Psychological & Brain Sciences 891W - S-Intro to Struct Equat Models

Structural equation modeling (SEM) refers to a family of methods that all involve tests of a priori statistical models of covariances. Another term for SEM is the analysis of covariance structures. SEM is most useful when the researcher has multiple indicators of many constructs and wishes to test the appropriateness of complex hypothesized networks among these constructs. SEM subsumes many more familiar techniques - for instance, simple and multiple regression analysis, and exploratory factor analysis. However, SEM also provides powerful generalizations of these techniques. Specific techniques in the SEM family include path analysis (PA), confirmatory factor analysis (CFA) and the creation of latent variables, and the evaluation of structural regression models (SR), which are hybrid models with features of both path and latent factor models. SEM can incorporate many outcomes and many predictors simultaneously and is particularly useful for modeling mediational processes. Additionally, SEM permits measurement error variance to be separated from the observed variance so that relationships among true underlying constructs can be examined. Thus, the methodology blends aspects of psychometrics with traditional statistical analysis. It is used in many areas in of the social sciences to test a wide variety of hypotheses, including those about mediation, moderation, measurement, and change over time. Psych 891W will provide you with the computer and statistical skills that you will need to perform structural equation modeling using Mplus software. In this course you will learn to plan, carry out, and interpret SEM analyses, with particular importance placed on learning to communicate results in writing, using formal reporting norms (APA style).

Open to Graduate Psychology majors only. Pre Req: PSYCH 640 and 641