I have model with 7 constructs. 6 of them are latent variables, and I use likert scales (1-5) to measure them, and one construct is e-marketing budget (F2) presented as percent of total marketing budget.
Relations in my model are:
Factor 1 -> Factor 2 (E-marketing budget)
Factor 1 -> Factor 3
Factor 2 -> Factor 4
Factor 3 -> Factor 4
Factor 2 -> Factor 5
Factor 3 -> Factor 6
Factor 4 -> Factor 5
Factor 4 -> Factor 6
My question is what kind of statistical analyse should I use to investigate relationships between factors (effect of one factor to another).
Is it a good approach to calculate scores for every scale and than calculate correlation between factors based on scores (how strong and how significant correlation is), and use results for interpretation.
If scores are way to go, my other question is what kind of scores should I use: mean score or sum of items. (I have seen different approaches how to do it, if someone can give me good reference how to do it in SPSS, that would be great).
Thank you.