Sources & method

2022 official data

We use the DOL Women’s Bureau National Database of Childcare Prices, selecting rows with STUDYYEAR = 2022. Prices are weekly county-level medians from state market-rate surveys. DOL standardizes and sometimes imputes values; “2022 official data” does not imply a new provider survey in every county that year. Counties may share a rate-zone price. We retain source aggregation flags in the data output.

Source fields: MCINFANT, MCTODDLER, MCPRESCHOOL and MCSA for centers; MFCCINFANT, MFCCTODDLER, MFCCPRESCHOOL and MFCCSA for home-based care. Infants are 0–23 months, toddlers 24–35 months and preschoolers 36–54 months in DOL’s aggregated definitions. School-age arrangements can be part-day during the school year. See DOL’s technical report, Appendix C for definitions and limitations.

We multiply weekly infant, toddler and preschool prices by 52 for a full-year equivalent. That assumes 52 paid weeks, not a provider’s actual billing policy. School-age prices are not automatically annualized. The budget calculator lets you choose paid weeks and care setting.

Inflation estimate, not a new local observation

We use BLS CPI-U series CUUR0000SEEB03, day care and preschool, U.S. city average, not seasonally adjusted.

Estimate = 2022 weekly price × 407.335 ÷ 330.304583.

The denominator is the arithmetic mean of all twelve monthly indexes for 2022. The numerator is the published index for September 2026. The multiplier is 1.233210. This applies national childcare price movement uniformly; local prices may have changed differently. We label the exact target month rather than treating it as a full-year 2026 observation.

State and national comparisons: a typical county

Our “state median” is the unweighted median of available county median prices for the same age, setting and year within that state. The national median uses available counties in the 50 states and District of Columbia. Each county counts once. These benchmarks describe a typical county, not the price paid by the typical resident or a pooled median across providers.

Benchmarks use all available source counties, not only the 500 county pages in the launch set. Coverage varies by metric and is shown on each result card. We suppress comparisons with fewer than 5 available counties. A missing baseline remains unavailable.

For example, the national infant-center benchmark covers 2,661 counties. No population weighting is applied. We apply the same CPI multiplier to both the county and its benchmarks, so price relationships are unchanged.

“About the same” means within ±5% of the benchmark, inclusive. Otherwise the chip says “lower” or “higher.” This is a display convention, not a statistical significance test, a measure of affordability or a quality verdict.

Population selection and family income

The launch set selects the 500 most populous counties with at least one available childcare price, using TOTALPOP from DOL’s 2022 rows. This is the source-vintage population, not a claim about today’s rankings. Each state and Washington, DC has a real hub page; a hub can have useful statewide comparisons even if none of its counties reaches the national 500-county population cutoff.

DOL includes Census ACS population (DP05) and median family income (MFI, DP03) in its workbook. We use those distributed fields without a Census API request. The historical income ratio is 2022 infant-center weekly price × 52 ÷ 2022 median family income. It compares county statistics and does not measure any individual family’s spending. We do not divide a 2026 estimate by 2022 income. Missing income yields no ratio; imputed income is labeled.

ZIP codes and county boundaries

Lookup uses the repository’s shared ZIP Code Tabulation Area to county mapping from Census. It is based on 2020 ZCTA boundaries and the Census 2023 Gazetteer and 2020 relationship files. ZCTAs approximate postal ZIPs; some PO-box and unique ZIPs are absent. The primary county has the greatest land-area share, not necessarily the most residents. We show every mapped county so you can choose where care is provided.

Lookup distinguishes a ZIP absent from the mapping, a county without source prices, and a county that has data but is outside the 500-page launch set. We never substitute a nearby county’s price.

Launch coverage and search indexing

The 500 pages are selected before applying the thin-data rule. A county needs at least four of the six full-day infant, toddler and preschool prices across center and home-based settings to enter the sitemap and search index. School-age prices do not count toward this threshold. The other selected counties remain accessible with a limited-coverage notice and noindex. This is an editorial coverage threshold, not a statistical confidence test.

A state hub is indexable when it includes an indexable county page, or at least three price categories each with five reporting counties in the full source. Empty or sparse state hubs remain linked from the map but noindex and outside the sitemap. We never replace missing data with national figures.

Nearby counties

We use the Census 2023 county adjacency file, the first available release after 2010. Nearby links are limited to adjoining counties in the same state that have launch pages, sorted by DOL source-vintage population, up to six. This is adjacency, not travel time. County definitions that do not match the 2023 geography, including Connecticut's historical counties, may have no nearby links; the state hub still lists the available pages.

Coverage, missing values and updates

The current source has 3,143 county rows in the 50 states and DC; 2,699 have at least one price and population data. Unknown prices stay missing. We preserve zeros only where a field legitimately permits them; zero or negative childcare prices fail validation rather than appear as free care.

Downloads are hashed and dated. Duplicate county/year keys, changed schemas, invalid prices, incomplete CPI base years and unexpected coverage changes block output replacement. Failed candidates are quarantined for review. DOL workbook retrieved 2026-11-03; BLS data retrieved 2026-11-03.