Data never speaks for itself
Every chart already contains choices: what to measure, whom to include, which scale to use and what to leave out. Good analysis begins with the question: what conclusion are we actually entitled to make?
Ideas
Thinking more precisely about data
Short essays on the ideas behind my courses, lectures and research-oriented teaching.
AI accelerates answers, not validation
A model can generate code, explanations and hypotheses quickly. It cannot take responsibility for data quality, a fair comparison or the consequences of a decision. This makes data expertise more important, not less.
Probability is a language of decisions
Probability is not only about coins and dice. It helps distinguish coincidence from evidence, reason about risk and recognize situations in which intuition fails with remarkable confidence.