Bianca De Stavola, Eduardo Fe, Rhian Daniel, and Andrea Aparicio Castro
This course will introduce participants to the two main approaches to estimating causal effects from observational data: those based on the assumption of no unmeasured confounding and those that exploit the availability of instrumental variables. The course will focus on settings where the exposure/intervention is time fixed but will also give an introduction to the more general case when exposures/treatments are time-varying (and hence may be affected by time varying confounding).
To develop an understanding of the assumptions most commonly invoked in causal inference and some practical expertise in applying estimation methods that rely on them.
There will be 5 on-line live lectures sessions and 3 on-line live discussion/computer practical sessions.
Date | Time | Title |
Monday 15 January | 10:00am – 1.00pm (regular breaks will be included)
| Live session: Lecture 1 Causal questions, causal estimands and their identification
Live session: Lecture 2 Overview of methods that rely on the no unmeasured confounding assumption and introduction to g-computation for time-fixed exposures
Guided computer practical 1 in Stata: Regression adjustment and G-computation for time-fixed exposures (R version will be made available)
|
Tuesday 16 January | 10:00am – 1.00pm (regular breaks will be included) | Live session: Lecture 3 Methods based on the propensity score
Live session: Lecture 4 Methods based on instrumental variables
Guided computer practical 2 in Stata: PS- and IV-based methods for time-fixed exposures (R version will be made available) |
Wednesday 17 January | 10:00am – 1.00pm (regular breaks will be included) | Live session: Lecture 5 Time-varying confounding: issues and overview of solutions With demonstration of computer practical applications of G-computation for time-varying exposures |
It I advisable that, before attending this course, participants attend both the module called “Addressing Causal Questions using real work data: an Introduction” and the short course called “Introduction to Causal Diagrams”, or equivalently be familiar with the first 2 parts of the book by Hernán and Robins “ Causal Inference: What If.” which is available here:
https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1268/2022/11/hernanrobins_WhatIf_13nov22.pdf
Free.
This is a UKRI funded project offering rigorous training in longitudinal data science. Please note that this training is NOT available to undergraduate or masters students.
