Detroit Water Shutoffs — Race, Income, and Time

Situation

Mass water shutoffs in Detroit raise equity questions about which households lose service. Cross-sectional and time-series views can disagree once race and income are correlated.

Decision question

Census-tract analysis of Detroit water shutoffs showing a persistent racial gap: tracts that are at least 75% Black have higher shutoffs per capita in every year from 2010–2016.

Data and constraints

See method and results below for sample, coverage, and documented limits.

Method

Built tract-level shutoffs-per-capita measures and plotted them against Black population share and median household income. Then compared time series (2010–2016) for tracts above vs. below citywide median income and for tracts that are vs. are not at least 75% Black.

Tools

• R
• tidyverse
• haven
• ggplot2
• Cross-section and time-series EDA

Deliverables

• Knitted R Markdown analysis (March 2026)
• Cross-sectional scatterplots and grouped time series

Outcome / decision value

• Cross-section: higher Black share and lower median income are both associated with more shutoffs per capita.
• Time series: tracts ≥75% Black have higher mean shutoffs per capita in every year; the income split runs the other way (above-median-income tracts higher), pointing to confounding with race and housing type.
• Descriptive evidence cannot rank race vs. income without regression controls; households in predominantly Black and lower-income tracts remain the most exposed overall.

Links