Part 1 Inference and the role of Confounders

 

You’ve probably heard the saying “Correlation is not causation.” But what does it really mean? Is correlation helpful in understanding causation? For example, you may want to know how changing the price of a product affects sales. This part of the book explains that the best way to figure out if one thing causes another is to do experiments. However, you can’t always do experiments. That’s where causal inference comes into play.

One big reason just looking at correlation may not help (or may even lead you in the wrong direction) when you don’t have experimental data is confounders. Confounders are factors that influence both the decision we’re evaluating and the outcome we’re interested in. They play a major role in causal inference. In chapter 2, you’ll learn how much a confounder can twist your analysis and how to estimate the effect on your decision by removing the influence of confounders using the adjustment formula.

Chapter 3 will teach you how to use graphs to model your analysis. Graphs help you clearly state your objectives, lay out your assumptions, and figure out which causes and effects you can estimate from your data.

Finally, Chapter 4 will show you how to use machine learning to work out the adjustment formula when you’re dealing with many confounders.