The National Academies of Sciences released a report on electronic voting today, “Asking the Right Questions About Electronic Voting.” One of us (Alvarez) was on this panel, and we will comment more extensively on this report soon.

The full report is available here.
• The “A Framework for Understanding Electronic Voting” website.
• A short summary of the report (from Herb Lin, Study Director).
• The press release for the report:

  Electronic Voting Systems Show Promise, But Require Bigger Commitment by
  Federal Government and States

  WASHINGTON — While electronic voting systems have improved, federal and state governments have not made the commitment necessary for e-voting to be widely used in future elections, says a new report from the National Academies’ National Research Council. To become viable, more funding, research, and public education on e-voting are required, said the report’s authoring committee, which was co-chaired by two former governors. And because electronic voting systems, like all complex computer systems, are fallible and may be compromised — either deliberately or by accident — backup systems will be needed in the event that allegations of fraud or malfunction arise.

  “Electronic systems are expected to improve election administration, but this will happen only if more resources are dedicated to understanding how these systems work and to educating election officials and the public on their use, ” said committee co-chair Dick Thornburgh, former Pennsylvania governor and U.S. attorney general, and now counsel to
  Kirkpatrick & Lockhart Nicholson Graham LLP, Washington, D.C. “In an election environment with great variability in state electoral law and in the qualifications of local election officials and poll workers, such an effort will be critical to realizing the full potential of electronic systems.”

  Election officials are concerned that the voting controversy in Florida during the 2000 presidential contest and other real or perceived voting mishaps have shaken the public’s confidence in the election process, and more locales are turning to e-voting as a presumably more-foolproof alternative. The report, however, emphasizes that decisions about using e-voting systems need to based on whether they will significantly improve the reliability and efficiency of elections, an obvious
  consideration, but one that often is overlooked in the public debate. Decisions about the ultimate desirability of e-voting systems should not be limited to assessments of their performance to date, the committeewarned, noting that e-voting technologies are improving over time. Election officials should ensure that these systems are reliable, user-friendly, and comply with election laws. In the end, elections that are trusted by the public should be considered the gold standard of
  election administration, the committee added.

  To help election officials gauge the robustness of an e-voting system, the committee posed a series of questions on matters such ascomputersecurity and voter privacy, system usability, and the life cycle of software and availability of upgrades. The answers should be made widely and easily available to officials and the public, who will need them to make sound decisions about when and if to adopt e-voting.

  Thornburgh’s co-chair on the committee was Richard F. Celeste, former governor of Ohio and U.S. ambassador to India, and current president of Colorado College, Colorado Springs. “The issues associated with electronic voting are not partisan issues,” Celeste said. “We’re asking questions that should be part of a rational and systematic investigation of the unresolved issues and ramifications of electronic voting. The public should be involved in asking these questions, and should pay heed
  to the answers. Our democracy demands nothing less.”

  The study was sponsored by the National Science Foundation. The National Research Council is the principal operating arm of the National Academy of Sciences and the National Academy of Engineering. It is a private, nonprofit institution that provides science and technology advice under a congressional charter. A committee roster follows.

It turns out here are additional street address standards, in addition to the URISA proposal that we discussed earlier. Joe Hall (UC Berkeley) emailed the following tip:

“… there is [xAL], the Extensible Address Language… for which a US-specific subset of the elements in that standard could be used (as a basis or as the standard itself).”

Joe also gave the url for xAL.

Thanks to Joe for the tip, and if there are others out there, please pass them along.

Anyone who has worked with real voter registration databases, voter history files, absentee voter files — or any sort of database that has street address information for voters — knows that working with street addresses can be messy and complicated. Streets have names, and sometimes they have typographical errors in them; sometimes numbered streets will be identified by number, sometimes by character strings (“7th Street” can sometimes be “Seventh Street”); there are apartment numbers and fractional street address numbers; and then there are lots of descriptive elements used for street addresses, that take different forms (“North” or “N”, “Drive” or “Dr”, etc.). These complexities make it difficult to use this data an any particular voter file, and certainly are going to cause headaches when folks start to try to compare data in new statewide voter registration files relative to other government databases (Department of Moter Vehicle files, for example).

An organization possibly unfamiliar to those in the elections arena ( the Urban and Regional Information Systems Association [URISA]) has initiated the development of draft street address data standards. These standards, while not being formulated for election administration, are of interest to those of use who are concerned about the need for election data transactions standards. The URISA documentation for their draft standards justifies the need as follows:

Street addresses are the location identifiers most widely-used by state and local government and the public. Street addresses are critical information for administrative, emergency response, research, marketing, mapping, GIS, routing and navigation, and many other purposes. Because they have evolved over many decades, under the control of thousands of local jurisdictions, in many different record and database formats, and to serve many purposes, different address formats and types pose a number of complex geoprocessing and modeling issues. As a consequence, government agencies struggle with these issues as they seek to integrate large, mission-critical files into master address repositories.”

Sound familiar?

We have only just begun to review the URISA draft standards, which are available for public comment until October 3, 2005. Like most folks in the elections area, our attention has been on the Voluntary Voting System Guidelines, which are also open for public comment until September 30. But based on our initial read, the URISA street address standards seem like a very good idea, and should be reviewed carefully by those thinking about election data transactions standards. There are points of overlap here between the URISA draft standards and what others are doing in this space, but specifically on elections:

• Hart InterCivic has proposed their “Election Data eXchange (EDX)” as a standard for the secure exchange of election data (recently submitted to the Institute of Electrical and Electronics Engineers [IEEE] for possible use in formulating such a standard [IEEE 1622]).
• The OASIS group has proposed “Election Markup Language” (EML)) as another data transaction standard.

Having street address standards clearly are necessary to implement the type of data transaction standards we recently proposed in our report, “The Next Big Election Challenge.” Standards for statewide voter registration databases are likely to be a hot topic in the near future, as more and more states come on line with their new systems, and the URISA approach to develop standards for street addresses might be helpful for those thinking about voter registration or general data transaction standards.

The Los Angeles Times on Sunday ran an interesting story that raises issues that Mike and I have discussed for much of the past week: what does Katrina mean for voting and elections? Consider the following questions:

First, Louisiana is scheduled to hold open primary elections on October 15 and a General Election on November 12. Where will the displaced New Orleans residents vote? Displaced residents who are still within the state of Louisiana should be able to re-register in their new place of residence and vote in these elections, according to state voter registation rules. This could have a dramatic effect on the outcome of elections across the state.

Second, will displaced New Orleans residents who are in Houston, San Antonio, or other places likewise be registered and vote? Most states have simple 30 day residency requirements, making these new voters a targeted population for the next elections.

Third, what does all of this mean for 2006 and the elections in Louisiana? New Orleans was a stronghold for Mary Landreau, and a very large percentage of her voters have gone off in a disapora. What will this mean for politics in the state? And as the previous item suggests, what does this mean for politics in surrounding states as well? As the LA Times reported,

The hurricane ruined so many people’s lives, and much of their suffering is because of failures of government–failures to act, failures to plan, failures to invest, failures to prepare–that we need to ensure that these people can vote so they can express to the government know exactly how they feel.

In an intriguing paper presented by Mary Fitzgerald (Harvard University) at the recent APSA conference, “The Triggering Effects of Election Day Registration on Partisan Mobilization Activities in U.S. Elections”, we have some evidence for why voter turnout is consistently greater in the six states currently using election day voter registration. As noted by Fitzgerald, there are a series of published studies that have systematically documented that voter turnout is consistently higher in states employing election day voter registration, a result that Stephen Ansolabehere, Catherine H. Wilson and I have documented as well in our research paper; we estimated that had all states used election day voter registration procedures in the 2000 presidential election “that voter turnout could have increased by 8.1%, from almost 63% to almost 71%.”

Fitzgerald’s innovation is to look at partisan mobilization strategies, in effect comparing these strategies across states using individual-level survey data from the American National Election Study [ANES] (she uses all of the survey data from 1972 through 2002). In the ANES data, a question is systematically asked of respondents as to whether they were contacted by one of the major political parties regarding voting in the upcoming election. In the terminology of social science research, she uses whether or not each survey respondent reported a party contact as the “dependent variable” — what she is trying to explain. She knows whether the respondent is in a state using election day voter registration (and other aspects of the state’s election procedures), and she knows other information about the respondent and their local political context that she can use as “control variables.’

Based on her statistical analysis, Fitzgerald is able to show that respondents in states using election day voter registration were consistently (and statistically!) more likely to report being contacted by a major political party, in both presidential and midterm elections, and by both Democratic and Republican parties. Fitzgerald summarizes her results:

“…EDR increases the estimated likelihood that an individual will be contacted by 12 percentage points in presidential elections, and by 14 percentage points in midterm congressional elections. Interestingly, when comparing the contact efforts of the Democratic and Republican Parties, individuals are slightly more likely to be contacted by the Democratic Party during both presidential and midterm congressional elections.”

But Fitzgerald finds that while the parties do contact a larger number of potential voters in EDR states, the parties are not clearly clearly targeting specific demographic groups in their EDR mobilization efforts (at least given her data).

In recent years, there has been a great deal of interest by social scientists and practitioners in non-partisan “get-out-the-vote” (GOTV) efforts; most significant has been the work of Donald Green and Alan Gerber in this area. Fitzgerald’s work pushes in a different direction, turning attention to how partisan mobilization efforts may be tailored to take advantage of electoral procedures like EDR. To the extent that these partisan mobilization efforts in EDR states have indirect effects on the friends and family members of those contacted to vote, and the efforts can be better targeted in the future at lower-turnout populations, it is also possible that partisan mobilization efforts may have even stronger effects than Fitzgerald finds in her paper.

This is a clever new take on the question of EDR and voter turnout. While I could complain at the margins about some aspects of the statistical model specification (for example, some of the specification and methodological issues raised by my NYU colleague, Jonathan Nagler in his 1991 and 1994 papers on voter turnout might need consideration here) and other methodological issues, I am not sure that those complaints necessarily would alter the central point of this paper. It will be interesting to see how Fitzgerald’s analysis may provoke some new studies of EDR, and perhaps some examination of partisan mobilization efforts.

Okay, when is enough enough? Today the EAC issued another two new RFP’s:

• RFP 05-08: Voter Information Public Access Portal Design Conference
• RFP 05-09: Voter Hotline Project

Developing voter hotlines we understand, especially given the success of some of those projects in the 2004 election (like 1-866-MYVOTE1). For those inquiring minds that just have to know, the EAC defines a voter information public access portal as “a Web site that disseminates voter education information.”

Why a design conference? Well, … “In June of 2005, staff at the U.S. Election Assistance Commission (EAC) undertook a survey of public access portals … The EAC found that there were several public access portals in operation for the 2004 Presidential election, and that their sponsorship ranged from locally-based, to independent sector and private corporations. many of the portals performed the same functions for voters, and efforts were found to be duplicative, disorganized, and oftentimes erroneous.” And the idea here is that whoever gets this contract is going to set up a conference to bring interested parties together to figure out strategies to solve these problems.

So if you are looking for some fun weekend projects, add these to your list. You can guess what election geeks like us are going to be doing this weekend!

In addition to the two “Requests for Proposals” (RFP) that the U.S. Election Assistance Commission issued earlier this week on poll worker programs, four more have come out:

1. RFP 05-04: Develop and Build a Legal Resources Clearinghouse
2. RFP 05-07: Best Practices on Vote Count and Recount Procedures
3. RFP 05-10: Improvement of Election Data Collection
4. RFP 05-11: Information Management Analysis and Records Management Policies and Procedures

Details are on the EAC RFP website. There may even be a few more issued shortly, they are coming out at a very fast pace!

For anyone interested in these RFP’s, read the fine print — these RFP’s all have very, very short deadlines: next week! Sharpen those proposal pencils!

In an op-ed piece in this morning’s San Jose Mercury News, California’s Secretary of State Bruce McPherson faulted current California legislation (SB 370):

My biggest objection in using the paper trail to check voting machine performance is that it denies many in the disabled community the ability to have the same level of confidence in the process that all other voters enjoy.

Secretary McPherson then went on to state:

Using paper receipts as secondary ballots at this point is too risky. They are designed for the voter’s review and are not printed on ballot-quality paper and might not retain their quality during the often-lengthy recount and legal challenge periods.

Additionally, to ensure voter confidence in California’s elections, manual recounts and paper audit trails should not be the only ways of verifying the accuracy of vote tallies.

Therefore, I have directed my staff to conduct parallel monitoring programs on Election Day in which we randomly select “live” voting machines, take them out of production, cast predetermined voting scripts on them, and check them against the totals produced by the machines.

Indeed, the issue of how California conducts the 1% manual recount needs examination and research, a subject that Jonathan N. Katz, Sarah A. Hill, and I recently studied in a working paper, “Machines Versus Humans: The Counting and Recounting of Pre-Scored Punchcard Ballots.” Not to sound like a broken record, but how recounts are conducted (both those that are mandatory and those that are initiated upon demand), needs much more study. The research here is very thin, consisting in only a few working papers that we know of, for example, Ansolabehere and Reeves’ study of New Hamphsire recounts (1946-2002), and the Herron and Wand paper on the 2004 New Hampshire Recount. This a woefully understudied area, and we hope that perhaps some new research might be spawned by research efforts like that of the Election Assistance Commission, under RFP 05-07, “Best Practices on Vote Count and Recount Procedures.”

Secretary McPherson also noted:

Before we hastily start treating the paper trail as a secondary ballot, we need to explore other, more inclusive avenues to reach the goal of voting verification. I am working with other states’ election officials on a national research and development effort to develop voting technology that assures independence and secrecy of the vote for disabled citizens.

This is a very good development, but other than this off-hand comment, we haven’t heard anything about this national research and development effort. If readers have any information about which states are participating and what the effort involves, we would appreciate the information!

At APSA, we presented some draft material from our upcoming book on the e-voting controversy. The paper was titled, “Policy Framing, Risk Amplification, and the Debate Over Electronic Voting.” We are planning on including this as chapter 4 of our book.

Here is the abstract of the paper:

Scholars of the policy process have long studied how language and stories shape the policy process. More recently, scholars of the sociology of risk have focused on the study of policy framing in the context of risk and the “risk society”. We employ these theoretical works to an analysis of the debate over election reform. In the aftermath of the 2000 election, there were strong claims made about the need for electronic voting. After all, the 2000 election debacle was occurred primarily on paper-based systems. However, over the last four years, opponents of electronic voting have been able to reframe the debate over electronic voting, creating a new “story” about elections to support their claims that an election can be “stolen” by DREs. Doubts about electronic voting are exacerbated by the strongly polarized political environment. America is evenly divided across “red” and “blue” states and across political institutions and many people thought that the outcome of the 2000 election was unfairly decided. Using a unique dataset examining media coverage of electronic voting, we present an analysis of the media coverage of electronic voting that shows how the critics of have instigated a dramatic shift in the way in which the media covers this issue. There has been a strong amplification through the media of the possible risks associated with electronic voting, with little consideration given to the other side of the equation. This shift has resulted in the media adopting an almost common view that electronic voting is problematic.

Comments welcome!

First off, the National Academies of Sciences project, “A Framework for Understanding Electronic Voting”, is likely to issue a report soon (the project had an approximate start date of 9/01/2004, and according to the NAS website “A final report will be issued at the end of the project in approximately 12 months from the start date”) … we will keep readers of our blog posted about the release of this report as we learn more!

Secondly, the current “Commission on Federal Election Reform” (better known as the “Carter-Baker Commission”) is poised to release their final report soon. According to the FAQ on their website, “The final report is scheduled for release in September 2005”). We will also keep Election Updates readers posted about this upcoming report. Based on comments made by Commission Executive Director Robert Pastor at the American Political Science conference last week, the Commission’s final report is likely to contain a series of specific, and possibly provocative, recommendations on voting technologies and election procedures.

Last, as discussed in our third podcast, our evaluation of the current situation in Congress is that the prospects for “opening up HAVA” are slim. Based on discussions with some Congressional staff last week, and with a wide variety of folks in Washington who routinely monitor election reform efforts, there does not seem to be much interest in Congress in revisiting or amending HAVA, at least until there is a stronger consensus that there has been sufficient time learn what does and does not work in the HAVA.