Civic Education in High School and Voter Turnout in Adulthood

Not recorded; a large nationally representative longitudinal adolescent survey with linked high-school academic transcripts, analysed with sibling family fixed effects

US adolescents followed into adulthood, with high-school social studies and civics coursework taken from academic transcripts rather than recall.

The methodological advance is measurement plus family fixed effects. Civics exposure comes from actual TRANSCRIPTS, not student recall, which removes the recall bias that inflates most of this literature. The authors then run sibling family fixed-effects models, absorbing every observed and unobserved influence shared by siblings raised in the same family - socialisation, parental politics, and, importantly for this archive's premise, a substantial share of shared genetic and environmental background. Sibling fixed effects are weaker than a discordant-MZ-twin design but stronger than covariate adjustment. LIMITATIONS: the turnout measure could not be confirmed from the paper and is most likely survey self-report rather than a validated voter file; and coursework is not randomly assigned even within family, so course-taking motivation within siblings remains a live confound. READ DEPTH CAVEAT: abstract only, via the publisher record; no coefficients extracted.

Civics coursework does not produce voters. In initial regressions civics courses appear to raise adult turnout. After individual and family controls the effect becomes 'fairly limited', with a few measures surviving. After sibling family fixed effects the authors conclude 'there is a null relationship between civic education and turnout; the best-case scenario is that any civic education effects that do exist are likely very small'. The specific null that matters most here: courses focused on SERVICE LEARNING, CIVIC SKILLS DEVELOPMENT or POLITICAL ISSUES - the three things the practitioner literature recommends over rote civics - showed no effect on adult turnout, which the authors note runs against the resource model of participation.

Medium. Sibling fixed effects absorb shared family environment and roughly half of additive genetic variance between siblings, which is why the association shrinks to nothing when they are applied - the raw association was substantially family selection. What remains unaddressed is within-family selection into civics courses by the more politically inclined sibling, which would bias the remaining estimate upward, not downward. So the null is if anything conservative.

Independently converges with Jung and Gopalan 2024 (state civics-test mandates, difference-in-differences, null on turnout) and Manning and Edwards 2014 (systematic review, little evidence on voting/registration). Three different designs, three different data sources, same answer.