My research topic is related to motivation regarding product usage and the negative items mainly consist of problems that hinder motivation. My dataset is composed of both positive and negative items (likert scale). I have 15 positive items are and 14 negative items. The sample size is 100. Conducting the analysis for the entire data set(both positive and negative items) results in a clumsy gathering of variables which can not be grouped easily. However, if I conduct EFA for positive items and negative items separately, it shows more a better organized result.
My questions are:
Whether I should conduct EFA separately for positive and negative items?
Can I separate the negative items and analyze frequency, means and standard deviations?
Is reverse coding for negative items OK or should I avoid it?
Do I have to select factors from the elbow of a scree plot or I can reduce the number of factors?
I am using SPSS for my problem.