Statistical Rethinking

About the book

Statistical Rethinking is Richard McElreath's introduction to Bayesian data analysis. It builds from basic probability and model fitting toward Bayesian regression and multilevel models, using the R programming language throughout. The subject is statistical modeling as a practical craft, with simulation and worked examples carrying most of the explanatory weight.

McElreath treats models as tools for reasoning about data. He asks readers to think about what a model assumes before fitting it, and the R code is integral: examples are meant to be run, modified and compared. The result is a working text that rewards active use.

How to read it

Read with R open and run the examples. The explanations make more sense when you can see the output change as you alter a model or a prior. Pause to check what each simulation demonstrates before moving on.

The difficulty rises steadily. Early material on probability and sampling is approachable for readers with basic statistics. Later chapters on regression and multilevel models demand patience and comfort with code. Some passages assume familiarity with likelihood and linear models, so readers without that background should expect to consult other sources alongside it.

Who it suits

This suits readers who know some statistics and want a grounded introduction to Bayesian modeling in R. It fits students and researchers who prefer learning by doing and who need to reason about uncertainty in real analyses. Readers coming from frequentist training will find the reframing useful, provided they are willing to work through the code.

Reading considerations

The book is tied to R and to the software conventions of its period, so some code and package details may need updating for current setups. Readers seeking a proof-heavy theoretical treatment should look elsewhere. The book teaches modeling practice, and its value depends on doing the exercises rather than reading passively.