Methods & sources

Love Census counts people. It does not multiply percentages from different places and hope for the best. Here is where every number comes from and what it can’t tell you.

Love Census is not affiliated with the U.S. Census Bureau. It uses public data the Bureau and other agencies publish.

Where the numbers come from

• Age, sex, marital status, living with a partner, race and ethnicity, income, education, kids at home, veteran statusU.S. Census Bureau, American Community Survey 5-year Public Use Microdata Sample (PUMS): 13 million adult records with weights2020–2024 (incomes in 2024 dollars)
• Splitting Census areas (PUMAs) into countiesCensus 2020 tract-to-PUMA relationship file + ACS tract adult counts (table B01001)2020 geography; ACS 2020–2024
• Metro areas (which counties make up each metro)OMB core-based statistical area delineation file, published by the Census Bureau (Bulletin 23-01)July 2023
• ZIP code lookupCensus 2020 ZIP Code Tabulation Area (ZCTA) to tract relationship file, weighted by ACS tract adults2020 ZCTAs; ACS 2020–2024
• Religion, how religious (importance, services, prayer), politicsPew Research Center, Religious Landscape Study (36,908 adults); party profiles from Pew's 2026 party affiliation fact sheet2023–24 (profiles 2026)
• TattoosPew Research Center, “32% of Americans have a tattoo” (8,480 adults)July 2023
• Smoking and binge drinking, county by countyCDC PLACES (county estimates) with the shape by age, sex, race and education from CDC BRFSSPLACES 2025 release (BRFSS 2023); BRFSS 2024–25
• Height and body mass index (BMI)CDC National Health and Nutrition Examination Survey (NHANES): measured, not self-reportedAug 2021 – Aug 2023
• COVID-19 vaccine: ever had a dose / never had oneCDC National Immunization Survey, adult respiratory virus moduleas of Jan 31, 2026
• COVID-19 vaccine: got the 2025–26 shotCDC adult COVID-19 vaccination coverage by seasonas of Feb 21, 2026
• LooksNo data, by design: “top 25%” simply keeps 25%—

How a count is made

1. Census traits are counted together. Age, sex, single, race, income, education, kids and veteran status are all on the same person records in the Census microdata. A record either matches every one of them or it doesn’t. Each record stands for a known number of people (its weight), so adding up the weights gives a real headcount.
2. Survey traits are matched to each person. Religion, politics, tattoos, smoking, drinking, height, BMI and vaccination aren’t in the Census. For each matching person we use the rate for their sex, age, race, education, marital status and state (county for smoking and drinking). Young women are far more likely to have a tattoo (56%) than adults overall (32%), so a single national rate would get it wrong.
3. Religion, politics, how religious and tattoos come from one joint model. “Christian + Republican + no tattoos” respects how those go together.
4. Looks is a plain multiplier, because nobody measures it.

What “single” means

Never married, divorced or widowed, and not living with a partner. People who are separated are still married, so they’re not counted as single.

The headline numbers

• The big percentage is the share of your comparison group (say, single women 25–35 in Orange County) who fit everything else you picked. “1 in N” is the same number turned around.
• Share of all adults divides the same headcount by every adult in the area.
• The headcount comes with a 90% margin of error from the Census replicate weights. It covers Census sampling only, not the uncertainty in the survey rates.
• How rare am I? reports how few people share all your traits. It is a count, not a score: rare doesn’t mean better or worse.

Counties, metros and ZIP codes

• The Census microdata identifies areas of about 100,000 people (PUMAs), not counties. Where an area spans several counties, its people are split between them by adult population, assuming they look alike.
• Big counties such as Los Angeles, Orange and New York are built from whole areas, so they are counted directly. About 2,600 small counties sit inside an area they share with neighbors; their numbers are that area’s profile scaled to the county, and we mark them with ≈.
• Metro areas are the counties OMB groups into each metropolitan statistical area (2023 definitions).
• ZIP codes are matched to counties with the Census ZIP code areas. A ZIP that crosses county lines goes to the county holding most of its adults.
• If fewer than 30 Census records match in a place, we show the state number instead (or the US number if the state is thin too) and say so.

“Where the odds are best”

Every US county and metro area is counted in one pass and ranked by the share of your comparison group who fit. A place is ranked only if it has enough data:

• at least 30 matching Census records,
• a 90% margin of error smaller than 30% of the share itself,
• at least 2,000 people in the comparison group, from at least 100 records.

We tried 20%, 30% and 40% for the margin rule. At 20%, real patterns disappear (military towns for young veterans, for example); at 40%, small noisy places start to win. “Biggest pools” ranks the same places by headcount instead. Counties that share one Census area with neighbors carry the same estimate, so only the largest of them is listed.

What these numbers can’t tell you

• Survey traits are estimates. Pew’s state samples are a few hundred people each (±5–8 points), and some religion figures by age were derived from published tables (±1–3 points).
• Some independence remains. Within a person’s group (sex, age, race, education, marital status, state), we assume a smoker is as likely to have a tattoo as a non-smoker, and that income doesn’t change religion. Religion, politics, how religious and tattoos are linked to each other; the rest are not.
• Self-reported traits. Smoking, drinking, religion, politics and vaccination are what people told surveys. Height and BMI were measured.
• Politics by state comes from Pew and sits noticeably off the 2024 vote in a few states (Hawaii, Maine, Mississippi).
• COVID vaccination is self-reported in a survey with a response rate under 25%. “Never got a COVID shot” means no dose ever, about 21% of US adults as of January 2026. It is not the same as skipping this season’s shot.
• Kids at home means their own children under 18 living with them. Census data misses some dads who live with their kids outside the usual household relationships.
• Who’s counted. Adults 18+, not counting people in prisons or nursing homes.
• Data years. The Census microdata pools 2020–2024 interviews. We’ll update when the 2021–2025 file comes out (expected early 2027).

Privacy

There are no accounts and we don’t save your answers. They are used to run the count, and a share link carries them in the URL so the calculator can fill them in for whoever opens it.

← Back to the calculator