Homeschooling's Invisible Children: Report of Findings 2024

nearly 500 cases of severe abuse, including 191 fatalities of homeschooled children since 2000 (133 with public-record evidence of chronic maltreatment)

Children in the United States identified in public records as homeschooled, or whose caregivers claimed to be homeschooling them, at the time of a severe or fatal abuse incident since 2000

A case database, not a study. Cases are collected from the public record, which the report states plainly means media coverage — so inclusion requires that a case be "almost always fatal, severe enough for criminal prosecution, or uniquely disturbing." There is no denominator (nobody knows how many children are homeschooled, because 11 states require no notification), no comparison group, and no sampling frame; the count is a numerator in search of one. The publisher is the Coalition for Responsible Home Education, which exists to campaign for homeschool regulation, and the report is written as advocacy with model legislation attached — hence `developer-involved` (the vocabulary has no slot for opponent-led; the load-bearing fact is that the authors have a declared stake). What makes it worth recording rather than ignoring is that the authors say all of this themselves: they write that their analysis is not representative of abusive homeschool environments and that this should motivate further research. That is a more honest limitations statement than several peer-reviewed items in this corpus. The report also broadens "homeschooled" to include families who merely claimed to be homeschooling without meeting state requirements, which is defensible for a policy argument about enforcement and fatal for any prevalence estimate.

The report's defensible content is about how abuse in these settings comes to light, not about how often it happens. Fewer than a third of database cases were resolved by someone reporting the abuse or by authorities intervening directly; of the reports that did happen, 82% (n = 105) came from non-professionals and only 18% (n = 23) from professionals; 61% of cases surfaced too late — the victim dead or dying — or through circumstances unrelated to the child, such as a parent's involvement with the law for something else. At least 148 cases involve abuse the authors judge to meet most definitions of torture; at least 31 children have died after being withdrawn from school under suspicious circumstances, and withdrawal cases carry higher prior-social-service history and higher fatality rates than other cases in the database. The policy backdrop: all 50 states permit a family recently subject to a social-services investigation to withdraw a child from school, 47 do not bar caregivers convicted of crimes against children from homeschooling, and no state requires a homeschooled child to have contact with a mandated reporter. None of this establishes that home-educated children are abused more often; it establishes that when they are, the ordinary detection machinery is absent — the same detection-versus-incidence distinction Luck et al. 2025 insist on.

Not applicable. The operative problems are the missing denominator, selection of cases on media salience, and reverse causation between maltreatment and withdrawal from school.

Not replicable by construction — a bespoke case database with a media-visibility inclusion rule. The one comparative claim built on it elsewhere (that observed homeschool fatality rates exceed expectation) depends entirely on an assumed denominator, and an advocacy source on the other side used the same database to argue the opposite (40% fewer fatalities than the national average). Two contradictory rates from one uncounted numerator is the whole problem in miniature.