Training & Retention in the Federal Workforce
People analytics · Personal project · Nov 2026
- Designed and directed a study of 15 years of OPM Federal Employee Viewpoint Survey data (1.65M responses, 29 agencies) asking whether employees who feel well trained are less likely to plan to leave. AI-assisted analysis in Python (DuckDB, statsmodels) through a reproducible pipeline.
- Found the raw ~23-point gap in intent to leave shrinks to about 5 points after controlling for supervisor support, workload and tenure. Year-to-year rises in agency development scores weren’t followed by less leaving (p = 0.81), so feeling well trained flags retention risk but isn’t shown to be a lever on its own.