ANALYSISApril 13, 2026
Should homeschooling be more strictly regulated?
Following a pandemic-era surge in homeschooling enrollment, lawmakers in multiple states including Illinois, New Jersey, and Connecticut have introduced bills in 2025 to increase oversight of homeschool families, citing child safety concerns. Strong grassroots opposition from homeschooling advocates has stalled or defeated most of these proposals, including Illinois House Bill 2827, which failed to pass the House.
When a parent pulls a child from the public system entirely, who's responsible for making sure that child is actually being educated — and protected? The debate over homeschool oversight forces a collision between family sovereignty and the state's duty to every child.
Not familiar with this story? Get context →
Connecticut 36% figure's actual meaning
C
The Connecticut finding that 36% of children withdrawn for homeschooling had prior abuse files is the most serious data point in this debate, and we should not wave it away. But the population it describes is not 'homeschoolers' — it is families already known to child welfare systems who then removed their children from mandated reporter contact. That sequencing points to a targeted problem, not a systemic one.
L
You're right that it's a self-selected subset, but that's precisely what makes it alarming rather than dismissible. A child with an open abuse file being legally withdrawn from the one setting where an adult is required by law to report what they see isn't a statistical artifact — it's a description of how the loophole works in practice.
C
Then the remedy should fit that specific mechanism: cross-agency data-sharing that triggers welfare checks when a child with an open file is withdrawn from school. That targets the actual harm. GED requirements and portfolio reviews do not — they burden millions of families whose children have no such file.
L
Cross-agency data-sharing only works if there's a registration requirement to generate the data in the first place — which is exactly what 42 states currently don't require, and what homeschool advocates have successfully blocked.
Whether oversight equals parental distrust
C
From Wisconsin v. Yoder forward, American jurisprudence has recognized parental direction of education as constitutionally grounded, not a state-granted privilege. The 83% of families citing school environment concerns and the 75% motivated by moral instruction are exercising exactly the kind of pluralism a free society should protect — not a risk factor requiring supervision.
L
Yoder was decided on the specific religious liberty claims of an intact Amish community — the Court was not issuing a blanket bar on all state interest in children's welfare. Requiring a GED and proof that education is occurring isn't treating parental motivation as suspect; it's the state meeting its parens patriae obligation to children who cannot advocate for themselves.
C
Parens patriae is the state's backstop for children with no functioning parent — invoking it to justify universal monitoring of families who have done nothing wrong inverts its purpose. The doctrine targets specific failure, not general parental autonomy.
L
The problem is that we currently have no mechanism to identify which families constitute 'specific failure' until a child is already seriously harmed — that's the gap a basic registration floor would close.
Data vacuum preventing evidence-based policy
C
The absence of a national reporting standard cuts both ways: it means we do not have evidence of a homeschooling abuse epidemic, not merely that we have failed to find it. The HSLDA's Will Estrada is correct that neither of the two peer-reviewed studies produced supports the abuse-rate claims being made legislatively — when data is genuinely disputed, the burden of proof falls on the party proposing to restrict liberty.
L
You're using the data gap as an argument against the very registration requirements that would close it — that's a circular defense. We can't calculate a reliable abuse rate when 42 states require no registration and generate no denominator. Basic registration wouldn't answer the empirical question, it would make the empirical question answerable.
C
That logic would justify registering any private activity where we lack complete data. The burden is on proponents to demonstrate a problem exists before building regulatory infrastructure — not on families to generate the data that might or might not vindicate the suspicion.
L
We already have a partial signal: the Connecticut 36% and 500-plus CRHE cases are the data we can see precisely because those children intersected with institutions. The unknowns aren't distributed randomly — they cluster where visibility is lowest.
Whether proposed remedies fit the harm
C
Abuse is already illegal. The proposed remedies — portfolio reviews, parental education minimums, curriculum alignment requirements — would not have stopped the specific cases being cited by proponents. Mandating a GED does not protect a child from a parent who holds a GED and intends harm. The regulatory architecture is being built to address a causal mechanism that hasn't been established.
L
You're setting a bar no policy could clear: if a regulation doesn't guarantee prevention of the worst-case abuser, it's invalid. But annual check-ins and basic registration don't need to stop the determined abuser to be worth doing — they create routine contact that makes isolation harder to sustain and generates the institutional record that lets intervention happen earlier.
C
Routine contact with families who have no risk indicators is surveillance in search of a problem. If the goal is earlier intervention for at-risk children, enhanced protocols for the families already in the child welfare system would accomplish that without treating every homeschooling parent as a suspect.
L
The 2025 Connecticut case — the 11-year-old whose abuse only surfaced after years of legal invisibility — had no prior welfare file. Your targeted-intervention model has no mechanism to find that child.
Democratic opposition as policy signal
C
The 40,536 witness slips against Illinois HB 2827 — compared to the typical few dozen for most bills — represent actual homeschooling families, not a lobbying apparatus. That democratic signal deserves to be taken seriously. When the people most affected by a policy mobilize at that scale, the presumption should be that the policy is getting something wrong.
L
Political mobilization tells you about organizational capacity, not policy merit. The HSLDA has been running this playbook since 1983 — they are genuinely good at turning out opposition. But the families who didn't file witness slips include every child currently being isolated in a home where no one is required to check.
C
The HSLDA framing dismisses tens of thousands of individual families by attributing their opposition to organizational astroturfing. These are parents who showed up because they believe their own rights are at stake — that's not a coordination problem, it's a constituency.
L
I'm not dismissing them — I'm noting that one side of this debate can organize and the other can't, because one side is adults with political agency and the other side is children with none.
Conservative's hardest question
The Connecticut finding that 36% of children withdrawn for homeschooling had prior abuse or neglect reports on file is the most difficult data point to dismiss — it suggests that in at least some cases, homeschooling is being used specifically to remove at-risk children from mandated reporter networks, which is a genuine and targeted harm that basic notification requirements alone may not address.
Liberal's hardest question
The HSLDA's Will Estrada correctly notes that only two peer-reviewed studies have examined abuse rates in homeschool settings, and neither conclusively establishes that homeschooled children face higher risk than public school peers — meaning the Connecticut 36% figure, while alarming, reflects a self-selected subset of families already known to authorities, not the homeschool population as a whole. A liberal case for regulation built on abuse prevention must honestly acknowledge that the causal link between homeschooling itself and elevated abuse risk remains statistically unproven.
Both sides agree: Both sides acknowledge that the Connecticut 36% figure reflects a self-selected population already known to child welfare authorities, not a representative sample of the broader homeschooling community.
The real conflict: The core factual dispute is whether the existing peer-reviewed literature is sufficient to establish elevated abuse risk in homeschool settings: conservatives treat the two available studies as exonerating, while liberals treat the data vacuum itself as evidence of a policy failure that forecloses any exoneration.
What nobody has answered: If basic registration requirements were implemented nationwide and a decade of data showed homeschool abuse rates statistically indistinguishable from public school rates, would pro-regulation advocates accept that as grounds to roll back oversight — and if not, what does that reveal about whether child safety is actually the load-bearing argument?
Sources
- Johns Hopkins Institute for Education Policy, Homeschool Research Lab — 2023–24 enrollment data
- Coalition for Responsible Home Education (CRHE) — Make Homeschool Safe Act (July 2024), abuse tracking data, quotes from Tess Ulrey and Jonah Stewart
- Home School Legal Defense Association (HSLDA) — quote from Will Estrada, opposition positions
- Illinois General Assembly — House Bill 2827 witness slip data and legislative status
- Connecticut state legislative record — Rep. Jennifer Leeper quote, 2025 abuse cases, 2018 abuse/neglect report
- New Jersey Assembly Bill 5825 — proposed homeschool curriculum and portfolio requirements
- Virginia Department of Education — homeschool enrollment figures, 4.3% and 6% annual increases
- 2023 parental motivation study — school environment, moral instruction, family life statistics