Limitations of RCTs (part 2.5): Green jelly beans linked to acne

The last in my three-part series on the limitations of randomized controlled trials (RCTs) is still in the works. In the meantime, here’s XKCD* on the dangers of atheoretical, post-hoc subgroup analysis.

Significant

* XKCD: The unofficial web comic of international development nerds. See also here (about statistics), here (about local ownership), and here (about physicists, but also economists).