Lecturer: Stijn Vansteelandt (University of Ghent, Belgium).
The course will be held on June 9th
to 18:00 and on June 10th
from 9:00 to 13:00.
Registration fee for the course (100 EUR for conference participants,
120 EUR for others) covers course materials, coffee breaks, dinner on June 9th
and lunch on June 10th.
Recent developments in causal inference within the statistical and artificial
intelligence literature have led to important new insights on how to address
problems of confounding and selection bias in a wide variety of settings.
The aim of this course is to review these developments and to provide
state-of-the-art statistical solutions for dealing with these problems.
The first half day of the course will focus on probabilistic graphical models
to express causal background knowledge and on d-separation to assess whether
a given data analysis suffers problems of confounding and selection bias
and whether/how this can be accommodated. This part of the course will introduce
the basic ideas, illustrate how graphical models can be used in a variety of
settings and discuss more advanced identification results (G-computation).
The second half day of the course will focus on statistical techniques
to adjust for measured confounding. Specifically, we will discuss limitations of
ordinary regression adjustment and focus on successful alternatives,
such as inverse probability weighting estimators in marginal structural models
and G-estimators. The motivating problem that will be discussed throughout the course
will be that of inferring whether a given exposure has a direct effect on a given
outcome other than through its influence on an intermediate variable.