Can I perform an EFA (oblique rotation) with variables that have undergone different types of transformations? Can I still interpret the factor loadings and other output coherently?
For questionnaire development I want to perform an EFA with 24 variables (5 point likert scale), of which 15 have significant negative and 2 positiv skew (probability of the z score of skewness <.01; although all skewness <|1|). Different transformations help best for different variables (log10, square root, inverse). Thus, my question (see above). Or do you have other recommendations? Thank you!
