Pharos
Shining light on what matters most.
A steady light on county-level suicide and mortality data, and the measures that move with it. It shows what the numbers say and, just as clearly, what they cannot: associations in a population, not individual risk or causation.
Map explorer
Every ingested covariate as a county map layer, hot/cold spot clusters, and model attributions.
Analysis dashboard
Animated correlation views over two decades, adjusted associations, multivariable models, and composite indices.
What this is
A research tool that assembles free, public, county-level data, including mortality outcomes and dozens of demographic, economic, health-access, and environmental measures. It lets you explore how they move together across US counties and over time. Every number describes a county population, never any individual in it.
What it cannot tell you
- Not individual risk. County-level associations do not transfer to people (the ecological fallacy).
- Not causation. Even adjusted correlations are associations among measured variables, not effects. Unmeasured confounding is always possible.
- Small counties are noisy. Rates from low-population counties swing year to year and are often not meaningful; the tools fade or shrink them and mark suppressed values as gaps, never zero.
Data & methods
Outcomes are age-adjusted death-certificate and modeled small-area rates (NVSS via County Health Rankings; IHME). Correlates come from Census/ACS, SAIPE, BLS, CDC PLACES, HRSA, EPA, and other free public sources. Every reported statistical test is corrected for multiple comparisons (Benjamini-Hochberg FDR); collinear predictors are flagged because their attributions are unstable. Full methodology and source list are inside each explorer.
Use & contact
Free to use and cite. Suggested citation: Pharos: county mortality statistics and correlates, accessed . Underlying data belong to their public sources (cite those for the numbers themselves). Found a data problem or have a question? Open an issue on GitHub (opens in a new tab).