Lois Downey posted on Friday, January 27, 2006 - 5:21 pm
I have run the following model: f1 by y1-y4; f2 by y5-y6; f1 f2 on x1; (y1-y4 are all ordered categorical, and x1 is a dichotomy)
In the Model Results, the estimate for f1 with f2 is 0.580, and the Std and StdYX both = 0.613. In the Tech4 output, the estimated covariance matrix shows f1 with f2 = 0.616, and the estimated correlation matrix shows f1 with f2 = 0.651. In laypersons' terms, what is the interpretation of these 4 coefficients?
Also, the Tech4 output shows the estimated means for f1 and f2 as 0.520 and 0.180. Are these the estimates for the combined values on x1, or just for persons who score 0 on x1? (If it is the former, is there any way I can compute the estimated f1 and f2 means for each category of the predictor?)
When a model has covariates, intercepts and residual covariances are estimated. When a model does not have covariates, means and covariances are estimated. Because you have covariates in your model, the value for your WITH statement is a residual covariance. The results show intercepts for f1 and f2 not means. TECH4 shows model estimated means, variance, covariances, and correlations.
Lois Downey posted on Wednesday, February 01, 2006 - 5:20 am
As nearly as I can tell, my results section doesn't show intercepts for f1 and f2. It gives only the estimated loadings of the indicators on the latent variables, the path coefficients for the latent variables on the covariate, residual covariances between the latent variables, threshholds for the categorical indicators, residual variances for the latent variables, and the R-Square table for the indicators and latent variables. Where should I be finding the intercepts?