We have been examining the risk for depression across different developmental phases. We were wondering if there was a way to model the data in MPLUS such that all developmental phases are in the same model/analysis. Right now we are running each separately but we would like to compare the different developmental phases to determine if the pathways are different.
earlpphg on pardepg ag; m9g on earlpphg pardepg ag;
MODEL INDIRECT: m9g via earlpphg pardepg;
We would like to compare child onset to adolescent to adult however the pathways are nonindependent i.e., if you are 18 and you don't get depressed until 17 you are still at risk for MDD and part of the denominator for those who get MDD in childhood (prior to age 13)(as in survival analysis).
Perhaps you want to consider doing discrete-time survival analysis in line with the Muthen-Masyn article on our web site. It should be possible to test different effects of background variables on the hazard for different time periods.