determination by December 7 (or even December 13). If that's right -- and I'm willing to be corrected if someone out there thinks that I've misstated the law -- it would be wise for the state (and perhaps others) to take another look at it's recount and contest statutes, lest they provide the fuel for a full-blown crisis to explode in some future election
Florida: Better Technology = Fewer Lost Votes
The replacement of punch cards and central-count optical scan voting equipment in Florida appears to have dramatically lowered the number of uncounted votes. The Miami Herald has this report on the reduction in residual votes (i.e., overvotes plus undervotes) in this year's election, arising from the switch to electronic and precinct-count optical scan systems, both of which prevent ballots from being rejected due to mismarking. (Thanks to Tova Wang for the pointer.)
The story starts by identifying a precinct in Miami-Dade County in which 113 of 868 punch card ballots showed no vote for President in 2000; this year, all of the ballots counted in that precinct showed a vote for President. While this is an extreme example, other places that switched from punch cards to electronic voting also saw a reduction in residual votes. Overall, the number of residual votes in Miami-Dade -- which now uses electronic touchscreen machines -- went from 28,000 to 4,227.
In Florida's 2004 election, the residual vote rate with electronic voting machines was about 0.5%, and with precinct-count optical scans about 0.25%. That compares to residual vote rates of 3.9% with punch cards and a whopping 5.7% with central-count optical scans four years ago. It's likely that most of the remaining residual votes are due to intentional undervoting -- i.e., voters deliberately choosing to cast no vote for president. Martha Kropf and Stephen Knack have found that about 0.3 to 0.7% of voters intentionally undervoted in past presidential elections.
The bottom line: This provides further support for what the ACLU and social scientists like Henry Brady (see here) have been asserting for some time -- namely that moving from punch cards and central-count optical scan systems to electronic and precinct-count optical scans can considerably reduce the number of lost votes. And punch cards have an especially negative impact on racial minorities, as Brady's colleagues Justin Buchler, Matthew Jarvis, and John McNulty explain here. Let's hope that election officials in Ohio, Missouri and the handful of other states that continue to use punch cards are paying attention.
The Election That Won't End
Like the energizer bunny, the Washington gubenatorial election keeps on going, now 20 days past election day. The Christian Science Monitor has this report, in which the Republican candidate Dino Rossi leads Democrat Christine Gregoire by 261 votes, with over 2.7 million votes cast. That triggers an automatic machine recount.
More Thoughts on the Berkeley/Hout Study
In this post last week, I noted my skepticism of the report by Michael Hout and his colleagues at U.C. Berkeley finding that Florida's electronic voting machines resulted in a 130,000 vote or more gain for Bush in 2004. Turns out I'm not the only skeptic. Mark Blumenthal's "Mystery Pollster" blog has this series of posts, some of which raise other questions about the Hout study. Patrick Ruffin finds here that, contrary to what the study suggests, Bush actually increased his advantage by more in optical scan counties than in electronic onces (2.54% to 2.25% respectively). Wisconsin political scientist Ken Mayer offers this view, posted on Rick Hasen's "Election Law" blog. And the "MyElectionAnalysis" blog has this post, suggesting that the variables that Hout and his colleagues used (Bush in 2000, Dole in 1996, county size, median income, hispanic population) are highly correlated with each other:
Why does this matter? Because independent variables have to be independent of each other for regression to work. If you use two correlated variables, they essentially resonate with each other, making it seems as if they explain a much greater amount of the variance than they actually do.My take: Some additional thoughts, provoked by the above and by my re-reading of the Hout study. The study asserts that, "net of other effects, electronic voting had the greatest positive effect on change in percent voting for Bush from 2000 to 2004 in democratic counties." A question worth asking is how this could possibly occur? The study notes that the greatest gains for Bush werein Broward, Miami-Dade and Palm Beach counties. Assuming this to be true, can it be that the chief election official in these primarily Democratic counties was in league with the Republicans? In Palm Beach County this would mean that Theresa LePore (a Democrat) was conspiring to promote Kerry's defeat. Of course, the argument might go, it could be that the voting machine companies pulled the wool over the eyes of the county election officials. The problem with that argument is that these counties use different vendors. For example, Broward and Miami-Dade us iVotronic according to the Division of Elections, Florida Secretary of State while Palm Beach uses a Sequoia system. Were both vendors involved in this scheme? Is that really likely?
In short, there seem to be at least two problems with the Hout study. First, even taking the findings at face value, it may be that there are other factors that caused the apparent relationship between the type of equipment used in the county and the increase in the Bush vote. Second, even if the statistical analysis is airtight, thought must be given to how the use of electronic voting equipment could cause more votes to be cast for Bush -- and in this case, that means seriously asking how likely it is that a conspiracy involving multiple voting companies and registrars (some of them Democrats) could occur.
Update: For more questions regarding the Hout study, see Crooked Timber and Andrew Gellman. Crooked Timber concludes "it seems more likely that these results show the Republican Party Machine was really, really well-organized in Palm Beach and Broward, and they were able to mobilize their vote better than the Democrats."
11/20/2004
Provisional Ballots in Illinois
The A.P. reports here that many of the provisional ballots cast in Illinois were counted. In DuPage County, only about one in four (962 of the 3,672) provisional ballots were counted. Cook County did somewhat better, with almost half (5,424 of 10,425) being deemed valid, but 2,414 were rejected because they were cast in the wrong precinct. Still waiting for statewide totals on the number of provisional votes counted in Illinois.
Wrong Precinct Ballots Rejected in Canton
The Stark County Board of Elections has rejected over 500 provisional ballots in Stark County, according to this story in the Canton Repository. Some of them were reportedly cast by voters who appeared at the right polling place, but went to the wrong table and therefore voted in the wrong precinct:
Janet Gross of North Canton went to the Knights of Columbus Hall to vote this year like she does every year.Another 59 provisional ballots were rejected because voters didn't sign the envelope. A total of 6032 provisionals were cast in Stark County this year, more than twice as many as four years ago. The rejected votes could affect a county commissioners race.
When she got there, the poll workers didn’t have her name listed.
They told her that she had to fill out a provisional ballot, which would be verified after the election.
Gross, who has lived on Circle Hill Road SE for 40 years, found out Friday that her vote won’t count.
Her vote was one of 505 provisional ballots rejected by the four-member Stark County Board of Elections because the vote was cast at the wrong precinct.
Gross may have been at the right polling place, but she went to the wrong precinct table. The Knights of Columbus had three precincts in the hall. Gross was supposed to vote in precinct 3D, but she voted in precinct 3C. The provisional ballot of her son, David Gross, also didn’t count because he was at the wrong precinct.
11/18/2004
Berkeley Report on Electronic Voting in Florida
Last week's conspiracy theory was that the election had been stolen in Florida counties using optical scan voting equipment, a paper-based system that functions like standardized tests. This week's may turn out to be the precise opposite: that electronic voting caused Bush to receive more votes than he should have.
As discussed here, the allegation made by some activists last week was that counties using optical scan equipment showed more votes for Bush than one would expect based on the party affiliation. By contrast, voters using electronic voting (the other type of system in Florida) had about the Kerry/Bush breakdown you'd expect, as shown in this chart. The allegation of foul play with optical scans was quickly debunked as noted here, when it was pointed out that the optical scan counties were more likely to have Dixiecrat voters (i.e., registered Democrats who vote Republican) while the electronic voting counties tend to be more urban ones (in which Democrats mostly vote for Democrats).
Now, a working paper from researchers at U.C. Berkeley asserts that electronic voting caused an increase in votes for Bush this year. In particular, the paper asserts that electronic voting resulted in 130,733 extra votes for Bush, primarily in three large counties (Miami-Dade, Palm Beach and Broward). The analysis is based on the increase in the number of votes for Bush in 2004, as compared with those he received in 2000 and those that Dole received in 2000. According to the paper, which can be found here:
Electronic voting raised Bush's advantage from the tiny edge he held in 2000 to a clearer margin of victory in 2004.... The state-wide impact of these disparities due to electronic voting amounts to 130,000 votes if we assume a "ghost vote" mechanism and twice that -- 260,000 -- if we assume that a vote misattributed to one candidate should have been accounted for the other.To be sure, the working paper doesn't go so far as to allege any foul play, much less that the election was stolen. Still, it reaches a provocative conclusion, one that electronic voting critics and those who've been alleging a conspiracy to steal this election -- and there's substantial overlap between these two groups -- are sure to seize on.
The big question is whether the conclusion the paper makes is supported by the evidence. I've previously asserted that we should be skeptical of allegations that the election was stolen with optical scan equipment. In my view, the latest allegations warrant the same skepticism. While this working paper should be examined carefully, it's not at all clear that the data upon which the paper relies warrants the bold conclusion that electronic voting caused the increased vote for Bush in 2004.
What's most noteworthy upon an initial read is that the paper only takes into consideration a limited set of factors that might have resulted in more votes for Bush this year. In addition to equipment type and Republican votes in prior elections, the report considers median income, and Latino population, overall change in voter turnout, and the size of county (measured by total votes for the two major candidates). But there are many other conceivable factors that might have caused people to vote for Bush in these counties.
For example, could it be that some more conservative Jewish voters in Miami-Dade or Palm Beach voted for Gore-Lieberman last time around, but switched to Bush in this election? And what about African-American voters, who are conspicuously absent from the paper's analysis? We know that the Republicans mounted an ambitious campaign of negative advertising directed at Kerry in this election. (See here for a story describing Kerry's response to those ads.) Could it be that some black voters in Miami-Dade or Broward county switched their votes to Bush? Finally, what about the possibility that the Republicans simply did a better job of persuading voters in these counties than they did in either 1996 or 2000? It's certainly possible that these things might have resulted in more votes for Bush in this year's election. Yet the analysis reported in the paper doesn't appear to take into consideration any of these possibilities.
To be clear, I'm not asserting that any or all of these things are responsible for the increase in Bush votes. What I am suggesting is that it's possible that these and other factors account for the larger number of Bush votes assertedly cast in the Florida counties examined. A press release announcing the working paper asserts that the likelihood of the reported "discrepancies" occurring by chance is less than 0.1% (one in a thousand). What that doesn't account for is the substantial possibility that the increase in Bush votes in Miami-Dade, Palm Beach, and Broward could be explained by factors other than the voting equipment used. Even taking at face value the findings that the paper makes, those findings don't support the conclusion that electronic voting caused the increase in votes for Bush.
The Berkeley working paper warrants more careful analysis. I'm sure that more able empiricists than I will, in short order, scrutinize their findings and conclusions. But as I've repeatedly warned, we should be wary of making accusations that the election was "stolen" before all the facts are in and have been analyzed carefully. In the end, such hasty judgments will do great harm to the vital objective of making our election system more fair and equal.
11/17/2004
Update on Washington Gubenatorial Race
As of this evening, Democratic candidate Christine Gregoire held a razor-thin lead of 28 votes over Republican Dino Rossi in the race for Washington Governor according to this A.P. report. A recount is required by state law where the margin is less than 2000 votes.
The A.P. also reports on the provisional voting case that came before King County Superior Court Judge Dean Lum. In an order issued yesterday, Judge Lum opens by stating: "We have arrived at the moment which all reasonable Washingtonians have dreaded for four years: the moment when the Court is asked to micromanage an election." The court had previously ordered King County (Seattle area) to make available to the Democrats the names of 929 provisional voters whose signatures had been questioned. The court declines to issue any further relief, including the TRO that the Republicans requested stopping King County from counting any provisional ballots brought in by third parties (i.e., someone other than the voters), finding that no evidence of fraud had been presented. More litigation seems almost certain.
Most Ohio Provisionals to Be Counted (So Far)
The A.P. reports here that approximately 80% of the state's provisional ballots so far have been deemed valid and will be counted. There does, however, appear to be some considerable variation among counties so. In Cuyahoga Couonty, one third of provisionals so far have been rejected. In contrast, in Geauga County only 77 of 669 provisional ballots (11.5%) have been rejected, according to the Cleveland Plain-Dealer. Such disparities tend to support claims being made by the Democratic Party and others that the absence of clear and specific standards violates the Equal Protection Clause, under the constitutional principle relied upon in Bush v. Gore.
11/16/2004
Dems Intervene in Ohio Provisional Voting Lawsuit
The Ohio Democratic Party has moved to intervene in the pending lawsuit, challenging the absence of clear standards for counting provisional ballots. The intervenor's complaint and motion to intervene can be found here. The original complaint in the cast (Schering v. Blackwell) was filed by a Cincinnati voter, alleging that the state's failure to set clear standards for determining which provisionals should count violates equal protection, under Bush v. Gore. The Democrats' complaint in intervention echoes these claims.
My take: This is interesting strategic gambit on the part of the Democratic Party. I tend to agree that the absence of clear standards for counting provisional ballots presents a serious equal protection problem, one that's more closely similar to the facts of Bush v. Gore than any case I've seen since 2000. On the other hand, the danger that they face is that Secretary of State Blackwell will set clearer standards, ones that may not be favorable to Democratic interests. For example, Blackwell could put the burden on voters to establish eligibility and require that provisional ballots be disallowed unless their names appeared on pre-election day registration lists -- a process that would come close to rendering provisional ballots a nullity.
The problem, in other words, is one that's been much discussed since November 2, 2004, and indeed since the 2000 election: partisan election administration. As long as states' chief election officials are partisan Democrats or Republicans, there's an enormous danger of them making rules that inure to the benefit of their party. That's true in Ohio with a Republican Secretary of State, just as it's true in California with a Democratic Secretary of State.
Of course, identifying the problem of partisan election administration is the easy part. Figuring out what to do about it is quite another. I wish I had the answer ... but I'm working on it.
Green and Libertarian Candidates to Request Ohio Recount
It's official. The Green Party and Libertarian Party presidential candidates will request a manual recount of the ballots cast in Ohio, according to this report from Bloomberg News. They've raised the $113,600 that it will cost to fund a statewide recount. The actual cost to the state will reportedly be around $1.5 million.
What exactly the Greens and Libs hope to accomplish through such a recount is difficult to say -- see here for my prior views on that subject. I understand the desire to keep the focus on the very real election problems that occurred in Ohio and elsewhere, such as provisional voting, long lines, and registration glitches. The problem is that a recount doesn't get you there.
In related news, see this exchange between Greg Palast and Salon's Farhad Manjoo on the election results in Ohio. Palast continues to argue that "Most voters in Ohio cast their ballots for John Kerry." He bases this assertion on uncounted provisionals and Ohio's use of punch cards, which disproportionately result in the denial of minority votes. Palast is right to point out the disparate impact of punch cards -- and in my view their continuing use violates both the Equal Protection Clause and the Voting Rights Act. But Manjoo is correct to point out that Palast's numbers don't add up. As I've noted here and here, the most net votes that Kerry could possibly have "lost" is around 91,000, well short of the 136,000 or so he'd need -- and that's applying the most generous assumptions (in Kerry's favor) that could remotely be considered plausible. If someone tells me my math is wrong, I'm more than happy to be corrected, but no one's done so yet, certainly not Palast.
One could perhaps argue that a disproportionate number of Democratic voters will discouraged by long lines, but that argument would depend upon an analysis of where the lines were concentrated. That's evidence we don't yet have. Finding out the answer to that question would be a much better use of $113,600.
The bottom line: Ohio's 2004 election revealed some serious problems that a recount won't begin to address and that we desperately need to explore. Oh yeah, and based on the votes that were cast in Ohio ... Bush won. (Sorry folks.)
Provisional Voting Problems
The A.P. has this story summarizing the problems that the states had in implementing provisional voting. It quotes my colleague Ned Foley as suggesting that Congress consider repealing HAVA's provisional voting requirement.
My take: I agree that there were serious problems in the way the provisional voting was implemented, particularly the lack of standards for determining which provisional votes should count. But I don't agree with my colleague Prof. Foley's view that Congress should consider getting rid of HAVA's provisional voting requirement, and certainly not at this early stage in the implementation of HAVA. The idea behind provisional voting is to provide a mechanism for votes to be counted when voter's names don't appear on registration rolls. It's a mechanism that has worked well for years in some states, including California.
In a world with perfectly accurate registration lists, we wouldn't need provisional voting. But we don't live in that world, and won't in the foreseeable future. The mandatory statewide registration databases, which will have to be implemented by 2006, may make things better in the long run. But in the short run, it's likely that it will aggravate the problems with registration lists. That makes provisional voting even more important.
What is important is that states enact clear rules, preferably through legislation, on how the provisional voting process -- especially counting -- should work. Also, as state and county election officials gain greater familiarity with how provisional voting is supposed to work, we can expect the process to improve.
11/15/2004
George Washington Law Review Conference
Blogging will probably be sporadic or nonexistent the rest of today and tomorrow. I'm in D.C. at an election law symposium hosted by the George Washington Law Review.
A Recount in Ohio?
The Green Party and Libertarian Party presidential candidates apparently plan to ask for a recount in Ohio. The A.P. has this report. In Ohio, only candidates can ask for recounts of candidate races. (For ballot measures, a group of five citizens may request a recount.) If they do ask for a recount, they'll have to pay $10 for each precinct, and there are some 11,000 precincts in Ohio. There were 92,672 ballots cast which showed no vote for president -- also known as residual votes -- in this year's election.
My take: The main question I have about the plan to ask for a recount in Ohio is ... Why? I understand the desire to keep a focus on the problems that occurred in this year's election. But a recount won't accomplish that. What recounts can accomplish is identifying ballots that should have been counted, but weren't, during the machine count. In Ohio, where about 70% of voters use the Votomatic punch card, that means identifying ballots with at least two corners of the chad detached.
A recount may result in a slight change in the vote totals, probably of not more than 10,000 in either direction. What it doesn't do is focus attention on the very serious problems that were revealed in Ohio's election such as the standards according to which provisional ballots should be counted -- an issue that's still very much up in the air. Concerned voting rights advocates would do far better to focus their attention in this direction than dropping $110,000 on a recount that's not likely to accomplish anything.
11/13/2004
More on Washington State's Provisional Ballot Fight
The Democratic Party won a small victory yesterday, in a court fight over the counting of provisional ballots in the Washington gubenatorial race. The Seattle Post-Intelligencer has this report. A state court judge ordered King County (Seattle area) to give the Democrats the names of 929 voters whose ballots had been initially rejected, on the grounds that the signatures on provisional ballot envelopes didn't match those in their electronic records. The Democrats want this information, so they can make an assessment as to whether the rejection of those ballots was proper. The Republican candidate leads by 1920 votes, with over 41,000 ballots still to be counted.
Democracy and the Blogosphere
Yesterday's N.Y. Times had this piece on the allegations of voter fraud circulating around the internet. It's headline was "Vote Fraud Theories, Spread by Blogs, Are Quickly Buried." It cites the suggestions of foul play in Ohio and Florida, discussed here, as among those that have been widely disseminated only to be proven false shortly thereafter.
My take: Blogs and other alternative websites are an essential channel for democratic discourse, one that's properly becoming more and more important. They provide a voice for concerned citizens without ready access to the mass media. In this election, they've played a vital role in focusing attention upon issues that the mainstream media can't or won't cover.
At the same time, it's important that those who see it as their mission to promote democratic equality and integrity -- and I count myself among this group -- take the responsibilities that come with that mission seriously. There are very serious problems that have arisen in this election, from voting machines to provisional voting to the I.D. requirement to long lines at the polling place. In addressing these problems, it's important to avoid making accusations of fraud or malfeasance that aren't supported by the evidence. When false allegations are debunked, it's much easier for proponents of the status quo to assert, quite incorrectly, that our election system worked just fine.
We should endeavor to shed light on the real problems that occurred in this election, and start building a consensus on the pressing need to make things better, but we must do so in a way that avoids hyperbole and exaggeration. Crying wolf before all the facts are in, as some have done, will make it more difficult to achieve this consensus and to effect meaningful improvements to our election system.
Early Returns on Electronic Voting
The N.Y. Times has this story reporting mostly favorable results from the use of electronic voting systems used by approximately one-third of voters. It does report a scattering of problems in some places, the two most serious of which are the loss of some 4500 votes in North Carolina and the mistaken addition (corrected almost immediately) of 3893 votes at a precinct in Gahanna, Ohio. The respected Caltech/MIT Voting Technology Project has found "no evidence, based on exit polls, that electronic voting machines were used to steal the election for President Bush."
We'll know more when data regarding the residual votes (overvotes and undervotes) with different types of voting technology becomes available. I'm sure the VTP and other empirical researchers are working on this as we speak. Of particular interest will be an examination of how the present generation of electronic voting technology compares to earlier electronic machines -- which, as the 2001 Caltech/MIT report noted, generated relatively large numbers of residual votes.
11/11/2004
Counting Ohio's Provisional Ballots
Although Kerry has conceded, county election officials throughout Ohio and other states are still proceeding with the process of determining which provisional ballots will be counted. The A.P. has this report on the process in Ohio. With precious little guidance on how to go about the process of determining which provisionals to count, it will be interesting to see whether there are wide variations among counties in the percentage that actually get counted.
Most provisional ballots in past Ohio elections have been counted, but that was under a system that was very different from the one that HAVA requires. Pre-HAVA, Ohio only allowed provisional ballots for voters whose address changed prior to election day but who didn't file change of address forms. But under HAVA, any voter whose name doesn't appear on the voting rolls for any reason should be allowed to cast a provisional ballot, if the voter affirms his or her eligibility to vote in that precinct.
How many of those provisional ballots will actually get counted? According to this report from the News-Messenger, Sandusky Counted has so far only rejected one provisional ballot (it's not clear from the story how many they've gone through so far). On the other hand, in Cuyahoga county, only 1749 of 2666 had been determined "valid." Most counties aren't keeping a running total of how many provisionals have been determined eligible and will therefore be counted. The deadline for counting them is December 1.
Settlement in Diebold Lawsuit
The California Attorney General and Diebold have reached a settlement in the state's lawsuit alleging that the company made false claims about its voting machine. The S.F. Chronicle has this report. Diebold has agreed to pay $2.6 million, $1.63 million to the state and $475,000 to Alameda County.
11/10/2004
Provisional Ballots Could Make Difference in Washington
Washington state, that is, where the Seattle Times reports that Democrat Christine Gregoire holds an 8700 vote lead over Republican Dino Rossi in the Governor's race, with some 213,000 votes still to be counted. In today's Seattle Times, it's reported that 31,700 provisionals were cast in King County alone, almost twice as many as four years ago. Washington has reportedly had provisional voting for more than 30 years.
Modesto Bee on Moving On
The Modesto Bee has this thoughtful commentary on the flaws in our system that this election revealed. It notes, correctly in my view, that the serious problems that have been identified don't justify the claims being made by some Democrats that the election was (again) stolen. For more on the continuing allegations that Kerry was robbed, see this Cox News report.
Washington Post Editorial on HAVA
Today's Washington Post has a great editorial on the election reform work that still needs to be done. It points out that part of the reason for the problems in implementing election reform result from the fact that Congress was late in providing the monies for voting system improvements authorized by HAVA. It recommends early voting as a part of the solution to the long lines that existed in some places. It notes the need to ensure that the statewide computerized registration databases, which must be in place by 2006, are accurate so that citizens are wrongly denied the right to vote. Finally, the editorial properly notes the need for clear standards on which provisional votes should count -- without which, Ohio would have faced a real mess had the vote been closer.
Democratic Congressmen's Letter to GAO
Six Democratic members of the House have written this letter to the GAO regarding alleged voting equipment problems in the 2004 election. It follows this one from November 5, also focused on voting equipment problems.
My take: While some of the problems that the Congressmen complain of are worthy of inquiry, the evidence available so far suggests that voting machines weren't the biggest problem in this election. Far more serious concerns surround problems in the implementation of provisional voting, registration list issues, and long lines at some polling place. There's a real danger that Democrats will get so caught up in the paranoia surrounding electronic voting, that they miss much more serious issues of vote denial and discouragement.
Creative Voter Education
The County of Los Angeles undertook some innovative methods of making sure voters understood how to use the county's "Inkavote" system. Up until March 2004, L.A. County used the Votomatic punch card system but was required to stop using that system by the court order in Common Cause v. Jones. Check out their "Got Dots" billboard along a California freeway. Other L.A. County ads can be found here on the Caltech/MIT Voting Technology Project's website.
Caltech/MIT Analysis of Voting Equipment & Exit Polls
The Caltech/MIT Voting Technology Project has issued this report assessing claims that the discrepancy between exit polls and actual vote in certain jurisdictions provides evidence of voting machine fraud. The report finds no evidence that electronic voting machines or any other type of equipment were used to "steal" votes. The report finds fault with the bluelemur.com article charging that the discrepancies between exit polls and the actual vote were more substantial in jurisdictions using electronic voting. Figure 3 on the last page debunks the notion that the exit poll/actual results discrepancy was largest in jurisdictions using electronic voting -- actually the opposite is true. To its credit, Blue Lemur has now removed the article.
