How we score schools

We score every public and private K-12 school in the country from 1 to 10, using only free public data. The score blends how much students improve, how they perform, how consistently that holds across groups of students, and a few things about the school itself. Different school types are scored on different signals, because different things are measurable for each.

Why we built our own

The well-known school ratings are commercial. Their numeric scores sit behind paid licences, and the terms do not allow a site like ours to republish them. One of the largest does not rate private schools at all.

So we built a rating from public data instead. That has a real advantage beyond cost: we can tell you exactly what goes into it, which none of them will.

Four groups, scored differently

A single formula for every school would be dishonest. What can be measured for a public elementary school is not what can be measured for a private high school. So schools fall into four groups.

• Public elementary and middle. Led by academic growth — how much students improve over a year, rather than where they started.
• Public high. Growth, plus college-readiness signals and graduation rates.
• Private elementary and middle. No state testing and no graduation outcome exist here, so this is the weakest of the four. See below.
• Private high. Anchored on where graduates actually go, plus reported test averages and school characteristics.

A school spanning several grade ranges gets more than one score. A K-12 school is scored once as an elementary/middle school and once as a high school, because the two use different inputs and are not comparable to each other.

Growth matters more than proficiency

This is the choice that most affects the ratings, so it is worth stating plainly.

Proficiency measures how many students clear a bar. It tracks the wealth of the surrounding neighborhood closely, which means a proficiency-led rating mostly re-describes local incomes. Growth measures how much students improve while they are at the school, which is much closer to what the school itself is contributing.

We lean on growth wherever a state publishes it. Where a state does not, the weight shifts onto what is available and the score carries lower confidence.

Comparing across state lines

Every state writes its own tests and sets its own bar. “Proficient” in one state is not the same achievement as “proficient” in another, so a 70% pass rate in one is not comparable to a 70% pass rate in the next.

Most of our other signals avoid this. Some are already relative to a state’s own schools, and some come from federal sources with one nationwide definition. Proficiency is the one that needed fixing.

We calibrate it against NAEP, the one test given the same way to a representative sample of students in every state. That gives a common yardstick, so a school in a high-scoring state and one in a low-scoring state can be placed on the same national scale. The difference between states comes from real NAEP results rather than from an adjustment we invented.

Private elementary and middle schools are our weakest score

We would rather say so than quietly pretend otherwise.

These schools sit outside state testing and have no graduation outcome to anchor on. There is no defensible measure of how well they educate that is publicly available. What we can see is structural: class sizes, accreditation, association membership, whether enrollment is growing or shrinking, and test results where a school publishes them.

Those describe the shape of a school, not its results. The score is flagged as low confidence on the map, and the honest advice for this group is the same one the sector gives: go and visit.

Where the data comes from

All of it is free and public:

• State education department report cards, for all 50 states and Puerto Rico
• The federal school and district data collections run by NCES
• The Civil Rights Data Collection, for absenteeism and discipline
• EDFacts, for graduation rates
• NAEP, for the cross-state calibration above
• Federal higher-education data, for college selectivity
• Accreditation rosters published by the accrediting bodies themselves

Confidence, and never guessing

Every score carries a confidence value showing how much of what we wanted to measure we actually found for that school.

When something is missing we leave it out and lean on what we do have. We never substitute a default or a typical value, because a guessed input is indistinguishable from a real one once it is baked into a number. A school with nothing at all to score is left off the map rather than shown with an invented rating. Around 98% of the roughly 141,000 schools in our data have at least one usable input.

Coverage varies by state. Some states publish rich report cards and some publish very little, so scores in the latter rest more on federal sources and carry lower confidence.

What we deliberately leave out

Two things are excluded on purpose, even though the data exists.

Religious affiliation and school type — the difference between a parochial school and a nonsectarian one, for instance. These describe what a school is, not how well it teaches. Treating them as quality signals would encode a preference that has nothing to do with education, so we do not read those fields at all.

What this score cannot tell you

• It is an average. Any school contains classrooms and teachers that are much better and much worse than its number.
• It measures the past. Leadership changes, and a school can move faster than annual data can show.
• It cannot see fit. Nothing here tells you whether a particular child will be happy somewhere.
• It says nothing about the families at a school, and should not be read that way.

How often it updates

Most state report cards refresh once a year, usually in late autumn. The federal sources run on annual or two-year cycles, some with a year’s lag. We re-run everything on that rhythm, and each school’s score records the dates of the sources behind it.

A note on trademarks

We do not republish, badge, or imitate any commercial school rating. Our score is computed independently from public data, and this page describes what goes into it.

You can see school scores for any area on the schools map. If something here looks wrong, we would like to know.