Training shows census-based method linking Lamont water contamination to racial exposure patterns

An online training hosted for water-board staff and partners demonstrated a reproducible, census-based method to identify racial disparities in exposure to a failing drinking-water system in Lamont, Kern County.

Instructor Hannah Cushman Garland, partner at DataMaid, led the session and used the Lamont case to show how decennial census block data, a public dataset that predicts private-well versus public-water use, and dot-density mapping can be combined to measure who is likely to be exposed. The exercise found that while Lamont’s total population is majority Hispanic, Hispanic residents were both numerically dominant and disproportionately likely to depend on the public water system that contained exceedances of maximum contaminant levels for arsenic and 1,2,3-trichloropropane.

The training taught a reproducible workflow: obtain a Census API key; pull 2020 decennial block-level counts with the tidycensus package; pivot and aggregate to the desired geography; fetch an EPA-derived dataset that estimates the probability a block relies on public water; join datasets by census block; and produce both choropleth and dot-density maps in R (using ggplot2 and a dots package). Garland said the working definition of racial equity for the module was that “race no longer predicts a person's access to water or the quality of resources they receive,” and emphasized that racial equity is an ongoing practice rather than a single outcome.

The instructors highlighted key methodological trade-offs. The demonstration used the 2020 decennial census because block-level counts are only available from the decennial product; the American Community Survey provides estimates at larger geographies and carries larger margins of error for small places. Garland warned that the 2020 census can still undercount “hard-to-count” groups and that analysts should choose between a count and a more recent estimate based on tolerance for margin of error and the need for small-area geography.

Participants and breakout-room reports stressed limits and follow-up steps. Breakout speakers recommended community outreach to confirm how water is used (drinking, cooking, agriculture or swimming), language-accessible communications, and coordination with existing programs. Maddie, a training participant, said the group planned to "connect with the Hispanic population that was the most affected and ... first figure out how that water is being used," to design immediate communications and health follow-up.

The training also showed how combining maps and weighted tables improves precision. After constructing dot-density maps (one dot per 10 people in the instructor's example) and overlaying the EPA probability-of-public-water-use layer, the class computed weighted means to estimate the share of each racial/ethnic group likely to be served by public water. The resulting table indicated a notably higher share of Hispanic residents reliant on public water (the instructor cited a figure of about 92% for the Hispanic group in the exercise) than for the other groups, producing a disparity beyond majority status alone.

Participants raised several assumptions to test before drawing policy conclusions: that everyone in the city limits uses public water; that block populations match current (2024) occupancy when the analysis used 2020 census counts; and that the census race/ethnicity categories used initially captured all locally relevant subgroups. The class expanded the initial three-group analysis to include all census racial/ethnic categories, and Garland noted forthcoming changes in Census race/ethnicity collection (for example the planned separate MENA category) that affect comparability over time.

Attendees also reviewed external data sources that can augment the analysis: EPA environmental-justice screening tools and mirror sites, CalEnviroScreen, the Division of Drinking Water (DDW) failing-list and Human Right to Water lists, SAFER program materials, and local public-utility records. Tanya, a participant who summarized a breakout group, urged checking the DDW “Human Right to Water” failing list and contacting SAFER staff or the Lamont Public Utility District if a project is not already underway; she also suggested applying for technical assistance through the appropriate funding or assistance channels.

The session ended with instructors and participants agreeing on practical next steps: add community outreach and language access to any technical plan; consult additional environmental and drinking-water data sources; and prepare local communications and interim solutions (for example, bottled water distribution) while remediation continues. The instructors said they would publish a slide update with resources gathered from the chat and recommended attendees reconvene for module 2 to examine source identification and deeper causal analysis.

The training was framed as a hands-on, reproducible approach to pair quantitative mapping with community engagement to surface disparities and plan equitable interventions.