Homeschooling and conventional school, compared honestly

Where the story comes from — and which claims survive checking

Almost every argument about "the school system" eventually reaches for the same handful of historical claims. We checked them against the primary documents — page scans, the statutes themselves, the original reports. Some hold up. Several do not. The corrections are the useful part, so they are printed here rather than quietly omitted.

One finding is worth stating up front, because it reframes the whole debate: the critics' single strongest piece of evidence is real, correctly attributed, and has been sitting in a 1916 Stanford textbook the entire time. That the popular version reaches for a fabricated Rockefeller quote instead is itself the most telling fact about how this story circulates.

Who actually pays for — and controls — K-12

"Follow the money" is good advice, and following it leads away from Washington. In a normal year the federal government supplies about 8% of K-12 revenue. The rest is state and local, and so is the authority over curriculum.

Read these numbers carefully, in both directions. The 2020–21 through 2022–23 federal shares are inflated by pandemic relief (ESSER), peaking at 13.8%; the 2010–11 figure is likewise inflated by post-2009 stimulus. Both spikes revert. Anyone citing a bare percentage without naming its year is either careless or selling something. The honest formulation: normally about 8%, temporarily as high as 14% during pandemic relief, now falling back.

Source: NCES Digest Table 235.10

How and where you actually homeschool

Homeschooling is legal in all fifty states and DC, but there is no federal homeschool statute. Compulsory attendance is state law, and each state decides how a family may satisfy it outside a public school. The result is a spectrum, usually grouped into four tiers.

• No notice 12 — The family does not have to tell any government body that it is homeschooling. Lightest regulation.
• Notice only 15 — A one-time or annual notice or affidavit of intent, but no testing or evaluation. Light regulation.
• Notice + assessment 9 — Notice plus periodic standardized testing or a portfolio or professional evaluation. Moderate regulation.
• Notice + assessment + approval 15 — All of the above plus parent-qualification requirements, curriculum approval, and/or official sign-off. Heaviest regulation.

“Verified” means the statutory citation was confirmed against the state’s own official code or legislature site — not a mirror such as Justia, Casetext or FindLaw. It does not mean every hours, subjects or assessment detail in the row was independently re-derived from primary text; those summaries also draw on state education-agency guidance and HSLDA state pages. Tier labels compress real differences between states, and several states offer more than one legal route — the alternates are in the notes.

One distinction that catches people out

A full-time virtual charter or public "school at home" academy is public school, not homeschooling. The child is enrolled, the curriculum is the district's, and the family is not exercising a homeschool exemption. This matters legally — and it matters in a co-parenting dispute, because moving a child between those two is a different act with different consequences than it looks.

In Wisconsin there is a further line: under Wis. Stat. § 115.001(3g), instruction that serves more than one family unit is not home-based private education. That is the boundary between a homeschool and a co-op or microschool.

The daily time budget

This is the section of any homeschooling comparison that is easiest to rig, so here is the rule we held ourselves to: every block is either sourced or explicitly labelled an assumption, both days sum to exactly 1,440 minutes, and the teaching parent's hours are carried as a line item rather than left out.

The most important thing on this page may be an absence. No federal or peer-reviewed measurement of homeschool instructional hours exists. The National Household Education Survey asks about reasons, providers, curriculum and subjects — not hours. The widely repeated “homeschool only takes 2–4 hours a day” traces to advocacy organisations and curriculum vendors, not to research. Rather than launder one of those numbers, the homeschool column here is built from Wisconsin’s statutory floor and clearly-labelled assumptions.

An illustrative day — 5th grader

The three numbers that survive scrutiny

1. Commute plus homework is about 1h40 a day — roughly 300 hours over a 180-day year. The categories are certain, but the commute duration is the weakest number on this page (the National Household Travel Survey gives mode share, not duration), so treat 300 hours as an order of magnitude rather than a total.
2. Adult-delivered instruction goes from roughly 1-to-25 to 1-to-1. The mechanism is real, but the honest magnitude is 0.37 SD pooled across 96 randomised trials — not the "2 sigma" figure the homeschool literature repeats, which has never once been reproduced. For parents specifically the effect is the weakest in that meta-analysis.
3. The teaching parent absorbs roughly 3.5 to 4 hours a day. That figure is modelled rather than measured — but its existence is certain. Homeschooling does not remove the hours. It moves them onto a parent.

Wisconsin's 875 hours is the only hard anchor. Wisconsin requires 875 hours a year of home-based private education — about 4 hours 52 minutes a day over 180 days — against its own public-school requirement of 1,050 hours in grades 1–6 and 1,137 in grades 7–12 (NCES Table 1.1, 2020). The homeschool floor is 17–23% lower. This is the only hard number available on the homeschool side of the whole comparison — and even it is softer than it looks, because Wisconsin does not define what counts toward the 875 hours, so compliance is largely on the honour system.

What the outcomes research shows

Homeschooled children, as a group that has been studied, tend to post above-average test scores and do well in college. But almost every study showing this draws on families who volunteered, and homeschooling families differ systematically in income, parental education, two-parent household rates and religiosity — all of which independently predict academic success. When researchers match or control for those factors, the advantage shrinks, splits apart, or disappears.

The single most defensible finding in the whole literature is not "homeschooling works" or "homeschooling fails." It is that structured and unstructured homeschooling produce sharply different results — and that the children most at risk are the ones the research systematically cannot see.

How many children — and why the numbers disagree by a factor of two

Three corrections. The often-repeated "homeschooling is now about 6% of US students" is not what the federal household survey says — NCES puts it at 3.4% for 2022–23. A ~6% figure is reached either by adding the overlapping full-time-virtual category (which is public school) or from an advocacy institute's independent estimate; those are different claims and should not be merged. The Census 11.1% figure is a household rate measured during school closures, not a student rate. And homeschooling declined from 3.4% to 2.8% between 2012 and 2019 — the "relentless growth" narrative is not supported by the federal series for that decade, though NCES itself cautions against comparing across those years, because the school-enrolment threshold changed from 25 to 24 weekly hours.

On child maltreatment there is one documented structural fact and almost no usable comparative data. The structural fact: homeschooling removes a child from routine contact with education personnel, who are the single largest source of maltreatment reports — 18.9% of reports in 2016 — and there are documented cases in which withdrawal to homeschool concealed severe abuse. What cannot honestly be said is anything about rates. The Coalition for Responsible Home Education, which advocates for more regulation, states plainly that no research has compared overall abuse rates between homeschooled children and children who attend school, because the data to conduct such a study does not exist. The widely-circulated 42% figure comes from a self-selected sample by its own authors' admission — the same defect that discredits the advocacy statistics pointing the other way. Anyone quoting you a rate, in either direction, is quoting something nobody has measured.

Notable people — categorised precisely

Bill Gates was not homeschooled. He attended Lakeside School, a private independent school, and then left Harvard. Leaving university as an adult and being educated at home as a child are two entirely different things, and conflating them is the single most common error in "famous homeschoolers" content. Of roughly two dozen names on the standard viral lists, four survived verification unqualified. The rest are partial, mis-categorised, or simply false.

Substantially home-educated

Home education for a substantial portion of their schooling. Each verified individually.

Partly home-educated

Real home-education periods bracketed by real school attendance. Calling any of these people simply “homeschooled” overstates the record — which is exactly what the viral lists do.

Failed verification

These appear on “famous homeschoolers” lists and the claim does not hold up. Showing the corrections is more useful than quietly omitting the names.

College dropouts — a different phenomenon entirely

Every person here attended conventional K–12 schools, several of them elite private ones. What they left was higher education, as adults. This list says nothing about homeschooling; it is here only because these names keep getting imported into homeschooling arguments.

Educated before mass schooling existed

Context, not evidence. Before compulsory-attendance laws spread in the mid-19th century, being taught at home was not an alternative to school — for most children there was no school to be an alternative to. Someone educated at home in 1790 made no educational choice.

Survivorship bias, stated plainly. A list of successful homeschoolers proves nothing about average outcomes — exactly as a list of successful dropouts proves nothing about dropping out. Both lists are selected on the outcome. What they are actually good for is narrower and still worth something: they demonstrate that the path is survivable at the top end, which is a real answer to "no serious person is educated this way," and no answer at all to "what will happen to my child."

When co-parents disagree about it

This is the part that belongs on a co-parenting site, and the framing is the most useful thing a separated parent can learn: in most states a schooling dispute is a legal-custody (decision-making) question, not a physical-placement question. Which box it falls into determines the standard the court applies, who carries the burden, and whether anything can happen at all without a formal modification.

Wisconsin makes it explicit

Wis. Stat. § 767.001(2m) defines "major decisions" as including — expressly — "choice of school and religion." So the answer to "is deciding to homeschool a major decision?" is yes, by the plain text of the statute. And under § 767.001(1s), joint legal custody means neither parent's rights are superior.

The practical consequence. Under joint legal custody, neither parent can unilaterally move a child into — or out of — homeschooling over the other's objection. Withdrawing a child from school to homeschool without the other joint legal custodian's agreement is the exact fact pattern that generates contempt petitions and emergency motions.

Wisconsin presumes joint legal custody is in the child's best interest (§ 767.41(2)(am)). There is also a middle path worth knowing about: because § 767.001(1s) and (2) let a court carve out specified decisions, an order can give one parent final authority over education only while the rest of legal custody stays joint. Where parents are deadlocked on schooling and functional on everything else, that is a recognised middle path.

What courts actually do

Appellate decisions converge on a two-step structure. First, what does the existing order already say about who decides? If educational decision-making is already allocated, the dispute is usually resolved by enforcing that allocation rather than re-litigating the merits of homeschooling. Second, if custody is joint and the parents are genuinely deadlocked — here the states split, and the split is the single most important thing to understand.

What courts weigh on the merits

• The child’s actual academic performance. Measured, documented evidence in the current arrangement — not either parent’s opinion of it.
• Socialisation and outside activities. Whether the child has meaningful contact and activity beyond the household.
• Each parent’s actual involvement. Courts have weighed it. In Staub the appellate court rejected the father’s argument that homeschooling inherently excludes the objecting parent, noting that on his own testimony he had been “relatively uninvolved in the home school program by choice,” and that there was “no evidence that Mother excluded Father from participating in the schooling process.” It cuts both ways: a parent who asked to participate and was refused has a record, and so does a parent who documented what they shared.
• Whether the parents can cooperate at all. A finding that they cannot may itself drive the remedy toward sole decision-making for one parent, which is rarely what either parent was asking for.
• Religion, treated neutrally. Courts decline both to favour and to penalise religiously motivated schooling.
• Compliance with the state’s homeschool law. A documented, demonstrably compliant program is much easier to defend than one that is not — which is an argument for keeping records even in a state that does not require them.

Three practical takeaways, stated neutrally

1. Check the order first. Whether "choice of school" or "education" is already allocated — jointly or to one parent — usually decides both who chooses and what the objecting parent has to prove.
2. Do not act unilaterally. In some states nothing moves until the parents agree or the order is modified, which in practice favours whatever is already happening.
3. Build the record. Attendance logs, hours, curriculum, work samples, assessment results and evidence of outside activities are precisely what courts weighed in Brown and Staub — and they are worth keeping even in a state like Wisconsin that does not require them.

Educational summary of published authority — not legal advice. Read the full opinions and consult counsel in your own jurisdiction.

Sources

Primary documents were read directly — page scans, statutes, and the opinions themselves — rather than summaries of them. Where a source could not be retrieved, the page says so instead of papering over it.

Primary documents

• Frederick T. Gates, The Country School of To-Morrow, GEB Occasional Papers No. 1 (1913) — page scan, p. 6
• Horace Mann, Seventh Annual Report, Mass. Senate No. 13 (1844) — p. 22
• Ellwood P. Cubberley, Public School Administration (1916) — p. 338
• The General Education Board: An Account of Its Activities, 1902–1914
• 20 U.S.C. § 3403(b) · 20 U.S.C. § 1232a
• Wis. Stat. § 767.001 · § 118.165 · § 115.001(3g)

Data and surveys

• NCES Digest of Education Statistics (2025 ed.), Table 235.10 — K-12 revenue by source
• NCES / NHES 2023 — homeschooling at 3.4% of K-12 students, 2022–23
• Census Household Pulse Survey — the 2020 household-rate spike

Research cited in the time budget

• Nancy Karweit, “Time-on-Task Reconsidered,” Educational Leadership (1984) — the “about half the day” figure, and her own warning against reading it literally
• Godwin et al., “Off-task behavior in elementary school children,” Learning and Instruction (2016) — 26–29% off-task
• Kathleen Cotton, “Educational Time Factors,” NWREL (1989) — synthesis of the allocated/engaged-time literature
• Nickow, Oreopoulos & Quan, NBER Working Paper 27476 (2020) — 96 randomised tutoring trials, 0.37 SD pooled; parent-delivered tutoring weakest
• National Household Travel Survey (2017) — school-travel mode share (33.2% bus, 54.2% car); it gives no duration, which is why the commute figure here is the weakest number on the page
• AASM / AAP consensus sleep recommendations — 9–12 hours at ages 6–12

Outcomes literature

• Martin-Chang, Gould & Meuse (2011), Canadian Journal of Behavioural Science 43:195
• Kunzman & Gaither, “Homeschooling: A Comprehensive Survey of the Research” (2020)
• Cardus Education Survey · NHERI research facts (advocacy-affiliated) · Coalition for Responsible Home Education (pro-regulation)
• Elizabeth Bartholet, “Homeschooling: Parent Rights Absolutism vs. Child Rights to Education & Protection,” 62 Ariz. L. Rev. 1 (2020)

Case law was read in full text via CourtListener; each case above links to its opinion. Two jurisdictions in the state table — Georgia and New Mexico — are flagged unverified because their official code hosts returned no statutory text to an automated request.