I have developed a model comprising two dimensions of aggression: (1) two contexts, real life and cyberspace; and (2) three forms, verbal, physical, and indirect, resulting in six cells, which comprise between 1 and 8 items.
The fit of the model including all five factors is better than the one taking into account only contexts or forms, respectively, which is a good thing. However, I have two questions:
1) Is it okay at all to use items as indicators of more than one dimension (here: context and form)? 2) Is there a problem if one of the six cells has only one item? (All factors are represented by five and more items, though.)