You cannot identify a growth model with both intercept and slope growth factors with two repeated measures. You would need at least three and we recommend at least four for modeling flexibility, that is, being able to free time scores and add residual covariances if needed.
Is there another way for mplus to help analyze a model with only two time periods?
I study whether dropping out of high school leads to increased alcohol use. I have variables for (1) drinking before drop out (2) drinking after drop out and (3) age at left school. I must control for unobserved heterogeneity (selection effects) as well as observed covariates.
What age range do you have in your data? If a student does not drop out, when do you measure the second alcohol use? Or do you do it at the same time for everyone? Are students all the same age at the first alcohol measure?