Wednesday, October 20, 2010

Selection Bias

I've been playing my iTunes on shuffle for the past three days, and I've been continually impressed with just how good the playlist has been. Owl City and Ingrid Michaelson and Imogen Heap and Kate Nash and Hellogoodbye and some great Tamil songs thrown in, along with some of the new Maroon 5 album. It made me really happy that such great songs were coming up on shuffle.
...
And then I realized that I'm loving all the songs that are coming up because they're, well, my library. Selection bias works its magic: I chose all the songs that went into my iTunes library, so of course a random playlist of them would be filled with all my favorites. It ruins the magic a tiny bit, but it also makes me smile. How circular our lives can be.

Another example of selection bias: patients on peritoneal dialysis are more likely to survive than those on hemodialysis, even though peritoneal dialysis is less effective at renal clearance than hemodialysis. Why, you ask? Well, because of many factors, including better retention of renal function in those on peritoneal dialysis, the lack of biocompatibility issues for peritoneal dialysis [unlike the ones rampant in hemodialysis], more cardiovascular complications on hemodialysis, and better clearance of middle molecules [which may be toxic to dialysis patients in the long run] in peritoneal dialysis. But most importantly, patients on PD might have a selection bias phenomenon: those who choose to do the more independent and demanding form of dialysis are likely to be better-educated (and thus take better care of their renal health overall), less sick, and have better family support.

^I wrote that entire paragraph to review something I studied today. Exams start in 6 days, and it's time for lockdown. Not long till I see the light!