VOTER FLOW
UX | UI
A tablet app for improving long lines on Election Day
Project Overview
BACKGROUND
For a number of reasons, many Americans find long lines at the polls when they go to vote in person, particularly in popular elections. This happens, in part, because there are a number of issues that can prevent a voter from receiving their ballot right away when they get to the check-in counter, from needing to verify their address to being in the wrong precinct.
Often, a “flagged” voter then must be told to wait in another line to see a poll worker who has been trained to address these issues and vote using a provisional ballot. This often results in frustration for both the voter and the poll workers.
To address this problem, this app gives poll workers a series of questions to ask voters while they are waiting in line so that they can be directed to the “flagged voter” line as soon as possible.
UX Designer
UI Designer
80 Hours
4 weeks
End-to-End App
Student Project
Figma
FigJam
a11y Color Contrast Checker
Canva
Google Sheets Sync
TIMELINE
ROLES
TOOLS USED
PROJECT TYPE
MAIN TASKS
Create a flowchart that encompasses the main voter flags
Design interface to be used by new users in all kinds of weather and lighting conditions
PROBLEM STATEMENT
How can we make long lines on Election Day shorter and more pleasant for both voters and poll workers?
Specifically, how can we use the time people spend waiting to address any issues before they get to the check-in table?
The Process
1
2
3
4
5
6
7
1
RESEARCH THE USER
Research on Long Lines
Having researched Election Day lines as a law student and having spent a few years working in election administration, I started by reviewing the academic research on the topic.
The research showed that bottlenecks at the check-in station are both a major cause of long lines and one that local administrators have the power to minimize.
KEY TAKEAWAYS
Interactions between regular pollworkers and flagged voters create check-in bottlenecks.
There is little a local administrator can do to reduce bottlenecks at the voting booth, but they can reduce bottlenecks at the check-in table.
Getting flagged voters to the pollworker who is specially trained to help them before they get to the regular check-in table may mitigate check-in bottlenecks.
The check-in table is where many bottlenecks occur. From The Michigan Bureau of Elections Election Officials’ Manual.
Based on this research, I wanted to dive a bit more into what check-in looks like from the pollworker’s perspective.
Competitor Analysis
To learn if there were other possible sources of delay at the check-in table, I looked at the common ePollbook programs that pollworkers use to look up and check in voters.
Most current ePollbook software will show pollworkers a list of voters who meet search criteria and relies on the pollworker to select the correct voter each time.
KEY TAKEAWAYS
While making the voter search function more error-proof may not reduce line length, it will improve the overall experience for voters and reduce the administrative cost of fixing the errors.
Most ePollbook software programs do not help pollworkers avoid the error of checking in the wrong person.
Pollworkers often check in the wrong voter by selecting someone with a similar name.
Now that I had an idea of what check-in looks like for pollworkers, I felt I needed to learn more about pollworkers in general- who are they and what is their day like?
Research on Pollworkers
I specifically wanted to focus on the other challenges pollworkers face that could cause delays.
Pollworkers are likely to be:
retirement age
working very long hours
working in a variety of lighting and weather conditions
on their “first day” on the job
KEY TAKEAWAYS
Since pollworkers work infrequently and have a myriad of responsibilities, pollworker software must be intuitive- users should be able to learn it with no prior experience or training.
Any software for pollworkers must be designed around the needs of older users.
Most pollworkers are over the age of 61. Photo from NC needs more poll workers, but pandemic making them hard to find.
With this research, I could now start to define who this product is for and what it needs to do.
2
DEFINE THE PRODUCT
User Overview
With my research and my background in election administration, I started to map out who this product needs to serve.
PRIMARY USERS: THE POLLWORKERS
Success Looks Like:
Providing all voters with excellent service while accurately checking them in as quickly as possible.
Current Challenges:
Election day for a pollworker is long and exhausting, particularly so for the 56% who are over the age of 61.
Every pollworker is essentially working on their “first day” of the job because of the relative infrequency of elections.
SECONDARY USERS: ELECTION ADMINISTRATORS
Success Looks Like:
Only the chief poll official, who is trained to handle flagged voters, is handling voters who will need to vote using a provisional ballot or fill out a voter assistance form.
Current Challenges:
When voters who will need to vote provisionally or who need assistance to vote check in at the regular check-in table, they are met with pollworkers who are not trained to help them and then need to be redirected to the chief poll official.
BENEFICIARIES: THE VOTERS
Success Looks Like:
Getting an “I Voted” sticker as quickly as possible.
Leaving the polling location feeling respected and enfranchised.
Current Challenges:
Long lines at the polls can make voting too costly for many would-be voters.
Frequently changing voting laws and redistricting efforts confuses voters and results in voters showing up in the wrong polling places or without required identification.
KEY TAKEAWAYS
Getting flagged voters to the pollworker trained to help them will make the voting experience more pleasant for both them and the voters who don’t have flags.
While this product might be used primarily by pollworkers, it will need to allow administrators to edit and add to the content.
The “I Voted” sticker is more important than one might think. Photo by Phillip Goldsberry
Knowing who my users are meant I could now brainstorm solutions to their challenges.
I wanted to learn if the challenges faced by the users might have a common solution, so I used a problem statement brainstorm to generate ideas.
Problem Statement Brainstorm
Most of the “point of view” statements in my problem statement brainstorm focus on the voters’ experience, but the solutions to these problems often involve providing information or prompts to the pollworkers at the right times.
KEY TAKEAWAYS
This tool can seamlessly serve other purposes as well- voters can be asked proactively if they need assistance or accommodation.
If given the right tool, pollworkers can help flagged voters avoid the regular check-in table and get them help earlier in the voting process.
Many voters are unaware that they are entitled to accommodations for people with disabilities. Photo by Steve Rhodes
Now I could define the product as a dynamic questionnaire using the voter registration database that pollworkers can use to help direct voters to the correct places and get them the help they need while they’re standing in line.
3
ORGANIZE THE INFORMATION
Question Flowchart
Before I could build a prototype, I looked at the most common types of flags that would need to be addressed before a voter can receive a ballot.
I sorted these flags into a flowchart and found that there are common categories of voter flags across jurisdictions, and they need to be asked about in the correct order.
KEY TAKEAWAY
There are three categories of issues and a clear order in which they needed to be asked:
voter is in the wrong place
voter needs to provide information or sign an affidavit
voter needs assistance
What a voter can be flagged for varies by state. From The State of Michigan’s Electronic Pollbook Election Inspector’s Manual
Next, I needed to sketch out my ideas for how this questionnaire needs to work and what it needs to look like.
4
GUIDE THE INTERACTION
Wireframe Sketches
I wanted to quickly document all of the ideas I had generated throughout the process so far.
The overarching idea throughout my sketches is that the pollworker should only be faced with one question or action at a time.
KEY IDEAS
The questions need to be binary and about verifiable facts the voter either knows or can easily get the answer for.
“Don’t make [the pollworkers] think.” The job is tough enough as it is, the product needs to break down the voter questionnaire into the simplest steps possible.
Apologies to Steve Krug.
With these ideas, I wanted to research what I can do to make sure this product works in the context of our users on Election Day.
Designing for Context
I not only needed to design around the likely age group of users but also around the fact that they would likely be standing outside in cold weather.
Big buttons, big fonts, and “dark mode” were all used with the hope that they would help older eyes see in bright sunlight, and help shaking hands select the right choices.
KEY TAKEAWAYS
Primary text is 90px and all text meets AAA web accessibility contrast requirements.
The primary buttons are very large (160px high) to accommodate older hands that may be shaking in cold November weather.
All screens meet 7:1 color contrast ratio recommended for people over 65.
The a11y Color Contrast Checker was crucial for ensuring my screens meet web accessibility guidelines.
Knowing how the questions needed to be asked and what guidelines the screens needed to follow, I could now begin designing the interface.
5
DESIGN THE INTERFACE
Logo Design
First, I needed to design a logo that communicates that this product is for US government elections.
Most companies that serve US government elections use some variation of red white and blue, while companies used for other kinds of elections do not.
Logos for companies that make products for government elections.
Logos for companies that make products for other kinds of voting.
To quickly design a logo, I searched Canva’s logo collection using terms like “waves” and “flow,” and then adapted the one with the most positive user response to meet the needs of the product.
Generating logo ideas using Canva.
Iterating the idea with the most positive user response until I had a winner.
With a logo I could now build a prototype.
First Prototype
I wanted to start testing as soon as possible, so I built a prototype incorporating a question flow that would encompass most voters.
My overall goal with this prototype was to see if testers would understand what they would need to do or ask at each screen.
Next, I was ready to test, but not quite ready to test with with my older participants.
6
TEST THE ASSUMPTIONS
Testing with UX Designers
Before I subjected my retirement-aged testers to using this app outside in November, I wanted to uncover as many issues as possible with fellow UX designers first.
AFFINITY MAP
My eight UX testers not only gave me valuable feedback on whether they understood what was needed on each screen but also some ideas on how to design a more elegant and usable experience.
KEY TAKEAWAYS
Testers often did not notice that there were two people with the same name in the list of voters- they tended to select the one on top.
There was a lot of confusion as to what the flags and the terminology meant.
Users needed more guidance as to what to ask of a voter at each screen.
One of the most interesting parts of this project was talking about how elections are run in other countries with UX designers who are not from the US. In Nigeria, where Tobi lives, a voter’s identity is verified with a thumbprint scanner.
With this feedback, I could now make some crucial iterations before testing with my target users.
Iteration #1:
Minimizing Error
The first fix was to design a solution to the problem of pollworkers mistakenly selecting voters with similar names- a problem that I witnessed during the first round of testing.
Most ePollbook software used at voter check-in displays a list of registered voters based on the criteria entered, and pollworkers then have to select the correct person from the list.
My first prototype followed this pattern.
PROTOTYPE #1
In the second prototype, the list of names is eliminated and replaced with a series of questions that narrows down the list of registered voters until there is only one possibility left.
PROTOTYPE #2
Iteration #2:
Minimizing Data
The second fix was to find a solution to the confusion caused when users were faced with terminology they did not understand or would have a hard time explaining to voters.
In the first prototype, if a voter runs into an issue where they have a choice of actions to take, the app asks for their decision, requiring the pollworker to explain the choices.
But asking for this information is unnecessary, and may create a record of the voter’s actions.
PROTOTYPE #1
In the second prototype, printed paper infosheets are used to give voters information about their options instead of asking for and recording their answer.
This not only reduces confusion, but it also ensures that:
voters are consistently presented with correct information that has been previously approved by administrators
the pollworker can move on to the next voter
the system is not collecting unnecessary data about the voter
PROTOTYPE #2
With these fixes in place, I was now ready to test with people over the age of 60 outside in cold weather.
Testing the Content
Another iteration was writing specific scripts for each screen to clarify what the pollworker needs to ask the voter, but I found during testing that these scripts came with their own problems:
Asking for a voter’s information before they get to the check-in table may be met with suspicion and allegations of intimidation, so it is important that this app is introduced to voters carefully.
However, especially with the long and repetitive scripts, testers started to paraphrase (inaccurately).
Testing The Interface
I had designed the interface in dark mode, but I discovered that it did not make the screens more readable in bright sunlight.
I had started this project with the assumption that dark mode is easier to see in bright sunlight, but testers preferred having the option to choose between dark and light mode depending on lighting conditions.
Users saw this dark mode interface while testing.
A quick mock-up of “light mode.”
Testing Holding the Device
“We’ve only done two voters and my arms are already tired.”
Thank you Nancy!
It became clear very quickly how untenable holding the tablet while using this app for hours would be for most people, let alone the majority of poll workers who are retirement age or older.
KEY TAKEAWAYS
Dark mode helps only marginally in bright sunlight- if this app is going to be used outside the device will need some sort of cover.
The scripts need to be reworked to provide flexibility to the pollworker while maintaining the accuracy of the content.
Holding a tablet, even for a few minutes, is exhausting.
The time it would take to help each of these voters is a long time to be holding a tablet. Pictured: an early voting line for the 2008 presidential election, Charlotte, North Carolina.