I am running a PCA analysis to create wealth index in a sample where people are living in both urban and rural area, one of the main variables I am using is livestock. This is quite common in rural areas and not so common in urban areas. In urban areas we feel that owning livestock somewhat indicates one not being wealthy, where as in rural areas it indicates if one is wealthy so this variable is working in two directions. Any good suggestions how we can tackle this situation without splitting our sample on urban and rural populations?
Is there an equivalent proxy measure for wealth in urban areas? Rather than trying to have livestock mean different things or split the sample can you create a new numeric item that may be informed by # of livestock in rural areas and # of something else in urban areas?
- Badges
- Science topic