Dear Drs. Muthen, I am thinking about using group-level slopes as indicators in a latent growth model. Measuring the relationship between effort and performance over some learning period (e.g., 6 paired measurements of effort & performance over 6 training sessions), I am interested in showing that group-level effort/performance relations occur (i.e., a LGM intercept factor) and that group-level effort/performance relations increase over time (i.e., a LGM slope factor) as a function of group-level influences (e.g., trainer characteristics, cohesion, communication, etc.).
Do you know of any problems with a slopes-as-indicators model? It appears that most people are stuck thinking in terms of the HLM-style slopes-as-outcomes models, but I can't come up with any reason why using slopes as indicators wouldn't be feasible. Have either of you ever heard of anyone doing this before? Can you imagine any special issues here with using Mplus for this?