I'm calculating distances for groups based on Age/Gender Compositions (to rank their similarity in demographic composition.)
I'm working with the following: Men 18-34, Men 35-49, Men 50-64, Men 65+, Women 18-34, Women 35-49, Women 50-64, Women 65+.
They're each a gender with roughly 15 years... I don't have demographics info by smaller breaks/continuous. Even though these demographic breaks are somewhat related and ordinal, is it okay to treat them like discrete probabilities for a Bhattacharyya calculation? Is there a more accurate way to do it? My resulting distance rankings make sense with the content.
Thanks!!
