Home Is Where the School Is: The Impact of Homeschool Legislation on School Choice

National Household Education Survey respondents spanning the 1982-1997 window in which 35 US states enacted explicit homeschool-rights legislation

US school-age children and their families, observed across states and years with individual, county and state-level covariates

The only credible policy-variation design located anywhere in the homeschooling literature. It exploits the staggered adoption of explicit "homeschool rights" statutes across 35 states between 1982 and 1997 and estimates a multinomial logit of school choice (homeschool / public / private) conditioning on individual, county and state covariates. The identification is the timing of statutory change, which is the right shock — but the OUTCOME is take-up, not learning. It tells us that legalisation moves families, not what happens to the children afterwards. Recorded here because its existence is the strongest available evidence for the claim this archive most wants tested (does the regulatory regime change outcomes?) and because it demonstrates that the data and the shock both exist: the same staggered-adoption design has never been pointed at an achievement or attainment outcome. Recorded from the abstract and the ERIC record; the article is behind a Taylor & Francis paywall that could not be bypassed.

Passage of homeschool-rights legislation raised the probability that a young child is homeschooled by 1.4 percentage points — a large relative effect against a base rate of roughly 1-2%, so legalisation roughly doubles take-up. Kunzman & Gaither's 2020 survey places this alongside Hail 2003, Levy 2009 and Stewart & Neeley 2005 as evidence that regulatory climate correlates weakly or not at all with homeschooling rates, growth, or homeschooler test scores; Bhatt is the one study in that group with a design capable of a causal reading, and it finds a real effect on take-up.

Low for the estimate reported — statutory timing is not a heritable family trait, and the comparison is across states and years rather than across families. The unaddressed threat is the ordinary staggered-adoption one: states legalised in response to organised homeschooling movements already growing within them, so some of the 1.4 points may be the movement causing the law rather than the law causing the movement.

Never replicated, and never extended to an achievement or attainment outcome — which is the single most consequential gap in this literature, because the shock, the staggered timing and the state administrative data all exist.