Wisconsin: the map versus the blood tests

Our lead-risk map is built only from public housing age and poverty data. It has never seen a blood test. We held it against Wisconsin's real childhood blood-lead, tract by tract. Here is the prediction next to what was measured.

0.70
rank correlation, predicted vs. measured blood lead (2018-21, ≥5 µg/dL)
208
census tracts with measured blood-lead data
8
states where the map checks out against real data

What we predicted

Risk from housing age + poverty. No blood data used.
lower
higher predicted risk

What was measured

Share of tested children with elevated blood lead.
lower
higher measured blood lead

Hover any neighborhood to compare both numbers. Where the maps share color, the prediction matched reality.

Why this matters

A risk map is only worth trusting if it agrees with where children were actually poisoned. In Wisconsin it does, at a moderate level. Most of the country has no public blood-lead map at all. A prediction from data every neighborhood already has brings the same warning everywhere.

Sources: Wisconsin childhood lead poisoning surveillance by census tract, Wisconsin DHS (ArcGIS open data). Predicted risk: U.S. Census ACS 2022 housing age + poverty, scored by DetectLead.

The map finds the risk. The test confirms it.

See all 8 statesOpen the national map