QLS Seminar Series - Julia Rohrer

Directed Acyclic Graphs as a Tool to Reason about Causality

Julia Rohrer, University of Leipzig
Tuesday April 16, 12-1pm
Zoom Link: https://mcgill.zoom.us/j/86855481591
In Person: 550 Sherbrooke, Room 189

Abstract: Correlation does not imply causation -- but researchers often do want to learn about causation with the help of correlations (even if they may not always be perfectly open about it). In this talk, I will introduce directed acyclic graphs (DAGs) as a helpful tool to reason about causality more systematically. For example, DAGs can be used to illustrate how a large class of methodological problems across fields of research can be explained by a single causal structure, the inverted fork. They can also be expanded in various ways to tackle problems such as missing data or generalization of effects across populations. I will try my best to convince you that everybody, including students, should learn about DAGs to improve their statistical inferences.