James Lindgren adds another very good discussion of sampling weights to Tim Lambert's ongoing document.
