A comment on D. V. Lindley's statistical paradox