Discussions of circular inference (a.k.a. biased selection, a.k.a. double-dipping) and how to avoid it during statistical analyses. It is about 15 minutes long.