Experiments & Causal Inference

A course on designing and evaluating the effect of a cause.

Description

Data is the evidence that is used in the process of statistical reasoning. Two major forms of such reasoning, causal and predictive, have their roots in correlations to begin with. A particular cause can produce two simultaneous effects which are strongly correlated (i.e., have similar variations). Those effects can predict each other very well, but we know that none of them is the cause for the other.

This course offers methods to answer whether a variable X causes Y or not. The ability to establish the existence of such causal relationship using data is very important. You can also make statements like "changing X by x% will change Y by y%". It also improves the quality of the prediction you make about Y because now you know that X is the source.

This course will contain concepts with examples and implementations. Basic probability and statistics will be helpful, but is not a prerequisite. References to essential concepts will be provided.

We will see different ways to estimate causal effects in experimental and observational scenarios. We will study RCTs, A/B tests, and similar experimental setups and their design. We will study impact evaluation methods using observational data. We will study heterogeneous treatment effects from recent methods. We will also cover causal mediations. We will see latest issues on identification of treatement effects in presence of spillovers and dealing with time.

Learning Objective

At the end of this course, you will know about widely used and latest methods to evaluate treatment effects. Whether you want to perform A/B tests online, RCTs, field experiments, or dig out causal evidence from observational data, you have all the required concepts covered. Overall, this course will boost you in a big way on understanding and evaluating causality. The importance of understanding causality is not limited to data-driven analysis alone, I believe this will enhance your perception on the observations you make in your daily life as well.