I just completed principal component analysis (PCA) and want to progress to multinomial logistic regression. I am wondering if I need to combine grouped items from my PCA into one variable in my regression analysis.
Normally ones does principal component regression on the PCA-scores as X-variables. Or did you see PCA only as exploratory analysis before regression on the raw-data itself?
It depends on the objectcive of your research. However, variables will be uncorrelated so may differentiate better among the two groups when doing logistics regression.