February 2023 Webinar: Introduction to Causal Inference and propensity modelling – Christel Swift – BBC

Introduction to Causal Inference and propensity modelling

• Wednesday 15th February 2023

With Christel Swift, Principal Data Scientist, BBC

About the webinar:

In an ideal world, when trying to assess the effectiveness of a treatment (e.g. a new medicine, a government policy, a marketing intervention), we would use a randomized controlled trial.  Unfortunately this is often not possible or practical and we have to make do with “observational studies”.  Directly comparing a treated vs a non-treated group can be problematic because the two groups may have very different profiles prior to being treated.  For example, how would you evaluate the effectiveness of programme trailers that are played before a viewer gets to watch their selected programme?  It’s important to control for factors that influence both treatment (e.g. being exposed to a trailer) and outcome (e.g. going on to watch the promoted programme) before drawing any conclusion.  This is what Causal Inference is all about.  Add to that the computational challenge of dealing with large datasets and some methods may be suitable than others.  

This webinar will walk you through some key concepts of Causal Inference, including counterfactuals, causal graphs, propensity modelling, inverse probability weights, love plots, and marginal structural models.