Mike Ion

I develop statistical methods to measure and simulate effective teaching at scale. Using natural language processing and Bayesian modeling on large-scale conversational data, I build classifiers that detect patterns predicting learning outcomes: breakthrough moments, scaffolding moves, and engagement dynamics.

My research combines measurement with simulation. I build statistical classifiers from observational data, then use validated simulation to test hypotheses about which teaching strategies work best. This approach enables testing at scales impossible with human participants alone.