A review of American Community Survey (ACS) data on income of Seattle residents and its relevance to future medical insurance enrollment outreach actions.
The Problem
The goal of this research is to answer the following questions:
Where in Seattle are health insurance enrollment rates lowest?
Can income patterns help explain these deficits?
The answers to these queries relate to the work of city planners, elected officials, and public health advocates interested in closing any enrollment gaps.
Key Findings
South Park stood out as a statistical outlier, as it contained the only census tract in the city with fewer than 78% of working-age residents enrolled (Figure 1). More broadly, the five census tracts with the lowest enrollment rates were all south of Downtown: South Park, North Delridge, Rainier Beach, Beacon Hill, and North Beacon Hill.
The Income Connection
Neighborhoods with higher median incomes do tend to have higher insurance enrollment rates (Figure 2), though income alone accounts for only about a quarter of that difference. What it leaves unexplained is spatially random; the gaps that remain are scattered across the city rather than concentrated in any one area. Taken together, these findings point to a relationship that is meaningful but not absolute, and suggest that other contributing factors remain unexplored.
What This Could Mean
These findings identify regions that King County could target for increases in ACA Navigator resources and community partnerships. They enable further investigation into efficiency efforts on this front by way of prioritizing geographically contiguous neighborhoods with low enrollment rates. It must be mentioned that the results of this analysis are associative in nature, rather than causal.
Methodology Notes
The ACS survey dataset available from the City of Seattle's ArcGIS Data portal linked here contains detailed counts regarding both health insurance enrollment and income partitioned by census tracts. As of this case study's writing, it aggregates results from 2015 to 2024 and contains 2,867 records across 26 fields. During the cleaning process, the dataset was reduced to focus solely on figures from the 2020 5-Year estimates. This was an independent analysis conducted outside of any professional or academic assignment, using real and publicly available data.
The geographic distribution of regression results was tested to confirm that the income findings were not an artifact of how neighborhoods happen to be arranged next to one another on the map.
All calculations were conducted using R. For technical details and process documentation, see the HTML Quarto notebook linked here or on the sidebar.
Future Directions
A natural next step would be re-running this study on more recent ACS data to see whether the geographic patterns have shifted post-pandemic. The dataset's multi-year scope would be conducive to studying change and trends in enrollment data over time. An alternate aggregation by city council district could prove to be useful for public health advocates wondering which elected officials should be prioritized for coalition-building.