When running an LCA with 8 categorical variables and 1492 cases, I keep getting the error:
Categorical variable INJSOM99 contains 127 categories. This exceeds the maximum allowed of 10.
I have looked at the data, and it is coded as 0,1,-99 (with -99 as missing in the mplus input). I have tried to recode the data using SPSS and by hand, in case there was an issue with the zeros being "O"s or something. But, that doesn't seem to fix it. I have also looked if there was missing data in any of the cells, but I don't see any.
You either have blanks in the data set and are reading it in free format causing the data to be misread or the number of variable names in the NAMES statement does not match the number of columns in the data set causing the data to be misread. If you can't see the problem, send the input, data, output, and your license number to email@example.com.
anonymous posted on Tuesday, April 12, 2011 - 12:37 pm
Hello, In terms of interpreting the best solution for a LCA, what do you do if the information criteria fluctuate so there are several "dips" in values? Do you take the first "dip" in the BIC value, even if a second "dip" shows a lower BIC value overall? thanks.
You choose the lowest BIC. Several "dips" might indicate using too few starts.
Djangou C posted on Friday, July 26, 2019 - 3:15 am
Dear Mplus team, I am aware that we can estimate an LCA and use the parameters to classified new people into the classes. Is it possible to improve the classification over time with the addition of new individuals without re-running the model from the start? Thank you in advance
To improve the classification, you would have to get new parameter estimates which does call for a new analysis. On the other hand, you can classify new people by using the same model and simply fix the parameters at values from the initial run.