LPA results PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Anne Chan  posted on Wednesday, December 29, 2010 - 2:47 pm
Hello! I run a LPA on three variables. Comparing 1 to 3 classes solution, the 3 classes solution is the best and it works perfectly fine.

However, when I go for the 4 classes solution. There is a error message:

"THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE FIRST-ORDER DERIVATIVE PRODUCT MATRIX. THIS MAY BE DUE TO THE STARTING VALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. THE CONDITION NUMBER IS -0.514D-18. PROBLEM INVOLVING PARAMETER 11.

From the output file, I see that 1 group only have 1 case in it.

May I ask what does this means? Is that the 4 classes solution does not fit the data?

Can I still continue my project with the 3-class solution in this case?

Thanks!
 Bengt O. Muthen posted on Wednesday, December 29, 2010 - 5:03 pm
If your 4-class solution has its best loglikelihood value replicated several times so you believe you have reached the best maximum, then this 4-class solution is not useful and you should proceed with 3 classes. A 4-class solution like that can happen for instance if you let the variances differ across classes, but also in other cases. The one person is like an outlier.
 Sandy lee posted on Monday, November 14, 2016 - 8:42 am
Hello! Hope you are having a great time today.

Anyhow, my question is about Latent Profile Analysis.
I’m conducting LPA with one Social Anxiety Scale which has 5 factors.

As I perform LPA to determine classes with the scale, profile comes out with high-medium-low classes.
But when I perform it with either +1SD samples(N=76), or upper 25% samples(N=110) different and interesting pattern comes out.

Thus, my question is, can I conduct LPA with either +1SD or upper 25% samples within one scale?
Thank you so much for your time. I really appreciate it.

Sincerely,
Sandy
 Bengt O. Muthen posted on Monday, November 14, 2016 - 5:52 pm
When you look at a subsample you can get more detail. But the variable distribution is different and the classes are affected by this so you have to have a strong substantive reason to focus on a subset.
 Sandy lee posted on Monday, November 14, 2016 - 6:30 pm
Thank you so much for the quick reply!

I really hope to focus on subsample but I'm not sure how to provide a strong substantive reason.

Please let me know if you know any study that focuses on subsample while doing LPA or a resource I can refer to.

Again, thank you so much for your precious advice.

Sincerely,
Sandy
 Bengt O. Muthen posted on Tuesday, November 15, 2016 - 10:27 am
I can't think of a ref. right now - ask on SEMNET.

Also check if it comes up in

Lubke, G., Muthén, B., Moilanen, I., McGough, J., Loo, S., Swanson, J., Yang, M., Taanila, A., Hurtig, T., Jarvelin, M. & Smalley, S. (2007). Subtypes versus severity differences in the Attention-Deficit/Hyperactivity disorder in the northern Finnish birth cohort. Journal of the American Academy of Child and Adolescent Psychiatry, 46, 1584-1593.
download paper contact first author show abstract
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: