Back to Top

Paper Title

Model fit and model selection in structural equation modeling.

Keywords

  • Structural Equation Modeling (SEM)
  • Model Fit
  • Model Selection
  • Path Diagram
  • Hypothesized Measurement
  • Observed Variables
  • Constructs
  • Causal Relationships
  • Hypothesized Mean Component
  • Parameter Estimation
  • Maximum Likelihood (ML)
  • Generalized Least Squares (GLS)
  • Confirmatory Factor Analysis (CFA)

Article Type

Book review

Issue

Volume : Chapter 13 | Page No : 209–231

Published On

March, 2012

Downloads

Abstract

One of the strengths of structural equation model ing (SEM) is the ability to test models that represent a complex set of theoretical hypotheses. The set of hypothesized relationships is specified and commonly represented graphically in the compact form of a path diagram. The model and its associated path diagram contain one or more of three components. It may con tain a hypothesized measurement component that re lates the observed (measured) variables to underlying constructs (Figure 13.lA). It may contain a structural (path) component that portrays the hypothesized causal relationships between the constructs (Figure 13.lB). It may contain a hypothesized mean component that por trays similarities and differences in the level of the con structs, potentially as a function of other variables (Fig ure 13.lC). Once a path model is specified, an important question arises: How well does the hypothesized model fit observed data on each of the variables? The path model diagram implies a set of algebraic equations whose parameters (e.g., factor loadings in Ay, factor variances and covariances in 'I') are estimated, typically through maximum likelihood (ML) or gener alized least squares (GLS) estimation procedures.

View more >>

Uploded Document Preview