Structure of data for LGC modeling PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
 Stace Swayne posted on Sunday, March 24, 2013 - 9:39 am
Dear Dr.'s Muthen,I have a question regarding the structure of

variables , my issue is the following:

I have 4 waves of data, in which the CES-D20 was measured, my question is---for each time point is it

appropriate to use the total score (sum across all 20 items) for the CESD at T1, T2, T3, T4, or is it more

appropriate to use the actual measurement model (latent variable for overall depression by v1 through

v20, across T1-T4).

Any advice would be greatly appreciated,

Warm regards,
 Bengt O. Muthen posted on Sunday, March 24, 2013 - 10:22 am
Using the total (unweighted) score assumes that the 20 items are unidimensional (with approximately the same measurement parameters) and that the measurement parameters are invariant across time. You might first want to do a factor analysis to confirm these assumptions. Different sub-dimensions (factors) may have different development over time and some items may not show invariance.

Note that Mplus makes it possible to do "longitudinal factor analysis", either using CFA or EFA (the latter via ESEM), that is, you can analyze all 80 items together.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message