Homeschooled adolescents in the United States: Developmental outcomes

pooled 2002-2011 National Surveys on Drug Use and Health, adolescents aged 12-17

Nationally representative US adolescents, home-schooled vs traditionally schooled, split by strength of religious affiliation

Repeated cross-sections from a nationally representative survey with demographic controls; the design contribution is splitting home-schoolers by religious tie rather than treating them as one population. Self-reported outcomes, no prior-achievement control.

Pooling home-schoolers hides the result. Home-schoolers with WEAKER religious ties were three times more likely to report being behind their expected grade level and two and a half times more likely to report no extracurricular activity in the prior year than traditionally schooled peers, and were more likely to report lax parental attitudes to substance use. Home-schoolers with stronger religious ties did not show this pattern. The author identifies the weaker-religious-tie group as an at-risk population.

High. Families with weak community ties who withdraw a child from school differ from families who do not in ways that are both heritable and directly causal, and the survey cannot separate reactive home-schooling (withdrawing a struggling child) from elective home-schooling.

The heterogeneity finding is corroborated by Hodge 2016's latent-profile analysis on overlapping NSDUH waves and by Schepis 2020's disaggregation of the comparison group; the pooled-positive picture (Vaughn 2015) comes from the same data analysed without those splits.