Friday, January 15, 2010

Concerns About New York's New Voting Machines

Study argues that new voting systems purchased in New York could lead to significantly higher rates of spoiled ballots.

As we all know, New York State and City are expected to replace our long used lever machines with new optical scan voting systems, in time for this year's statewide primary. With optical scan systems, voters fill in a bubble next to their choices with a pen, much the way they'd fill out a lottery ticket or SAT exam.

At least one of the two systems, the ES&S DS200 (to be used in New York City, among other jurisdictions) has an unusual feature: unlike most optical scan systems, if a voter makes a mistake (voting for two candidates when she was only entitled to vote for one, for instance), the machine doesn't "spit" the ballot out and direct the voter to correct her mistake. Instead, it keeps the ballot and gives the voter the opportunity to either cast the ballot as is (in which case, the mistake will remain, and the voter's vote in that contest not counted) or request the ballot back, to make a correction.

We can imagine reasons why election officials may want systems that keep the ballot rather than automatically sending them back to the voter for correction. Automatically sending the ballot back would likely slow the process down (everytime a voter made a mistake, the machine would send back the ballot, the voter would have to review, and all the while other people would be waiting on line), and there are legitimate concerns about voter privacy (will other voters or pollworkers see the filled out ballot when it comes back out of the machine?).

But in a state like New York, where voters are used to machines that won't allow them to overvote (it is impossible to vote for more candidates than you are allowed to vote for on lever machines), where there is fusion voting (the same candidate listed twice or three times under different parties in the same contest), and there are very often contests where voters can choose up to two or three or four candidates, there is reason to be concerned that a signficant number of people will overvote ballots, and as a result, won't have their votes counted. This is particularly worrisome because, as a Brennan Center analysis has shown, New York's current requirements for what must be on a paper ballot makes them more confusing than necessary to use.

A recent study out of Florida, confirms that this is a potentially serious concern. The authors of this study argue that counties which used the ES&S DS200 had an overvote rate on Election Day 2008 that was as much as 18 times (!) that of the systems used in other Florida counties (if a ballot is overvoted, it will not be counted). They have also told me in an interview that they believe the Dominion Imagecast (the other system to be used in New York State) has the same feature and will result in similarly high overvote rates.

Paul Malischke, a voting rights advocate in Wisconsin, tells me that his state conditioned its purchase of the DS200 on a promise from ES&S that the machine be reconfigured to immediately return a ballot to the voter if it detects an overvote.

It's not clear to me that this is possible in New York, and no doubt, given their privacy and voting time concerns, many election officials would be opposed to such a reconfiguration. But at the very least, the State Board and legislature should be looking at this issue and figuring a way to address it.

In this letter, the Brennan Center, Usability Professionals Association, Center for Plain Language and Design for Democracy, urge the State Board to conduct usability testing to figure out how best to educate voters and poll workers about this potential problem, and to figure out what changes to the state's ballot layout and design requirements would make it less likely that people will overvote. We have yet to receive any response.

In a close election, as Florida and a few other states could tell New York, large numbers of overvotes can be an election nightmare, resulting in recount litigation, and shaking the faith of the public in their voting systems.

The State Board and State legislature should be taking action, soon, to address this problem. It should be addressed before these machines are deployed statewide, not after a close election, when the problem could lead to a post-election meltdown.

2 comments:

Anonymous said...

It’s great that this important issue is being discussed. Regarding this comment in the blog:

“Automatically sending the ballot back would likely slow the process down (everytime a voter made a mistake, the machine would send back the ballot, the voter would have to review, and all the while other people would be waiting on line),”

Actually, the fastest way to handle the overvote situation is to send it back to the voter immediately. Then that voter can step aside and be counseled by a pollworker, while other voters can use the tabulator. This method has worked well in Wisconsin for over a decade with the Optech Eagle optical scan tabulator.

The way the DS200 is configured for New York, when it detects an overvote, the ballot is held by the machine. No one else can insert his or her ballot into the tabulator while the voter is deciphering the error message. If there are other people waiting behind the voter, there will be pressure for the voter to make a hurried decision without understanding the consequences. Also, the voter must make this decision before the voter has a chance to review the ballot, since it is out of view in the top of the machine! This was detailed in the superb report by Mary Garber of the Florida Fair Elections Center on page 13.
http://www.ffec.org/documents/Invalid_Vote_Report_Revised_23June2009.pdf

The Florida report was based upon 2008 data from the Florida Secretary of State’s office, at http://election.dos.state.fl.us/reports/ under the heading “Legislative Report on General Election Overvotes and Undervotes”. This is a great quality control tool and all states should issue a similar report.

The U.S Voluntary Voting System Guidelines should be changed to require optimum handling of overvotes. See my letter at http://vote.nist.gov/ecposstatements/Fair-Elections-Wisconsin.pdf

More information on overvotes with optical scan tabulators is at
http://www.fairelectionswi.com/Overvote/Overvote.html

Paul Malischke
malischke@yahoo.com
Madison, WI

Anonymous said...

For heaven's sake ... at some point doesn't the user have _some_ responsibility?