Statistical Rethinking (2022 Edition) (github.com)

Article summary

The article discusses the 2022 edition of the book 'Statistical Rethinking' by Richard McElreath, which focuses on Bayesian data analysis and scientific modeling. The book is accompanied by online lectures, problem sets, and solutions, and is available in various formats, including R, Python, and Julia. The course teaches data analysis with a focus on conceptual, causal models and precise questions about those models. The book and course aim to connect scientific models to evidence using Bayesian data analysis.

Main themes

• Bayesian vs Frequentist methods
• Statistical modeling
• Data analysis
• Scientific inference
• Philosophy of statistics
• Methodological debates

What commenters say

• Some commenters criticize the book's introduction for being overly critical of frequentist methods, while others defend it as a necessary critique of their misuse in science.
• Bayesian methods are not inherently better than frequentist methods, and the choice between them depends on the specific problem and context.
• Frequentist methods have their own strengths, such as providing guarantees about long-run frequencies, which can be important in certain applications.
• The debate between Bayesian and frequentist methods is often fueled by emotional and philosophical discourse, rather than a focus on the results and suitability of each approach for different cases.
• The importance of prior information and model selection is highlighted, with some arguing that Bayesian methods can be problematic if the prior is inaccurate or the model is misspecified.
• Some commenters argue that frequentist methods are more 'objective' because they do not rely on prior information, while others counter that this objectivity is an illusion and that all statistical methods involve some degree of subjectivity.
• The choice between Bayesian and frequentist methods should be based on the specific goals and requirements of the analysis, rather than on ideological or philosophical grounds.