Intro to SEM P. Soukup
Literature Kline.2005. Principles and practice of structural equation modeling. New York : Guilford Press Byrne. 2001. Structural equation modeling with AMOS :basic concepts, applications, and programming. New Jersey: Lawrence Erlbaum Maruyama.1998. Basics of structural equation modeling. Sage Publications Raykov and Marcoulides.2006. A first course in structural equation modeling. Mahwah : Lawrence Erlbaum Associates Schumacker and Lomax.2004.A beginner’s guide to structural equation modeling. Mahwah : Lawrence Erlbaum Associates Articles: Journal Structural Equation Modeling
Exploratory FA Confirmatory FA
Why SEM (CFA)? Testing of hypothesis(es) Complex model of relashionships between latent and manifest vars Also possible: compare groups, longitudinal analysis etc.
Correlation, regression and path analysis Before CFA Correlation, regression and path analysis
Correlation and simple regression Both sided relationship=correlation One sided relationship = regression (simple). E
Multiple Regression analysis More ind. vars Y' = a + b1X1 +b2X2 +b3X3
Some statistical notes Nr. of estimated parameters (what are these) Correlation Regression Nr. of individual pieces of info from data Degrees of freedom Test of model: chi-square Examples in regression and correlation
Path analysis=more regressions Two or more regressions at once Dependent and independent vars – necessary to exchange by exogenous and endogenous Direct and indirect effects Measurement error (E) E E
Path analysis-example Duncan’s model Evaluation of different models Constraining of parameters The best model? (AIC or BIC for selection)
Intro to CFA
Exploratory FA Confirmatory FA
CFA – equations (I) Equations for vars: E E
CFA – equations (II) Equations for covariances: We have: covariance matrix for manifest vars in our data (Σ) We estimate covariance matrix of latent vars (Ψ), measurement errors (Θ) and factor weights (Λ) E E
CFA – estimates Many techniques max. likelihood generalized LS unweighted LS generalized LS max. likelihood E E
CFA – evaluation Overall evaluation – test and criterias Individual parameters – statistical and substantive significance Change in model – modification indeces E E
Software for SEM AMOS EQS LISREL MPlus SAS – CALIS Statistica - SEPATH