Drs, I am interested to test whether theorized relationship is invariant across two time points (T0 & T1). I would like to make sure that I use appropriate modeling approach to test the hypothesis. I have 3 latent variables (4 factors each). An assumption is that exercise climate relates to perception of autonomy and that, in turn, relates to exercise behavior. Specifically, my questions are: 1. Should I test measurement invariance where all variables are in one model: I.e. climate -> autonomy->behavior together with regressive paths between climate (T0 & T1), autonomy (T0 & T1), and behavior (T0 & T1). And then testing partial and full measurement invariance using normal procedures. 2. OR should I test, firstly, measurement variance separately between each variable i.e. climate (T0 & T1), autonomy (T0 & T1), and behavior (T0 & T1). And, secondly, adding the regressive paths after measurement variable is established. My primary interest is whether theorethical model is invariant across time. Thank You very much.