Fun stats lesson for the day.

Simpson's paradox, and how it can might make LPing look harder than it really is under some circumstances.

No idea if this happens in practice. Just a cautionary note about aggregating individual LP data into an average LP for display purposes.