The article explains how researchers distinguish correlation from causation, emphasizing that while randomized controlled trials are the gold standard for determining causal relationships, ethical and practical limits often require alternative study designs — such as natural experiments, instrumental variables, and difference-in-differences approaches — to analyze real-world data. It concludes that understanding causality, especially in complex issues like autism risk, is an ongoing process that relies on combining evidence from multiple rigorous methods over time. Originally published at StatNews.

Elizabeth Stuart co-wrote the piece.