The research question
Do colleges and universities respond, at the program level, to state policy that funds education outside district-run schools? The unit of analysis is the education unit — the college, school or department of education at a degree-granting institution.
What we do
- Structured search of each institution's education unit for: microschool, homeschool, school choice, education savings account, private-school leadership, education entrepreneurship, and "non-traditional settings."
- Direct verification — live institutional program pages are read in full, not summarised from search results. The Checked date on every row is the date that page was read.
- Course-level enumeration where feasible: complete catalog listings, not keyword sampling.
- Operating moves are coded, not just programs. This is the methodological difference. Where universities have entered this sector they have often done it by running or authorizing schools — acts that never appear in a course catalog.
- The state's own sponsor or authorizer registry is checked before any institution is coded Tier 0. Added August 2026 after we coded a registered charter sponsor "nothing found" — see Correction 02. A program roster cannot tell you what a board of trustees has authorized.
The coding scheme
| Built for sector | Stated purpose is to prepare, credential, support or operate education outside district-run schools. |
| Names sector | Institutional text explicitly names private, classical, charter, academy or "non-traditional" settings — or a dedicated school-choice course exists. |
| Incidental | Reaches these educators only as a side effect, or is explicitly built around district partnership. |
| Nothing found | No offering identified. Absence of evidence, not evidence of absence. |
Limits that bind hard
- Comparison states are not a census. Most-likely-case sampling can prove something exists in a state; it cannot prove absence. The West Virginia, California and New York nulls are weak and should not be cited as evidence.
- Cases were found by searching for engagement — sampling on the outcome. This biases every explanation the Index offers, and we say so before anyone else has to.
- Course-level enumeration covers one institution of 31.
- Michigan is now complete — and that is itself a limit. All eight public university authorizers are coded as of 8 August 2026. But completeness here means the authorizing side; the coding of each education unit rests on published program and catalogue text, not on interviews or internal proposals. An institution that considered a choice-sector program and declined would be indistinguishable from one that never considered it.
- Florida's college-sponsored charter pathway is now mapped; the colleges themselves are not. As of 8 August 2026, exactly three Florida College System institutions are approved to sponsor charter schools — Tallahassee State, Miami Dade and Pasco-Hernando — all three now coded, alongside the University of West Florida in the State University System. The remaining 25 state colleges are not sponsors, but their education units are still unscanned, and a college can build a program without ever holding a charter. The coverage gap has narrowed and changed shape; it has not closed.
- The Nova Southeastern retrenchment is confirmed, but not dated. The archives were checked on 8 August 2026 and the program is documented in NSU's own institutional repository. What remains unknown is when it ended — the repository dates the record only approximately and holds no discontinuation notice. Cite the existence and the current absence; do not cite a year.
- Single point in time. Coded July 2026. This space moves fast.
On neutrality
Engagement is treated as an observable institutional behaviour, not a virtue. Building a program for this sector and declining to build one are both coded observations. A reader who believes portable-funding policies harm public education can read the same rows as institutions declining appropriately; a reader who supports those policies can read them as institutions lagging. The categories are written to allow both readings. That is a design requirement, not a disclaimer.
Corrections
When we are wrong, the correction runs on the front page and stays in the record permanently. We do not make silent edits.
Correction 02 · August 2026
We coded the University of West Florida Tier 0 — nothing found. That was wrong. UWF is listed in the Florida Department of Education's own Directory of Charter School Sponsoring Entities as an approved sponsor, and its Board of Trustees approved the university as authorizer of a Somerset Academy charter elementary school. The row is recoded Tier 1.
Our July pass read only UWF's School of Education program roster. That reading was accurate as far as it went — no program built for the sector was found, and that finding still stands — but it was the wrong place to stop. The Index's central methodological claim is that catalog scans miss the main event, because universities enter this sector by operating and authorizing schools rather than by listing courses. We then made the catalog-scan mistake ourselves, in Florida, the state we cover most heavily.
The error was caught by reading the state's own sponsor directory, which is now a standing part of the method: for every state in the Index, the official authorizer or sponsor registry is checked before any institution is coded Tier 0. That single change also surfaced Miami Dade College and Pasco-Hernando State College, both since coded.
Correction 01 · July 2026
An early draft of our baseline claimed that no one had systematically scanned colleges of education for choice-sector programs. That was wrong. EdChoice published exactly such a scan — Surfing the Pipeline by Michael Q. McShane, October 2023, with catalog research by Hanover Research across Florida, Arizona, Wisconsin and Vermont.
It is also the origin of a statistic our draft described as untraceable. It was traceable. Our baseline now builds on that work with attribution. The correction changes what is novel here: the first scan is not our contribution — the questions left unasked after it are.
What would move this forward fastest
- Explain Central Michigan. Michigan is complete, and the answer was no — six more authorizers, no further programs. The open question is no longer whether the pattern repeats but why it happened once. That means going past published text: who proposed the CMU program, when, and what made it survive.
- Take the second reading. Re-measure the 2023 four-state baseline inside the Index, with full attribution to the original.
- Read CAEP standards and two state licensure codes to establish whether a choice-sector credential is even permissible. The licensed-boundary finding makes this test sharper: it now predicts exactly where programs can and cannot exist.
- Look at the mature systems. The Netherlands has funded school choice constitutionally since 1917; Sweden since 1992; Chile since 1981; England moved teacher training out of universities deliberately. The hundred-year answer already exists and is unexamined in this literature.
Corrections and challenges welcome
If a row about your institution is wrong, incomplete, or out of date, we want to hear about it — particularly anything marked unverified. Corrections are published, not quietly absorbed. Get in touch.