Methodology

How the CensusEasy Scores are calculated

The CensusEasy Scores are four original indices we compute from raw US Census Bureau data. Each answers a question the Bureau does not publish directly, and each is built from the same source tables that drive the rest of the site. This page documents exactly how they are made so anyone can check, cite, or reproduce them.

Diversity Score Afford Score Segregation Score Gentrification Score Detailed origins & ancestry

Diversity Score

The chance that two residents picked at random are of a different race or ethnicity.

Range: 0 to 100. Higher means more mixed; 0 means everyone shares one group.
How it is calculated: A Simpson diversity index over the five race and ethnicity shares: 1 minus the sum of each group's share squared. We rescale so a perfectly even split across all five groups reads as 100 (the raw maximum for five groups is 0.8, so we divide by 0.8 and multiply by 100).
Inputs: The five ACS shares we already publish: White (non-Hispanic), Black, Hispanic or Latino, Asian, and all other races combined (ACS table B03002).
Coverage: Every place type (states, counties, cities, metros, ZIP codes, tracts), computed for every year we hold race data, back to the 1990 and 2000 decennial censuses for states, counties, and cities.
Notes: A within-place measure: it describes the mix inside one place, independent of how those residents are distributed across neighborhoods (that is the Segregation Score).

Most diverse cities →Most diverse counties →Most diverse metros →

Afford Score

How many years of a typical local income it takes to buy a typical local home.

Range: A multiple of income (for example 4.2x). Lower is more affordable; higher means homes cost more years of local pay.
How it is calculated: Median home value divided by median household income, for the same place and year.
Inputs: Median home value (ACS table B25077) and median household income (ACS table B19013).
Coverage: Every place type, every year both inputs exist, back to 1990 and 2000 for states, counties, and cities.
Notes: A price-to-income ratio surfaces affordability pressure a raw home-value list hides: college towns, resort towns, and coastal metros rank far higher by ratio than by sticker price alone.

Least affordable metros →Least affordable cities →Most affordable cities →

Segregation Score

How sorted a place's residents are into separate neighborhoods by race.

Range: 0 to 100. It reads as the share of residents who would have to move neighborhoods for every neighborhood to match the area's overall racial split. 0 is perfectly evenly mixed.
How it is calculated: The classic index of dissimilarity between White (non-Hispanic) residents and all other residents, measured across the census tracts inside each county or metro: one half times the sum, over all tracts, of the absolute difference between each tract's share of the area's White population and its share of the area's non-White population, times 100.
Inputs: Tract-level White (non-Hispanic) share and total population (ACS), rolled up to the parent county and to the metro (CBSA) via its component counties.
Coverage: Counties and metros only, for 2020 and 2024, the years we hold tract-level race data on a consistent 2020 tract geography. An area needs at least two tracts and both groups present to receive a score.
Notes: Diversity and segregation are different questions: a place can be very diverse yet highly segregated (a mixed population sorted into separate neighborhoods). This is the standard measure used by researchers and the press.

Most segregated metros →Most segregated counties →Least segregated metros →

Gentrification Score

How fast a place has moved upmarket over the most recent decade.

Range: 0 to 100. Higher means faster upmarket change relative to its peers.
How it is calculated: We measure three changes over each place's widest available window of at least five years ending in the latest data year: growth in inflation-adjusted median household income, growth in inflation-adjusted median home value, and the rise in the share of adults with a bachelor's degree. Each change is percentile-ranked against other places of the same type, and the three percentiles are averaged.
Inputs: Inflation-adjusted (real 2024 dollar) median household income and median home value, plus the bachelor's-degree-or-higher share (ACS), across the multi-year series.
Coverage: Cities, counties, and metros, which carry the full multi-year series needed to measure change. Stored at each place's latest data year.
Notes: A relative measure of economic momentum, not a judgment about displacement. Because it is ranked within place type, a city is compared to other cities, not to counties or metros.

Fastest gentrifying cities →Fastest gentrifying counties →Fastest gentrifying metros →

Detailed origins & ancestry

The number and share of residents in each specific national-origin and ancestry group, beyond the five broad race and ethnicity buckets.

What it covers: About 118 groups across four families: Hispanic or Latino specific origin (Mexican, Puerto Rican, Cuban, Salvadoran, Colombian, and more), Asian groups (Chinese, Korean, Vietnamese, Filipino, Asian Indian, Japanese, and more), Native Hawaiian and Pacific Islander groups (Native Hawaiian, Samoan, Chamorro, and more), and reported ancestry (German, Irish, Italian, Polish, Nigerian, Lebanese, and more).
Two metrics per group: An absolute population count ("Korean population") and that count as a share of all residents ("Korean share" = count divided by total population). The count powers the "largest population" rankings; the share powers the "highest concentration" rankings.
Source tables: American Community Survey 5-year detailed tables: B03001 (Hispanic origin by specific origin), B02015 (Asian alone by selected groups), B02019 (Native Hawaiian and Other Pacific Islander by group), and B04006 (people reporting ancestry). All from the same ACS 5-year vintage that drives the rest of the site.
How places aggregate: Counts sum across component areas; a metro's count is the sum of its component counties, and the national count is the sum of the states. Shares are recomputed from the summed count over the summed population, so a metro or national share is exact rather than an average of averages.
Coverage: States, counties, cities, metros, and the nation, for the latest ACS year (plus a 2020 snapshot where available). The Asian and Pacific Islander group tables begin in 2017, so they carry no earlier history. Detailed origins are not published at the ZIP-code or census-tract level here, where small-group estimates are largely suppressed or carry very large margins of error. A group is stored for a place only when its count is above zero.
Notes: Ancestry is self-reported and a person may report more than one, so ancestry shares across all groups can exceed 100 percent. The Asian and Pacific Islander group tables tally each person in every group they report. These are Census Bureau estimates with sampling error; treat small counts as approximate.

Largest Korean population, metros →Largest Mexican population, cities →Highest Italian share, cities →

Age distribution & marital status

How a place's population splits across age bands, and how adults are distributed across marital statuses by sex.

Age bands: Twelve bands (Under 5, 5 to 9, 10 to 14, 15 to 19, 20 to 24, 25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 to 74, 75 to 84, and 85 and over). The boundaries are chosen so that every Census vintage we hold can be folded onto the same axis, which lets the age chart move through the decades on a single comparable scale.
Age source tables: Sex by age, summed across men and women: ACS table B01001 for the latest year, the 2000 and 2010 decennial Summary File 1 table P12, the 2020 Demographic and Housing Characteristics table P12, and the 1990 Summary Tape File 3A table P13. The bands sum exactly to total population.
Marital status: Men and women aged 15 and over, each split across never married, married, widowed, and divorced ("married" combines all now-married categories). Shares on the chart are within each sex. Source: ACS table B12001 for the latest year and the 2000 decennial Summary File 3 table P018.
Coverage: States, counties, cities, metros, and the nation. Age bands cover 1990, 2000, 2010, 2020, and the latest ACS year. Marital status covers 2000 and the latest ACS year only, because the 2010 and 2020 short-form censuses do not collect it; on an age snapshot that has no marital data the place page shows the nearest available year. Neither is published at the ZIP-code or census-tract level.
How places aggregate: Every band and marital count is a raw count, so a metro is the sum of its component counties and the nation is the sum of the states. Shares are recomputed from the summed counts, so an aggregated share is exact rather than an average of averages.

Names: surnames & first names

How we build the last-name and first-name data, the race composition, and the full-name estimates.

Surname source: The US Census Bureau genealogy name files: the 2000, 2010, and 2020 Frequently Occurring Surnames releases, plus the 1990 surname file. Each lists a name's rank, the number of people who carry it, and its frequency per 100,000 people. The 2020 file covers 156,621 surnames held by 100 or more people.
Race and Hispanic-origin composition: For 2000, 2010, and 2020, the Census tabulates each surname across six groups: White, Black, American Indian and Alaska Native, Asian and Pacific Islander (a single combined group), two or more races, and Hispanic or Latino origin. These are Census-tabulated shares, not modeled estimates. The 2020 file publishes them as exact counts, which we convert to percentages; 2000 and 2010 publish them as percentages directly. The 1990 file carries no breakdown. Small cells the Census suppresses for confidentiality are shown as blank.
First names: The 2020 Census first-name file, which lists each name's count split by sex, and the 1990 male and female first-name files. Counts are living-population totals, not births. The popularity-over-time series back to 1880 comes from the Social Security Administration and is documented where it appears.
Most common vs most distinctive: "Most common <group> surnames" ranks names by the count of people in that group who carry them. "Most distinctively <group>" ranks by the share of a name's bearers in that group, among names held by at least 2,000 people, so the list surfaces names strongly associated with a group rather than the largest names overall.
Full-name estimates: Full first-and-last name counts are never published (the Census surname and first-name files are deliberately separate for confidentiality). The estimate multiplies each name's frequency: the number sharing a full name is approximately P(first) times P(surname) times the US population. It assumes the two are independent, which undercounts names that tend to go together (for example Jose Garcia or Wei Zhang). Every full-name and celebrity figure is a statistical estimate, not a Census count, and is labeled as such.

Source data. All four scores are derived from the US Census Bureau: the American Community Survey 5-year estimates and the decennial censuses (1990 through 2020). Dollar figures used in the Afford and Gentrification Scores are adjusted to constant 2024 dollars using the BLS Consumer Price Index.

Reuse. Every ranking offers a full-list CSV download, and the scores appear on the relevant place pages. If you cite a CensusEasy Score, a link back to the ranking or this page is appreciated.