A

B C

As shown in Figure 4, in the first study, when presented with the control interface, 63 percent of the time participants selected the checkbox when choosing their candidate. Thirty-one percent of the time participants selected the button, and only 6 percent of the time participants selected the candidate’s name. However, when the checkbox was removed in the experiment interface, 61 percent of the time participants selected the candidate’s name, followed by participants selecting the button itself 39 percent of the time.

In the second round of studies, 86 percent of the time participants selected the checkbox when it was provided as an option. When the checkbox was removed, 84 percent of the time participants chose the candidate’s name.

In total, 55 people participated in the studies. Out of the 156 recorded touches in the control studies, 72 percent touched the checkbox, 6 percent touched the candidate’s name, and 22 percent touched the button. Out of the 153 recorded touches in the experiment studies, 31 percent touched the button, and 69 percent touched the candidate’s name.

Interestingly, regardless of the presence or absence of a checkbox, a percentage of the time ( 5 to 6 percent) study participants elected to touch the candidate’s name as opposed to the supplied checkbox. In a real election, if 5 percent of the voters touched the candidate’s name and some of these selections resulted in unintentional voter selection, this could be more than enough votes to change the outcome of a close election, given the significant number of people who do not verify their ballot choices [ 2, 3].

The findings suggest that ballot
design may influence how voters
interact with controls on an elec-
tronic ballot. In both studies, more
voters selected the checkbox when
it was present, followed by select-
ing the button itself, with an even
smaller number touching the can-
didate’s name. However, when the
checkbox was removed, more vot-
ers selected the candidate’s name
than the button.

 

• Figure 5. Design recommendations for electronic ballot buttons. a) Design option 1 before selection b) Design option 1 after selection c) Design option 2 before selection.

November + December 2012

 

interactions

Improving Design, Minimizing Error

The results of this study suggest that the use or absence of checkboxes may influence user interactions with an electronic ballot. However, some users still elect to select a candidate’s name. We therefore suggest that candidates’ names be centrally located on buttons, both vertically and horizontally to minimize unintentional voter selections. If checkboxes or other images are used, they should be vertically centered as well.

Due to limited screen real estate, ballot designers may minimize gap space between buttons to accommodate all candidates in a contest. This design choice provides the opportunity for finger roll, as a voter’s finger may roll up, down, or sideways depending on the positioning of the touchscreen. The recommendations presented here are based on the use of sufficiently large buttons within the design. Sufficiently large buttons are buttons that are large enough to provide space for a voter’s finger to roll on a selection without rolling into the area of the adjacent or surrounding buttons. Illustrations of recommended designs are presented in Figure 5.

In design option 1, the text is horizontally and vertically centered within each button (Figure 5a). Centering text in this way reduces unintentional voter error