Books I like
These are some books which I really like - both fiction and non-fiction.
The ultimate in sci-fi\ fantasy. The story follows Paul Atreides, prince of House Atredies and heir to the planet Dune - the source of the most valuable substance in all the known universe. The characters and plot are great and have aged very well. The underlying theme is the danger of blindly following heroes. It's a tragic, amazing story. I read it almost every year.
Bayesian data analysis
Bayesian Data Analysis is pretty much the definitive text on the modern use of Bayesian statistics. The examples are clear, useful, and intuitive. The only downside is they don't talk much about Stan (the newest software for fitting Bayesian models), but the general knowledge is excellent. I'd recommend it to people who aren't even interested in Bayesian modeling because it covers lots of frequentist territory as well.
Richard McElreath's book is a great introduction to advanced statistics. It's a Bayesian book, but it covers lots of general statistical techniques very well (interaction terms, evaluating models, hierarchical modeling). If I ever teach a graduate statistics course I would strongly consider this.
Data analysis using regression...
This book by Andrew Gelman and Jennifer Hill is one of the best general books on statistical modeling and inference. While it's a little bit dated (although Andrew Gelman says he's working on a new version), the book very clearly outlines the basics (regression, analysis of variance) and then moves into the importance of hierarchical modeling.
Causal inference for statistics...
The definitive text on causal inference. Imbens and Rubin exhaustively list the importance of constructing proper counterfactual units for observational studies. They go way, way beyond the usual "you need to use propensity score matching" in most statistics books. A very good read.