Homeschooling statistics 2026 — and why every number you read is wrong

Every homeschooling statistics article opens with a confident number. Almost all of them are quoting each other, and the original source usually measured something narrower than the sentence implies.

So this piece does two things: gives you the best available figures, and explains why they disagree — because understanding the disagreement is more useful than memorising a number you will later have to defend.

Why the numbers are a mess

Four structural problems, none of which are going away.

1. Eleven US states require no notification of any kind. Alaska, Connecticut, Idaho, Illinois, Indiana, Iowa, Michigan, Missouri, New Jersey, Oklahoma and Texas have no register to count. Any state-data-based national figure is missing those families entirely.

2. Household surveys and administrative data measure different things. A survey asks families to self-identify; a registration count records those who filed a form. Families using an umbrella school, a charter programme or a hybrid model may appear in one and not the other — or in both, or in neither.

3. The definition is contested. Is a child in a public charter homeschool programme home educated? A child in a two-day-a-week hybrid school? A child registered with a distance provider? Different sources answer differently, and the answer moves the total by hundreds of thousands.

4. Some of the most-cited research uses volunteer samples. Studies recruiting participants through homeschool organisations reach engaged, organised, often long-term homeschooling families. Those are exactly the families most likely to produce strong outcomes — which is precisely why the results cannot be generalised.

The United States

Scale. Credible mid-2020s estimates cluster between roughly 3 and 4 million children — approximately 5–6% of the school-age population. A decade earlier the comparable figure was around 3%.

Trajectory. The shape matters more than the level:

The genuinely significant finding is the last row. The pandemic spike was temporary; the new floor was not. A large share of families who tried home education under duress chose to continue.

Who. The demographic shift is the most under-reported part of the story. Twenty years ago the median US homeschooling family was white, religious and rural. Growth since 2020 has been fastest among Black, Hispanic and Asian families, among secular families, and in urban areas. Reasons given have shifted correspondingly: concern about the school environment now outranks religious motivation in most surveys.

The United Kingdom

For years there was no national figure at all. The Department for Education’s census of local authorities produced the first systematic counts: approximately 92,000 children in elective home education in autumn 2023, rising to roughly 111,700 a year later.

Both are undercounts, and knowingly so. A child who never attended school need not be known to the council, and registration was not mandatory when these figures were collected. The direction of travel — a substantial year-on-year increase — is more reliable than the level.

The most commonly cited reasons in UK local authority data are mental health, dissatisfaction with special educational needs provision, bullying, and philosophical preference — a noticeably different profile from the US.

Elsewhere

Reliable comparative data barely exists, but the direction is consistent:

• Canada — provincial registration data shows large post-2020 increases, with Alberta and British Columbia the most documented because both fund registered home educators.
• Australia — state registration numbers have risen sharply, roughly doubling in several states over the last decade.
• Ireland, New Zealand, the Philippines and the Gulf states have all recorded substantial growth.
• Germany, Sweden, Greece and Brazil remain the significant restrictive cases, so the practice is largely invisible there by design.

See homeschooling around the world for the legal picture behind these numbers.

What the outcome research does and does not show

This is where the most misleading claims live, in both directions.

Frequently claimed: homeschooled students score 15–30 percentile points above the national average. The problem: most such studies use volunteer samples, and volunteer samples over-represent organised, engaged, resourced families. The finding tells you about those families, not about home education.

What better-designed studies suggest: smaller or no significant differences once family background is controlled for. The defensible conclusion is that home education does not, on average, harm academic outcomes — which is a weaker claim than advocacy usually makes and a stronger one than critics usually allow.

Where the evidence is more solid: socialization. Multiple studies using different methods consistently find home-educated children socially typical or better on most measures, with the important caveat that this depends on families actually arranging regular social contact.

What this means for your family

Almost nothing, and that is worth saying.

National averages describe populations. Your child’s outcome will be determined by whether they are read to, whether the work is at the right level, whether they have friends, and whether the adult teaching them is well enough to keep going. None of those appear in any of these tables.

The statistics are useful for exactly two things: reassuring a worried grandparent, and arguing about policy. For deciding what to do on Tuesday morning, they are irrelevant.

Where to go next

• What the research says about homeschooling outcomes — three specific studies, examined
• Homeschool socialization: the real answer — where the evidence is genuinely strong
• Homeschooling around the world — the legal picture behind the numbers
• Homeschool laws by state — including which states count and which do not