An empirical investigation into the utility of the not proven verdict