- No small samples.
- Analyze covariance, not correlation matrices.
- Simpler models are better.
- Verify distributional assumptions (multivariate normality)
- Consider theoretical and practical significance, not just statistical significance.
- Report multiple fit statistics.
- Use two-step modeling for structural regression models. (Test the measurement model then the structural model.)
- Consider theoretically plausible alternative models.
- Respecify rationally. (Test the respecified model on new or split-halves data.)
- Acknowledge equivalent models.
- Issues
- Sample size and missing data
- Normality of sampling distributions
- Outliers
- Linearity
- Adequacy of covariances
- Identification
- Path diagram---hypothesized model
- Estimation method
- Major analyses
- Assessment of fit
- Residuals
- Model chi square
- Fit indices
- Significance of specific parameters
- Variance in a variable accounted for by a factor
- Assessment of fit
- Additional analyses
- Lagrange Multiplier test (not endorsed in my SEM course!)
- Tests of specific parameters
- Addition of parameters to improve fit
- Wald test for dropping parameters
- Correlation between hypothesized and final model or cross-validate model
- Diagram---final model
- Lagrange Multiplier test (not endorsed in my SEM course!)
Kline, R. B. (2010). Principles and practice of structural equation modeling (3rd ed.). New York: Guilford Press.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Pearson.
Thompson, B. (2000). Ten commandments of structural equation modeling. In L. Grimm & P. Yarnell (Eds.), Reading and understanding more multivariate statistics (pp. 261-284). Washington, DC: American Psychological Association.