Methodology

How Ozipedia is made

The About page says what the sources are. This page says how the page in front of you got made.

The rule everything else follows

Data can be thin, but nothing is made up.

Every claim on an Ozipedia page traces to a value in a named dataset. Where a value does not exist for a ZIP code, the sentence that would have used it does not get written, and the section it belonged to does not get published. No gaps are filled from general knowledge, regional reasoning, or inference about what a place is probably like.

This is the whole basis on which a reference work is worth reading, so it is worth saying plainly rather than burying it.

How a page is assembled

1. Data collection. Each dataset is ingested on its own schedule and joined to ZIP Code Tabulation Areas. ZCTAs are the Census Bureau’s statistical approximation of USPS ZIP codes; they are the unit federal data is actually published against, which is why Ozipedia uses them.
2. Section gating. A section is generated only when its required fields are present. Pages therefore vary in length, and a shorter page means less data existed — not less care.
3. Templated figures. Every number you see is inserted from the dataset. Numbers are never typed by hand or restated from memory.
4. Comparisons. A figure is only compared against a benchmark when the two are the same measure from the same kind of source and were published at broadly the same time. A market index is never compared against a survey median, and a current figure is never compared against a years-old one. Where no valid benchmark exists, the page shows the local figure alone.
5. Narrative. Section prose is written from the figures for that section and nothing else. Narratives are generated from each ZIP code’s own dataset values using a language model, under a closed-payload constraint — the model receives only that ZIP’s figures and cannot introduce outside facts.
6. Review. Pages publish in waves, and each wave is sampled and reviewed before the next one goes out.

Public safety

Crime is reported to the FBI by law-enforcement agency, not by ZIP code. A police department’s jurisdiction is the smallest area for which anyone in the United States publishes verified counts. ZIP codes are postal delivery routes; they were never drawn to match a patrol boundary, and mostly they don’t.

That leaves everyone who wants to publish crime by ZIP code with a choice. Most services take an agency’s counts and spread them across the ZIP codes that overlap it, weighted by population or by a model of their own. The result looks precise. It is not measured. Two such services will give you different numbers for the same ZIP code in the same year, and neither will show you the working.

Ozipedia does not do this. We publish the agency’s own reported counts, and we tell you what area they cover. Where a ZIP code and its policing agency cover substantially the same ground — which is true for a great many American towns, islands and small cities — that count is a ZIP-level figure, and we say so. Where a ZIP code is one small part of a larger jurisdiction, we give the agency’s figure and name the area it describes. The rule is the same in both cases: publish what was measured, state the area it was measured over, and never close the gap with an estimate of our own.

What is on the page

• The agency that polices the ZIP code, named. Often a city police department; frequently, in rural America, a county sheriff’s office.
• Reported offence counts for the most recent complete year, violent and property, plus individual categories where they are notable.
• The agency’s rate per 100,000 residents, calculated from its own reported count and its own population — arithmetic on two published figures, not an estimate.
• The national and state rates for the same period, published by the FBI. This is the comparison that carries the meaning; a count on its own tells you almost nothing.
• The ZIP code’s share of that agency’s jurisdiction, so you can judge how closely the figure describes where you actually live.

What is not on the page, and why

There is no safety grade, score or rating. We removed the one we had. A letter attached to a named residential area is a verdict, and a verdict invites two things we want no part of: a reader who stops thinking, and a resident whose property is marked down by a number nobody can audit. Counts and comparisons are more useful and more honest.

We do not describe places as safe or dangerous. The data supports a rate. It does not support an adjective.

The limits, stated plainly

Not every agency reports. Around 96% of the United States population is covered by agencies submitting to the FBI’s Uniform Crime Reporting programme, which means some are not — and a few of them are large. Where the agency policing a ZIP code does not report, this section does not appear. An absent section is our answer to missing data everywhere on Ozipedia, and public safety gets no exception.

Reported offences are offences reported to and recorded by police. Crimes that go unreported are in no dataset anywhere, and reporting rates differ between communities and between offence types. Counts also move with population, policing practice and recording changes, so a year-on-year shift is not automatically a change in conditions.

An open invitation to law-enforcement agencies. If your department can supply incident data at ZIP-code or sub-jurisdiction level, we would like to publish it, attributed to you. That is the single change that would most improve this section, and it is the one thing we will not manufacture in your absence. Write to us at /corrections.

Source: FBI Crime Data Explorer, Uniform Crime Reporting programme. Figures are for the most recent complete reporting year and are refreshed when the FBI updates them.

Our own scores

Some measures on Ozipedia are ours rather than a federal agency’s. Where one appears it is labelled as an OurZip metric and linked here. Two honest caveats apply to all of them: they are estimates, and they are comparative — designed to place a ZIP code relative to others, not to be read as an absolute statement about a place.

Where a composite disagrees with its own components, the page says which component is pulling it. A health composite sitting below average above four better-than-average disease measures is not a contradiction; it means something else in the composite — usually air quality — is doing the work, and the page says so on the card rather than three paragraphs later.

OurZip Step Score

A walkability composite of points-of-interest access, transit access and street connectivity, derived from OpenStreetMap and ACS commute data.

The composite runs 0–100 and is reported out of five. Its three sub-scores — points-of-interest access, transit access and street connectivity — are shown individually on each page, because they frequently disagree and the disagreement is the useful part.

It is measured per square mile, across the whole ZIP code area — and that has a consequence worth stating plainly. A ZIP code with a compact, genuinely walkable centre sitting inside a very large rural area will score near zero, because the score describes the average density of the whole area rather than the best part of it. ZIP 99801 covers Juneau, Alaska across roughly 2,200 square miles of mountain and inlet; its downtown is walkable and its Step Score is 0.0. Both of those things are true.

Across the country the distribution falls out roughly as: rural ZIP codes at 0–1, suburban at 2–3, dense urban at 4–5. A low Step Score in a large or rural ZIP code is a statement about how spread out the area is, not a judgement about whether its town centre is pleasant to walk in. Read it alongside the land area shown at the top of the page.

OurZip Piggy Rating

Affordability relative to local incomes: home value and rent measured against median household income for the same ZIP.

It is a local measure: housing cost relative to the incomes of the people who live there, not relative to national cost. A ZIP can be expensive in dollars and unremarkable on this rating, or inexpensive and strained on it. Reported out of five.

OurZip Economics Score

A roll-up of affluence, business density, housing wealth and market opportunity. Affordability is scored separately and is not part of this composite.

Affordability is deliberately excluded from this roll-up and scored separately, which is why a ZIP can carry a strong economics score and a low affordability rating at the same time. Both are true; they measure different things.

OurZip Health Score

A roll-up of chronic-disease prevalence, preventive care, risk factors and environmental burden, including air quality.

Environmental burden — principally air quality — sits inside this composite alongside disease prevalence. That is the usual reason a health score reads lower than the disease measures printed beside it.

Sources and vintages

Datasets used across Ozipedia, with their update cadence.

What it covers	Source	Cadence
Population, income, housing tenure, commuting, education	U.S. Census Bureau — American Community Survey	Annual, 5-year estimates
Home values and rents	Zillow Research — ZHVI and ZORI	Monthly
Chronic disease prevalence	CDC — PLACES	Annual
Air quality	U.S. Environmental Protection Agency	Annual summary
Climate normals	NOAA — 1991–2020 U.S. Climate Normals	30-year normals
Natural hazard risk	FEMA — National Risk Index	On FEMA’s release cadence
Reported crime	FBI — Crime Data Explorer	Annual, by law-enforcement agency
Boundaries and geography	U.S. Census Bureau — TIGER/Line	Annual
Streets, transit and points of interest	© OpenStreetMap contributors, ODbL	Continuous
Public schools and enrolment	NCES — Common Core of Data 2024	Annual
Private schools and enrolment	NCES — Private School Universe Survey 2022	Biennial, published with a lag
Tax rates	Tax Foundation	Annual
Federal and state representation	Congress.gov and OpenStates	Continuous

Every page carries the vintage of the data it uses and a last-updated date, and lists the specific sources it drew on. Where a dataset does not cover a ZIP code, the corresponding section does not appear. We would rather publish a short page than a padded one.

What we deliberately don’t do

• No third-party proprietary ratings. No licensed school ratings, no licensed walkability scores. Where Ozipedia shows a composite, it is one we compute and document.
• No scores presented as bare verdicts. Where a score appears it is labelled as ours and linked to how it is calculated.
• No invented detail. If the data for a section does not exist, the section is absent — not stubbed, not padded, not filled in from general knowledge about the region.
• No paid placement. Nothing on Ozipedia is for sale and no ZIP code can pay to be described differently.

Errors and corrections

We publish corrections rather than quietly editing. If something here is wrong we want to know — email corrections@ozipedia.com with the ZIP code, the section, and what you believe is incorrect. A source, if you have one, helps.

Most figures are reported directly from federal datasets. If the ACS says a ZIP has a median age of 43.9, that is what we publish, even where local knowledge suggests otherwise — but tell us anyway, because a mismatch between the data and the ground is often a sign we have mapped a boundary wrong, and that is very much our error to fix.

We do not remove or soften accurate figures because they are unflattering to an area. Crime rates, hazard ratings and affordability measures are reported as the data has them.