Meetup : [Boston] Using Causal Graphs to Understand Bias in the Medical Literature

This talk is a very gentle introduction to modern causality theory, as developed by Jamie Robins and Judea Pearl. We will use as little mathematics as possible to introduce some of the central ideas in the field, with examples from medicine and epidemiology. In the one hour talk, we will cover:

(1) Causal Directed Acyclic Graphs and the rules of D-Separation

(2) Definitions of confounding and selection bias

(3) Methods to adjust for confounding

(4) Examples of situations where standard methods are always biased.

Cambridge/Boston-area Less Wrong meetups are every Sunday at 2pm in MIT's building 66 at 25 Ames St, room 156. Room number subject to change based on availability; signs will be posted with the actual room number.

Our default schedule is as follows:

—Phase 1: Arrival, greetings, unstructured conversation.

—Phase 2: The headline event. This starts promptly at 2:30, and lasts 30-60 minutes.

—Phase 3: Further discussion. We'll explore the ideas raised in phase 2, often in smaller groups.

—Phase 4: Dinner. It's about a ten minute walk to the usual restaurant.