A Systematic Review of the Empirical Research on Selected Aspects of Homeschooling: Updated 2017–2026

24 studies on academic achievement, 25 on social and emotional development, 24 on adult success

Studies of US (and some international) homeschooled students and homeschool graduates. Inclusion is restricted to peer-reviewed studies using representative samples — a stated tightening over the author's 2017 review, which is already in this archive.

A vote-count review, not a meta-analysis: outcomes are tallied as positive, mixed or negative with no effect sizes pooled, no quality weighting, no risk-of-bias assessment and no design hierarchy. The inclusion filter is "peer-reviewed and representative-sample", which screens on sampling but not at all on identification — a nationally representative cross-sectional correlation counts the same as a within-student design would. Under this archive's rule that a review inherits the design quality of what it aggregates, and given that the underlying literature contains no randomised, lottery, discordant-sibling, regression- discontinuity or difference-in-differences study of homeschooling achievement, the review cannot rise above grade D whatever its arithmetic. Brian Ray is founder and president of NHERI, the advocacy research institute; recorded from the published abstract, the full text being behind a Cloudflare-protected paywall.

Fifteen of 24 studies (62%) on academic achievement reported a positive result for homeschooled students versus institutionally schooled students, with "a few" mixed or negative; 16 of 25 (64%) on social and emotional development positive; 13 of 24 (54%) on adult success positive. Two things in that tally matter more than the headline. First, a 54-64% positive vote count is a weak result even taken at face value — roughly a third to a half of studies fail to find the advantage. Second, the review's own recommendation is that the literature be enhanced by "well-controlled NON-EXPERIMENTAL design research" — the field's most committed advocate, writing in 2026 with the whole literature in front of him, does not propose a causal design and does not claim one exists. This archive treats that as the strongest available concession that the counterfactual has never been constructed.

High, inherited wholesale. Every study aggregated compares self-selected homeschooling families to institutionally schooled children without prior achievement, without random or quasi-random assignment, and without within-family comparison. Vote-counting confounded designs does not reduce confounding; it multiplies it and hides the variance.

Updates the same author's 2017 systematic review (already in this archive) and reaches the same conclusion by the same method, which is continuation rather than replication. The independent counterpart review, Kunzman & Gaither 2020, surveys much of the same literature and reaches a markedly more cautious verdict about what the designs can support.