Residual vote forecast

Following the 2000 presidential election, the Caltech/MIT Voting Technology Project coined the term “residual votes” to describe ballots that did not record a vote for president, for whatever reason — the ballot was outright blank, the vote cast on the ballot wasn’t counted, or the ballot contained an over-vote.  Although the residual vote rate measures only one dimension of election equipment performance, it has become the most common metric used to evaluate the quality of election systems.

For the states where we can compare residual vote rates in 2000 and 2004, the rate fell from 1.9% in 2000 to 1.1% in 2004.  In my 2006 Election Law Journal article, I estimated that changes in voting equipment from 2000 to 2004 accounted for a considerable fraction of this reduction, resulting in 1 million more votes being counted in 2004 than would have been, had HAVA not forced/facilitated upgrades.

What to expect for 2008?  It is weeks until we will have all the data available to begin making firm reports, but here are some thoughts:

  1. Watch (out) New York. The Empire State remains the one state that still uses an antiquated system that has been shown to produce more residual votes than either DREs or optical scanning.  Still, NY improved considerably from 2000 to 2004, even without changing equipment.  Score one for election administration!  Still, it will now be the one outlier in 2008, as far as modernization is concerned.  We’ll see if the state’s slow-as-molasses upgrade process is vindicated or not.
  2. Technology improvements should reduce the residual vote rate. The period from 2004 to 2008 represents the most extensive shift in voting equipment usage in American history.  Most of those changes were along paths that should reduce residual vote rates.  If 2008 was like 2004, expect another 1,000,000 new votes to be counted due to these upgrades.
  3. Turnout is not our friend. Research done by my colleague Steve Ansolabehere and I showed that a surge in turnout is bad for residual vote rates.  New voters overwhelming precinct judges will lead to more miscast ballots.  All these reports of record turnout make me worry that the votes won’t make it into the tally.
  4. The DRE backslide. As the Election Data Services 2008 Voting Equipment Study reports, the period 2006 to 2008 represented the first time in American history that the number of jurisdictions and voters using DREs dropped.  This provides an opportunity to study whether the Opscan-to-DRE switch is symmetrical.  That is, jurisdictions dropping Opscan ballot and adopting DREs saw a decline in residual vote rates in previous years.  What we now need to see is whether dropping DREs and going to Opscan will produce the symmetrical effect of increasing residual vote rates, if only slightly.
  5. The dual-edged sword of convenience. With so many voters using “convenience” methods, such as early voting and on-demand absentee ballots, we’ll finally be able to answer the question about whether these new methods increase residual vote rates.  Holding everything constant, they should, since mail-in absentee ballots provide no chance to catch and correct an over-vote.  As well, there seems to be an odd trend of jurisdictions moving to central-count optical scanning, which will also increase residual vote rates.

On net, I expect the residual vote rate to go down in 2008, compared to 2004 (and certainly 2000).  However, all factors are not pointing in the same direction.  Most worrisome are the predictions that many localities will be indundated with chaos due to voter registration problems.  Because I would rather vote in a well-run county on a lousy voting machine than in a poorly-run county using the best voting machine, anything that adds to Election Day chaos will be bad for making sure that all the votes are counted.