Homeschooling statistics: how many families, and who they are

In the 2022-23 school year, 3.4 per cent of US students aged 5 to 17 were homeschooled, according to federal survey data. An independent estimate for 2024-25 puts the figure at roughly 3.4 million students, or 6.3 per cent. Both numbers are defensible, and the gap between them is the most interesting thing in homeschooling statistics.

If you have seen homeschooling described as both "a small but growing minority" and "larger than Catholic school enrolment", you have met this gap. Here is what each source actually measured.

The federal numbers

The National Center for Education Statistics runs the Parent and Family Involvement survey as part of the National Household Education Surveys programme. Its most recent release reports, for 2022-23:

• 3.4 per cent of students aged 5 to 17 with a K-12 grade equivalent were homeschooled.
• 2.5 per cent were enrolled in full-time virtual education, regardless of whether the parent called it homeschooling.
• 5.2 per cent received instruction at home when both groups are counted together, up from 3.7 per cent in 2018-19.

The longer series in the NCES digest tables shows the trajectory before that: 1.7 per cent in 1999, 2.2 per cent in 2003, 3.0 per cent in 2007, and 2.8 per cent in 2019.

The strength of this source is its method: a large, nationally representative household survey with published sampling error. Its weakness is that it depends on a parent identifying their child as homeschooled, and the virtual-school boundary makes that genuinely ambiguous.

The independent estimate

The National Home Education Research Institute builds its figure differently: it combines state registration and enrolment data from 23 states with Census Bureau survey data, then adjusts upward for families it believes the data misses. Its 2024-25 estimate is about 3.408 million school-age homeschoolers, or 6.262 per cent of the school-age population, with a stated range of 3.07 to 3.75 million.

For scale, that estimate puts homeschooling at roughly twice US Catholic school enrolment and approaching public charter school enrolment.

The strength here is that registration data counts actual families rather than survey responses. The weaknesses are that states with no registration requirement are invisible to it, that the 23 states are not a random sample, and that the upward adjustment for undercounting is a judgement rather than a measurement.

Why the gap is so large

None of these is a mistake. They are different questions wearing the same words. The useful habit is to ask, of any homeschooling figure: who was counted, in what year, and by what method.

What is not in dispute

Three things hold across every source.

Homeschooling grew sharply after 2020 and did not return to its previous level. The federal series and the independent estimates disagree on magnitude and agree on direction.

The population diversified. The post-2020 growth was not concentrated in the groups that dominated homeschooling historically, and the reasons parents give have broadened well beyond the religious motivations that characterised earlier decades.

Home instruction is more common than homeschooling alone suggests. Once full-time virtual schooling is included, roughly one US student in twenty was being taught at home in 2022-23.

The claim to be careful with

You will see it asserted that homeschooled students score far above average on standardised tests. Treat this carefully. Most studies supporting it recruit volunteers through homeschooling organisations and compare them with national averages. Families who volunteer to be tested are not a random sample of homeschoolers, and homeschooling families differ systematically from the general population in income, parental education and parental involvement. That design cannot separate the effect of homeschooling from the effect of who homeschools.

This cuts both ways: it does not show homeschooling is worse either. It shows the question is open, and anyone quoting a clean number in either direction is overselling the evidence.

Who homeschools now

The federal survey asks parents why, and the answers have broadened considerably since the 1990s, when religious instruction dominated. Concern about the school environment - safety, bullying, peer pressure - now ranks at or near the top in most years, alongside dissatisfaction with academic instruction and a child's specific needs.

Two changes since 2020 are worth stating carefully, because they are frequently overstated:

The population diversified. Post-2020 growth was not concentrated in the demographics that historically dominated homeschooling. The federal data shows growth across groups rather than within one.

Reasons broadened. Families who began during school closures and continued are a different cohort from those who chose it in 2015, and their stated reasons reflect that.

What the data does not support is a confident claim about the political or religious composition of the current homeschooling population. Survey categories are coarse and the composition is genuinely contested.

State-level variation

The national figure hides enormous range. The independent 2024-25 estimate drew on registration data from 23 states, and the per-state share of all students varied from about 1 per cent to nearly 10 per cent.

That variation is partly real and partly an artefact of what each state requires. A state with a registration requirement produces a count; a state with none produces an absence. Anybody comparing two states should check whether they are comparing two measurements or one measurement and one gap.

How to read a homeschooling statistic

A short checklist, because almost every figure in circulation is missing at least one of these:

1. Who was counted? Homeschooled only, or including full-time virtual school?
2. Which year? The period from 2019 to 2023 saw real change; a 2019 figure is not current.
3. What method? Household survey, state registration, or an adjusted estimate.
4. Who adjusted it, and by how much? An upward correction for undercounting is a judgement, and it should be stated rather than folded in silently.
5. Is this a share of students or a count? Both move, and they do not move together when the school-age population changes.

Where TruLearn fits

We build for the practical problem behind these numbers: a growing group of families teaching subjects they did not plan to teach. TruLearn teaches AP Biology and introductory college biology by voice, asks the student to explain each idea back, and grades that against a rubric written by a person.

We have no outcome data of our own yet, and we say so on our research page rather than adding another unverifiable statistic to this field.

More reading: how to teach a subject you don't know, and where homeschooling is heading.