IMPROVING ACCESS TO ELECTION DATA TO STRENGTHEN PUBLIC TRUST
PROJECT DURATION
May - October 2024 initial Launch + 2025 iteration
ROLE
Design Strategy , UX Research
PROJECT OVERVIEW
In partnership with the Michigan Department of State (MDOS) , Center for Civic Design designed and launched an interactive online election data dashboard to centralize public election data to improve transparency in the electoral process.
The Michigan Voting Dashboard was launched in October 2024, ahead of the 2024 presidential election.
Learn more about the design approach in a resource published by Center for Civic Design.
KEY CONTRIBUTIONS
As Lead UX Researcher, I oversaw strategic planning and design, collaborating closely with a UI designer and a cross-functional product team at MDOS. I conducted user research from initial concept through launch and dashboard refinement.
IMPACT
The dashboard made data readily available, which contributed to:
-
Fewer public record requests Nearly eliminated Freedom of Information Act (FOIA) requests for early voting and absentee ballot data.
-
High public engagement 345,000+ views leading up to the presidential election.
-
Nationally recognized Received a Spotlight Award in the 2025 ClearMark Awards from Center of Plain Language for impactful communication.
CHALLENGE
Problem
Michigan’s election data was technically public, but it was spread across various state and local web pages and formats, making it difficult to find quickly and efficiently. This fragmentation unintentionally created space for mis/disinformation, which could potentially undermine public trust in the electoral process.
Opportunity
Research goals
How might we create a centralized user friendly interactive dashboard that:
-
Improves the findability of election data ahead of an upcoming election
-
Supports real-time reporting and research
-
Simplifies complex election data and policies for both experts and non-experts
RESEARCH APPROACH
Research Design
CONSTRAINTS
This project required balancing research depth with the delivery of a high-quality public dashboard quickly before the 2024 election.
-
Rapid Product Timeline We had 3 months to design and launch the dashboard.
-
Cross team collaboration The team transitioned from a project-based model to a product team, requiring time to align workflows and processes.
-
Election cycle capacity We prioritized speedy research activities to inform design with minimal burden on team.
RESEARCH PHASES
Given our constraints, we planned for research in two phases:
Phase 1 Content + Structure (Pre launch)
Methods
-
Audited election dashboards
-
Conducted informal interviews with internal stakeholders
-
Synthesized learnings into personas and user stories
Phase 2 Validate and Refine (Post Launch)
Methods
-
Moderated usability sessions
Phase 1 Pre-Launch
Content audit
ELECTION DASHBOARD AUDIT
Our design team audited state election websites, including dashboards and Michigan’s existing PDFs, to learn what election statistics were commonly reported.
From this, we learned:
-
Level of data granularity expected
-
Key metrics
-
Design patterns in election dashboards
Key Learning
Most election dashboards assume users are highly knowledgeable about the election process and policies, yet few provide guidance and context for the data.
Interviews and Personas
PERSONAS
The team conducted informal, unstructured interviews with internal MDOS partners and stakeholders. I synthesized the findings and leveraged Center for Civic Design's extensive research on voter needs to develop personas. I identified 3 primary user groups who rely on election data in different ways.
Dedicated public servant
BEHAVIORS + ACTIONS
-
Invests in research and data to guide decision-making
-
Regularly coordinates with county teams on election initiatives
PAINPOINTS
-
Uncertainty about when datasets are updated
-
Has to cross-check multiple sources to confirm accuracy, which is time-consuming
Election Expert
BEHAVIORS + ACTIONS
-
Conducts in-depth research on voting patterns
-
Leverages professional networks to access hard-to-find datasets
PAINPOINTS
-
Data requires extensive cross-checking across counties or precincts, which increases research time
-
Limited access to structured datasets
Reengaged young voter
BEHAVIORS + ACTIONS
-
Finds election information primarily through social media
-
Prefers visual, easily digestible content
PAINPOINTS
-
Finds election websites confusing
-
Struggles to start exploring data without knowing exactly what to search for.
DESIGN
Design Decisions + Tradeoffs
From interviews, persona development, and alignment with the product team, I created user stories to guide dashboard structure and feature prioritization. Based on the research, I partnered with a UX designer to develop the initial wireframes, prototypes, and later high-fidelity designs. The following decisions reflect key trade-offs made during development.
Election data is dense and multi-layered. How do we break up election content to allow for easy navigation?
The dashboard was organized into 4 primary tabs (Home, Absentee & Early Voting, Voter Registration, Election Results). This structure helps separate complex datasets into manageable views, allowing users to easily navigate between tabs.
Segmenting content reduces cognitive load and allows users to focus on one dataset at a time, while still supporting deeper analysis across tabs.
Users cannot see all election data in a single view making it difficult to compare for analysis.
01 Structuring the dashboard
Desktop: Wireframe of tab navigation
Election data is segmented into four tabs, allowing users to explore one dataset at a time without being overwhelmed.
Users need some guidance on how to use the dashboard. Where should orientation content live so it’s helpful but not overwhelming?
We placed onboarding content above the dashboard into the CMS website layer rather than embedding into the Power BI dashboard experience. Expandable accordions and side navigation provided quick access to supporting content without taking up valuable screen space. This made updates easier for the web team to manage.
Embedding static content inside the dashboard created clutter and made it longer to get to key data metrics. Separating it allowed users to reach live election data faster while still supporting learnability.
Users, especially election novices, may miss helpful context if they skip the introductory content.
02 SEPARATING ONBOARDING CONTENT FROM THE DATA EXPERIENCE
Desktop: Home tab
LEFT IMAGE Tab version with heavy onboarding content inside the dashboard
RIGHT IMAGE New version with onboarding content on CMS website
Separating onboarding static content outside the dashboard reduced the number of tabs and allowed users to quickly access live election data.
Separating onboarding static content outside the dashboard reduced the number of tabs and allowed users to quickly access live election data.
Election data is highly nuanced and easily misinterpreted. The processes, timing, and terminology involved in elections are complicated. How to create a dashboard that presents complex policy and data calculations in a clear, concise, and user-friendly way?
We applied plain language principles and framed headings as questions to make them easily scannable and maintained consistent naming conventions across tabs.
To provide deeper explanations, we used click triggered popups instead of hover tooltips to provide non disruptive contextual information, such as terminology and data calculations, that remained visible until the user chose to close them.Users needed to reference definitions and methodology while reviewing live data. Keeping contextual information close to the metrics reduced misinterpretation and supported both expert and non-expert users.
Unfortunately, the original mobile interaction did not meet accessibility standards, as some screen readers did not recognize the pop-up trigger.
To meet launch timelines, we introduced a temporary fix: added a link-out button for the mobile version of the dashboard. The desktop kept the pop up pattern, while mobile displayed the explanatory text inline. While this created a longer and denser mobile experience, it preserved important context for all users.
03 Providing Contextual information WHILE KEEPING CONTINUITY
Desktop: High fidelity screens of the Absentee and Early Voting tab
Pop-ups allow users to reference election definitions and how metrics are calculated without losing their place in the data.
Mobile: High fidelity screens
The desktop kept pop-up patterns, while mobile displayed explanatory text inline. This made the mobile experience denser but preserved important context for all users.
Separating onboarding static content outside the dashboard reduced the number of tabs and allowed users to quickly access live election data.
Phase 2 Post Launch
User testing
RESEARCH GOALS
The primary goal of user testing was to validate the dashboard’s design and identify gaps in supporting various types of expert users.
HOW WE TESTED
Moderated usability tests using task-based scenarios.
Participants
6 election experts
Method
-
Think-aloud protocol for task-based scenarios
-
Task completion tracking
-
Misclick observation to identify confusion
WHAT WE ASKED
WHAT WE WERE OBSERVING
Task 1 Understand voting methods
Identify the voting methods used most during the 2024 general election.
Navigation of dashboard hierarchy and interpretation of summary metrics.
Task 2 Analyze county-level data
Find the total number of absentee ballots sent in [county].
Use of filters and movement between website and Power BI dashboard.
Task 3 Explore graphs
Identify the number of early voting ballots cast in [2 townships] .
Interaction with visualizations and interpretation of detailed data views
LEARNINGS
Key Insights from Usability Testing
Usability testing revealed that the dashboard improved access to election data and usability compared to multi website pages. However, several usability issues prevented users from fully leveraging the data.
Insight 03
Transparency about how election data is calculated and updated is essential for trust.
Election data carries high stakes for professionals who rely on it. Accurate and credible data is essential for these users.
The more granular the data, the more users wanted clarity about how it was produced.
Insight 02
Users struggled to quickly locate the interactive dashboard.
Users encountered barriers in 2 sections:
Above the fold
Most participants missed the interactive dashboard. Instead, they expected to find a button or link on the top section of the website.
Interactive dashboard
The multiple scrollbars and nested scrolling within the embedded Power BI interface were difficult to navigate and prevented users from seeing all the content.
Insight 01
A centralized dashboard dramatically improves the experience of accessing election data.
Participants found the dashboard easier to use and more organized than the election data sources they normally rely on. Having access to county and municipality-level data in one place saved them time, and the ability to switch between views supported quicker analysis.
The highlighted color indicates where participants expected to find the dashboard button or link during usability testing.
Multiple nesting scrolls bars were difficult for users to navigate
Prioritized Improvements
Dashboard findability was the highest priority opportunity due to its impact on task completion and its frequency among participants. If users cannot easily locate the dashboard, it delays access to the product's primary value.
We explored multiple concepts to improve findability and streamline access to the interactive dashboard. The final concept provided a clearer visual entry point through content restructuring, making the dashboard easier to find while preserving helpful context.
Additional product Improvements identified were incorporated into the product backlog for future releases.
-
Enable data export functionality to support reporting workflows
-
Clarify data definitions and update timing for key metrics to improve trust
-
Enhance interactivity on visualizations to support analysis
Initial Election Dashboard Launched
Post-Launch Election Dashboard
After improvements
REFLECTIONS
Personal Takeaways
01 Trust is built through the small details.
This project reinforced the idea that civic design is about clarity and building trust. Small design decisions regarding labeling, plain-language explanations, and metric definitions can meaningfully shape how people interpret public information.
02 In civic products, neutral data can become political.
Even factual metrics can carry political meaning depending on how they are framed, compared, or positioned in the interface. In this case, I had to understand the content, sometimes doing a deep dive on very specific election policies and processes while also understanding the political narrative surrounding that topic.
03 Trusting the iterative process
I was a bit concerned about launching a public-facing dashboard before conducting extensive user testing, since it’s relatively high stakes. This project taught me the value of working iteratively under constraints: prioritizing what mattered most for launch, learning from users quickly, and improving the product over time.