Guide
How to read census figures for a city
Most figures for a city, town or census tract come from the American Community Survey, a sample, so each is an estimate with a 90% margin of error: the true value lies within estimate ± margin in 90 of 100 samples. Austin's median household income is $93,658 ±$1,670; for Altoona, AL (948 residents in 2020) it is $43,056 ±$17,183.
What does the margin of error tell me?
Read the estimate as a range. Austin's median household income lies, with 90% confidence, between $91,988 and $95,328: a narrow range, because the survey reached many thousands of its households over five years. Altoona's lies between $25,873 and $60,239, because a town of 948 people gives the survey a few dozen households.
When is an estimate too uncertain to use?
Divide the margin by 1.645 to get the standard error, then by the estimate: that is the coefficient of variation (CV). Austin's income has a CV of 1.1%; Altoona's 24.3%. PlainCities follows a widely used convention for survey estimates: a CV under 12% is reliable, 12-40% is usable with care, and over 40% is unreliable. Figures over 40% carry a ring and are left out of every ranking; the Census Bureau publishes them, and so does PlainCities, with the margin beside them.
How do I tell whether two places really differ?
Two estimates differ when the gap between them is larger than 1.645 times the square root of the sum of their squared standard errors (the Census Bureau's test at 90% confidence). In practice: if their ranges overlap a lot, the survey cannot tell them apart. The comparison tool runs this test and marks only the differences it confirms.
Can I compare 2015-2019 with 2020-2024?
Yes, because the two periods do not overlap; comparing overlapping five-year periods (2019-2023 with 2020-2024) mostly compares the same responses. Dollar figures must first be put in the same year's dollars: PlainCities multiplies 2015-2019 dollars by the ratio of the BLS R-CPI-U-RS averages for 2024 and 2019, the index the ACS uses. A change is then called only when it passes the same 90% test. How often incomes really changed.
Which population figure should I quote?
For a full count rather than a sample, the 2020 Census. For the latest figure, the Census Bureau's yearly estimate (July 1, 2025 for incorporated places and counties). For population characteristics (age, income, housing), the ACS 2020-2024 estimate, which describes the five-year period, not one year.
Why a median, and what does "or more" mean?
Income, home value and rent are given as medians: half the households are above the figure and half below. A median is not pulled up by a few very large incomes or very expensive homes the way an average is, so it describes a typical household better. The survey does not publish medians beyond a top value: a median household income above $250,000 appears as "$250,000 or more" and a median home value above $2,000,000 as "$2,000,000 or more". Places at that cap are tied in a ranking, whatever their true order.
Why do my city's figures differ from another website's?
Usually for one of four reasons. The period: a one-year estimate (published only for places of 65,000 or more) or an older five-year release gives a different figure from the 2020-2024 estimate used here. The area: a metropolitan area, a ZIP code or a school district is not the same area as the city limits. The dollars: some sites adjust incomes to a later year and some do not. The measure: median and average income, or household and family income, are different figures. PlainCities names the survey, the period and the table beside every figure, so any of them can be checked at data.census.gov.
Source: American Community Survey 2020-2024 5-year estimates and published margins of error, U.S. Census Bureau · Census guidance on comparing ACS data