A book to increase statistical problem solving skills I would like to have from you any suggestion about a specific book on statistics. I am a physicist, which means I already have a background in statistics and many other quantitative disciplines. My problem is that statistics has never been the main topic of any exam I did. Which means I already have a knowledge about this discipline but I fill a little failing in problem solving.
The book I believe is the right one for me should be mainly about exercises, in order to increase my problem solving skills, should have a somehow intriguing character, as at the moment statistics is not much more than an hobby, but should also be correlated with theory, in order to properly fill the gap I may have in the theory.
Any suggestion about that?
 A: One fun book to start with is

See https://bookshop.org/books/fifty-challenging-problems-in-probability-with-solutions/9780486653556.  To solve this problems you need to think in a statistical or probabilistic way!
A: *

*I would suggest a great book about theoretical statistics with methods which have lots of intriguing problems in various topics on Statistics:
Mathematical Statistics: Basic Ideas and Selected Topics (Chapman & Hall/CRC Texts in Statistical Science) by Bickel and Doksum https://www.amazon.com/dp/1498740316/.
One of our teachers suggested this book. I have only read Volume 1 though - I don't know much about Volume 2.


*Otherwise, you can have the classic book: Statistical Inference by Casella and Berger.


*If you want to know more about probability, you may look at An intermediate course in probability or Probability: A Graduate Course by Allan Gut or other classic probability books.
The books I mentioned have many intriguing problems, but might be little 'new' to the newcomers in statistics. But as you have some knowledge of Statistics, I would strongly recommend the first book.
A: As a former Physicist myself, I very much sympathise with your desire to increase your statistical problem solving skills.
I'd like to recommend the brilliant

Don't let the playful tone of the book fool you: Despite (?) the sparsity of mathematical derivations in this book, it covers topics & models that lead to complex and highly non-trivial results. In my opinion McElreath manages to redefine and teach statistical problem solving in a way that was entirely missing from the stats courses I took as a Physics graduate 20 years ago.
McElreath's focus is on statistical modelling, model validation and exploring/questioning possible data generating processes, with plenty of examples & exercises to work through in R.
The book also comes highly recommended by Andrew Gelman and Stan colleagues.
