We are processing voting mode information in Florida and Peter Miller (my RA, off to UC Irvine in a month) has produced this graphic comparing older and younger voters and whether they vote early in person, at the precinct place, or by mail.

Pretty obviously, older voters much prefer voting by mail, while moderately old (50-65 year old) voters prefer early in person. We're puzzling over the reasons right now--it could be that the oldest voters have mobility problems, and that's why they vote by mail. I expected that the early in-person voters would cluster in the middle age groups, 35-50, because these voters tend to have the busiest lives with the most outside commitments. James Hicks, another EVIC RA, points out that these older voters are more committed partisans, on average, which we know is positively related to use of early voting methods.

A provision of the National Voter Registration Act of 1993 (Sec. 1973gg-6(i)(1)) provides for public inspection of data measuring the performance of "Motor Voter":

Each State shall maintain for at least 2 years and shall make available for public inspection and, where available, photocopying at a reasonable cost, all records concerning the implementation of programs and activities conducted for the purpose of ensuring the accuracy and currency of official lists of eligible voters, except to the extent that such records relate to a declination to register to vote or to the identity of a voter registration agency through which any particular voter is registered.

To meet the requirements of this section, jurisdictions may track the source of a registration (e.g., from the motor vehicles department) as a field on their voter registration files and report aggregate statistics in public reports. The "shall" and "except" clauses in this section provide a clear prohibition on the public release of these individual data, for good reason, since the privacy of individuals who register through a public assistance office should be maintained.  However, researchers and public advocates might want questions answered that jurisdictions may not ask. For example, these registration source data provide a means to evaulate claims that registration bounty programs needlessly inflate the registration rolls or they may be used to make an independent evaluation of the efficacy of various voter registration programs that can lead to better targeting of scarce election resources.

Many people, myself included, have undergone special human subjects training and have handled confidential data for various projects. A simple change to the NVRA granting states the right to approve access to registration source data for persons who can demonstrate their integrity to work with such data would help greatly with the improvement of election administration.

During the beginning of 2008, I and others of Forsyth County felt that we had to have significantly higher numbers for our Advance Voting in order to ensure that the lines for voters would be shortened on Election Day. The best way to do this is to get more voters to the polls during the week before the day of the election. Georgia allows the counties to open up Advance Polling locations during the week before. Forsyth County opens for the entire week and has five sites that are open from 8 a.m. to 4:30 p.m.

The Census report on voting and registration in the 2006 election has been posted.    http://www.census.gov/population/www/socdemo/voting.html. The general consensus on places like the Election Law Listserv is that this report, based on the Census' Current Population Survey (CPS) vastly overstates voting, and probably understates registration. For instance, Kim Brace estimates that there were almost 11 million fewer votes in 2006 than the Census reports here, which is an overestimate of more than 10%.  Michael McDonald and Lorraine Minnite chime in with similar skepticism, and there is a productive conversation about overreporting problems and how we can validate vote reports in surveys.

 "Just Democracy" (Katrina vanden Heuvel of the Nation) calls for a new wave of election reform following the 2008 election.  At least some of these measures (election day registration, the frequency of voter fraud) could be informed by a high quality set of "data for democracy."

At least two others--and end to popularly elected statewide elections officers, and an "end to party duopoly"--may be less well-received by Secretaries of State and some elections officials. 

There are a number of surveys either in the field or being administered in the 2008 cycle targeted at elections officials, pollworkers, etc.  It would be very interesting to find out how these "street level bureaucrats" think about these kinds of reform proposals.

From the conference program of the Joint Statistical Meetings in Denver:

My colleague Doug Chapin came across this recent paper, from the Eighth Annual Conference on State Politics and Policy:

Modeling Problems in the Voter ID - Turnout Debate

Robert Erickson and Lorraine Minnite
Columbia University

Abstract: In April 2008, the U.S. Supreme Court upheld Indiana’s controversial new voter identification
law.  Adopted in 2005, the law requires voters to show a current, government-issued photo
identification.  Opponents worry the new rules will place an undue burden on the voting rights of
elderly, low income and minority voters, disputing the need for these laws.  Nevertheless, over
the last five years, new voter ID requirements have been adopted on party line votes in more than
a dozen states.  Stimulated by the pressing policy debate, recent scientific research on the turnout
question is largely inconclusive: different datasets, measurement rules and statistical models
produce different and contradictory findings.  Voter identification requirements do not appear to
affect voter turnout in a straightforward way, and the complexity of electoral rules and how they
influence voting behavior make statistical outcomes quite sensitive to research designs.  We see
problems with existing designs.  Our paper evaluates recent research and investigates the
strength of the statistical arguments used to support findings.  We use difference-in-difference
techniques and Current Population Survey data to test for an impact in the states, comparing
turnout in the 2002 and 2006 midterm elections.  Our findings suggest that the data are not up to
the task of making a compelling statistical argument.

I'm sure my colleague Doug Chapin, who is sadly attached to the Washington Nationals, will know about this website: http://www.fivethirtyeight.com/.

Readers familiar with Bill James and the emergence of sabermetrics may be familiar with Nate Silver's
"PECOTA" rating of baseball players.  Now this University of Chicago econ grad is applying his forecasting skills to elections.

The site is fun for political junkies, but why should it be interesting for readers of this blog?  After all, Silver's forecasting method primarily relies on polls.  If you read further, however, you'll see that Silver recently posted a long comment about voter history files and how they are used by campaigns.  The reactions to his post also reveal how campaigns can use some state files and face difficulties with other state files.

Chris Mann's article (posted here, look at the PDF from the conference) makes some of the same points.

With the advent of early voting and the large numbers of voters using this relatively new process, it has been incumbent upon election officials to ensure that their sites are not discriminatory. We have used geocentric data to evaluate our early voting sites in Forsyth County Georgia; below are some graphs that show early voting by location and by political party.

A recent story in the Chicago Daily Herald  indicates that Kane County officials are considering a big push for early in person voting in 2008 as a way to reduce costs.

I know there are studies from Oregon that claim cost savings when going fully vote by mail, but the anecdotes I have heard from elections officials is that a "mixed" system (partially in person, partially absentee) is actually the most expensive of all.