There is an interesting new research paper by David Kimball and Martha Kropf, recently published in Public Opinion Quarterly. Those of you with research university access can get the full text of the paper. Otherwise, here is the abstract of the paper:

The 2000 presidential election focused attention on the problem of unrecorded votes, in which a person casts a ballot but fails to record a valid vote for a particular contest. Although much recent research has evaluated voting technologies and their effects on unrecorded votes, there has been little research on the effects of ballot design. We argue that the same theories used to design and evaluate self-administered surveys can be used to analyze ballot features. We collect and code paper-based ballots used in the 2002 general election from 250 counties in 5 states. We code the ballots in terms of several graphic design elements, including the content and location of ballot instructions and the layout of candidate names and office titles. Our analysis suggests that several ballot features are associated with unrecorded votes (both overvotes and undervotes) in the gubernatorial contests. We also find that ballot design features exacerbate the racial disparity in unrecorded votes. Ballot design can be an important factor in determining whether voters are able to cast a ballot accurately, which can influence the legitimacy of elections.

Interesting research, and more on this general issue later after I’ve had a chance to digest this (and some other) new research that has just been published.

As was reported today, the New Orleans city elections have been postponed until the Summer or Fall of 2006. This delay raises an interesting question about the role of government in facilitating elections during times of crisis.

When Mayor Ray Nagin suggested that the elections in New Orleans should go forward, because “voting during our regular cycle would further bring a sense of normalcy and empowerment to our citizens,” he was in part reflecting an attitude that has existed in America since the Civil War, when President Lincoln decided not to postpone federal elections during a time of internal war. Several historians have noted that Lincoln’s decision was not what would be expected; nations historically had not held elections in the midst of a civil war.

The difference between the Civil War and today is that, during the Civil War, the state and federal governments worked to adopt new voting schemes to address the problem. Expanded absentee voting and remote voting for military personnel in most states dates to the Civil War. Some states set up actual voting precincts in forward state militia positions so that soldiers could vote. Other states allowed military personnel to give a proxy to a third party.

In New Orleans, it is clear that the government response, especially at the federal level, was not working to facilitate the February election date. As the Washington Post noted, Louisiana Secretary of State Al “Ater laid much of the blame for the delay on the Federal Emergency Management Agency, which he said has not provided any of the $2 million his office requested to repair voting machines damaged in the Aug. 29 storm and to upgrade New Orleans’s absentee voting system.”

The key issue now is how to facilitate voting for residents who, even 9 months from now, are still unable to live in New Orleans but plan to go back. One solution is to create, for this election, highly liberalized absentee voting rules, much like those used for military personnel and civilians stationed overseas. These voters–called UOCAVA voters for the Act that enfranchised them fully–can vote absentee in their last place of residence before they left the U.S. Perhaps former New Orleans residents should be given the same liberal absentee voting rights, even if they now are residents of another state.

It’s nice to know that others agreed with my conclusion about the “Voting Systems Testing Summit”, held at the beginning of this week in Sacramento, that there is a growing sentiment for states (especially California) to play a stronger and perhaps leading role in the testing of voting systems. Writing in this morning’s Oakland Tribune, Ian Hoffman (who was at the summit) summarized the current situation of voting system testing in this story lead:

For 11 years, most states have relied on voting systems tested to minimal federal standards, the results withheld from public scrutiny and given the green light by a nongovernmental agency working on a shoestring budget.

After much more detail regarding the current situation, and comments from some participants in the summit, Hoffman noted later in his story that:

But states such as California are moving toward more rigorous testing, perhaps in league with other states. To explore those ideas, California’s Secretary of State Bruce McPherson hosted a conference on voting-systems testing that was the largest West Coast gathering of major players in the debate on voting technology in at least two years.

…

California uses both the national testing and its own, which under McPherson has grown in rigor.

His office now requires every voting-system maker to supply dozens of machines for a massive, mock election to ferret out manufacturing or reliability problems. McPherson also has agreed to let a computer expert try hacking into a Diebold system, and state officials are weighing whether to require the same kind of security testing for all voting systems.

No doubt, California is taking the lead in implementing a wide set of testing protocols, especially for electronic precinct voting machines. Developing a more coherent process of testing in California is a good idea, especially if it can combine the testing protocols now being used into one program that all devices used for balloting and tabulation (precinct, absentee and early; electronic and optical scanning) must go through. And then we need to get going on testing protocols for other parts of the election process, especially the procedures and technologies used for new statewide voter registration systems.

This was my argument, made during my presentation to the summit. I began my noting that security planning and testing must be focused on the entire voting system, from end-to-end. I then talked about both security and contingency plans, and about the need for new and more comprehensive testing practices (especially talking about field tests, and testing for other important aspects of voting system performance, like usability).

I’m sure we will hear more discussion of these issues early next week, when a number of the participants at this week’s summit will head to Washington for two days of discussion and debate about the future of the electoral process, in a forum sponsored by the National Academy of Sciences. These are the same folks who brought us the recent “A Framework for Understanding Electronic Voting”, and the final report, “Asking the Right Questions About Electronic Voting.” Should be another productive NAS event and project!

For readers who have long memories of election reform debates in the United States, you will no doubt remember that the Carter-Ford (yes, Carter-Ford, not Carter-Baker!) election reform commission recommended that “Congress should enact legislation to hold presidential and congressional elections on a national holiday.” Furthermore, the Carter-Ford commission recommended “that in even-numbered years the Veterans Day national holiday be held on the Tuesday next after the first Monday in November and serve also as our Election Day” (see page 41). Those of you who remember this will also remember that this was one of the most controversial recommendations the Carter-Ford commission made, and that it was soundly opposed by many groups, especially veterans organizations.

Interestingly enough, there is some chance that California voters may be asked to decide whether state general election days should be made state holidays. On November 17, 2005, a proposed initiative statute entered circulation for signature collection (needing 373,816 signatures of registered voters to qualify) to enact this change. The official summary language of the initiativie (as written by the California Attorney General):

Provides that statewide general election days be state holidays. Further provides that public schools and community colleges shall close on statewide general election days, and that certain public employees, including public school and community college classified employees, may be entitled to a paid holiday on statewide general election days pursuant to collective bargaining agreements …

Given that this might soon be on the ballot in California, we’ll have to take a look back at the research literature to see what evidence exists that making election day a statewide holiday has positive effects on important outcome variables like turnout, polling place availability, and easier recruitment of polling place workers. I’m not aware of much in the research literature on this subject, and my examination of the research materials developed by the Carter-Ford commission shows no mention of research on what sorts of effects election-day holidays have on election administration or outcomes.

As an interesting parallel, on February 21, 2003, AB 1472 was introduced in the California state Assembly by Jackie Goldberg. As introduced, AB 1472, was summarized:

Under existing law, if local elections officials so request, the governing boards of public schools must allow schools to be used for polling places on election days.

This bill would require public schools to be closed on election days, and their facilities, including computer systems, to be
available to local elections officials. Schools would be open an extra day at the end of the school year to compensate for election day.

The bill would further provide that public school employees would not be paid for election days, and would instead be required to work an extra day at the end of the school year.

Because this bill requires public schools to be closed on election days and to be made available for use as polling places, it results in a state-mandated local program.

However, as I recall, AB 1472 encountered some significant opposition, due to the requirement that it would close schools on election day. That opposition came from a variety of sources, but I recall from discussions with legislative staff at that time that many parents were upset about having school closed during the work week, with questions raised about how they were supposed to take care of their kids on a work day.

AB 1472 was heavily amended, and when passed, the school-closing requirement was heavily watered down:

Under existing law, if local elections officials so request, the governing boards of public schools must allow schools to be used for polling places on election days.

This bill would authorize governing boards to exercise the options, in response to such a request, of: (1) keeping schools used for polling places in session but identifying to the elections official the specific areas of the school buildings not occupied by school activities that will be allowed for use as polling places, (2) designating the election day for staff training and development, or (3) closing the schools to students and nonclassified employees.

This bill would require that elections officials requesting the use of school buildings submit requests sufficiently far in advance to permit the governing board to determine whether or not to keep affected schools in session before school year calendars are prepared and distributed to parents.

I wonder if the same questions may arise about this potential ballot measure mandating that state general elections be state holidays, given the parallels here with the experience of the original AB 1472.

The two-day “Voting Systems Testing Summit”, sponsored by the California Secretary of State, was a wonderful opportunity to hear a lot of ideas and perspectives on how voting systems are tested today, and how we might improve testing to insure that future voting systems have increased accuracy, security, accessibility, usability, and other important attributes. Once I pull my notes together, I’ll have some more commentary about some of presentations and interesting ideas discussed at the summit.

But many of the participants in the summit kept asking the same question — exactly what was the motivation for the California Secretary of State to host this event? Again, it was great to hear from election officials, vendors and academics about how we can improve testing processes — but why now, and why in Sacramento?

One interesting hypothesis circulated in discussions with many participants, based on two themes that clearly resonated throughout many of the summit presentations and discussions. The first theme was that the current testing and certification process is not working well, as vendors, election officials, and the academics seemed to agree that the existing process is broken. The second theme suggested a possible solution — a stronger role for states in the testing and certification process? There were many different suggestions made along these lines, ranging from states (California?) developing strong and independent testing regulations and procedures. Others suggested that perhaps there might be some ways for states — perhaps in a consortium — to collaborate on stronger testing and certification procedures. And
others argued for a consolidation of state and federal testing, to save money, time and prevent duplication of effort.

So this leaves open the question — will California take the lead on an initiative to develop a strong and independent testing authority within the state? Will California work with a consortium of other states to develop some strong and independent joint testing authority? If California doesn’t do it, then who will?

But many inquiring minds, at least by the conclusion of this summit, left with the conclusion that there is a strong possibility that, based on the presentations and discussion in this summit (and the simple fact that the California Secretary of State hosted and organized this summit that produced these discussions), California might lead some new initiative to develop a strong and independent testing authority that will either augment the existing federal testing and certification process or possibly work jointly with the federal process. If so, this could be a revolutionary new development in the testing and certification of voting systems.

I skimmed through the recent press about the November elections and there is a wealth of information about problems at the polls that are directly poll worker related. The stories below encapsulate why effective poll worker training and polling place management are such important issues.

In the Detroit mayor’s race, there was controversy about voting. As the Detroit Free Press reported, “A final tally released Tuesday by the city Board of Canvassers indicated that Kilpatrick beat Hendrix by a vote of 123,140 to 108,600. Even those results, however, came with a caveat as elections officials said the number of ballots cast in nearly 100 precincts don’t match the number of voters who signed poll books or the number of votes recorded on receipt tapes from election machines. Suspicions rose after the election when the Free Press reported that Detroit officials had lost track of ballots in nine precincts and did not count them until two days after polls closed. In one case on election night, a poll worker took home two computer data packs containing ballot information and didn’t turn them in until the following day.

In North Liberty, Iowa, people initially thought one person was elected mayor when in fact that was not the case. As the Iowa Press Citizen reported, “There was some initial confusion about who was to be North Liberty’s next mayor following the Nov. 8 election. A poll worker reported the wrong total number of write-in votes from one precinct. The mistake, which was cleared up after tape from the voting machines was brought into the auditor’s office, gave Bahl the lead with 36 percent of the votes.

In the Queens, NY Chronicle, The Asian American Legal Defense Fund reported possible violations of the federal elections laws at eight polling stations in Queens including: missing voting signs at one school, missing interpreters, and rude behavior by poll workers.

“The preliminary reports show that Asian Americans needed to overcome several obstacles to exercise their right to vote,” Glenn Magpantay, staff attorney for the legal defense fund, said in a statement. “As in the previous elections, poll workers were hostile and some did not understand their obligations under federal laws. This is unacceptable.”

For example, “a poll worker rushed one Chinese-American voter to finish voting even though she had not finished reading the instructions. The poll worker stopped when a supervisor arrived. Another voter reported that a poll worker at the site was very rude and tried to belittle her. At one poll site, a Spanish interpreter was available but poll workers did not seek the interpreter’s assistance when a voter needed it. One election district voting table had bilingual materials stacked in a pile in unopened packages.

Recently I received an inquiry from the California Secretary of State’s office for research regarding open source software in voting systems. I’ve been hearing rumors for months that there is some sort of task force or study committee that has been in the works, though no formal announcement that I know of has been made about a study on the open source software issue.

The inquiry I received stated:

California Secretary of State Bruce McPherson is leading a fact-finding, impartial study regarding the use of open source software in voting systems in California. Secretary McPherson is responding to Assembly Concurrent Resolution (ACR) No. 242, filed with the Secretary of State on August 31, 2004. ACR 242 asks the Secretary of State to investigate and evaluate the use of open source software in all voting systems in California. It also requests that the Secretary of State report his findings and recommendations to the Legislature by January 1, 2006. I have enclosed a copy of ACR 242 for your review.

In order to provide thorough research on the issue of open source software in voting systems, Secretary McPherson is asking for public input, including but not limited to reports, studies, summaries and other written documentation relating to open source software in voting systems. He is seeking a broad range of opinion and information because he recognizes that open source software is a complex and technical issue and it deserves a careful, balanced and in-depth study over time.

Secretary McPherson is convening a group of experts, representing a variety of viewpoints and expertise that will have regular meetings and hold public hearings if necessary. One of Secretary McPherson’s top priorities is to ensure the security and integrity of the elections process and this group, along with public input, allows for a careful and thoughtful examination rather than a hasty superficial review.

And for those of you interested, here is information on AC 242:

BILL NUMBER: ACR 242 CHAPTERED
BILL TEXT

RESOLUTION CHAPTER 171
FILED WITH SECRETARY OF STATE AUGUST 31, 2004
ADOPTED IN SENATE AUGUST 24, 2004
ADOPTED IN ASSEMBLY JULY 1, 2004
AMENDED IN ASSEMBLY JULY 1, 2004
AMENDED IN ASSEMBLY JUNE 28, 2004
AMENDED IN ASSEMBLY JUNE 23, 2004

INTRODUCED BY Assembly Member Goldberg
(Coauthors: Assembly Members Berg, Calderon, Chan, Chavez, Chu,
Cohn, Corbett, Correa, Diaz, Dutra, Dymally, Firebaugh, Hancock,
Jackson, Kehoe, Koretz, Laird, Leno, Levine, Lieber, Longville,
Lowenthal, Mullin, Nation, Negrete McLeod, Nunez, Oropeza, Pavley,
Reyes, Salinas, Steinberg, Vargas, Wesson, Wiggins, Wolk, and Yee)

JUNE 3, 2004

Assembly Concurrent Resolution No. 242–Relative to ballot tally
software.

LEGISLATIVE COUNSEL’S DIGEST

ACR 242, Goldberg. Ballot tally software.
This measure would request the Secretary of State to investigate
and evaluate the use of open-source software in all voting machines
in California and report his or her findings and recommendations to
the Legislature by January 1, 2006. The measure would state that the
Legislature is committed to exploring and reviewing all available
programs and features, including open-source software, that may
enhance the security of election results and increase voter
confidence in the integrity of the voting process.

WHEREAS, The State of California has a compelling interest to
ensure public confidence that election results are complete, honest,
and accurate; now, therefore, be it
Resolved by the Assembly of the State of California, the Senate
thereof concurring, That the Legislature hereby requests the
Secretary of State to investigate and evaluate the use of open-source
software in all voting machines in California and report his or her
findings and recommendations to the Legislature by January 1, 2006;
and be it further
Resolved, That the Legislature is committed to exploring and
reviewing all available programs and features, including open-source
software, that may enhance the security of election results and
increase voter confidence in the integrity of the voting process; and
be it further
Resolved, That the Chief Clerk of the Assembly transmit copies of
this resolution to the author for appropriate distribution.

Hopefully this study will produce some productive recommendations on the open source issue. Unfortunately, given the very short time frame given here in ACA 242 (January 1, 2006), there is not much time for a detailed study nor for much public input on this issue.

Reliable information on the Diebold “hack” that was reported on late last week in a number of major media outlets has again become elusive. I wrote about the media reports in a recent blog entry, but over the weekend received a number of email messages from folks noting that new coverage of this story indicates that the “hack” is still under negotiation, and may be on hold for some time.

Writing in the Oakland Tribune yesterday, Ian Hoffman noted that the “hack” was planned for this coming Wednesday, “but is likely to be delayed until mid-December.”

Hoffman’s story also provided some details:

…there have been extensive, ongoing negotiations between Black Box Voting and the California Secretary of State’s office, which also is talking to Diebold, over conditions of the test, confidentiality of the results and measures of success. Those talks continued this weekend, but state officials said they remain committed to performing the test.

“Secretary (Bruce) McPherson takes testing these systems very seriously,” said his spokeswoman Nghia Nguyen Demovic. “He wants safeguards in place so that every vote cast is secured. He’s doing his due diligence to assure voter confidence.”

Still no official word I can find on the California Secretary of State’s website on this “hack”. Many questions remain, especially about the precise nature of the “hack”, who will be involved, what methods will be used, whether voting systems other than Diebold’s will be tested, and most importantly, what information on the results will be released. I’m also curious as to why the Secretary of State’s office isn’t soliciting proposals from academic groups and from other organizations that routinely undertake projects like this (i.e., “tiger team” testing); might a better approach be to open this up to competitive proposals from all interested and knowledgeable groups and organizations?

Reports have been circulating since earlier this week that California’s Secretary of State, Bruce McPherson, has asked Harri Hursti (from Finland) to try to “hack” a randomly selected Diebold voting machine. I first started hearing of this earlier this week, and at that time could not find any confirmation of this request. But this morning the story started to circulate on the newswires, and through a variety of news outlets. For example, here is the Associated Press story, as provided online by MercuryNews.com. There is a longer version of the story provided by the San Francisco Chronicle. I could not find any confirmation of this request on McPherson’s website, nor any official statement or elaboration of this test.

Thus, at this time, details of the test are a bit sketchy. The wire reports indicate that one Diebold precinct voting device, from one of the California counties using these voting machines, will be selected randomly for the test. But exactly what methodology the “hack” will involve is not known; whether it will focus primarily on technology or procedures (or perhaps both) is not known; nor is it clear what information will be provided to the public about this “hack” once it is completed (for example, will we know how many attacks are attempted, what types of attacks they are, and about the outcome of each attack?). It is unclear from the available information whether the vendor will be involved in the “hack” (other than the report that the vendor was not allowed to pick the exact voting device that will be used in test); also unclear is the extent of election official (state or local) involvement in the “hack”. It is also not clear whether the same type of “hack” will be initiated for other voting devices, either those now certified for use in California or those pending certification.

In any case, perhaps more about this “hack” will be revealed in next week’s “Voting System Testing Summit 2005”, sponsored by McPherson’s office. That might be a good opportunity to hear more discussion about the role of “hacks” (also known as “red team” or “tiger team” attacks), as part of a broader process of assessing the overall operational security of a voting technology. I’ll be attending this conference, and I am moderating a session on “Security/Paper Trails/Accountability”. Panelists on our Tuesday afternoon session include Kim Alexander, Henry Brady, David Dill, Avi Rubin, and Michael Shamos. We have been asked to address five different questions:

1. Evaluating security — how to do it best
2. Recovery process/contingency plans
3. Source code — how much should be reviewed? By whom?
4. Standards for AVVPAT
5. For what purpose should AVVPATs be used?

It will be an interesting discussion, and I’ll do my best to try to summarize this session, as well as the entire conference, next week.

I recently gave at talk to my colleagues at the University of Utah about the impact that poll workers have on public confidence in elections. The powerpoint slides from the presentation are here.

This summer I did a study with two colleagues from BYU–Quin Monson and Kelly Patterson–where we implemented an internet survey of voters who had cast ballots in the 2004 general election at a polling place in Utah. Quin and Kelly run an exit poll project and in 2004, a random sample of individuals did not complete an exit poll; instead, they provided their email address and basic demographics. Utah has the nation’s highest internet penetration, making it an excellent population to survey online.

In the survey, we asked four types of questions: (1) their political preferences about candidates and issues, (2) their experiences at their polling place in 2004, (3) how confident they were that various aspects of the electoral process were fair, and (4) demographic information.

What we found was that poll workers have a very significant impact on how confident people are that their ballot was counted accurately, that the election was run fairly, and that democracy works in the state. It was the second most important factor in the analysis, after partisanship. (Not surprisingly, Republicans were more confident than Democrats in a state run by Republicans!)

If you study polling places regularly, as Mike and I both do, you often see instances where poll workers implement the rules in their own way. In fact, researchers at Loyola-Marymount in Los Angeles found this to be the case in a report published in 2004. For example, poll workers were asking voters for ID in a small but sizable number of polling places, even when doing so was counter to California law. Not surprisingly, problems occurred more often in low income and minority communities.

Poll workers can be seen, therefore, as having the ability to add to the cost of voting for some voters. If a voter has a bad experience at the polls, it may not make voting so costly that they do not vote again, but it could. It could also drive them to vote in another way–such as by absentee ballot–or it may merely make them less confident in the way government works.

One problem with studying elections today is that we know so little about poll workers. However, this study does help us see that the need to study poll workers is great, because the voter-poll worker interaction clearly affects the way people view important aspects of the electoral process.