On a statistical H-index