I am so glad I have this book now. It's changed the way that I do statistics. To me reading this book has been like taking the "red pill" in a world where everyone takes the "blue pill" and lives in a 'blissful' world where they don't have to worry about model uncertainty in their research. I know this statement sounds drastic but I truly believe what I am saying. You will fully understand what I mean if you read the book.I'm a PhD graduate student in economics who does lots of applied statistics research to examine causality. One thing that economists and econometricians RARELY discuss is model uncertainty and pretesting. The common thing to do is to pick a model without much justification, or pretest to select a model. Then they assume that the selected model is the "true" on (the blue pill in action). Now that I know that there are problems with this approach, it's changed the way I see statistical inference for the better.The section on model averaging is great as well. Model averaging is something that really needs to be picked up by applied statisticians. It has only recently been considered by macroeconomists. This book, and the related literature, have led me to work on my own paper on model averaging in causal inference, where the choice of your model is pretty important. So that's an added bonus.This book covers model selection and model averaging in depth. The approach is both intuitive and rigorous, so it should appeal to applied statisticians (like me) and more "pure" statisticians. The examples in the book are very eye opening, interesting, and relevant to various research interests. The examples show how poor statistical inference can be when model uncertainty is ignored. Looking at a graph of coverage probabilities for a post AIC pretest estimator kind of blew my mind. It's insane how bad they are.I highly recommend that all economists who use data get this book. It goes without saying that this book is excellent for statisticians as well.