This folder contains a curated set of empirical Jupyter notebooks showcasing how RobustiPy can replicate classic results and explore robust inference in applied economics and related fields. The examples emphasize bootstrapping, specification variation, and model‑agnostic diagnostics, using modern Python tooling (e.g., statsmodels, pandas) together with RobustiPy’s robust pipelines.

Brief summaries of each notebook:

• empirical_type1_acemoglu.ipynb — Replicates The Colonial Origins of Comparative Development; uses statsmodels + RobustiPy to probe specification sensitivity.
• empirical_type1_ehrlich.ipynb — Replicates Participation in Illegitimate Activities: Ehrlich Revisited (1960) using RobustiPy for robust inference.
• empirical_type1_union.ipynb — Uses union.dta (NLS Young Women 1968–1988; missing 1974, 1976, 1979, 1981, 1984, 1986) to study wages vs. union membership with robustness checks.
• empirical_type2_mrw.ipynb — Revisits MRW (1992) with PWT v10.01; tests Solow covariates log(n+g+δ) and log(I/Y) via two RobustiPy runs and inspects adjusted R² distributions.
• empirical_type3_asc.ipynb — Adult Social Care example based on Zhang, Bennett & Yeandle (BMJ Open). Data are not public due to UKHLS–ASC linkage; contact the authors for replication materials.
• empirical_type3_ukhls.ipynb — Example using Understanding Society: Longitudinal Teaching Dataset (Waves 1–9, 2009–2018). Registration is immediate; documentation is provided with the dataset. Demonstrates applying RobustiPy to longitudinal survey data.
• empirical_type4_airline.ipynb — Predicts airline customer satisfaction (129,880 rows, 22 features). Factors include punctuality, service quality, and other operational metrics; demonstrates robust modelling for service improvement insights.
• empirical_type4_framingham.ipynb — Uses the Framingham heart disease dataset (4,240 records, 16 columns) to predict 10‑year CHD risk. Demonstrates the LRobust class for binary outcomes, leveraging a large feature space and focusing on out‑of‑sample accuracy.
• empirical_type5_gino.ipynb — Shows use of RobustiPy with multiple dependent variables by replicating the retracted Gino et al. (2020) study. References Data Colada’s blog discussions of the case.
• empirical_type5_orben.ipynb — Multiple dependent variables; replicates Orben & Przybylski (Nature Human Behaviour, 2019). Data via the UK Data Service; reference code available on Amy Orben’s GitHub.

To run these notebooks, you must have the Jupyter Notebook stack installed alongside RobustiPy and its dependencies. This typically includes:

• jupyter / notebook (or jupyterlab)
• robustipy