WORKSHOP ON DEVELOPING A RESEARCH AGENDA FOR ELECTRONIC
VOTING TECHNOLOGIES
September 17-18, 2004
Convened by the American Association for the Advancement of Science
Main | Program | Participants | Synopses
Paul S. Herrnson and Michael Traugott
A SYNOPSIS OF PRESSING RESEARCH QUESTIONS PERTAINING TO ELECTRONIC VOTING TECHNOLOGIES
A major lesson derived from the 2000 presidential election is that the manner in which voters cast their ballots is important. Voting technology and ballot design can influence election outcomes, affect how voters feel about their ability to exercise their right to vote, and influence voters' willingness to accept the results of an election as legitimate. The 2000 election also showed that most polling places nationwide employ outdated technology, including unreliable punch-card ballots and mechanical lever machines; only a third of the electorate used modern computerized technology, such as optical scanning systems or direct recording electronic (DRE) systems with "ATM-style" touch-screen voting.
States have responded to the problems associated with the 2000 elections by commissioning studies, revamping election administration, redesigning ballots, and, in some cases, by investing heavily in new voting equipment. These efforts have been supported by passage of the Help America Vote Act (HAVA) through which the federal government is prepared to help financially in this transition to new technology. Although most of the attention of researchers and the media have focused on issues related to ballot security, the Florida experiences of 2000 and 2002 indicates that the major problems occurring thus far have involved the interface between voters and voting systems (including different voting machines and ballot designs).
For this reason, we believe that human-computer interaction as it relates to voting systems constitutes a pressing area for further research. We are currently part of an interdisciplinary team that draws on the expertise of political scientists, human factors psychologists, design engineers, computer scientists, and election officials to conduct a multifaceted assessment of voting systems. We believe that some general subjects for research regarding electronic voting involve determining which voting machines, ballot designs, and combinations of the two meet the following criteria:
1. Result in the most accuracy in voting.
2. Result in the smallest number of spoiled ballots and residual votes.
3. Most faithfully record voter intent.
4. Are most easy to use.
5. Produce the greatest citizen satisfaction with the voting experience.
6. Provide the greatest flexibility in terms of allowing election administrators
to meet the legal requirements of their states or localities, including providing
voters with the opportunity to cast a straight-ticket vote, vote in elections
that allow voters to cast ballots for more than one candidate, cast ballots
for one party (as is done in closed primaries) or for both parties (as is done
in open primaries and general elections)
7. Provide the greatest flexibility in terms of allowing voters to easily cast
votes for all candidates or ballot issues (these traditionally appear at inconvenient
places including the end or the top of the ballot).
In addition to these general subjects, further research questions should focus on specific interface designs, population groups, and combinations thereof, including:
1. Whether voters make fewer mistakes with, are the most comfortable using,
and have the most confidence in voting machines that rely on ATM-style touch
screens, other pressure activated switches, knobs and buttons, or paper ballots
used in conjunction with computerized devises that scan for mistakes.
2. Whether voters make fewer mistakes with, are the most comfortable using,
and have the most confidence in ballots that are organized by office, party,
or include a mechanism for straight-party voting.
3. Which combinations of voting machines and ballots produce the fewest mistakes,
result in the greatest levels of comfort, and encourage the most confidence
among voters.
4. Whether voters who have less experience with computers, including the elderly
and less-well educated, will make more mistakes, be less comfortable using,
and have less confidence in electronic voting systems than will other voters.
5. Whether voters for whom English is a second language will make more mistakes,
be less comfortable using, and have less confidence in electronic voting systems
than will other voters.
One counter-intuitive hypothesis worthy of testing is that voters who report being extremely comfortable using a particular voting system are more likely to make errors when voting than those who are less comfortable. The underlying idea of this hypothesis is to dispel the notion that voter comfort is synonymous with accuracy. It may be that voting systems that require significant, but not overwhelming, mental or physical effort require more focus than do others, and therefore result in fewer mistakes.
Our study explores these questions using four different research approaches:
1. Review by human-computer interaction experts.
2. Usability experiments conducted under laboratory conditions.
3. Usability experiments conducted under field conditions similar to those faced
by voters on election day.
4. Natural experiments involving cross-sectional time-series analysis of aggregate
data comprising votes cast on different voting systems and in different elections.
(It is important that appropriate data to support such investigation is collected
during the election process. This implies individual ballot images so that studies
of behavior such as straight and split ticket voting, roll off and drop off
can be assessed when election administration and technology changes, as well
as small unit (precinct-level) data and higher levels of aggregation such as
cities and counties. Regular survey data collection about citizens' attitudes
about the electoral process and confidence in it are also crucial.)