Elina Dale posted on Tuesday, May 07, 2013 - 9:04 pm
Dear Dr. Muthen, Am using mixture randomized trials modeling with CACE estimation. I ran my model 3 times with increasing number of STARTS and STITERATIONS. It seemed to work fine and my best loglikelihood values were replicated. However, I keep getting this warning message (and based on TECH1 I see it's about my Alpha (F) parameters for my Class 2):
"These parameters were fixed to avoid singularity of the information matrix. [It] is most likely because the the model is not identified, or because of empty cells in the joint distribution of the categorical variables in the model."
I also see in the Model results that Intercepts for latent class 2 (F1, F2, F3, F4) have p-values=999.000, i.e. they couldn't be computed. As I understand this is the result of the above warning message, b/c for class 1 I got coefficients and p-values.
I do not understand where the Program has "fixed these parameters" as it says in the warning message. Where do I find them? Are they those values that are provided under the Intercepts? If so, then why are their p-values impossible to compute? Thank you!