Confounding and Collapsibility in Causal Inference

Abstract

Consideration of confounding is fundamental to the design and\nanalysis of studies of causal effects. Yet, apart from confounding in\nexperimental designs, the topic is given little or no discussion in most\nstatistics texts. We here provide an overview of confounding and related\nconcepts based on a counterfactual model for causation. Special attention is\ngiven to definitions of confounding, problems in control of confounding, the\nrelation of confounding to exchangeability and collapsibility, and the\nimportance of distinguishing confounding from noncollapsibility.