confounding
-
Why Machine Learning Can’t Answer ‘What If?’
Machine learning can predict with stunning accuracy and still mislead about cause and effect. See how causal inference separates prediction from causation.
-
Causality and Study Design: RCT vs Case-Control vs PSM
Which study design supports causality? See how case-control, propensity score matching, and RCTs meet the Bradford Hill criteria, and where each falls short.
-
Propensity Score Matching: Causality Without a Trial
Propensity score matching can support causal claims from observational research, if you design first, check balance and test for hidden confounding.