A 5-layer adversarial-audit engine (dependency-free Python core) plus the two papers that formalise it — managing epistemic circularity, the survivor gate, and three scientific ledgers. It never reports "validated" on internal grounds; only a human does.

Monte Carlo simulation study comparing ANOVA, Welch's ANOVA, and Kruskal–Wallis under small samples

Statistics for Data Analysis | Sample Mean vs. Population Mean and P-Values

Bridging Cost-sensitive and Neyman-Pearson Paradigms for Asymmetric Binary Classification

Console applications for distribution functions and simulations.

Sample R code for paper "Design and analysis of cluster randomized trials with time-to-event outcomes under the additive hazards mixed model" by Blaha, Esserman and Li (Under Review).