S1E16: Missing Data: (IF EPISODE=16 THEN EPISODE=-999)
Quantitude
00:00
The Paradox of Missing at Random
The best test of whether or not your data are missing completely at random is if a unicorn canters by with a talking narwhal on its back. If you don't have anything on the table, something on the table that's relevant, then I think the test is not particularly useful. The name is absolutely horrible, but the idea is that if you condition on this variable, then it does become random. It's kind of like being completely pregnant or being pregnant.
Transcript
Play full episode
