I have a question I was hoping this group could help answer or provide more insight on.
I'm looking to construct an overall composite index for counties. Really a simplified version of the County Health Rankings http://www.countyhealthrankings.org/
I have 49 county-level health outcome variables such as prevalences of smoking, various cancers, obesity, I have U.S. Census commute, fruit veggie intake, and I have SES poverty/education, and population density data.
I've completed descriptive stats, looked at correlations, conducted principal components analysis using PROC FACTOR in SAS, completed rotation, and have decided upon three PC Factors to include in my analysis that explain about 74% of the variance. Roughly, the factors can be thought of as cancer prevalences, all over disease prevalences(smoking, obesity,physical activity), and SES factors.
My question is as follows: how do I combine these all into one index? Or how do I make separate indices for each? I realize the factors are all uncorrelated but they are all pertinent factors into rating the healthscape of a county which is what I'm trying to index.
In past threads such as this Creating a single index from several principal components or factors retained from PCA/FA which references this publication http://www.cup.ualberta.ca/wp-content/uploads/2013/04/SEICUPWebsite_10April13.pdf (pg. 19), the strategy was frowned upon in the comments by @amoeba and @ttnphns. Can anyone elaborate?
