A causal-inference version of a statistics problem: If you fit a regression model with interactions, and the underlying process has an interaction, your coefficients won’t be directly interpretable. He’s looking for probability puzzles In criticism of criticism of criticism A question about physics-types models for flows in economics What I got wrong (and right) about […] The post On deck this month appeared first on Statistical Modeling, Causal Inference, and…