Message/Author 

Jon Heron posted on Monday, July 08, 2013  6:58 am



Hi Bengt/Linda I see we have a new bit of output, and I'm struggling to work out exactly what it is. Please can you explain the difference? I can recreate the first but not the second in Stata using the cprobabilities Average Latent Class Probabilities for Most Likely Latent Class Membership (Row) by Latent Class (Column) Classification Probabilities for the Most Likely Latent Class Membership (Row) by Latent Class (Column) thanks, :) 


The new version of Web Note 15 has the details I believe. 


The revised Web note 15 also describes the Lanza et al. method for distals that we recently incorporated (option names DCON/DCAT). 

Jon Heron posted on Monday, July 08, 2013  7:59 am



ace, thanks Bengt 

Jon Heron posted on Tuesday, July 09, 2013  8:25 am



Ahh, the second is Vermunt's D matrix. 


Is the auxiliary option not available for GMM that use numerical integration to handle ordinal data? Do you have suggestions on potential alternatives? 


Only Auxiliary types E and R are available for Mixture with numerical integration. None of the new types are available. 


From page 4 of Webnote 15, it seems like the table “Classification Probabilities for the Most Likely Latent Class Membership (Row) by Latent Class (Column)” represents the probability that the most likely class will be c1 given that the true class is c2 (P(S = c1  C = c2)). Is that true, and if so, why don't the columns sum to 1? Thank you! 


Each row looks at the observations with most likely class membership in a particular class. The columns represent the average of the posterior probabilities for those observations for each class. 


Hi Linda, I have the same question as Jonathan about the table "Classification Probabilities for the Most Likely Latent Class Membership (Row) by Latent Class (Column)". Webnote 15 says it represents P(S=c1C=c2). Since the columns represent latent class (C) and the rows represent most likely class (S), shouldn't the columns sum to 1? Thank you! 


Good that you asked  the output heading is off. For both the Classification probabilities table and the Logits table, the words "row" and "column" should be switched. This means that it makes sense for the Classification table that for a given row the probabilities sum to 1 over columns. 


Thank you! 

Back to top 