The Lasting Legacy of Redlining by FiveThirtyEight

When 2020 Census data was released, FiveThirtyEight visual journalists Ryan Best and Elena Mejía used the new data to explore the extent of which racial segregation caused by redlining was still present in 138 major American cities. The challenge was two-fold: first, to find a fair way to analyze how formerly redlined neighborhoods look today, and, second, to find a clear and compelling way to present the complicated results.

Best and Mejía’s analysis started with the original redlined maps drawn by the now-defunct Home Owner’s Loan Corporation (HOLC) from 1935 to 1940, collected as part of the University of Richmond’s Mapping Inequality ( project. They aggregated block-level population counts from the 2020 census to compare modern populations across geographic lines in these archival maps. Even with this granular data, making direct comparisons between neighborhoods proved to be of limited value in understanding the influence of redlining in today’s racial segregation. Through experimentation and reporting Best and Mejía developed a strategy of comparing each HOLC zone to the greater metropolitan region in which it is situated.

The key statistical concept at work is something called a location quotient: a small-area measure of relative segregation calculated at the census tract level. An LQ of 1 means a tract’s proportion of a race is even with the surrounding area. Above 1 is overrepresentation of a race, below is under.

Best and Mejía iterated the design of maps and visuals for months to find exactly the right solution. (Best details some of the concepts in this Twitter thread Some of the issues: maps of redlining zones paired with beeswarms representing population didn’t scale well to smaller areas. One version called for radar charts of the Location Quotients for each race and ethnicity, but this required a heavy explanation of the metric. Chloropleth maps didn’t work. Finally, they landed on population dot-density maps. The creative design of these maps allow the reader to understand the takeaways from the location quotient metric without requiring an in-depth understanding of the complicated statistics behind it.

This visual, too, went through many, many iterations. To fairly represent each race and ethnicity in the analysis, each color needed to feel equal hierarchically. They tried a dot representing many people or exactly one person; different basemaps, maps that zoomed in and showed only one spot at a time, small multiples showing each redlining rating.

The solution is a side-by-side approach that uses a dropdown to allow users to select between the types of redlined neighborhoods paired with a map that shows the greater context of the surrounding area. Bar charts below each map give the share of each race in the zone versus the surrounding metro area. The project yielded an in-depth analysis with a complex level of granularity that successfully demonstrated how redlining repercussions are cemented across the country.