Questions about the data

Every figure comes from a federal statistical release — the Census Bureau, the Bureau of Labor Statistics, the Bureau of Economic Analysis, the FBI, the CDC, the EPA, HUD, the Federal Housing Finance Agency, the IRS, NOAA, FEMA, the FAA and the Department of Education among 19 active sources. Nothing is bought from a data broker, scraped from a listings site, or modeled to fill a gap. Every source is listed on the data sources page with its terms, coverage and update cadence.

It depends on the source, and each page says which period its figures cover. Labor and building permits are read monthly. Deaths, reported crime, rents, regional price parities and house prices are checked quarterly, though several are published only once a year, and the page says which period each figure covers. Census, school, hazard, air-quality, climate, airport and tax-migration data are read annually. A page shows when its data was last updated, and that stamp reflects new source data rather than the day the page was rebuilt.

Because too few of the underlying measures were available for it. A score is withheld rather than calculated from partial data — 48 published profiles currently show measurements without one. The methodology sets out exactly which signals must be present.

Reported crime is published by police department, not by town, and only where a department reported every month of the year. 2,960 cities have a municipal department whose returns were complete. Many smaller communities are policed by their county and have no department of their own, so there is no figure to show — a partial year is never scaled up to look like a whole one.

No. It is a weighted average of standardized signals, and the weighting is a general-purpose guess about what people care about — not a judgment about you. Someone moving for schools and someone moving for weather should not use the same weights, which is why every profile lets you set your own and see the ordering change.

9,533 profiles are published, and a place is published only when enough authoritative data exists to describe it fairly. Communities below roughly ten thousand residents are left out of publication by choice: below that size the census estimates carry margins wide enough that a comparison says more about sampling than about the place. Their pages still load if you follow a link, marked as unpublished and with the figures that exist, but they are not in search results or the sitemaps, and a ZIP code search still finds the ZIP. That is a threshold AreaProfile sets, not a limit of the data, and the editorial policy states it.

It gets checked against the archived copy of the source that produced it. Every figure is traceable to a stored file, so a correction is a matter of looking rather than arguing. See the corrections policy, or use the report link at the foot of any profile.

The underlying federal data is public. AreaProfile’s own calculations — the scores, signals, percentiles and rankings — are its work; the terms set out what is permitted. If you need the figures behind something for research, ask and they will usually be provided.

The sentences on a profile are produced by rules with explicit numeric conditions, from figures already stored and checked — a statement appears because the numbers behind it crossed a threshold, and the values that triggered it are kept alongside it. No figure anywhere is invented, estimated or inferred to fill a gap.