Discovering Statistics Using R

Honestly such a helpful book. Of course we are in the era of AI but it doesn’t explain everything well. This book explains the stats, the code, and why you’re doing xyz. It’s a great book and puts it all in simple terms. Here I am in my post doc and I still refer to it!

This book fills a niche that very much needed to be filled. It is both a review of basic statistical concepts and directions as to how to perform the corresponding analyses/tests in R. It's light on theory of course, but supplying proofs and in-depth descriptions isn't what this book is about. Although I'm a bit rusty, I've had a great deal of graduate level statistics, none of which emphasized application. This book is an excellent guide as to how to actually apply statistics. Extremely welcome is its emphasis on underlying assumptions. In my theoretical statistics classes, the Central Limit Theorem was the answer to almost all questions involving assumptions. As the authors point out, even with a sample size that's sufficiently large, the CLT does not always guarantee normality. I also like that the authors give complete steps in each chapter. Thus the entire coding to accomplish something is present and you don't have to go looking for how to accomplish some preliminary step before you can do the current procedure. At the end of each chapter is a list of what R packages and functions have been used. The authors do include some sophomoric humor, maybe to make this more palatable to undergraduates, but this doesn't become annoying. Finally the authors appear to like cats, a mark in their favor. One word of warning, Field may not provide a context for something—a test, a transformation, etc. Readers are advised to look at the references he provides at the ends of the chapters. For instance, his later presentations on bootstrapping will make a lot more sense if you’ve read the paper by Wright, London, & Field he suggests. This can be found online. When presenting the Fisher transformation of Pearson’s r to a z-score in Sect. 6.3.3, he doesn’t tell you that it should be used only in tests of null hypotheses rho = some constant not = to 0 or to 1; where .3 < |rho| < 1, r’s sampling distribution will tend to be skewed, making the Fisher transformation necessary. Not knowing this context, given in Chen and Popovich, one of the references at the end of Chapter 6, could cause a reader to use the Fisher transformation inappropriately.

The writing style is highly accessible, fun, varied, and rich in detail. Simply a superb way to get going quickly in R AND in statistics, but even if you have considerable stat under you belt, as I do, it provides an excellent review of concepts, and their implementation in R. I am pleased in every way with this massive survey of the field. With this in hand I know I can go off in whatever direction of specialization I require. There is simply no question in my mind that this the best starter book for both stat and R (and learning the two together, these days, just makes sense). It turns out to be far better than I expected. Loaded with extra information, plenty of fine-grained detail, well worked-out examples, and unexpected humor, this makes its subject just about as accessible as can be done. A great value!

I chose this book because I wanted to learn R through a progression of examples, and for that it has served well. I would recommend it for that purpose. My understanding of statistics, though, has not progressed as much as my R programming. Specifically, I struggled in almost every chapter with "assumptions" - the criteria that must be met to justify the use of a particular statistical method. How important is each one? Why is it important? What exactly is it testing for, or trying to prevent you from doing?Close reading of the text was usually unhelpful to my understanding, and I frequently had to turn to outside sources. After a fair amount of struggle with the topic, here's what I've learned: Basically, this issue is confusing because the whole idea of assumption checking is a simplification born out of the rise of statistics software. Which allows people like me, with limited mathematical literacy, to blithely run lots of analyses using wildly inappropriate and mis-specified models, then report the results as if they were something other than meaningless noise. If I actually understood how the models work, when they fail, and how to choose a meaningful specification, then I wouldn't need strict guidelines for assumption checking. The problem is that the simplification is an oversimplification. A checklist of assumptions is no substitute for an understanding of the reasoning behind the modeling techniques. There will be cases where the model is appropriate even if you fail an item on the checklist, and cases where the model is inappropriate even if your data happens to check all the boxes. A checklist will never tell you that there's a better technique for your purposes. And it will always be hard to get a straight answer to questions like "how much is too much?" because the checklist thresholds are arbitrary in the first place, and therefore constantly open to debate. Field is very little help in understanding how the models work or why they fail. Instead he takes a cookbook approach that mostly amounts to glorifying checklists. To the point that I suspect at least one of his chapters - on analysis of covariance - uses a mis-specified model as its core example (though it checks all the checklist boxes). It certainly doesn't look much like the appropriate situations for the technique according to outside sources, including some sources he cites in his bibliography. I learned quite a bit of R. But if I want more than a surface understanding of statistics, I'm going to need another book.

日本にはこのような本がないので、大変参考になります。

Perfecto si eres principiante en el mundo de R y la estadística. Un libro muy útil y ameno para leer. Lo recomiendo.

I highly recommend this book for beginners. Although the explanations are a little lengthy they are very clear and pertinent. This book is aimed at readers with a moderate proficiency in maths (for example matrix calculus is avoided). However, it explains the history, backgound and main insights behind statistics reasonning very well. As a result, from an intuitive viewpoint, this book would also be helpful background reading to more mathematically minded readers such as Engineering and Physics students. My hearty congratulations to the authors for having produced such a helpful and well thought out teaching book.

It is a very clear introduction to the use of R for conducting the main statistical techniques. It appears very useful for psychologist, since the book covers a number of techniques used in psychological research. A strength of the book is that it covers a number of techniques that are little known, also among specialists. For example, the robust linear models and multilevel analysis are explained very clearly. The book is written in a very slight style, so the reader is not frustrated by the difficulties of some arguments. A very good manual!

tiene ejemplos sencillos (El libro está en inglés)