|Home What we do Storytelling for User Experience Articles and downloads About us|
by Whitney Quesenbery
Unfortunately, we know very little about the user experience of voting and what affects it. In the absence of substantive research to report on, this segment of the workshop will focus on what we do know about usability in official public access systems and what we have learned from other Internet applications. These issues will be discussed in an open forum, exploring how the legal, technical and security requirements for a voting system interact with usability issues.
The Usability Challenge of Voting
Voting may be one of the most difficult usability challenges because it is a task completed by virtually anyone, it is done infrequently, it is never exactly the same because the actual ballot differs for each election, and privacy requirements make it difficult for voters to seek help in using the voting system.
We have taken few opportunities to study the usability of voting systems. This is, in part, because the system was generally regarded as working. Errors rarely made a statistical difference in the election results and the task itself is not seen as difficult. In addition, the privacy issues which surround an election can make it difficult to construct meaningful situations for observation. It is not impossible, however. Susan Roth King, the author of one of the few papers on voting and usability, reports however that the people in her studies were "willing and able to provide constructive and valuable feedback" - if someone was just willing to ask.
What Is Usability?
Usability describes a quality of a product, whether that product is a web site, a software tool, a physical object or some combination. The ISO standard on usability (ISO 9241:1998) defines it as:
This definition reflects the origins of usability practice in the ergonomics
of work. Except for the relatively weak final term, "satisfaction,"
it does not address the emotional aspects of the user experience, focusing
instead on measurable metrics such as time on task or error rates. Colloquial
definitions tended to use softer terms, such as "user friendly"
or "easy to use" as a way of shifting attention to the user
and their perception of the experience, especially in regard to initial
barriers to effective use. Professional discussions have suggested several
means of broadening the definition to take into account different types
of products. I summarized these as "the five E's" or the five
dimensions of usability. They are:
Each of these dimensions are important in varying degrees for all products. They are of interest as a design tool because a clear statement of how they apply to the domain acts as a guide for decision making, especially when trade-offs are needed.
For example, in a voting system, an analysis of the dimensions of usability might be:
In 1990 (and revised in 1994), Jakob Nielsen published a list of heuristics for usability. These guidelines are important because they are not specific to any one application or context, but provide a common base for discussing the usability of a variety of products. They can be refined into more specific guidelines for voting systems, and can provide a means for comparing the usability strengths and weaknesses of different voting systems. The most important guildeins for a voting interface are:
Why is Usability Important
Many people I have talked to have wondered why usability is important in a task as simple as voting. One answer is that it is only deceptively simple: the CalTech/MIT Voting Project concluded that there are opportunities for error at every step of the process, including voter authentication, ballot preparation by voters, verification and deposit, tabulation.
Part of the answer lies in the question of what error rate we are willing to tolerate in each step. Perhaps the goal should be a system with no user errors, rather than an acceptance of designs that disenfranchise a percentage of the electorate.
A saying in the usability community is that "to the user, the interface is the product." This means that the presentation and interaction layers are just as important in voter perception (and acceptance) as the underlying technology. Usability is important because it provides guidance as to how standards should be met, and can be the link between standards and operational guidelines. Too often, systems are designed without a clear model of the user experience in mind. The result is an interface that is patched together on top of a collection of sub-systems, rather than a system in which all of the elements come together to create a seamless, usable experience.
Topics for Discussion on Internet Voting Usability
Each participant in the workshop was asked to suggest topics for discussion by the workshop.
How can we guarantee an appropriate level of usability in the initiation and authentication stage of the voting process? What kind of design process and usability testing is needed?
Even when the actual application is easy to use, the installation and
individual system setup can still cause usability problems. For example,
in the Voting Over the Internet Project, "Of the 157 digital certificates
issued to volunteer voters, nearly half (75) had to be replaced. Approximately
one-third of these digital certificates were reissued for expired documents;
the other two-thirds were reissued because of forgotten passwords."
With just 91 people registered to vote in the project, and 84 votes cast,
the project help desk logged 71 support calls.
What are the requirements for readability of the on-screen ballot
to ensure that voters of all ages and physical abilities can read the
The current resolution of a typical computer screen is vastly inferior to paper (72 dpi vs. 2400 dpi). As part of the design research for an online testing service, readability tests were conducted at a range of screen resolutions from 800x600 to 1600x1200. The results showed dramatically fewer comprehension errors at the higher resolution. The California Task Force recommends that "No contest, either for an office or a proposition, should be split across two pages." With large numbers of candidates in some races, screens may be very crowded.
How can we determine the differences in effectiveness (a dimension of usability: how accurately the vote is cast) between full-face and paged ballots?
The use of a computer screen most likely implies a design that uses a
aged ballot rather than a full-face ballot because of the relatively low
screen resolution. Although several current voting technologies used paged
systems, there may be added usability issues because of the addition of
navigational elements to the voting interaction, as well as making it
more difficult for the voter to see, at a glance, all of the votes. The
California Task Force recommends "Multi-page ballots should be easily
navigable by voters, with no way to get lost or leave the balloting process
What usability testing is needed for the general ballot format; and what for each specific ballot, and what is the right methodology for this testing.
In general, ballot designs are examined for accuracy, but not for usability - quality tested, rather than usability tested. After Florida's 2000 election, the Palm Beach County Supervisor of Elections, Teresa LaPore said, "I was trying to make the print bigger so elderly people in Palm Beach County can read it. We sent out sample ballots to all registered voters, and no one said a word."
The California reports suggests that, "It will be essentially impossible (to) offer phone support to help voters who have technical difficulties voting from institutional computers. Vendors should prepare and counties will need to distribute an extensive instruction sheet for voters." We have evidence from the events in the 2000 election in Florida that voters do not ask for assistance at the polls. Caroline Jarrett, a usability consultant for the Inland Revenue, suggests that people "don't realize that they have a right to complain if they can't understand official forms" and that they will only complain when "the effort of making a complaint has some sensible relationship to the chance of the complaint making a difference."
For more information...
This paper was presented at the Georgia
Tech Research Institute Internet Voting Workshop, held November 13-14,