Description du livre
This new, updated second edition of In All Likelihood explores the central role of likelihood in a wide spectrum of statistical problems, ranging from simple comparisons-such as evaluating accident rates between two groups-to sophisticated analyses involving generalized linear models and semiparametric methods. Rather than treating likelihood merely as a tool for point estimation, the book highlights its broader value as a foundational framework for constructing, understanding and computational implementation of statistical models. It emphasizes how likelihood perspectives inform model development, assessment, and inference in a cohesive and intuitive way.
While grounded in essential mathematical theory, the book adopts an informal and accessible approach, using heuristic reasoning and illustrative, realistic examples to convey key ideas. It avoids overly contrived problems that yield to theoretically clean and closed-form solutions, instead embracing more realistic and complex real-world data analysis made tractable by modern computing resources. This perspective helps focus attention on the statistical reasoning behind model choice and interpretation.
The text also integrates a wide range of modern topics that extend classical likelihood theory, including generalized and hierarchical generalized linear models, nonparametric smoothing techniques, robust methods, the EM algorithm, and empirical likelihood. Suitable for students, researchers, and practitioners, this book provides both foundational insights and contemporary perspectives on likelihood-based statistical modelling.