Creating multiple grouping variables ... PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Dylan Johnson posted on Tuesday, April 18, 2017 - 11:11 am
My data is currently set up in a flatfile whereby each row (case) has three time points (T1, T2, and T3). I am running a latent profile analysis on each of the three time points and it looks something like this:

varA_T1
varB_T1
varC_T1
varD_T1

varA_T2
varB_T2
varC_T2
varD_T2

varA_T3
varB_T3
varC_T3
varD_T3

I am aware of the Save = cprobs command, but will this work when I technically have three profiles for each case: one at T1, one at T2, and a third at T3?

What is my best option for this?
 Bengt O. Muthen posted on Tuesday, April 18, 2017 - 6:15 pm
It sounds like you should analyze these 12 outcomes in a single analysis using a latent class variable at each time point, which means that those 3 latent class variables should be related as in an LTA.
 Dylan Johnson posted on Wednesday, April 19, 2017 - 8:06 am
My full planned analysis is to conduct a latent profile analysis at each time point then to conduct an LTA across each time point. The indicator variables for each profile are as follows:

internalizing_T1
externalizing_T1
ADHD_T1
indirect_aggression_T1

internalizing_T2
externalizing_T2
ADHD_T2
indirect_aggression_T2

internalizing_T3
externalizing_T3
ADHD_T3
indirect_aggression_T3

I have already ran a latent profile analysis at each time point and the best fitting model consisted of 4 profiles. This was the case for all three time LPAs, at each timepoint. What I would like to do next is bring my data back into SAS and conduct an LTA, followed by a logistic regression. I plan to group my cases based on their trajectory over time and then use these trajectory groups to predict outcomes using a logistic regression.

So I need to create three new variables for each case in MPlus. One variable for their most likely profile at time 1. One variable for their most likely profile at time 2. And one variable for their most likely profile at time 3. Is this possible to do?
 Bengt O. Muthen posted on Friday, April 21, 2017 - 6:07 pm
Yes, use Save = Cprobs and look at the last columns.
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: