Client: Temple University’s Center for Public Health Law Research
Prompt: How might we help CPHLR increase Open Data adoption of their LawAtlas datasets, democratize policy surveillance data, and ultimately help policy analysts build stronger cases for laws that improve population health?
Background: LawAtlas rests on a profound insight: With proper analysis, the gray areas of the law can be distilled into simple sets of yes / no questions. Even though the text of a law may vary from state to state, these discrete attributes, as judged by a group of experts in the field, can be compared across jurisdictions, supporting a wide range of public policy analysis, and ultimately smarter policy decisions.
The Problem: With 137+ datasets, and more added every year, it’s difficult for policy researchers to discover the datasets that are most relevant to their research.
Goals / Challenges:
- Help users evaluate a dataset by answering a variety of possible questions. The most common: Showing policy adoption distributions (e.g. “How many countries ban or restrict abortion?”), showing policy groups and intersections (e.g. “During Covid, how many states closed public OR private schools AND required indoor masking?”) and showing policy changes over time (e.g. “Which areas adopted Good Samaritan laws first?”).
- Support a variety of jurisdictional levels, including cities, states, territories and countries.
- Dataset Quirks. While datasets are similarly structured, they’re not identical, so any visualization needs to gracefully handle outliers.
- Dataset Search & Discovery. With 100+ datasets, across a number of different topics, finding data of interest can be hunting for a needle-in-a-haystack.
- While choropleth maps are a popular choice for similar clients, they’re limited to displaying only a single variable. To show policy groupings and co-occurrence, we’d need to show multiple variables at once. Some research suggests as many as 6 or 7 different variables can be overlapped on a single surface, so we explored overlapping patterns to represent different (combinations of) policy attributes.
- Geographic maps benefit from viewers’ spatial memory when looking to specific areas, however for this use case, the priority was visualizing an overall tally of adoption and supporting multiple variables, both of which benefit from tile maps’ fixed size per jurisdiction.
- Interactivity can help manage complexity. Multi-dimensional datasets are inherently complex and quickly exhaust viewers’ cognitive capacity. The tool should offload this burden to the user’s mouse: additional context and reminders can be just a click (or hover) away.
- All roads (searches) lead to Rome (data directory). Depending on users’ needs, they might favor different ways to identify datasets of interest, therefore search and filtering mechanics should support multiple paths (without introducing undue clutter or UI complexity).
- Policy Map Visualization. This tilemap approach makes it easy for users to see both overall adoption of a single policy and the tapestry of policy combinations across states by overlapping 4 distinct markers on each tile. The tilemap also supports the full range of geographies found in the datasets.
- Dataset Explorer Tool. The Policy Map Visualization is manipulated through a 4-variable toggle for each policy attribute, where each variable corresponds to a color that appears on the corresponding tiles.
- Universal Instant Search. To find datasets of interest, users can search for a specific dataset by keyword or filter datasets by topic. From the dataset directory page, users can filter by further criteria identified by the team.