Ronald Christensen

This book examines statistical models for frequency data. The primary focus is on log-linear models for contingency tables but also includes extensive discussion of logistic regression. Topics such as logistic discrimination, generalized linear models, and correspondence analysis are also explored. The treatment is designed for readers with prior knowledge of analysis of variance and regression. It builds upon the relationships between these basic models for continuous data and the analogous log-linear and logistic regression models for discrete data. While emphasizing similarities between methods for discrete and continuous data, this book also carefully examines the differences in model interpretations and evaluation that occur due to the discrete nature of the data. Numerous data sets from fields as diverse as engineering, education, sociology, and medicine are used to illustrate procedures and provide exercises. A major addition to the third edition is the availability of a companion online manual providing R code for the procedures illustrated in the book. The book begins with an extensive discussion of odds and odds ratios as well as concrete illustrations of basic independence models for contingency tables. After developing a sound applied and theoretical basis for frequency models analogous to ANOVA and regression, the book presents, for contingency tables, detailed discussions of the use of graphical models, of model selection procedures, and of models with quantitative factors. It then explores generalized linear models, after which all the fundamental results are reexamined using powerful matrix methods. The book then gives an extensive treatment of Bayesian procedures for analyzing logistic regression and other regression models for binomial data. Bayesian methods are conceptually simple and unlike traditional methods allow accurate conclusions to be drawn without requiring large sample sizes. The book concludes with two new chapters: one on exact conditional tests for small sample sizes and another on the graphical procedure known as correspondence analysis.

This textbook on Bayesian Statistics can be used to teach an upper division statistics course, a master's level course, a Ph.D. level course or used independently by researchers. It emphasizes the art of data modeling and analysis over technicalities and is rich with technical demonstrations available to Ph.D. students who need them and to others who are interested in them. The new edition features new material and new exercises and shifts the software focus from WinBUGS to OpenBUGS.

This textbook provides a wide-ranging introduction to the use and theory of linear models for analyzing data. The author's emphasis is on providing a unified treatment of linear models, including analysis of variance models and regression models, based on projections, orthogonality, and other vector space ideas. Every chapter comes with numerous exercises and examples that make it ideal for a graduate-level course. All of the standard topics are covered in depth: estimation including biased and Bayesian estimation, significance testing, ANOVA, multiple comparisons, regression analysis, and experimental design models. In addition, the book covers topics that are not usually treated at this level, but which are important in their own right: best linear and best linear unbiased prediction, split plot models, balanced incomplete block designs, testing for lack of fit, testing for independence, models with singular covariance matrices, diagnostics, collinearity, and variable selection. This new edition includes new sections on alternatives to least squares estimation and the variance-bias tradeoff, expanded discussion of variable selection, new material on characterizing the interaction space in an unbalanced two-way ANOVA, Freedman's critique of the sandwich estimator, and much more.

Now in its third edition, this companion volume to Ronald Christensen’s Plane Answers to Complex Questions uses three fundamental concepts from standard linear model theory—best linear prediction, projections, and Mahalanobis distance— to extend standard linear modeling into the realms of Statistical Learning and Dependent Data. This new edition features a wealth of new and revised content. In Statistical Learning it delves into nonparametric regression, penalized estimation (regularization), reproducing kernel Hilbert spaces, the kernel trick, and support vector machines. For Dependent Data it uses linear model theory to examine general linear models, linear mixed models, time series, spatial data, (generalized) multivariate linear models, discrimination, and dimension reduction. While numerous references to Plane Answers are made throughout the volume, Advanced Linear Modeling can be used on its own given a solid background in linear models. Accompanying R code for the analyses is available online.

Analysis of Variance, Design, and Regression: Linear Modeling for Unbalanced Data, Second Edition presents linear structures for modeling data with an emphasis on how to incorporate specific ideas (hypotheses) about the structure of the data into a linear model for the data. The book carefully analyzes small data sets by using tools that are easily scaled to big data. The tools also apply to small relevant data sets that are extracted from big data. New to the Second EditionReorganized to focus on unbalanced dataReworked balanced analyses using methods for unbalanced dataIntroductions to nonparametric and lasso regressionIntroductions to general additive and generalized additive modelsExamination of homologous factorsUnbalanced split plot analysesExtensions to generalized linear modelsR, Minitab®, and SAS code on the author’s websiteThe text can be used in a variety of courses, including a yearlong graduate course on regression and ANOVA or a data analysis course for upper-division statistics students and graduate students from other fields. It places a strong emphasis on interpreting the range of computer output encountered when dealing with unbalanced data.

This textbook provides a wide-ranging introduction to the use and theory of linear models for analyzing data. The author's emphasis is on providing a unified treatment of linear models, including analysis of variance models and regression models, based on projections, orthogonality, and other vector space ideas. Every chapter comes with numerous exercises and examples that make it ideal for a graduate-level course. All of the standard topics are covered in depth: estimation including biased and Bayesian estimation, significance testing, ANOVA, multiple comparisons, regression analysis, and experimental design models. In addition, the book covers topics that are not usually treated at this level, but which are important in their own right: best linear and best linear unbiased prediction, split plot models, balanced incomplete block designs, testing for lack of fit, testing for independence, models with singular covariance matrices, diagnostics, collinearity, and variable selection. This new edition includes new sections on alternatives to least squares estimation and the variance-bias tradeoff, expanded discussion of variable selection, new material on characterizing the interaction space in an unbalanced two-way ANOVA, Freedman's critique of the sandwich estimator, and much more.

Now in its third edition, this companion volume to Ronald Christensen’s Plane Answers to Complex Questions uses three fundamental concepts from standard linear model theory—best linear prediction, projections, and Mahalanobis distance— to extend standard linear modeling into the realms of Statistical Learning and Dependent Data. This new edition features a wealth of new and revised content. In Statistical Learning it delves into nonparametric regression, penalized estimation (regularization), reproducing kernel Hilbert spaces, the kernel trick, and support vector machines. For Dependent Data it uses linear model theory to examine general linear models, linear mixed models, time series, spatial data, (generalized) multivariate linear models, discrimination, and dimension reduction. While numerous references to Plane Answers are made throughout the volume, Advanced Linear Modeling can be used on its own given a solid background in linear models. Accompanying R code for the analyses is available online.

Analysis of Variance, Design, and Regression: Linear Modeling for Unbalanced Data, Second Edition presents linear structures for modeling data with an emphasis on how to incorporate specific ideas (hypotheses) about the structure of the data into a linear model for the data. The book carefully analyzes small data sets by using tools that are easily scaled to big data. The tools also apply to small relevant data sets that are extracted from big data. New to the Second EditionReorganized to focus on unbalanced dataReworked balanced analyses using methods for unbalanced dataIntroductions to nonparametric and lasso regressionIntroductions to general additive and generalized additive modelsExamination of homologous factorsUnbalanced split plot analysesExtensions to generalized linear modelsR, Minitab®, and SAS code on the author’s websiteThe text can be used in a variety of courses, including a yearlong graduate course on regression and ANOVA or a data analysis course for upper-division statistics students and graduate students from other fields. It places a strong emphasis on interpreting the range of computer output encountered when dealing with unbalanced data.

This textbook provides a wide-ranging introduction to the use and theory of linear models for analyzing data. The author's emphasis is on providing a unified treatment of linear models, including analysis of variance models and regression models, based on projections, orthogonality, and other vector space ideas. Every chapter comes with numerous exercises and examples that make it ideal for a graduate-level course. All of the standard topics are covered in depth: ANOVA, estimation including Bayesian estimation, hypothesis testing, multiple comparisons, regression analysis, and experimental design models. In addition, the book covers topics that are not usually treated at this level, but which are important in their own right: balanced incomplete block designs, testing for lack of fit, testing for independence, models with singular covariance matrices, variance component estimation, best linear and best linear unbiased prediction, collinearity, and variable selection. This new edition includes a more extensive discussion of best prediction and associated ideas of R2, as well as new sections on inner products and perpendicular projections for more general spaces and Milliken and Graybill’s generalization of Tukey’s one degree of freedom for nonadditivity test.

As the new title indicates, this second edition of Log-Linear Models has been modi?ed to place greater emphasis on logistic regression. In addition to new material, the book has been radically rearranged. The fundamental material is contained in Chapters 1-4. Intermediate topics are presented in Chapters 5 through 8. Generalized linear models are presented in Ch- ter 9. The matrix approach to log-linear models and logistic regression is presented in Chapters 10-12, with Chapters 10 and 11 at the applied Ph.D. level and Chapter 12 doing theory at the Ph.D. level. The largest single addition to the book is Chapter 13 on Bayesian bi- mial regression. This chapter includes not only logistic regression but also probit and complementary log-log regression. With the simplicity of the Bayesian approach and the ability to do (almost) exact small sample s- tistical inference, I personally ?nd it hard to justify doing traditional large sample inferences. (Another possibility is to do exact conditional inference, but that is another story.) Naturally,Ihavecleaneduptheminor?awsinthetextthatIhavefound. All examples, theorems, proofs, lemmas, etc. are numbered consecutively within each section with no distinctions between them, thus Example 2.3.1 willcomebeforeProposition2.3.2.Exercisesthatdonotappearinasection at the end have a separate numbering scheme. Within the section in which it appears, an equation is numbered with a single value, e.g., equation (1).

The third edition of Plane Answers includes fundamental changes in how some aspects of the theory are handled. Chapter 1 includes a new section that introduces generalized linear models. Primarily, this provides a defini­ tion so as to allow comments on how aspects of linear model theory extend to generalized linear models. For years I have been unhappy with the concept of estimability. Just because you cannot get a linear unbiased estimate of something does not mean you cannot estimate it. For example, it is obvious how to estimate the ratio of two contrasts in an ANOVA, just estimate each one and take their ratio. The real issue is that if the model matrix X is not of full rank, the parameters are not identifiable. Section 2.1 now introduces the concept of identifiability and treats estimability as a special case of identifiability. This change also resulted in some minor changes in Section 2.2. In the second edition, Appendix F presented an alternative approach to dealing with linear parametric constraints. In this edition I have used the new approach in Section 3.3. I think that both the new approach and the old approach have virtues, so I have left a fair amount of the old approach intact. Chapter 8 contains a new section with a theoretical discussion of models for factorial treatment structures and the introduction of special models for homologous factors. This is closely related to the changes in Section 3.3.

Emphasizing the use of WinBUGS and R to analyze real data, Bayesian Ideas and Data Analysis: An Introduction for Scientists and Statisticians presents statistical tools to address scientific questions. It highlights foundational issues in statistics, the importance of making accurate predictions, and the need for scientists and statisticians to collaborate in analyzing data. The WinBUGS code provided offers a convenient platform to model and analyze a wide range of data.The first five chapters of the book contain core material that spans basic Bayesian ideas, calculations, and inference, including modeling one and two sample data from traditional sampling models. The text then covers Monte Carlo methods, such as Markov chain Monte Carlo (MCMC) simulation. After discussing linear structures in regression, it presents binomial regression, normal regression, analysis of variance, and Poisson regression, before extending these methods to handle correlated data. The authors also examine survival analysis and binary diagnostic testing. A complementary chapter on diagnostic testing for continuous outcomes is available on the book’s website. The last chapter on nonparametric inference explores density estimation and flexible regression modeling of mean functions.The appropriate statistical analysis of data involves a collaborative effort between scientists and statisticians. Exemplifying this approach, Bayesian Ideas and Data Analysis focuses on the necessary tools and concepts for modeling and analyzing scientific data.Data sets and codes are provided on a supplemental website.

This is the second edition of Linear Models for Multivariate, Time Series and Spatial Data. It has a new title to indicate that it contains much new material. The primary changes are the addition of two new chapters: one on nonparametric regression and one on response surface maximization. As before, the presentations focus on the linear model aspects of the subject. For example, in the nonparametric regression chapter there is very little about kernal regression estimation but quite a bit about series approxi­ mations, splines, and regression trees, all of which can be viewed as linear modeling. The new edition also includes various smaller changes. Of particular note are a subsection in Chapter 1 on modeling longitudinal (repeated measures) data and a section in Chapter 6 on covariance structures for spatial lattice data. I would like to thank Dale Zimmerman for the suggestion of incor­ porating material on spatial lattices. Another change is that the subject index is now entirely alphabetical.