Student Seminar in Statistics

Topic: Causality

Description:

Many traditional statistical techniques focus on associations between variables. In practice, however, one is often much more interested in causal relationships between variables. In this seminar, we study ways to make such causal inference. Special emphasis is placed on the assumptions that underly causal inferences, the languages used in formulating those assumptions, and the conditional nature of causal claims inferred from nonexperimental (i.e. observational) studies. These include the control of confounding, the assessment of causal effects and the interpretation of counterfactuals.

The talks must be given in English.

Literature:

Judea Pearl. Causality - Models, Reasoning, and Inference. Cambridge University Press, NY, 2008.

Judea Pearl. Statistics and causal inference: A review PDF

For additional material please check the program.

Program:

The assignment of the talks will take place on Monday, February 16.
List with possible topics for a talk + further references.
Schedule of talks.

Synopses:

• Design of Experiments

• Introduction to Graphs and Causal Models

• Functional Causal Models

• SGS- and PC-Algorithm

• Identification of Causal Effects

• Calculus of Intervention and Graphical Tests of Identifiability

• When is the effect of an Action identifiable?

• Graphs, model testing, and Identifiability

• Simpson's Paradox: an Anatomy