Causal Inference Engine
Rigorous RCT analysis with heterogeneous treatment effects (CATE)
Causal Effect Estimation
Runs high-powered RCT (n=400) with randomization to intervention vs active control. Analyzes both average and subgroup effects.
Treatment effect (ATE)
+18% mastery
[+14%, +22%]
Average gain for all children
Heterogeneous effect (CATE)
+32% (High IQ) vs +8% (Low IQ)
Interaction p=0.04
Effect depends on baseline IQ
Compliance effect
ATE corrected: +22% (vs +18% intent-to-treat)
[+17%, +27%]
Adherence matters
Placebo control vs training
+15% placebo, +33% training
[+27%, +39%]
True effect ≈ +18%
Statistical Methods
- • Primary: Intention-to-treat (ITT) analysis
- • Secondary: Per-protocol (PP) adjusted for non-compliance
- • Subgroup analysis: IQ, age, comorbidity interactions
- • Sensitivity analysis: Multiple imputation for missing data
Publication Standard
Pre-registered on ClinicalTrials.gov. Follows CONSORT 2010 reporting guidelines. Submitted to high-impact journal (JAMA, Pediatrics, Child Development).
