From The Ergoweb® Learning Center

Ergonomics and Voting Systems: Voter Interface, Voting Machines at Odds in 2008

No one will ever describe the 2008 United States election as ho hum. For the first time, a black American is running for the highest job in the land and a woman for the second highest. At the same time, the country’s economic turmoil is driving Americans to the polls in record numbers. From an ergonomics and human factors standpoint, the big numbers are unfortunate.
The record-setting turnout promises to exacerbate still-present turmoil in the nation’s voting system—a mish-mash of procedures, rules and regulations put together by individual states and counties. A human factors researcher who has been exploring kinks in electoral systems finds little to suggest the country’s election officials have learned from the "hanging chad" debacle in Broward County, Florida, in 2000. Those problems shook a nation’s faith in one of its most prized democratic institutions – elections.
In an October interview with the Ergonomics Report™, Tiffany Jastrzembski, Ph.D., a cognitive research psychologist at the Air Force Research Laboratory, Mesa, Arizona, outlined some of the issues, and suggested ways certain kinks could be prevented.
Voter Interface Losing Out

She isn’t optimistic about voters’ ballot-casting experiences in 2008, and explained that states and counties seem "overly concerned with the type of machines used, rather than actually making the voter interface easier to use."
Several widely reported accounts of problems during the first wave of early voting in Virginia, Tennessee and Texas make her point. In Tennessee, vote flipping had been reported. The term describes a situation where the voter selects one candidate or party and the machine registers the vote for another candidate or party. In a report in the Decatur County Herald, Election Commissioner Rick Box blamed an area of the screen where the buttons for "President" are extremely close together. "The way the machine is set up, when you are standing in front of it and seeing it at a certain angle, it looks like you are touching the middle [of the button] when you are actually touching the line above it," he explained.
Bloodied by the publicity over its 2000 election mess, Florida replaced the flawed punchcard system with direct recording electronic devices in time for the 2004 election. The changeover introduced a new set of kinks during the 2004 vote. As a result, the state mandated another change—to optical scanners—for all Florida precincts in 2008. Dr. Jastrzembski described the latest change, in so many words, as the wellspring of other kinks.

The state argued that the optical scanner interface would be simpler for voters to use and would provide a paper trail in case of a recount, she explained. The paper trail is a nice feature of the technology, and one that is particularly attractive to Florida after its infamous 2000 election, but she faults the argument that optical scanners will be easier for voters to use. In fact, she said, recent analyses of residual rates show that direct electronic recording devices actually result in fewer errors than optical scanners (Stewart, 2006). The research suggests this may be because optical scanners do not provide notification to a voter if they have failed to submit a vote for a particular candidate. On an electronic voting machine ballot, she said, the voter typically will have to select the candidate they want to vote for, and the machine will give them the opportunity to confirm the choice or go back to change it.

A recent news report makes the distinction clear. Conducting an exit poll at a Florida polling station, a reporter asked a middle-aged man what his experience was like in the booth. “Well, I think I completely missed a race I intended to vote in,” the man responded.  “I guess it just blended in with the rest of the form.” Dr. Jastrzembski said his response illustrates the fact that with optical scanners, undervoting may prove to be a problem, since the voting device will not alert the voter that they missed casting a vote for a particular election. She warned that the voter must exercise caution before casting a ballot that is not fully completed.
Dr. Jastrzembski is also concerned that optical scanners require fine manual dexterity, which is in short supply in some populations. "When a voter wants to vote for a candidate, they need to actually fill in the space with a single line, to complete the arrow which corresponds to the candidate they select." Unfortunately, she added, constructing that single line may be a taxing issue for a good portion of the voting population—for example, older adults suffering from arthritis or visual impairments.  

This design weakness feeds a nationwide lack of voter confidence that a vote will actually be cast as it was intended, she added. Early voters have already raised concerns that they must be incredibly diligent to ensure  the machine will be able to read their ballots.

A recent CNN report on early voting in Broward County manifested the design weaknesses in the ballot—and introduced another of its consequences. An elderly voter was concerned her line would not be straight enough next to the candidate she wanted to vote for, and that her vote would be spoiled. "She spent an inordinate amount of time making sure that her line was just right" so her vote would be accurately counted, said Dr. Jastrzembski. Given the limited number of voting booths and machines to tally votes, such focused attention to detail means fewer voters passing through a polling station—and longer and more sluggish lines for them to endure. 

Human Factors Oversights

She noted that half of Florida’s voters are going to be voting on this entirely new device, requiring them to interact with a new system. "Changing the voting machine itself does not necessarily imply that the ballot is going to be easier to cast, and so this is one of the biggest problems I foresee in this election—election officials have focused on the machine that voters will interact with, but have largely ignored the importance of good ballot design."  She pointed out that with respect to older voters in particular, ballot designers should be concerned with issues of clear instructions, font size, contrast, clear distinctions between races, and clear language regarding how many selections the voter may choose for each race to avoid spoiling the vote by over-voting. Ballot language is often confusing, she noted, as in the instruction: “Vote for not more than 2”). 
As evidence of some of the human factors oversights in this election, the psychologist explained that some Florida election officials have reported ballots coming in with the candidates’ names circled rather than selected by a corresponding line. The circled names spoil the ballot for automated readers, and require time-consuming and perhaps error-prone human intervention. Breakdowns are also slowing the pace. "Machines were malfunctioning, and could not read or count the ballots. … Rather, they would spit the ballot back out when the voter … attempted to put it in. Election officials were simply forced to set them aside and place them in a box to be read later." This problem also sets back a voter’s trust that the ballot would ever be counted correctly, she added.
The Potential for Long, Sluggish Voter Lines
All of these factors will have repercussions for how quickly the voters can move through the polls, and how long the wait times are going to be, Dr. Jastrzembski explained. "In the spirit of the human factors notion that milliseconds matter—small design choices … can lead to tiny differences in the completion times at the individual level, but when summed over thousands of voters in a single election day, can lead to major differences in the amount of voters that can move through the polls in each precinct."
Reports from states and counties that set up early voting bear out the milliseconds precept—eight hours in Georgia, two hours in parts of Florida, 90 minutes in Houston and Chicago and an hour in Charlotte, North Carolina. "This is the early voting stage, too," Dr. Jastrzembski, said, "so Heavens forbid what it’s going to be like come election day [with] voter turnout that’s probably going to be one of our largest ever."
The close and contested result in 2000 and electoral turmoil in Florida prompted increased scrutiny of the whole system, and many other problems were found. A commission headed by the former Presidents Jimmy Carter and Gerald Ford was set up. It recommended a major overhaul of the voting system. Legislators and election officials addressed the ailments with guidelines and a rush to new technology. The National Institute of Standards and Technology (NIST) issued guidelines and the Help America Vote Act (HAVA) were passed in 2002.

Dr. Jastrzembski sees value in both measures. The Act provided funding to update voting equipment to states and counties to help ensure voting accessibility to all voters, she explained, including those with disabilities like blindness and deafness; problems associated with aging; life visual deterioration or limited manual dexterity. She described the NIST guidelines and ACT as "great," but pointed out that they are not mandatory for states and counties. "They haven’t really taken off the ground and made any significant or substantial changes to how we are doing elections still today."

For something as important as this 2008 election, she said, you would think that system information and ergonomics issues would be hammered out well before November 4. "The kinks still aren’t out," she noted, describing herself as a little confused about why it is taking so long for the changes to happen.
In a recent article about election issues from the Human Factors and Ergonomics Society (HFES) community, she complained that insufficient objective usability testing has been done for all voting technologies and ballot designs. "Although counties seeking to upgrade systems often move towards electronic voting machines, there are plenty of states and counties that still use other technologies – like levers, punch-cards, optimal scans or even paper ballots. Usability testing needs to take place to make recommendations for all of these systems."

She also sees the lack of standardization and regulation of voting systems as evidence that existing recommendations for preventing kinks might not have reached counties making design decisions—or that "they simply will not be employed." Some counties that upgraded to electronic voting machines in 2004 have reverted to devices used prior to that election, she noted, due to voters’ problems with poor interface designs. "I do not believe that scrapping the technology is the right course of action. Rather, better design of the interface would be a better, less-costly way to go."
In the commentary she also argues that designers and engineers need to produce more usable systems at the very earliest stages of design—by analyzing the system using performance-based measures and by applying validated models of human processing to critical pathway analyses of competing designs.
In a recent human factors study, Dr. Jastrzembski and Neil Charness, Ph.D., a professor of psychology in the Psychology Department at Florida State University, tested ballot and machine usability with a particular focus on older voters. Because of reduced vision and motor control, seniors tend to have more problems using computers, according to the researchers, especially under time pressure. The researchers adopted a gerontological approach, which implies that when systems are made easier for older people to use, younger users’ performance also improves. The results of their study were published in the fall 2007 issue of Ergonomics in Design, and were widely reported in the nation’s newspapers.

The researchers found the pure touchscreen format with one ballot per screen produced the most accurate results, but the pure touchscreen with full ballot on a single screen showed the fastest completion times. The findings exposed a trade-off between accuracy and speed: even a small percentage of errors could potentially result in hundreds of thousands of miscast ballots, which must be weighed against the need to reduce waiting times at the polls.
The Role for the Ergonomics and Human Factors Community

Dr. Jastrzembski told the Ergonomics Report™ that she believes something needs to be done on a national level to make sure that we can accommodate all voters on election day: “The United States is the only industrialized Western democracy that doesn’t govern its own national elections. Rather it gives the power to decide how elections are run to the states and counties. As a result, we have this tremendous variability from state to state and county to county, even in the laws, and procedures and the number of voting machines that are necessary to get a number of eligible voters through a voting precinct, [and in] the ballot design itself. So I really believe that some national standardization or regulation of state and county election practices is going to be critical in this day and age."
She advocates more involvement by the human factors community, which "can and should have a large say in helping congressmen decide exactly what regulations, and standards and mandates need to be set forth." She pointed to the human factors community’s valuable involvement in other applied domains, such as ATM design and web browsing. "I believe that this community really can have an impact and lead the way in helping America vote in the 21st Century—but I think it’s going to be up to us to take the initiative and actually get our good science and good information out to the lawmakers in Washington to make some sweeping changes in the way our elections are conducted."

Decatur County Herald/Salon
HFES unpublished commentary 2008
Stewart, Charles (2006). Residual vote in the 2004 election. Election Law Journal, 5(2), 158-169
Jastrzembski, T., & Charness, N. (2007). What older adults can teach us about designing better ballots. Ergonomics in Design, 15, 6-11

This article originally appeared in The Ergonomics Report™ on 2008-10-29.