Hi, In your mails you give a reference to my old mail which was a reply to : http://root.cern.ch/root/roottalk/roottalk02/1393.html I think I need to confess something. Although the "facts" that I mentioned in my reply above are correct (i.e. in general log-likelihood behaves better than chi-square in low statistics cases), the particular problem, that the original author had, was related to something else. (In other words, my "early" diagnosis was a "false" one. Thank Goodness, and some colleagues who corrected me, the patient is still alive, I hope.) As the problem might be interesting for many people, I am just trying to find the "old" mails about this subject. If we succeed, we will post a summary to the RootTalk. Jacek.
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:12 MET