Dear Scholars

I am going to test a complex model (11 latent constructs with 65 items). I did pooled measurement model in AMOS and reached good model fit with the help of modification indices. Out of 11 latent constructs (in my theoretical model to be tested), there are one exogenous variable, six mediating variables, and four endogenous variables. Again, same mediating variables will be tested as moderating variable as well. Being complicated model to test path analysis or structural model with measurement model at a time, I have imputed the data from pooled measurement model (Latent factor scores to imputed factor scores) in AMOS. Again, being complex paths, I drop the idea of drawing paths as well from the imputed data in AMOS. Now, I am thinking to test hypothesis from ordinary least square method (regression analysis in SPSS) from the imputed data (where imputed data will be treated as observed variables). Therefore my questions are: (1) is this scholarly accepted process to test hypothesis? If yes, it would be very useful for me if I get the relevant literatures to support (especially published articles) this idea. (2) Is there better options than using OLS for such a complex model (complex in the sense it is difficult to test full SEM at a time being many items and latent constructs as well as many paths)?