You need to impute latent variables at dimension level. You may run two separate analyses for this i.e., one for overall IV to DV and other for dimension level analysis. In alternate, you can test it in a single model. For that you have to drawn model according to your requirements.
In CFA, the criteria for goodness of fit of the data to the model is assessed through CFI, GFI, TLI, RMSEA and RMR.
Thank You So much for valuable answer *Wali Ur Rehman. Will you help me further if i needed? Actually i am on initial stage of AMOS. will be thanked for help.
As you mentioned you are a beginner in SEM (AMOS). I would advise to check the assumptions for AMOS first. For example AMOS is a co-variance based technique which is sensitive to sample size (n-200+), contrary to this Smart-PLS is a non-parametric technique which does not require a specific sample size. You might also need to address outliers and fulfill the assumptions of normality and linearity. If you have addressed all these assumptions than you may proceed with confirmatory factor analysis and model estimation.
I would recommend you to please read
Chin, W. W., and Newsted, P. R. 1999. "Structural Equation Modeling Analysis with Small Samples Using Partial Least Squares." In Statistical Strategies for Small Sample Research. Ed. R. H. Hoyle. Thousand Oaks: Sage, 307-341.
Hair, J. F., Ringle, C. M., and Sarstedt, M. 2011. "PLS-SEM: Indeed a Silver Bullet." Journal of Marketing Theory and Practice 19 (2): 139-151.
You may also find the AMOS user Guide useful at this stage (Attached for your help)