Each notebook solves a specific analytical problem using methods appropriate to the data structure and identification challenge.
Methods for estimating causal effects when randomization isn’t possible:
| Notebook | Use When | Key Technique |
|---|---|---|
| Heterogeneous Treatment Effects | Effects vary across subgroups | Causal Forests, Double ML |
| Synthetic Control | Single treated unit, no parallel trends | Weighted donor pools |
| Regression Discontinuity | Policy threshold exists | Local polynomial regression |
Production-ready templates for common policy questions:
| Notebook | Policy Domain | Data Sources |
|---|---|---|
| Labor Market Intelligence | Workforce development | BLS, Census, QCEW |
| Opportunity Zone Evaluation | Place-based programs | Census, IRS, HMDA |
| Workforce Development ROI | Program evaluation | Administrative earnings |