The Fine Print on Small-Town Census Numbers

Pull up the profile of a town of two thousand people and you’ll see a median income, a poverty rate, a set of racial percentages — each printed to the dollar or the tenth of a percent. That precision is real about the arithmetic and misleading about the certainty. Nearly every demographic figure for a small place is a survey estimate carrying a margin of error that often runs into the thousands of dollars or many points of percentage. Reading that data well means reading the fine print the headline number hides.

The ACS is a sample, not a headcount

There are two very different “censuses.” The decennial census, taken every ten years, tries to count every single person — the tally that reapportions Congress. Almost everything else, including all the income, education, housing, and detailed demographic data on this site, comes from the American Community Survey (ACS), which is a sample. Each year the Bureau surveys a few million addresses — a small fraction of all households — and uses statistics to infer the whole.

Sampling is what makes the ACS possible: you couldn’t ask every household detailed income and commute questions every year. But sampling means every estimate is exactly that — an estimate, with a knowable amount of uncertainty attached. The Bureau is fully transparent about this; it publishes a margin of error (MOE) next to every figure. The MOE is the part that gets stripped off when a number travels into a headline or a listing, and stripping it off is where readers get misled.

Why small places wobble and big ones don’t

The margin of error isn’t the same size everywhere — and that’s the single most useful thing to understand about it.

In a big city, the sample includes tens of thousands of households, so estimates are tight; a city’s median income might carry a margin of plus-or-minus a few hundred dollars, small enough to ignore. In a town of 2,000, only a few dozen households may land in the sample. The same statistical machinery now has very little to work with, and the margin of error balloons. A small town’s median income might be published as $61,000 with a margin of ±$14,000 — meaning the true figure could reasonably sit anywhere from the high $40,000s to the mid-$70,000s. The point estimate is the center of a wide range, not a precise fact.

This is why the Bureau leans on 5-year estimates for smaller geographies. Pooling five years of responses is the only way to gather a sample large enough to say anything at all about a small place. The trade-off is baked in: 1-year estimates are more current but only exist for areas above 65,000 people; 5-year estimates cover everywhere down to the smallest town and rural tract, but they describe a five-year average rather than this exact moment. Every figure on this site is a 5-year estimate, for exactly this reason — it’s the only vintage that reaches all 31,555 places.

Why some numbers are simply blank

Sometimes the honest thing to publish is nothing, and the Bureau does.

When a place is so small that an estimate would be hopelessly unreliable, the Bureau suppresses it rather than print a number no one should trust. That’s why you’ll occasionally find a blank income or home-value field on a very small town’s profile — not an oversight, but a deliberate refusal to imply precision that isn’t there. The same instinct explains top-coding at the other extreme: household incomes above $250,001 and home values above $2,000,001 are capped to protect the privacy of a handful of households, so a figure sitting exactly at the cap means “at least this much,” not “precisely this.”

A blank field and a capped field are both the data being honest about its own limits. Read them that way and they stop looking like gaps.

Reading small-place data without being fooled

None of this means small-town data is useless — it means it should be read as a range and a direction, not a pinpoint. A few habits keep you honest:

• Round hard. For a small place, treat the median income as “around $60,000,” not “$61,214.” The extra digits are arithmetic, not knowledge.
• Distrust tiny gaps. If Town A shows $58,000 and Town B shows $61,000, that difference is almost certainly inside both towns’ margins of error. They’re effectively tied. Small differences between small places are usually noise.
• Trust the big signal. A town estimated at $130,000 really is far richer than one estimated at $40,000 — that gap dwarfs any margin of error. The data is reliable for large contrasts even where it’s shaky on fine ones.
• Lean on the county. When a town is too small to say much, its county — with a far larger sample and tighter margins — often tells the surrounding story with much more confidence.

Read this way, the “fine print” isn’t a reason to distrust the data — it’s what lets you use it correctly. The numbers on this site are as accurate as a national survey allows, and we publish them straight from the Census Bureau’s own estimates. Knowing which figures are solid bedrock and which are a best guess within a wide band is simply the difference between reading demographic data and being fooled by it. Our full approach — vintages, suppression, top-codes, and how we map every place — is written up on the methodology page, with definitions in the glossary.