The Brennan Center released a report this week detailing the tens of thousands of votes that were lost in New York because voting machines read their choices as “overvotes” – the invalid selections of more than the allowable number of candidates. Instead of returning the ballot, as is done in many other jurisdictions, the ballots were retained and the machine displayed a screen message using complex election jargon that gave voters misleading cues about their options. In the 2010 election, this confusing message led to as many as 20,000 lost votes in the governor’s contest alone and as many as 60,000 lost votes across all contests. The New York Times, Daily News, WNYC, and Politico have all done a great job covering the story.
The Brennan Center and others warned state election officialsabout the potential problems that would arise due to this confusing message. Represented by the Brennan Center, the NAACP New York State Conference, the National Coalition on Black Civic Participation, Families United for Racial and Economic Equality, and several individual plaintiffs filed a lawsuit against the New York State and New York City Boards of Elections in June of 2010 over the discriminatory impact on minority voters.
Our new study confirms that people of color were disproportionately likely to lose their vote. One percent of black and Hispanic voters in New York City did not have their votes for governor counted. In two predominantly Hispanic election districts in the South Bronx, nearly 40% of all votes were not counted in 2010; despite our repeated requests for an investigation into the overvotes in the South Bronx, we are not aware that one has been conducted. Our report also details problems with ballot design, finding that voters were more likely to cast an overvote in a race where candidates for the same office were displayed across two rows of the ballot (such as in the governor’s contest and in Sen. Kirsten Gillibrand’s U.S. Senate contest).We estimate that if no revisions are made to the overvote message, over 100,000 votes in New York could be lost in the 2012 election, when turnout will be much higher.
Fortunately, as detailed in the report, the State Board of Elections has agreed to adopt a better overvote warning in time for the 2012 election. But more steps can and should be taken to prevent lost votes. Election officials should make election results available by precinct; those results should report the number of overvotes in each contest on the ballot. When problems are discovered, election officials should be empowered required to investigate the reasons for high overvote rates. Ballots should be treated as public records to allow members of the public and voting experts to determine if ballots were in fact overvoted or simply recorded as overvotes because of a machine error. And states should reexamine their ballot design requirements and provide election administration officials with the guidance and flexibility they need to create voter-friendly ballots.
The recommendations are not specific to New York and can serve as models for jurisdictions across the nation to ensure that votes are counted as they were intended to be cast. The new report is available through the Brennan Center’s website.
The Brennan Center submitted testimony today before the New York City Council Committee on Governmental Operations. The Committee held a hearing on a resolution calling on the New York State Legislature and the Governor to pass legislation to improve paper ballots in New York State. One of the bills supported by the City Council’s resolution is the Voter Friendly Ballot Act of 2011 – the product of collaboration between the Brennan Center, the Usability Professionals Association, AIGA, and Assemblyman Brian Kavanagh.
Late last week we were happy to announce a big win for New York's minor parties and all of New York's voters, as we settled our lawsuit on behalf of three of New York's minor parties against the state Board of Elections. The lawsuit challenged a discriminatory vote-counting practice called "double voting."
Essentially, when a candidate ran for two different party lines for the same office, some voters were choosing to "double vote": that is, mark in the paper ballot with multiple votes for the same candidate on the different party lines. Under the old practice, New York's new voting machines accepted such ballots, counted these ballots as having a single vote for the candidate on the major party line (e.g. Democrat or Republican), and ignored the voter's attempt to vote on the minor party line as well. Now, voters will be notified of the problem and given opportunities to make sure their vote counts for the party of their choice.
Sam Roberts at the New York Times City Room reported:
Under a consent decree signed approved Thursday by Judge Jed S. Rakoff of United States District Court in Manhattan, the board agreed to reprogram voting machines before the 2012 general election to prevent people from voting more than once for the same candidate.
“They have to put a very bold message on the screen that says basically: ‘You voted incorrectly. Your candidate preference is clear, but your party preference is not,’ ” said Daniel Cantor, executive director of the Working Families Party. “Then you get to press a button and get your ballot back so you can fix it. Our view is that voters who mean to vote for us will want it to count.”
The improvements that will come from this settlement will help improve elections for all New Yorkers.
We've blogged before about serious concerns the Brennan Center and other New York voting rights groups had about policies adopted by the State Board for New York's new electronic voting machines, which will be used statewide for the first time this September. After extensive research, we have concluded that these policies violate New York State law and should have been precleared with the Department of Justice, pursuant to Section 5 of the Voting Rights Act. This apparently was not done. We have alerted the Department of Justice, and the New York State and City Boards to our findings. A copy of the letter detailing our findings, and a press release can be found here and here.
Last week's election presents
more evidence, if any was needed, of the potentially disenfranchising effects of
poor design. As a political blog in Seattle noted, a poorly-designed ballot probably caused as many as
40,000 King County voters to miss a property tax State Ballot Initiative.
As you can see from this picture of the ballot:
The contest was placed
immediately below the instructions and to the left of all other contests -- very easy for voters to miss. What can
election officials do to avoid these kinds of mistakes in the future? Well, one thing is to use design checklists, like those provided by the Design
for Democracy and the Brennan Center. But I'm not sure that in this case, either of those checklists would have
alerted officials in King County to the problem. (While both checklists emphasize
the importance of consistency in presentation -- and having all contests except
one to the right of the instructions is certainly inconsistent -- I'm
afraid this direction would have been too general to provide sufficient warning
for many officials).
And while it's easy, in retrospect, to say this
problem should have been obvious, I don't think that's fair. Such
problems are almost never obvious beforehand. Election officials and
others working on forms are usually on tight deadlines, trying to get
the ballots to fit into limited space and ensuring that everything and every
name is correct. Even if they are only focused on how a design
might confuse voters, they are often so familiar with the design that they're blind to problems; for the very same reason that it's often so difficult to
spot one's own typos.
What probably would have alerted
officials to this problem ahead of time, and at little or no cost, would have
been a simple usability test: observing ten or fifteen King County citizens as
they "voted" on the ballot before the design was
finalized. This solution is simple, easy and cheap. The Usability Professionals
Association has a great explanation of how it's done.
If county officials watched a dozen people fill out
the ballot, at least a couple might have accidentally skipped the
ballot initiative. And, with that, officials would have been alerted to
the fact that their ballot contained a serious flaw.
The ballot eventually got it's
usability test, of course...but on Election Day. And
approximately 40,000 voters showed -- a little too late -- that this particular ballot
design failed.
Jackson Pollock? Perhaps Brice Marden? Afraid not. Unfortunately, those frenetic abstract gestures to the right were made by a Minnesota voter on a ballot rather than a canvas (click image to see ballot).
The Senate contest between Al Franken (D) and Norm Coleman (R) is still undecided, as the recount has now stretched well into holiday shopping season (the widget posted below should provide the most up-to-date tally). Interestingly, a number of media outlets have posted features on their sites that allow anyone to examine all the contested ballots. The first time I scrolled through the Star Tribunes' version, "Ballot Challenge," one thing quickly became clear: usability, the ease at which voters are able to correctly navigate and mark a ballot, appeared to be a significant issue for some voters. As a result, there's a chance their vote won't count in a race that could very well be decided by a handful of votes.
This morning, I turned to Dana Chisnell, a researcher who studies "how people interact with technology specializing in design for older adults, ballot usability, and plain language," for insight as to why some of these challenged ballots have been miss marked, mangled, and must now go before a canvassing board to be interpreted. The analysis below is by no means exhaustive, but merely a sampling of her genius. For more on her work, click here to read her postings on the Usability Professional's Association blog.
The ballots we discussed can be found on the Minnesota Public Radios feature, here. Click on the ballots to see them larger.
A couple of weeks ago, I blogged about some counties in Ohio that split the presidential contest into two columns on their paper ballots. We noted that this kind of split has caused problems for voters in the past.
More recently we've found out that some counties that use touch screen machines have split the presidential contest onto two pages -- meaning on the first screen voters might see Barr and Nader, and on the second screen McCain and Obama. For reasons explained in a memo we sent to county officials, that's likely to confuse and frustrate a fair number of voters. So far we have confirmed that Portage (home of Kent State), Hancock, and Wayne Counties have split the presidential contest this way. Others like Montgomery, Stark, and Butler Counties have managed to get the presidential contest (eight candidates and a write-in slot) onto one screen.
We've asked officials to change this ballot to make sure all candidates for the same office are on the same screen -- or, short of that, to make sure there is good public education at the polls to reduce voter confusion and frustration.
Some of you may have read about North Carolina's very confusing straight-party voting rules and ballots. If history is any guide, this has the potential to have a huge impact on the outcome of the presidential contest in North Carolina.
North Carolina, like 14 other states, has an option on the ballot that allows voters to cast a straight-party ballot with one mark. Straight-party voting allows voters to vote quickly and easily, and often results in more votes for down-ballot contests. But it presents some serious usability challenges for voters. And In North Carolina, straight-party voting is particularly tricky: a straight-party vote (counter-intuitively) does not include a vote for president—voters must make a separate mark under the presidential contest (this is in contrast to other states, like Texas, where attempting to vote for President after choosing straight party voting could nullify one's vote).
Perhaps not surprisingly, in the last few elections, North Carolina has had an unusually high number of voters who did not cast votes in the presidential election. Dr. Justin Moore of Duke University estimates 3.15% of voters did not have votes recorded for President in 2000, and 2.57% did not have votes cast for president in 2004.* As we noted in our recent report Better Ballots, having more than 1% of voters fail to cast a vote in the presidential contest is unusual and often an indication of a ballot design flaw or other problem.
Permalink