ngraetz / g_formula

Testing environment for generalizable g-formula development.
0 stars 1 forks source link

Test more complex autocorrelation structure within-individuals (e.g. latent growth curves) #6

Open ngraetz opened 5 years ago

ngraetz commented 5 years ago

I'm pretty sure "latent growth curves" are just a dumb SEM way of estimating random intercepts by individual and random slopes on age by individual. We should just do that. The more complex step would be adding a functional form to the random slope, i.e. Gompertz on the censoring/mortality model or whatever. Adding this latent individual structure also makes us not have to worry about unbalanced panels.