8 Advanced tools with the DoubleML library
This chapter covers
- Estimating the effect of a continuous decision variable with double machine learning
- Learning to calculate confidence intervals for the ATE
- Combining two estimators to get a better one with doubly robust techniques
In the previous chapters, we covered the basic theory and practice of removing the effect of confounders using the adjustment formula. In this chapter, we will go one step further and explore three separate techniques that will improve the accuracy and reliability of our causal estimates. These techniques, double machine learning, confidence intervals, and doubly robust techniques, are used when precision is paramount. For instance, people working in healthcare or finance may apply these approaches.
Even though learning these techniques takes a while, using them is straightforward with open source libraries. Fortunately, the methods discussed in this chapter have already been implemented in the DoubleML library (https://docs.doubleml.org), available for both R and Python. After delving into the theoretical foundations, we will explore how to effectively employ these methods using this library.
NOTE You don’t need to read this chapter to be able to follow the rest of the book. If these topics are not a priority or are too technical, you can skip this chapter and return later when you need to learn about these tools.