The potential outcome approach to causal inference will be introduced and statistical methods for inferring causal effects from randomized clinical or observational studies will be presented. Examples and practical sessions will be based on case studies in biostatistics, epidemiology, and public health.
- Introduction to causal inference from the potential outcome perspective
- Design and analysis of randomized clinical trials (RCT): Fisher’s exact p-values, estimators of average causal effects, regression, imputation-based approaches
- Practical session 1: Analyzing an RCT on the effect of statins on cholesterol
- Design and analysis of observational studies under confoundedness: the role of the propensity score; matching, weighting, regression estimators
- Practical session 2: Analyzing an observational study on the effect of statins
- Beyond RCTs. Intercurrent events: challenges and opportunities. Presentation of a case study with discussion
Lecture notes, slides, data, articles and other reading material will be distributed before the course.
Practical sessions will be in R but no a priori knowledge of R is required.
Imbens G., Rubin D.B. (2015) Causal Inference for the Statistics, Social and Biomedical Sciences: An Introduction, Cambridge University Press
Statisticians, health professionals with statistical background, master and PhD students.
Statistical inference, multivariate analysis are taken for granted.
Lectures with some practical sessions/examples