What Should a Researcher Do If He or She Fears

What should a researcher do if he or she fears that a nominal variable's categories are not sufficiently homogeneous?

A) convert the variable to interval-ratio form
B) revert to descriptive analysis
C) delete some of the categories for the variable
D) generate extra categories so that cases are classified with enough specificity for the researcher's purposes