Jan Ivanouw posted on Tuesday, October 02, 2012 - 7:48 am
When computing FS for 1st and 2nd order factors in a 2nd order factor model with categorical indicators and the WLSMV estimator, Tihomir Asparouhov told me at a consultation in Utrecht that both are calculated using the procedures described in tech app 11. While I get the right results for a simple first order model, and also for the 2nd order factor in a 2nd order model, I do not get the rigth results for the 1st order factors in the 2nd order model. I wonder if this is because some scaling problem, using logit or probit scale parameters?
Can you tell me which scaling is used for entering the parameters in the eq’s 229-231 from app 11 when calculating 1st order factors in a 2nd order model?