Causal Effects

What is a treatment effect and how to compute it?

This article is a continuation of a 3-part series on causality. Previous posts focused on causal inference and causal discovery. This post builds upon these ideas and describes three different causal effects. In future posts, we will discuss computing causal effects from observational data and evaluating causal effects using the do-operator.

Key Point:

• Causal effects quantify the impact of a treatment by comparing the outcomes of different treatment values

Why?

The notion of cause and effect plays a central role in how we (i.e. humans) understand the world around us. For example, “the sidewalk is wet because it rained last night”, enables us to make inferences in new situations, e.g. why is the patio furniture wet?

This comes down to a fundamental question, why? Why did this happen? What is the cause of this? Or, where is this going? What will be the effect?

In many practical settings (e.g. business, healthcare, education) it is not only advantageous to know what caused what, but by how much? In other words, what degree of the patio furniture’s wetness is due to the rain from last night…