Structural Equation Modeling

Structural Equation Modeling (SEM) is a method for determining the extent to which data on a set of variables are consistent with hypotheses about causal association among the variables. Reults are produced by creating a path diagram and letting the statistical package determine the covariances, regression coefficients, and factor loadings (for latent constructs) that apply to the arrows drawn in the diagram. Benefits to using SEM include:1) use of maximum likelihood to make use of cases with missing data, 2) confirmatory factor analysis for latent constructs,3) measures of model fit and diagnostics to determine the source of ill fit, and 4) ability to apply assumptions to the model (such as 0 correlation between two variables).