Mingnan Liu posted on Saturday, April 11, 2015 - 12:23 pm
I have a mode like the following where I force the factor loadings for f1 to be 1. When I am checking the standardized model results, I noticed that the standardized factor loadings for f1 are not identical? My understanding it that they should be the same. Am I right?
Model: c1 by tr1a* tr2a tr3a tr4a; c2 by re1a* re2a re3a re4a; f1 by tr1a@1tr2a@1tr3a@1tr4a@1 re1a@1re2a@1re3a@1re4a@1; c1@1; c2@1; f1 with c1@0c2@0; c1 on black hispeng hispspa female age2 age3 age4 age5 age6 highsch somecoll female inc; c2 on black hispeng hispspa female age2 age3 age4 age5 age6 highsch somecoll female inc; f1 on black hispeng hispspa female age2 age3 age4 age5 age6 highsch somecoll female inc;
The stand'd values are different for different variables because the variables have different variances.
Mingnan Liu posted on Sunday, April 12, 2015 - 10:28 pm
I was trying to replicate the model in
Billiet, J. B., & McClendon, M. J. (2000). Modeling Acquiescence in Measurement Models for Two Balanced Sets of Items. Structural Equation Modeling: A Multidisciplinary Journal, 7(4), 608–628.
In Table 4, the unstandardized lambda's are all 1 and standardized lambda's are all the same (0.169). I tried to apply their model in my data set. Did I model it in the wrong way or did I mis-interpret their table?
They analyze a polychoric correlation matrix for ordinal outcomes which means that the variances of the DVs are all the same and equal to 1. This then gives the same standardized factor loading values for the loadings fixed at 1. You probably don't analyze polychoric correlations.