The Evidence on Teaching

Behavioral Risk Profiles of Homeschooled Adolescents in the United States: A Nationally Representative Examination of Substance Use Related Outcomes

Hodge DR, Salas-Wright CP, Vaughn MG · 2016

grade Clongitudinalindependentreplicated
Sample
1,321 home-schooled respondents aged 12-17 (NSDUH 2002-2013)
Population
Nationally representative US home-schooled adolescents
Design
Latent profile analysis within the home-schooled population, then multinomial regression on substance-use outcomes. No comparison to schooled adolescents; the contribution is to show the exposure category is not one thing.
Key findings
Four latent subgroups of home-schoolers: (1) highly religious and engaged, (2) limited parental monitoring, (3) high parental warmth and support, (4) secular permissive. Only the highly religious and engaged subgroup was distinctly unlikely to use substances. The authors conclude that pooled home-schooling estimates obscure the variation that actually predicts outcomes.
Genetic confound
High for the outcome levels; the finding used here is about within-category heterogeneity, which is a description of the exposure rather than a causal claim.
Replication notes
Converges with Green-Hennessy 2014's religious-tie split on the same survey series using a different method.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Latent subgroups within the home-schooled populationnumber of profiles4 distinct profiles; substance-use risk concentrated outside the religious/engaged groupself-reportconcurrent, ages 12-17nonenot-applicablebehaviour

Cited by