This is all true. I originally added a disclaimer to my post that said "assuming you have enough data to support the level of granularity" but I removed it for

This is all true. I originally added a disclaimer to my post that said "assuming you have enough data to support the level of granularity" but I removed it for brevity because I thought it was implied -- small sample size isn't part of Simpson's paradox. My apologies for being unclear