User blog comment:Kitahara/Recording drum expedition success rates/@comment-25779449-20150129053933/@comment-25730832-20150129074635

I would want 0 sparkle to have as much data as possible since that will be our baseline. As you mentioned, the baseline doesn't seem to be 50% at all. My current record is 1/15 for Exp21 which is around 7%. However, I'm only on my 15th run  so that will probably change like 1 sparkle case.

From what I see is...I'm pretty confident that 50% is just an arbitrary number someone throws in.

I want to get to 38 runs of 0 sparkle for exp21 since that should give equal sample size between mine and yours which (error and C.I. aside) might make it more acceptable to just ignore FS and combined level and finally get one big data.