The Four Sectors
Comparing public, charter, private and homeschool education across the 50 states and DC, built from federal data with its gaps shown rather than hidden.
Where it started
I've spent a lot of time around people in public, charter, and private schools, and all of them had opinions about school funding. They were all sure they were right, and their stories all conflicted.
I wanted to know whether the public data could back up any of those sides. That turned into a rabbit hole: where the money comes from, where it goes, and how much any of it shows up in how students actually do.
The question
American schooling happens in four sectors: traditional public, charter, private, and homeschool. Comparing them state by state sounds like a matter of lining up four numbers per state. It isn't, because the four sectors are not measured the same way. Public and charter schools file an annual census. Private schools answer a survey every two years. Homeschooling is counted only where a state requires families to register, and twelve jurisdictions don't.
A site that put four numbers side by side without saying that would be misinformation. So this one shows the coverage map first.
The rule
Where a source reports nothing, the site shows nothing. Nothing is modelled, imputed, or interpolated, and every chart has a table behind it.
That rule decided more than any design choice did. Homeschool counts are registrations, a floor rather than a total, so they are shown per state with their year and source attached and kept out of every market-share calculation. Special education spending varies twentyfold between states for reasons that turn out to be accounting practice, so the site says that instead of ranking it.
Nine sources, one dataset
The data comes from nine sources: eight federal collections (NCES's school census, private school survey, NAEP, household survey and IPEDS, plus Census and Bureau of Labor Statistics files) and a Johns Hopkins compilation of state homeschool registrations. Most of the work was making them agree on what a number means:
- Enrollment is K–12. The public school census counts prekindergarten; the private survey doesn't. Leaving PK in understates the private sector by close to a point of market share.
- Charter sits beside traditional public, not inside it. States without a charter law get an explicit zero rather than a gap, so they still appear in every ranking.
- Private figures are survey-weighted, and the weighted school counts land within two schools of NCES's own published totals.
What the data said
A few findings that hold up once the confounders are named:
- Charter growth is concentrated and slowing. It went from 1.0% of the public system in 2000–01 to 8.2% in 2024–25, but five states account for 57% of the 2013–2021 growth.
- It didn't come out of private schools. Charter growth and private enrollment change correlate at r = +0.17, weakly positive.
- Administration isn't where the money goes. District central administration is 2.0% of operating spending, about $285 per pupil.
- Poverty explains about half the difference in state test scores; spending explains almost none of it. The site says plainly that this is a limit of comparing state averages, not evidence that funding doesn't matter.
The comparison it was most tempting to make, school spending against college selectivity, turned out to be an artifact. The site walks through why, with a whole chapter on it.
How it's built
A Python pipeline fetches each source, caches every response, and composes one dataset. The site itself is a static React app that reads that file. Charts give each sector one fixed color, use a single y-axis, and ship with a table view, because a bare correlation coefficient invites exactly the over-reading the site is arguing against.