I need to construct a CI for use in move selection. I can't use the Wald CI as used in SPRT because it has poor coverage in many cases, it's inappropriate if the number of samples is small, and if the mean is an extreme value e.g. close to zero or one, the average interval coverage is very bad.
Someplace, maybe on this board, I read a discussion on how draws are handled when using the Wald interval. My memory isn't the greatest. It needs to be refreshed often for me to retain the information. I waited too long and have forgotten what I read. I've looked for this discussion again with out luck. Could someone give me a brief explanation of this or direct me to a paper or discussion that deals with this topic.
Regards,
Zen
Constructing a Confidence Interval
Moderators: hgm, Dann Corbit, Harvey Williamson
-
Zenmastur
- Posts: 919
- Joined: Sat May 31, 2014 8:28 am
Constructing a Confidence Interval
Only 2 defining forces have ever offered to die for you.....Jesus Christ and the American Soldier. One died for your soul, the other for your freedom.