I am working on data of monthly pain assesment among workers. I would like to use the 14 measurements to define classes of workers based on their pain level at each assesment. Therefore LCGA seems to be a way to do so. I am workinig in mplus, but a colleague is using latentgold. We get different results when comparing our analysis. We thought that is was due to the fixed within class variability. Therefore I used the following syntax to free the variance: ANALYSIS: TYPE = MIXTURE; estimator is mlr; MODEL: %OVERALL% i s | INT1@0INT2@1INT3@2INT4@3INT5@4INT6@5INT7@6INT8@7INT9@8INT10@9INT11@10INT12@11INT13@12INT14@13 ; %c#1% i s ; %c#2% i s; %c#3% i s; %c#4% i s ;
Is that correct?
However, we still get different results. My colleague gets a class that contains only 0 all the way (since the data is on a working population, many do not experience pain throughout the year that the Measurements were taken), whereas I get a class with very low pain levels. How can that be?