State Water Board subcommittee launches racial-equity data training with DataMaid demonstration

The State Water Board's Racial Equity Data Subcommittee held its first webinar to introduce practical methods for analyzing racial equity in water programs, featuring a presentation by Hannah, a partner at DataMaid, who demonstrated census retrieval, spatial-join workflows and exploratory analyses.

The session matters because staff across the Water Board will increasingly rely on program and public-data analysis to assess whether policies and operations produce disparate outcomes by race and ethnicity. Molly Williams, an environmental scientist in the Office of Public Participation, said the subcommittee's goal is to "build and operationalize data resources from across the Water Board and across the state so that we can better understand and promote equity in all of our programs and policies." She invited staff interested in equity-data work to join and said slides and a recording would be shared after the webinar.

Hannah, the presenter, framed racial equity using a definition from racialequitytools.org as the condition "that would be achieved if one's racial identity no longer predicted in a statistical sense how one fares," and recommended three procedural steps: define the analytic question and necessary programmatic variables, seek input from subject-matter experts and community members, and add historical or social context to interpretation.

On methods, Hannah demonstrated programmatic (API) and point-and-click approaches for obtaining demographic baselines. She recommended the U.S. census products for race/ethnicity and socioeconomic variables and highlighted complementary indices such as the CDC Social Vulnerability Index and FEMA's National Risk Index. For state work, she noted CalEnviroScreen and the EPA's EJScreen as alternative baselines to consider depending on the variables of interest.

For reproducible, programmatic retrieval in R, Hannah showed how to use the tidycensus package to call decennial and American Community Survey (ACS) tables (get_decennial/get_acs), to list variables (load_variables), and to request geometries for mapping. She showed that calling the method for each year is the standard approach for multi-year comparisons and emphasized use of the summary-variable option to simplify percentage calculations.

Hannah then walked through a spatial-join example using the sf (simple features) package. She described converting complaint-point coordinates to spatial objects (st_as_sf), projecting both datasets to a common California Albers CRS (st_transform), and performing point-in-polygon joins (st_join) to attach tract-level demographics to each complaint. She stressed that mismatched projections or coordinate-reference systems can produce incorrect joins.

Illustrating analytic pitfalls, Hannah cautioned that "data are not people" and that raw complaint counts can mislead. Using the Water Board's complaint and investigation records as an example, she filtered for discretionary dismissals and computed a dismissal rate by tract, then compared tracts with higher-than-average dismissal rates to others. That analysis reduced apparent racial differences seen when counting filings alone and highlighted that outcome-based measures (dismissals, investigations completed) can tell a different equity story than counts of filings.

During Q&A, attendees asked how the CDC SVI differs from CalEnviroScreen and EJScreen, how to assemble multi-year series from the census and ACS, and how to handle overlapping census geographies when aggregating to custom boundaries (for example, subbasins). Hannah and Forrest described pragmatic options: drawing custom GeoJSON boundaries, area-weighted allocation of tract values to a boundary, or using smaller block-level geographies and population-weighted allocation when precision is needed.

The subcommittee will post the slides, recording and resource links and has invited staff to submit data-project requests for individual coaching by DataMaid. Williams closed by reiterating how staff can get involved via the subcommittee page and the equity-data help email.

The webinar provided concrete, reproducible starting points (tidycensus, sf, spatial-equity tools) and emphasized careful question framing, community input and caution when interpreting programmatic counts as proxies for impacts.