STRUCTURAL EQUATION MODELING, 6(2), 216-217 Graphical Multivariate Analysis with AMOS, EQS, and LISREL: A Visual Approach to Covariance Structure Analysis (in Japanese). Yutaka Kano. Kyoto, Japan: Gendai-Sugakusha, 1997, 235 pages. Reviewed by Kentaro Hayashi Department of Psychology University of California-Los Angeles This book examines how to analyze data using three major software programs for structural equation modeling (SEM): AMOS 3.51, EQS 5.4, and LISREL 8.14. The book is written in Japanese by one of the leading scholars in theoretical aspects of structural equation models. Although the author is originally trained as a mathematical statistician, the book is oriented more to applied researchers. The book consists of seven chapters. The first three chapters are fairly introductory, whereas the remaining chapters provide more in-depth discussion of SEM issues. Chapter 1 introduces regression analysis as a transition model to SEM analysis. In chapter 2, the author explains in detail how to conduct a data analysis with AMOS, EQS, and LISREL. The chapter focuses on two modeling examples: path analysis (i.e., a model without latent variables) and multiple indicator analysis (i.e., a model with latent variables). The step-by-step descriptions are easy to follow, even for readers who are inexperienced with these software packages. Chapter 3 focuses on exploratory and confirmatory factor analysis and chapter 4 examines the foundations of covariance structure analysis. These chapters include excellent discussions on such topics as identification, methods of estimation, goodness of fit indexes (e.g., RMR, GFI, AGFI, RMSEA, NFI, NNFI, and CFI), analysis of categorical data, and violation of normality assumptions (e.g., discussion of elliptical distributions). Chapters 5 and 6 are a little more advanced. In chapter 5, methods for modifying proposed models are given, with discussion of Lagrange multiplier and Wald tests in EQS and modification indexes and Wald tests in AMOS and LISREL. In chapter 6, the topic of analysis of multiple groups is presented and different kinds of factorial invariance models are explained. In addition, the chapter provides a discussion of mean structures associated with the analysis of multiple groups. Finally, chapter 7 provides a brief summary of SEM and discussion on some unsolved issues. Although the title of the book is "graphical" multivariate analysis, it is a reasonable name because many path diagrams, output results, and command files appear throughout the book. Despite the fact that the phrase "graphical modeling" has in recent years been used to describe conditional independence models for continuous and categorical data, these are not the same as structural equation models, and the author does not use the word with this particular meaning. Although the intended readers of this book are novice users of SEM, more advanced researchers will find the presentation quite interesting. In fact, a lot of background information is scattered throughout the book. Unfortunately, because the book is written in Japanese, it is not accessible to most of the otherwise potential readers throughout the world. Nevertheless, it should be of great interest to readers of this journal to know that SEM is becoming widely accepted in Japan. |