Computational Aspects of Psychometric Methods with R
本书介绍社会科学中测量模型的构建，涵盖经典测试理论、项目反应理论等，并用R演示应用，适合想了解测量方法实际操作的读者。
This book is a practical and concise text on the construction of measurement models in the social sciences.Many of the variables used in the social sciences cannot be directly observed unlike objects in the physical world.In Chapter 1, the authors introduce the idea measurement based on a model that relates an unobserved variable to observed indicator variables.Because unobserved variables cannot be directly measured, researchers need to provide evidence that the interpretations based on measurement models are useful.The authors discuss the use of linear regression and ANOVA to support claims of measurement validity in Chapter 2 while the use of cluster analysis and exploratory factor analysis to provide evidence on the internal structure of measurement scales is discussed in Chapter 3.Chapter 4 introduces Classical Test Theory which aims at estimating the influence that random measurement error, or reliability, has on test performance.Reliability is not sufficient for accurate measurement because there is a need to know whether items are adequate for measurement or need to be added or removed.Because Classical Test Theory was not concerned with these issues researchers usually rely on ad hoc methods for these purposes which are discussed in Chapter 5.Chapter 6 introduces Item Response Theory using regression models assuming that variables are directly observable and measured without error.In Chapter 7, the authors then extend this approach to models for binary items which predict the probability of success on an item depending on the person's location on the latent variable and the item difficulty.Chapter 8 expands this discussion to include models for categorical items.The final two chapters in the book discuss applied aspects.Chapter 9 discusses differential item functioning which occurs different groups with the same level on the unobserved variable have different probabilities of success on an item.Chapter 10 discusses the use of Item Response Theory models in adaptive testing where subjects answer items selected using an algorithm.The book is written in a tutorial style with numerous applied examples and exercises providing an accessible introduction to the basic ideas.Each topic is discussed at an intermediate technical level before the authors demonstrate the procedure in R. The book uses existing R packages and anyone buying the book expecting a detailed discussion of how to write code to fit the models will be disappointed.The book focuses on the application of measurement theory not the theory itself and would not be suitable as a companion for an ordinary psychometrics course.