Kung Fu ‘Metrics

I assign Angrist and Pischke’s Mostly Harmless Econometrics in virtually all of my graduate courses in economics and political science, largely because it’s one of the best, most practical, and most readable guides to causal inference out there. But it’s still a very hard book, with mathematical passages I myself struggle to follow once in a while.

What’s been needed for some time is a more casual introduction. And it has arrived.

Mastering Metrics is a more intuitive, example-strewn introduction to methods for figuring out causality in statistics. Just as the previous book used the Hitchhiker’s Guide to the Galaxy as a cute narrative device, this one uses Kung Fu. And more prose than math.

I have only skimmed the book (a review copy from the publisher) and it looks very good. A colleague in comparative politics walked into my office, saw it on the table, and got very excited. I expect this to be the general reaction. I also listened to Russ Roberts’s interview of Angrist on EconTalk a couple of weeks ago. It’s a nice overview, but sadly few deep insights or personal stories you won’t find in the book.

The real test will come from teaching with the new book, which I plan to do in the fall. But I expect advanced undergraduates, master’s and PhD students all to find this useful, especially alongside the previous book.

Personally I would like more international examples, and more on matching (and how it’s often misused). But these are small things.

More importantly, I build my courses around tearing apart new papers, and running replications and new data analysis. The Angrist and Pischke books are limited on their own.

Indeed, a quote, attributed to the Kung Fu Master Tan Soh Tin:

Never forget that, at the most, the teacher can give you fifteen percent of the art. The rest you have to get for yourself through practice and hard work. I can show you the path but I can not walk it for you.

8 thoughts on “Kung Fu ‘Metrics

  1. I like Morgan and Winship’s Counterfactuals and Causal Inference myself, but I’m going to get this one ASAP. Wonder if its any better?

    Btw, for those interested in causal inference, see:


    This is the CMU philosophy workshop on causal inference. The clips are terrific, especially the intro on how to use TETRAD and the “All of Causal Discovery” one. I’m still trying to work my way through these.