Summary

In this chapter, we investigated the complexities of causal inference, focusing on confounding and associations. We explored this by understanding various types of associations – positive, negative, and null – and their implications in different contexts, such as traffic patterns. We also emphasized the fundamental challenges in causal inference, notably in scenarios with infinite data where statistical associations still fail to reveal causality’s direction or nature.

We dealt with ITEs in a population, using a neighborhood-moving scenario to illustrate how different people react uniquely to the same treatment. This led to a discussion on ATEs and the difficulties in their calculation, especially due to missing data in causal inference. We differentiated between confounding and association, using a fitness program’s impact on weight loss as an example. We elaborated that a link between treatment and outcome doesn’t necessarily imply causation...