BGonline.org Forums

Question on combining ?

Posted By: David Rockwell
Date: Monday, 19 November 2007, at 7:59 p.m.

In Response To: Question on combining ? (neilkaz)

I have forgotten my statistics which I last touched in 1977. But, I believe common sense is sufficient to find an answer in this case. A rollout of 46656 is simply 36 different rollouts of 1296 lined up one after another. To get the combined results of all 46656, you add them all together. The standard deviation of component parts is not used to get a combined result. If that was correct, you could get different results from the same set of rollouts depending on how you segmented the sample data. (For example you could split the 46656 data points into 23328 and 23328 or you could split them into 1296 and 45360. Any method that gets two different answers depending upon the segmentation is not working.) The same principle holds for the combination suggested in the previous post. If you want combined results, you perform a simple weighted average based on rollout length.

Messages In This Thread

• 21slot/ 11 and 21slot/31
  bob koca -- Sunday, 18 November 2007, at 7:53 a.m.
  • 21slot/ 11 and 21slot/31
    Stick -- Sunday, 18 November 2007, at 8:05 a.m.
    • 21slot/ 11 and 21slot/31
      bob koca -- Sunday, 18 November 2007, at 9:46 a.m.
      • 21slot/ 11 and 21slot/31
        Stick -- Sunday, 18 November 2007, at 9:49 a.m.
        • 21slot/ 11 and 21slot/31
          neilkaz -- Sunday, 18 November 2007, at 3:55 p.m.
      • Question on combining ?
        neilkaz -- Sunday, 18 November 2007, at 4:18 p.m.
        • Question on combining ?
          Maik Stiebler -- Monday, 19 November 2007, at 12:40 p.m.
        • Question on combining ?
          David Rockwell -- Monday, 19 November 2007, at 7:59 p.m.