Skip Navigation

Overvotes: Phantoms of the Ballot Box

Election officials should take steps to detect overvote problems when they arise in order to prevent lost votes in the future.

  • John Travis
May 10, 2012

Crossposted at ReformNY

The New York State Board of Elections, New York City Boards of Elections, and voting machine manufacturer ES&S each released reports yesterday detailing the results of an investigation into the abnormally high numbers of lost votes attributed to “overvoting” in the South Bronx in 2010. The upshot is that a machine defect led to “phantom votes” on at least one machine used in the 2010 election, resulting in some candidates receiving more votes than they should have, and the choices of many more voters being voided when the machines detected both actual and phantom votes in the same contest. Now that the reports on how this happened are out, election officials must make sure that what happened in the Bronx in 2010 does not happen again in the future.

Voting machines record overvotes when they detect more than one candidate selected for a contest. In such cases, no vote is recorded for any candidate in the overvoted contest, regardless of the voter’s actual intent. The Brennan Center first uncovered a high number of overvotes in the South Bronx while reviewing documents produced for discovery in a litigation it brought against the State and City. It published its findings in Design Deficiencies and Lost Votes; the report notes that in some election districts up to 40% of the votes cast did not count.

The investigations conducted by the City, State and ES&S conclude that the unusually high overvote rates were not due to voter error, but rather a malfunction in the voting machine once it became heated after a couple hours of use.  The malfunction resulted in a distortion of the ballot images as read by the machines, causing blank ovals to appear darker than they should have. The machines registered these darker images as votes. These “phantom votes,” either led to some candidates getting extra votes (if no candidate had been chosen by a voter) or overvotes (if the voter had filled out a different oval for another candidate in the same contest). 

While the machines in New York provide voters with a warning when ballots cannot be read because of overvoting, the warning used complex election jargon that gave voters misleading cues about their options. Voters in these predominantly Hispanic South Bronx districts apparently chose to override this message without understanding the result was that their votes were not counted. Fortunately, as part of a settlement agreement reached with the State, New York’s voting machines will be reprogrammed before the presidential election in November with an overvote warning message that uses plain language that more clearly explains to voters if the machine is having problems reading their ballot.

We applaud the State and City Boards for conducting a thorough investigation of this matter. The State Board of Elections has forwarded their report to the U.S. Election Assistance Commission so that it can be distributed to other jurisdictions across the country using the ES&S DS-200.

However, more steps need to be taken to prevent lost votes in the future by detecting these problems when they arise. Election officials in New York should publish election results by precinct and report the number of overvotes in each contest. Rockland County already does this. The only reason the Brennan Center was able to discover this anomaly was by reviewing documents obtained in the course of litigation. Had we not done so, the problems in the South Bronx would have likely gone undetected and the machines would continue to be used election after election. It should also be noted that we did not receive complete data from New York City or from other jurisdictions in the state that use the DS-200. As a result, there is no way of knowing where else these kinds of problems may have happened.