Here is how I currently understand the relationship between correlation and causality, and the collective findings of meta-scientific research: a shockingly large fraction of psychological research and other fields is simple random noise which cannot be replicated ‘everything is correlated’—even things which seem to have no causal relationship whatsoever most efforts to change human behavior and sociology and economics and education fail in randomized evaluation in every field from medicine to economics, when we directly ask how well correlations predict subsequent randomized experiments, we find that the predictive power is poor all variables are part of enormous dense causal graphs ‘folk causality’ often performs badly, especially in extremely complex fields with ambiguous long-term outcomes

• Uncategorized