A note on an unexpected anchoring bias in intuitive statistical inference