Field Guide · Free

How we count: the methodology behind Oahu area data

Evergreen; kept current with the platform · updated 2026-07-05 HST

Every number this platform produces is descriptive: what was recorded, reported, noticed, and sought, in a stated past period. This is the plain-language version of the full methodology.

Four signals

Recorded is official HPD open data. Reported is community submissions to the Stolen Stuff Hawaii network, unverified and alleged. Noticed is public attention: views of public posts. Sought is search demand. They are different instruments measuring different things, so they are shown side by side; where they disagree is information, not error.

The denominator that makes rates fair

Per-capita rates divide by daytime population (residents and workers present from U.S. Census LODES and ACS, plus daily visitors in visitor areas from State of Hawaii DBEDT), not residents alone. Commercial and tourist districts fill up by day; dividing by residents alone would overstate their rates several-fold. Using the population actually present follows the criminology literature on ambient populations.

Honest uncertainty

Counts under 5 are suppressed. Every rate carries a 95% Byar's Poisson confidence band, and rates built on fewer than 20 incidents are labeled small-count rates: one or two records visibly move them, so the band is the honest read.

Glossary

Recorded: An incident in HPD's public open-data feed. The most recent month is the least complete; records settle over about a month.
Reported: A community submission to SSH. Unverified, alleged, and counted on its own axis.
Noticed: Views of public SSH posts. An attention measure shaped by audience size, not an incident count.
Sought: Search demand: what people looked for on the public search site and on Google.
Daytime population: Residents and workers present in an area by day (Census LODES + ACS), plus daily visitors in visitor areas (State of Hawaii DBEDT). The denominator for every rate.
Byar's Poisson 95% band: The interval a rate is statistically consistent with; smaller areas get wider bands.
Suppression: Counts under 5 are hidden as "<5" so no individual incident or person is inferable.
The dark figure: Crime that happens but is never reported or recorded. Absence in the data can mean "not reported," not "did not happen."

How to read this. Everything we publish is descriptive: what was recorded, reported, noticed, and sought, in a stated past period, with caveats. Figures are aggregate and de-identified; person-linked counts under 5 are suppressed; per-capita rates use daytime population with 95% confidence bands. Never a forecast, score, ranking, or safety judgment. Full detail: the methodology page.

Official crime figures derive from data.Honolulu.gov (HPD), modified for use from the original source. The City & County of Honolulu does not vouch for the quality and timeliness of the data presented herein, any analysis of the data presented herein, nor any content, viewpoints, products, or services presented herein. Not endorsed or certified by the U.S. Census Bureau.