Causal Inference Using Cohort Data

Description:

The aim of this three-day in person short course is to teach methods of causal inference with an emphasis on application. In our experience, many causal inference courses focus on specific methods which, although elegant in theory, require data that do not appear much in practice. We instead adopt a data-first approach: given the abundance of cohort data and the fact that confounding is the norm rather than the exception, how can we make the most of cohort data? Accordingly, this short course focuses on the most widely applicable methods, while also giving a general grounding in causal inference theory. Teaching is focused on the intuition, rather than mathematics.

The course content includes:

•  An Introduction to Directed Acyclic Graphs
• Missing data
• Hypothetical interventions using the Target Trial Emulation framework
• Inverse Probability Weighting to address Time-Dependent Confounding
• Natural Experiments, Instrumental Variables, and Regression Discontinuity Designs
• Fixed Effects Approaches
• Elaborate Theories and Other Quasi-Experimental Devices (e.g., negative controls designs, cross-context analysis for causal inference, and multiple comparison groups)
• Quantitative Bias Analysis

The course consists of a mix of lectures and applied practicals to cement knowledge. Attendees should have at least working knowledge of basic statistical methods for data analysis (e.g., regression). Students can complete practicals in Stata or R.

Any questions about the course content can be directed to the teachers, Michalis Katsoulis (m.katsoulis@ucl.ac.uk) and Liam Wright (liam.wright@ucl.ac.uk).

The course will be run at or near the UCL’s London Bloomsbury Campus. Precise location details will be emailed to registered participants.

Cost:

Cost: £400 (PhD Students) or £800 (Otherwise)

Website and registration:

Region:

Greater London

Keywords:

Quasi-Experimental Research, Missing data, Weighting, Directed Acyclic Graphs, Hypothetical interventions, Inverse Probability Weighting, Natural Experiments, Instrumental Variables, Regression Discontinuity Designs, Fixed Effects Approaches, Elaborate Theories and Other Quasi-Experimental Devices, Quantitative Bias Analysis

Related publications and presentations from our eprints archive:

Quasi-Experimental Research
Missing data
Weighting

Back to the training database