Sean Oneile posted on Wednesday, August 14, 2019 - 11:39 pm
Following the latent growth curve model with a distal outcome presented in Muthen and Curran (1007, Fig3), I would like to estimate a linear LGC model with the observed x being measured at three time points (3 weeks apart) and then predict x measured at 32 weeks later.
Would it be possible to include two dummy-coded covariates (intervention A vs B; and intervention A vs C) in the above model to examine potential different intervention effects on the growth factors as well as the distal outcome?
I thought about the multiple-group approach, but the sample is only N=450 (roughly equally distributed across three intervention groups). I would appreciate your inputs.