Some people used EFA on the data even the scale is adopted and sometime adapted. Is this logical? or what are the reason behind even they are not developing scale?
Some authors, such as Joseph Hair, suggest using EFA to "explore" the data, observed if some items had "abnormal behavior" (biased towards an answer, for example), but EFA maybe a "meat grinder", i.e., mix variables de others componets (or constructs).
If you have a factorial model (scale already validity) the use CFA seem more logic.
See:
NUNALLY, J.C.; BERNSTEIN, I.H. (1994) Psychometric Theory. 3ª ed. New York: McGraw-Hill Inc.
The authors point that the basic difference between exploratory and confirmatory factor analysis is: EFA --> data driven - without fatorial model and CFA --> Theroy driven - with factroial model.