I'm writing my master thesis and in order to measure the level of receptiveness to new technology-based assessment methods, I used a survey with 14 Likert-items based on the 3 main constructs of an existing model and a new latent integrated into the first model based on another existing model. In that way, I have created a new extended version of the first model with 4 constructs (4 factors), by combining these 2 instruments. I have retrieved the items from existing validated studies, but I have adjusted them to fit the purpose of my study. I have developed 4 assessment scenarios, and to assess the receptiveness to each of them I use the same 14-item survey. Now my question is, do I need to perform a PCA to see the relationship between the items and the number of the factors needed? Of course when I do that, in each assessment scenario I get different loadings even if I use 4 factors in my extraction. Or should I move to a confirmatory factor analysis (skipping the PCA) since I already know that there are 4 constructs and where each item should belongs? Thank you in advance.