Sources, methodology and vintage

A gap is shown as a gap

Where we have no value, the page says which kind of nothing it is — not reported, suppressed, or not yet loaded — and hovering gives the full reason. Never a blank, never a zero, and never a dash that could be mistaken for one. If an invented value is ever published it carries a dotted underline; there are currently none on the site.

Suppressed cells are never imputed

States withhold figures for small groups to protect student privacy. Where that happens we label the cell as suppressed and say why. We do not estimate, interpolate or model the missing value, and a suppressed cell never becomes a zero.

Each state is modeled its own way

States do not organize their schools the same way, so we do not force them into one shape. We group each state’s districts the way that state does: Georgia by RESAs; Kansas by counties. Where that grouping is a voluntary cooperative rather than a geography — membership optional, overlapping, and sometimes spanning most of the state — we do not put it in a URL at all, and record it as an affiliation on the district instead. Each state also carries its own list of indicators, drawn from what its own agency publishes. That is why no table column is ever shared across two states.

Regional figures are rolled up, not re-reported

Each metric carries its own rule. Counts are summed. Rates are weighted by enrollment, so a 400-student district does not pull a regional average as hard as a 40,000-student one. Metrics that are only meaningful at district level are not rolled up at all.

Coverage

Live: Georgia, Kansas — every district in both, 466 in all. Some measures are not yet loaded for every state, and those are marked on the page rather than omitted; DATA_PROVENANCE.md in the repository is generated at build time and lists exactly what is sourced, what is a gap, and why. Corrections are welcome at tj@standardeducation.com.