Zero to UX: Broward County Ballot Design
Overview
​
The 2018 Broward County, Florida election saw a significant undervote in the Senate race. Experts state that the design of the ballot was to blame since the placement of the contest was below the instructions.
​
Objectives:
-
Compare Control and Treatment ballot designs and see which yields fewer instances of undervoting.
-
Identify usability issues with either ballot design.​
​
​
​
My RolE
​
I worked as a UX researcher on this project. Some of my responsibilities included :
-
Planning, defining scope and strategy
-
Arranging and conducting a usability test and semi-structured interview
-
Analyzing data
Process
​
Planning
​
Methodology
-
Desk research: app reviews, market research, sales, support team
-
1-1 in-person moderated usability test
-
Semi-structured interview
​
Research Questions
-
Which ballot yields a lower instance of accidental undervoting? Why?
-
Which do users think is easier to fill out? Why?
-
What are the opportunities for improvement in the ballot designs?
​
Study Limitations
The study was conducted over video calls due to the COVID19 Pandemic. This environment may not reflect natural human behaviour, but was a reflection of how participants were living at that time.
​
Usability Test
​
I was studying how people viewed the ballots, so a usability test was used in conjunction with a semi-structured interview to gather detailed information about the users' thoughts and feelings about the ballots.
​
I used the original ballot (control) and a version of the ballot with a slightly different, more spaced out design (treatment).
​
Participants completed both ballots as they would usually during a vote. Participants were asked to 'think aloud' during the task. They then provided feedback. To avoid bias, 3 participants viewed ballot A first, 3 viewed ballot B first.
​
Semi Structured Interview
​
Open questions were asked as warm ups before and prompts during the usability test to gather feedback, as well as some post-test clarification questions. Since participants respond in different ways, and spoke about the ballots individually, different questions were utilized.
Example questions:
What is your experience of voting in the past?
Can you explain how you feel about that?
What can you tell me about completing ballot A?
Participants:
-
6 Participants
-
4 Female, 2 Male
-
Age range 18-65
-
All participants are registered voters.
​
Recruitment:
Recruited via social media, using a screener questionnaire to ensure I captured the right users.
​
Findings
​
Key Insights
​
-
The Treatment design yielded significantly fewer instances of undervoting for the Senate race compared to the Control design.
-
Participants found the Treatment design to have a clearer distinction between the instructions and voting section.
-
Participants found the Control design to be visually cluttered as the voting section was so close to the instructions section.
​
​
Opportunities
​
-
Distinguish the voting section from the instructions.
​​
-
Consider using a clear indication to show where the voting section begins.
​
​
Further Findings
​
Some participants felt the ovals to complete would have been better if they were circular as they would be easier to complete.
Recommendation: Consider the shape of the check boxes and if possible adapt them for ease of use.
“ I want the lowest amount of effort to fill this in”
Participant #2
“Seems straightforward if you have good literacy but if you don’t then that first bit could be confusing”
Participant #6
“I prefer this one. It’s better cos the sections are all blocked together”
Participant #3 re: Ballot B
“Also why is it an oval? Why is it not a circle? Circles are easier to fill in for me. What an awkward shape”.
Participant #3
Reflection and Learning
​
During this project I found that by listening to users work through a task in real-time, it was easy to observe their pain points and the things they liked about each ballot. It was also clear that if the design of a form, or product has not been tested by users, it can reveal itself to be confusing, and also have a detrimental effect at a later date. By testing as early as possible with users, we can ensure the product suits the needs of the folks that use it.
​