The problem with statistics