Week 6

Julia Rohrer — Lecturer, Leipzig University

Thinking Clearly about Correlations and Causation with the help of Directed Acyclic Graphs

Correlation does not imply causation and thus, psychology usually privileges experiments over other designs, both in research and in teaching. But correlations are often all that we can get—for example, because experiments are unethical, unfeasible, or simply impossible when the cause of interest cannot be directly manipulated. In such situations, Directed Acyclic Graphs (DAGs) provide a powerful tool to reason about causality in a principled manner: What assumptions are necessary to arrive at causal conclusions; which third variables should (or should not) be adjusted for? In this talk, I will provide a brief introduction to DAGs and highlight how they are also useful for reasoning about causality in the experimental context.