Texas Justice Initiative
UX research and consultation in January 2019
Conducted research and presented insights to Texas Justice Initiative in support of their data portal redesign.
Texas Justice Initiative (TJI) is an online resource for data about the Texas criminal justice system for journalists, legislators, advocates, and more. After participating in a Design Jam meetup that focused on TJI’s design portal, I approached the co-founder to see how I could help with the redesign of their website as they changed platforms.
I conducted five interviews with prisoners’ rights advocates, a public defender, and a Master’s student in information sciences. I hoped through these interviews to validate and further inform TJI’s plan for their redesign. In my discussion guide for these interviews, I wanted to learn more about users’ research practices and their use of the current data portal in order to make users’ experience of the new data portal more intuitive and the data visualizations more digestible.
The week that TJI’s co-founder and I planned to reach out to potential interviewees was the first week Texas legislature went back into session, so we didn’t receive a lot of response to our initial call for interviewees. In an effort to still provide value to TJI by our established deadline, I reached out to my own network to find TJI users and others who fit their user base.
Based on the feedback we collected from our Design Jam meetup and my first interview with the information sciences student, I expected a much larger trend toward issues with the data visualization aspect of the site. Specifically, designers and data analysts seem to take issue with donut graphs.
Surprisingly, all other interviewees easily grasped the information presented through with the donut graphs. Interviewees were more concerned with the language of the filters and the sources of the data. This was a relief for me as I’m much more comfortable with content and transparency than I am with data visualization, but it was bittersweet, as I was excited to spread my wings into new territory.
- User Persona
- Insights from research
- Heuristic analysis
- Design recommendations and sketches
I had some concerns that because my interview pool wasn’t drawn from the list originally given to me by the client that this persona might not be the primary persona for TJI. I let the client know that this persona represents one user persona based on TJI’s users and recommended further research to develop other personas based on a broader user base.
- Existing users go straight to the data portal with specific information they’re looking for.
- With few minor setbacks, such as confusion over color choices and distinctions between colors, pie charts were easy to read.
- An explicit acknowledgment of the limitations of the data and links to other resources would instill confidence in users about the veracity of TJI’s data sources.
Heuristic Analysis of Current Site
A Landing Page That Tells a Story
One of the main talking points during the Design Jam meetup was how the landing page could be improved. Indeed, two of my interviewees pointed out that much of the content of the home page, which really lends itself to the story TJI is telling about its data, is lost beneath the fold and users could easily mistake this content for the footer and ignore it.
Two users I interviewed suggested that the color choice for the first block of the information below the fold might lead users to overlook this information, despite a barely-perceptible and non-clickable chevron urging visitors to explore it. For inspiration for TJI’s home screen, I looked at other data-focused sites, such as NASA’s Open Data Portal and USGS, to see how progressive disclosure encouraged exploration of the site’s data outside of the official data portal.
A Consistent Color Key
One of my interviewees echoed another point that came up during the Design Jam; a wish for a clearer, more consistent color key for the donut graphs in the data portal. Some colors were very similar to each other, which made it difficult to interpret information. Additionally, the same colors were used to represent pieces of information across all donut graphs, which caused some confusion.
Acknowledgment of Data Sources and Limitations
All of the users I interviewed commented on the way the filter criteria were named and categorized; for example, sex and race. Everyone quickly assumed that these naming conventions came from TDCJ or whatever forms were used to source the data. Still, users hoped for a more explicit in-context acknowledgment of the source of these names.
Sketches and Recommendations
Reorganize Landing Page and Menu
For the home screen, I wanted to put both “About TJI” and “About the Data” on the main menu to make more clear where users could find more information about the data. In the sketches of my suggestions for the redesign, I uncollapsed the “About” sub-menu and moved “Donate” and “Publications” into contextual links. With this, I hoped to encourage more exploration of the homepage and TJI’s data stories.
Data Portal Menu
For the data portal itself, I brought the filtering menu beneath the drop-down menu to select one of the three data sets to make more clear that the filtration options were specific to the selected data set.
To address the feedback about the donut graphs we heard from designers during the Design Jam and one of my interviews, I included a drop-down menu to specify give users the ability to view the information using the same variety of data visualization methods displayed on the home screen, including histograms, bar charts, and line charts.
The founders don’t want to use Tableau, because it isn’t as responsive as they prefer. This might make creating multiple visualizations for the various filtration options require some work, but coupling a sensitivity to varied learning styles with newfound accessibility compliance (another insight from the Design Jam) could really put TJI on the map.
The next step would be to test these initial designs and iterate before diving into the logistics of programming this. Having the filtering menu out in a drawer that can be retracted when on mobile is a great way to save space, so I’d be interested to see whether that’s the best way to handle having three data sets with three different filtering menus. For example, we might try a/b testing this design against making the process more clear in the copy or creating a unified filtration menu with an explanation of how the data criteria have been modified in the “About the Data” section or in the “Explore the Data” introduction.
I’m also looking into potential color schemes that could answer to the great variety offered by the filtering menu and offer users a consistent experience across all graphs.
Other Case Studies