Bayesian Mixed-Effects Models

All models will control for baseline symptom levels and baseline dysfunctional cognition levels. The models will include random intercepts for trials and random slopes for key predictors. Missing data will be handled through multiple imputation under a missing-at-random assumption within the Bayesian framework. Between-study heterogeneity will be estimated through variance components with 95% credible intervals. Bayesian linear mixed-effects models will test whether change in dysfunctional cognitions predicts follow-up outcomes. Analyses will be conducted separately for child-reported symptom severity and clinician-rated clinical severity.