Abstract
As highlighted in Dawid (2000) and Pearl & Mackenzie (2018), deducing the causes of given effects is a more challenging problem than evaluating the effects of causes in causal inference. For the case with a single causal variable, the probability of causation and the probability of necessity have been used to assess causes of effects. For a case with multiple causes that may affect each other, we propose the posterior causal effects based on observed evidence, as a measure of causes of effects. Since posterior causal effects involve probabilities of counterfactual variables, their identifiability requires assumptions of no confounding and monotonicity beyond those needed for traditional causal effects; we present these assumptions and provide the corresponding identification equations. We further extend this framework to settings with multiple effect variables. The proposed approach applies broadly to causal attribution, medical diagnosis, and the assessment of blame and responsibility in studies with multiple effect or outcome variables, and we illustrate it through numerical examples.
Speaker Bio
Dr. Zitong Lu is a Postdoctoral Fellow in the Department of Statistics and Data Science at the Chinese University of Hong Kong. He received his Ph.D. in Systems Engineering from City University of Hong Kong and a B.Sc. in Statistics from Peking University. His research focuses on causal inference, particularly causal attribution and individual treatment effects.