Alex He posted on Saturday, February 07, 2015 - 11:52 am
I wonder what the best way would be to analyze data collected from infants that were measured at different time points.
For example, time 0 for some infants were 2 months of age, and for some were 4 months of age. Time 1 for some infants were 6 months of age and for some infants were 8 months of age, so on and so forth (there are more time points)
I suppose one way to do it is to use TYPE = RANDOM MIXTURE, and create time scores. To do so, I would need to treat all infants' first measurement times as 0, despite the fact that not all infants were of the same age at time 0.
Developmentally, however, infants that are 2 months of age can be quite different from infants that are 4 months of age, so I am not sure if this is the best way to go about it.
An alternative (which I am not sure if it is OK), is to treat data collected at 2 months of age as time 0 data. For those infants that were not measured when they were 2 months old, their data are considered missing. And I do the same thing for all the subsequent time points (e.g., Month 4, Month6, Month 8, etc.).
I wonder if the 2nd solution is a possibility. If not, I wonder what might be a good way to deal with this scenario.
Alex He posted on Saturday, February 07, 2015 - 12:02 pm
Just to follow up: would another alternative be that I use option 1 as described above, and include the infants' ages at time 0 as a time-invariant covariate?
You should use age as your time axis and use the individually-varying time scores approach of Mplus. See the TSCORES and AT options.
Alex He posted on Saturday, February 07, 2015 - 3:32 pm
Thank you for the quick response! I have a couple of follow up questions. The model below is based on the model in Ex6.12
TITLE: Individually varying time GMM DATA: FILE IS ex6.12.dat; VARIABLE: NAMES ARE y1-y4 x a21-a24 a11-a14; TSCORES = a11-a14; CLASSES = c(2); ANALYSIS: TYPE = RANDOM MIXTURE; MODEL: %OVERALL% i s | y1-y4 AT a11-a14; st | y1 ON a21; st | y2 ON a22; st | y3 ON a23; st | y4 ON a24; i s st ON x; c on x;
(1) Does the model look reasonable? (2) How can one obtain the probability of class membership for each individual?