What techniques can social scientists use when an outcome variable for a sample is not representative of the population for whom they would like to generalize the results? This book provides an introduction to regression models for such data including censored, sample-selected and truncated data. Richard Breen begins with a discussion of Tobin's Tobit model and examines issues such as maximum-likelihood estimation and the interpretation of parameters. He then discusses the basic sample selection model and the truncated regression model. Breen then elaborates on the modeling of censored and sample-selected data via maximum likelihood, showing the close links between the models introduced in this volume and other regression models for non-continuous dependent variables (such as the ordered probit). The book concludes with an exploration of some criticisms that have been made of these approaches and the difficulties associated with them.