|
|
Counts/proportions for level-2 in ML ... |
|
Message/Author |
|
|
Dear professors, I am running a two-level mixture model (repeated measurements on level-1 and persons on level-2) with the latent class on the between-level, and I would like to know the number of persons in each class. It seems that the "FINAL CLASS COUNTS AND PROPORTIONS" in the output concerns the number and proportions of level-1 observations, but since my design is unbalanced it is hard to know how many persons this is. Is there any way of getting these estimates? Best, Fredrik Falkenström |
|
|
You can use this code model: [C#1] (a1); model constraints: new(p1); p1=1/(1+Exp(a1)); This will give you the model estimated class probability. If you want to see the posterior probability for each person you can use savedata: file=d.dat; save=cprob; |
|
|
Thanks, but how do I interpret the p1 estimate? Is it the proportion of Level-2 clusters (persons in my case) belonging to class2 as opposed to class1? |
|
|
Class 2. |
|
Back to top |
|
|