Dynamic treatment effects on socio-emotional skills I Notes: This figure shows the treatment effect of the PATHS intervention on children's socio-emotional skills from ages 7 through 15. The dependent variable in Panel (a) is ADHD symptoms (disruptiveness and impulsiveness). The dependent variable in Panel (b) is opposition and defiance. All dependent variables are indices standardized to mean zero and a standard deviation of one. All models include controls for baseline child and household characteristics. Child controls include the age and gender of the child, having Swiss citizenship, measures for anxiety and depressivity, ADHD, non-aggressive externalizing problem behaviour, non-aggressive conduct disorder, opposition and defiance, prosociality, four measures of aggressive behaviour, and four measures of overall behaviour. Household controls include household income, mother's and father's education level, age of the mother, indicator variables for the mother's having Swiss citizenship and being born in Switzerland, and indicator variables for a single-parent household, a household that received financial aid, and a household that experienced financial problems. To provide evidence on balance across the treatment and the control groups, we do not include individual controls in the estimation of the treatment effect at age 7. For each SBQ measure, we combine measures from teacher and parent reports by taking the average of the two standardized indices and standardizing the resulting index again. For measures at ages 8, 10, 12, 13, and 15, we rely solely on teacher reports, as there are no parent surveys at these times. Details on the SBQ items and construct reliability are provided in Supplementary Appendix B. Shaded areas indicate the baseline and the intervention periods. The dashed vertical line shows the time when tracking into secondary schools takes place. All models include strata fixed effects for the level of randomization. Each point estimate is shown with the respective 90% and 95% confidence intervals calculated based on standard errors clustered at the school level.