Latent class analysis PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Alice Frye posted on Wednesday, October 08, 2008 - 12:36 pm
LCA uses the MLR estimator.
Does this mean that there is no additional value to be gained by using a negative binomial or zero inflated negative binomial regression for categorical variables within a latent profile analysis, because the MLR is sufficiently impervious to violations of distributional assumptions?

I'd be grateful for any comments or references people might have on this topic.
 Linda K. Muthen posted on Wednesday, October 08, 2008 - 4:38 pm
MLR may not be sufficient for continuous variables where deviations from normality are extreme and for categorical and count variables with strong floor or ceiling effects. Special modeling like the negative binomial model for count variables may be needed in these situations.
 Jeongwook CHOI posted on Thursday, January 09, 2020 - 11:45 pm
Hello professor.

I wanna know to check local independence in latent class analysis.

I used polytomous items in latent class analysis.

I saw post that local independence could be identified by looking into Tech 10 result when researcher used binary items.

when I used polytomous items, can I check local independence as the same way?
 Bengt O. Muthen posted on Friday, January 10, 2020 - 1:03 pm
Yes.
 Jeongwook CHOI posted on Sunday, January 19, 2020 - 10:13 am
Hello professor.

I analysed latent profile analysis using 10 continuous variables about dementia.

Local independence was not met.

So, I added direct effect between some variables but the problem was remained.

I added direct effect between all variables to the model.

is this way fine?

Does way that I did affect validity of result?



I analysed model chi-square test to test improvement between 3-class model and 4-class model.

the chi-square test was significant.

But when interpretability of 4-class model was bad, can I ignore the chi-square result? and can I select 3-class model?

thank you.
 Bengt O. Muthen posted on Monday, January 20, 2020 - 8:03 am
Q1-Q2: See UG ex7.22. The references given there show that this is an acceptable model.

Q3-Q4: You cannot use a regular chi-square difference test to check on the number of classes because the assumptions behind that test are not met. See

Nylund, K.L., Asparouhov, T., & Muthén, B. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling. A Monte Carlo simulation study. Structural Equation Modeling, 14, 535-569.
download paper show abstract

I would go by BIC.
 Kätlin Peets posted on Wednesday, September 02, 2020 - 6:09 am
I am running a mixture model. When comparing class 2 to class 3, I am somewhat puzzled. It seems that odds/ratios and standard errors are huge (and p value is not significant). Any idea why might that be the case? As I would expect the classes to differ with regard to the specific item.

Latent Class 2

CBR3N
Category 1 0.051 0.038 1.321 0.187
Category 2 0.949 0.038 24.826 0.000

Latent Class 3

CBR3N
Category 1 0.980 0.021 46.855 0.000
Category 2 0.020 0.021 0.951 0.342

Latent Class 2 Compared to Latent Class 3

CBR3N
Category > 1 926.222 1213.739 0.762 0.446

Thank you!
 Bengt O. Muthen posted on Friday, September 04, 2020 - 3:55 pm
See the FAQs on ORs on our website.
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: