Behavioral Data Analysis with R and Python

Part V. Advanced Tools in Behavioral Data Analysis

This is the last part of the book, where everything comes together. We’ll see three powerful tools of behavioral data analysis, first moderation in Chapter 11 and then mediation and its offspring instrumental variables (IVs) in Chapter 12.

Moderation is a versatile mathematical tool that allows us to understand interaction effects, as well as effectively and transparently segment our customer population. Mediation allows us to peek into the black box of causal relationships and understand how a variable affects another. Finally, I’ll use instrumental variables to make good on my promise to measure the impact of customer satisfaction on later customer behaviors.

Moderation, mediation, and IVs are meaty topics and are the subject of lively methodological debates and entire books. But the tools we introduced earlier in the book will make the journey much easier. First, using causal diagrams will give intuitive interpretations for all three of them. Second, using the Bootstrap will allow us to directly build confidence intervals and completely bypass the methodological complications associated with p-values. As a result, we’ll be able to draw deep and actionable behavioral insights with one-liners of code.