I sometimes get asked what is a “good” book for learning econometrics or statistics. To avoid me giving an incomplete or ill thought-out answer, I list a few of my favourites here,

• “Mastering Metrics” by Josh Angrist and Jörn-Steffen Pischke. This is the best introductory text on causal inference that exists. Its chapters guide the student through the five main paths to causal inference, including: randomisation, regression, instrumental variables, regression discontinuity designs and differences-in-differences. The text is written in a style that means it has much to offer researchers whilst still being accessible to aspiring inferers.
• “Mostly Harmless Econometrics” (again) by Josh Angrist and Jörn-Steffen Pischke. This is the best non-introductory text on causal inference that exists. Don’t let the slim size of this book fool you. This isn’t a Jo Nesbo novel. Behind its meagre paper weight, it comprises seriously weighty content. It covers the predominant paths to causal inference, including all those topics discussed in Mastering Metrics, and more, like propensity score matching. Owning this book is essential for all serious practitioners of inference from observational data.
• “Bayesian Data Analysis” by Andrew Gelman et al. (be sure to get the third edition). This is the most comprehensive text on Bayesian analysis that exists. Whilst cover-to-cover reading is perhaps not encouraged, I find myself dipping into individual chapters time and time again. Note: this book is quite mathematically advanced. This is part of the reason I wrote my book (see below).
• “Introductory Econometrics” by Jeffrey Wooldridge. This is the book that ignited my interest in econometrics. It is written in a very accessible way and – whilst I would argue is a little bit dated now – is probably the best introductory text on classical econometrics.
• “A Student’s Guide to Bayesian Statistics” by me (sorry for this somewhat shameless plug). Hopefully does what is says. The text does not assume any previous knowledge of statistics – classical or Bayesian – nor probability. The content is about as un-mathematical as I could make it without turning it into a novel (about statistics). The book, instead of focussing on the maths, highlights the intuition behind Bayesian inference and MCMC algorithms, working up to a practical introduction to Stan and hierarchical models.