jrfiedler/causal_inference_python_code

Python code for part 2 of the book Causal Inference: What If, by Miguel Hernán and James Robins

Python Code for Causal Inference: What If

This repo contains Python code for Part II of the book Causal Inference: What If, by Miguel Hernán and James Robins (book site):

Hernán MA, Robins JM (2020). Causal Inference: What If. Boca Raton: Chapman & Hall/CRC.

This Python version roughly corresponds to the Stata, R, or SAS programs found at the book site, and was also translated into Julia, here.

The code in this repo has been checked against the 30 March 2021 version of the book.

Python dependencies

Required Python packages:

• numpy
• pandas
• statsmodels
• scipy
• matplotlib
• linearmodels
• tqdm

If you use the Anaconda distribution of Python, you'll have most of those packages already, and you'll only need to install

• linearmodels
• tqdm

Data

The data can be obtained from the book site.

The notebooks all assume that the Excel version of the data has been saved in the same directory as the notebooks.

Author

James Fiedler, with contributions from Petty PY Chen and Piyush Madan

causal_inference_python_code's Issues

Why GEE gives the right coefficients and better standard errors?

It is mentioned in Chapter 12 and section 12.2 that WLS gives the right coefficients but the standard error is off and GEE gives better standard errors. Can you provide more detail about this statement?

Cell 13 of IV notebook

Hi,

I am not sure, but I believe that cell 13 of Chapter 16 is not correct. The regression on the outcome (Y) should be done on the predicted value of the treatment (A) and not on the actual value of it. If I am mistaken, can you clarify this to me please?

Best,
Sergio