Using LISREL for Structural Equation Modeling: A Researcher′s Guide

摘要: Structural equation modeling (SEM) is one of the most rapidly growing analytic techniques in use today. Proponents approach have virtually declared advent a statistical revolution, while skeptics worry about widespread misuse complex and often poorly understood methods. Despite interest structural models, few individuals using these benefitted from any formal training. Indeed, graduate programs provide no courses on SEM. Individuals interested acquiring skills this technique must eider attend expensive training seminars or plow through technical books manuals their own.The two new under renew are therefore timely. Both valuable, but differ important ways. Kevin Kelloway's book directed at researcher with little knowledge intricately linked to more popular programs, LISREL. For researchers keen begin analyzing data quickly, an invaluable resource that will speed one's introduction SEM.On other hand, volume written by Rex Kline represents comprehensive available introductions application, execution, interpretation technique. The for both students who do not extensive quantitative background. It especially attentive issues common applications.Kelloway's designed unfamiliar software. Chapter 1 provides brief overview differentiates among historical concepts such as path analysis latent variable model. Although focus LISREL, offers clearly concise introductory chapters SEM I ever read. They ideal relevant basic technique.The theory behind steps reviewed 2 includes model specification, identification, estimation, testing fit, respecification. author emphasizes importance specifying fundamental step allows test hypotheses relation number variables, makes inherently confirmatory How specified influences identification fit. Currently, there over 20 indices fit computed programs. 3 three general classes indices, namely those assessing absolute comparative parsimonious Absolute assess ability reproduce accurately manner which observed variables actually covary. Comparative proposed account relative less restricted Parsimonious recognize better usually achieved simply increasing parameters estimated. compensate evaluating benefit achieved, given cost estimating additional parameters.Chapter 4, chapter book, explains various algebraic components matrices required fitting understanding associated needed run recent versions EQS, AMOS, useful. has prudently avoided directing towards "point-and-click" users. This sufficient information novice users appreciate complexity without discouraging them.Chapters 5, 6, 7 devoted applications SEM, factor analysis, analysis. …