Regression Models for Categorical Dependent Variables Using Stata

About this book

Although regression models for categorical dependent variables are common, few texts explain how to interpret such models.

This book fills this void, showing how to fit and interpret regression models for categorical data with Stata. The authors also provide a suite of commands for hypothesis testing and model diagnostics to accompany the book. The book begins with an excellent introduction to Stata and then provides a general treatment of estimation, testing, fit, and interpretation in this class of models. It covers in detail binary, ordinal, nominal, and count outcomes in separate chapters. The final chapter discusses how to fit and interpret models with special characteristics, such as ordinal and nominal independent variables, interaction, and nonlinear terms. One appendix discusses the syntax of the author-written commands, and a second gives details of the datasets used by the authors in the book. Nearly 50 Per cent longer than the previous edition, the book covers new topics for fitting and interpretating models included in Stata 9, such as multinomial probit models, the stereotype logistic model, and zero-truncated count models. Many of the interpretation techniques have been updated to include interval as well as point estimates.

New to the Second Edition: Regression models, including the zero-truncated Poisson and the zero-truncated negative binomial models, the hurdle model for counts, the stereotype logistic regression model, the rank-ordered logit model, and the multinomial probit model Stata commands, such as estat, which provides a uniform way to access statistics useful for postestimation interpretation. Expanded suite of programs known as SPost Inclusion of confidence intervals for predictions computed by prvalue and prgen.

Contents

Preface PART I GENERAL INFORMATION Introduction What is this book about? Which models are considered? Whom is this book for? How is the book organized? What software do you need? Where can I learn more about the models? Introduction to Stata The Stata interface Abbreviations How to get help The working directory Stata file types Saving output to log files Using and saving datasets Size limitations on datasets Do-files Using Stata for serious data analysis Syntax of Stata commands Managing data Creating new variables Labeling variables and values Global and local macros Graphics A brief tutorial Estimation, Testing, Fit, and Interpretation Estimation Postestimation analysis Testing estat command Measures of fit Interpretation Confidence intervals for prediction Next steps PART II MODELS FOR SPECIFIC KINDS OF OUTCOMES Models for Binary Outcomes The statistical model Estimation using logit and probit Hypothesis testing with test and lrtest Residuals and influence using predict Measuring fit Interpretation using predicted values Interpretation using odds ratios with listcoef Other commands for binary outcomes Models for Ordinal Outcomes The statistical model Estimation using ologit and oprobit Hypothesis testing with test and lrtest Scalar measures of fit using fitstat Converting to a different parameterization The parallel regression assumption Residuals and outliers using predict Interpretation Less common models for ordinal outcomes Models for Nominal Outcomes with Case-Specific Data The multinomial logit model Estimation using mlogit Hypothesis testing of coefficients Independence of irrelevant alternatives Measures of fit Interpretation Multinomial probit model with IIA Stereotype logistic regression Models for Nominal Outcomes with Alternative-Specific Data Alternative-specific data organization The conditional logit model Alternative-specific multinomial probit The sturctural covariance matrix Rank-ordered logistic regression Conclusions Models for Count Outcomes The Poisson distribution The Poisson regression model The negative binomial regression model Models for truncated counts The hurdle regression model Zero-inflated count models Comparisons among count models Using countfit to compare count models More Topics Ordinal and nominal independent variables Interactions Nonlinear models Using praccum and forvalues to plot predictions Extending SPost to other estimation commands Using Stata more efficiently Conclusions Appendix A Syntax for SPost Commands Appendix B Description of Datasets References Author Index Subject Index

Customer Reviews