Hi Kay,

I think your factors are correlated. Try oblique rotation methods. Such as direct oblimin or quartimin.

Good Luck

In addition to what Abdullah said, you should choose among factors to find the most appropriate factor for any item. And the criterion is to choose the factor with bigger factor loading, i.e. the factor that item is loaded on it more than other factors. Recently, I heard (not read) that the chosen factor should be loaded on at least two times than any other factors. Of course, it's a very harsh criterion.

Dear Kay,

As Abdullah said, it is appropriate to use the oblique rotation technique (I don't know what rotation technique you used.) if the correlations between dimensions are high. However, if you have items' high loadings that more than one dimention, you may do the following. You can delete if the difference between the loadings of the items that load the multiple dimensions is less than about .10. If the difference between the loadings of items is more than 0.10, these items can be assigned to the dimension with high loading. Of course the construct you are trying to measure may not be simple, but complex. In this case it is also possible to associate the items with more than one dimension. At this stage it is important to know (or be able to describe well) the psychological consturct that is being measured.

thank you very much for your kind help

Article Korean Patient-Perceived Satisfaction Scale of Community-Bas...

I hope that it is not too late. You can defend your decision on the issue based on the article I attached above. If you have difficulty in finding the rationale, please send me a message in the ResearchGate.