Childhood Skill Development and Adult Political Participation

891 children randomized (445 treatment, 446 control) in 55 cluster-randomized high-risk schools; 812 analysed

High-risk kindergarteners identified as at risk of conduct problems in four US communities (Durham NC, Nashville TN, rural central Pennsylvania, Seattle WA), three cohorts starting 1992. Largely low-income children of low-participation parents; only ~26% of the control group ever voted 2004-2012. Outcomes measured at ages 26-29.

Cluster-randomized (55 schools) multi-year, multi-component psychosocial intervention running through elementary school: social-skills training, peer pairing, parent groups, home visits, tutoring. It contained NO political content whatsoever - Fast Track was designed to prevent conduct problems, not to make citizens. Prior CPPRG publications establish that it moved self-control, self-efficacy and empathy while leaving COGNITIVE ability essentially unchanged, which is what makes it a clean test of the noncognitive channel. In July 2014 Holbein matched participants to state voter files on first name, last name and birth date; attrition and mover rates were balanced across arms (91.7% vs 93.3% had a current address, p=0.38). Outcome is VALIDATED turnout in federal elections 2004-2012, not self-report. Programme attendance decayed over six years (77%, 72%, 71%, 64%, 59%, 53%). The multi-component design means the active ingredient cannot be isolated - the author says so explicitly.

Assignment to a childhood psychosocial-skills programme with no political content raised adult validated voter turnout by 7.3 percentage points (ITT, 95% CI 0.02 to 0.13) against a control base rate of about 26% ever voting - a 30-40% relative increase. Treatment-on-treated was 11.1 points (95% CI 0.03 to 0.20). On share of elections voted, ITT was 9.4 points (95% CI 0.0004 to 0.19, p=0.049) and TOT 14.0 points (p=0.052). The programme moved people from never voting to voting at least once. Mediation analysis (descriptive, not causal) points away from cognitive ability, income and social status and toward the psychosocial skills the programme actually targeted. THE POINT FOR THIS CLUSTER: the one clean school-based RCT that has been shown to move validated turnout taught NO CIVICS AT ALL. It taught self-regulation.

Low. School-level randomization with balanced attrition, so heritable traits are balanced in expectation across arms. Residual risk is voter-file match quality, which the author shows is balanced.

Holbein re-analysed data he did not generate (the Conduct Problems Prevention Research Group's Fast Track trial), so the voting result is an independent secondary analysis of a randomized trial. A 2021 Prevention Science paper reports converging effects of early-childhood preventive interventions on turnout, and Cohodes and Feigenbaum independently identify the same noncognitive mechanism in a lottery design - but neither was read here, so this is recorded as unclear rather than replicated.