GP posted on Wednesday, September 02, 2009 - 4:00 pm
I'm running LCGA with two classes and 6 timepoints with data values ranging from 17 to 92. The plot of the sample and estimated means shows that these means are far apart at timepoint 1 (about 20 points), but not at later timepoints. There is no missing data at timepoint 1, while there is afterwards. I'm a little puzzled, and I would like to hear your thoughts. Thank you in advance.
Perhaps you use a linear growth model, but should use a quadratic or piecewise model. In some studies, such as with mental health, the first time point often has a much different value than subsequent time points which cannot be well approximated by linear development.