Nonlinear Statistical Methods A. Ronald Gallant Describes the recent advances in statistical and probability theory that have removed obstacles to an adequate theory of estimation and inference for nonlinear models. Thoroughly explains theory, methods, computations, and applications. Covers the three major categories of statistical models that relate dependent variables to explanatory variables: univariate regression models, multivariate regression models, and simultaneous equations models. Includes many figures which illustrate computations with SAS(R) code and resulting output. 1987 (0 471-80260-3) 610 pp. Exploring Data Tables, Trends, and Shapes Edited by David C. Hoaglin, Frederick Mosteller, and John W. Tukey Together with its companion volume, Understanding Robust and Exploratory Data Analysis, this work provides a definitive account of exploratory and robust/resistant statistics. It presents a variety of more advanced techniques and extensions of basic exploratory tools, explains why these further developments are valuable, and provides insight into how and why they were invented. In addition to illustrating these techniques, the book traces aspects of their development from classical statistical theory. 1985 (0 471-09776-4) 672 pp. Robust Regression and Outlier Detection Peter J. Rousseeuw and Annick M. Leroy An introduction to robust statistical techniques that have been developed to isolate or identify outliers. Emphasizes simple, intuitive ideas and their application in actual use. No prior knowledge of the field is required. Discusses robustness in regression, simple regression, robust multiple regression, the special case of one-dimensional location, and outlier diagnostics. Also presents an outlook of robustness in related fields such as time series analysis. Emphasizes "high-breakdown" methods that can cope with a sizable fraction of contamination. Focuses on the least median of squares method, which appeals to the intuition and is easy to use. 1987 (0 471-85233-3) 329 pp.
Applied Linear Regression, Second Edition is a comprehensive guide to the methods of applied linear regression. Focusing on model building, assessing fit and reliability, and drawing conclusions, it develops estimation, confidence, and testing procedures mostly using least squares. Throughout, the importance of assumptions and their relevance in specific problems is stressed. Updated to reflect the enormous progress in the area of linear regression since the First Edition in 1980, the Second Edition cites more than 60 references, and includes several new problems, figures, and a totally new chapter that introduces students to nonlinear, logistic, and generalized linear regression models. Containing more than 20 worked examples, real data is used to illustrate variable selection, new predictor construction and dummy variables, model validation and other topics. Applied Linear Regression, Second Edition provides the most in-depth coverage available on transforming variables, finding problems with assumptions, and identifying influential cases. It discusses the special problems of inference and prediction from regression models. And throughout, graphical methods are generously discussed and illustrated. Additional topics include:
- Standard results for simple and multiple regression.
- The difficulties of using and interpreting regression models and estimates.
- Model building, variable selection, adding polynomials, and choosing transformations.
- Regression diagnostics, assumptions, and influence of cases.