Why not? It's useful to have a study at a higher level of abstraction with the tradeoff that it is less certain.

Should a physicist be making approximations? Probably not, but a biomedical researcher probably should.