Iain
Thanks for this - the s/ware solved my problem. The 2x2 tables for test
evaluation data work very well & gave me all the stats I needed. I couldn't
do the same with SPSS9.
===
Alan Hassey (mailto:[log in to unmask])
Dear Alan,
A likelihood ratio is a ratio of two binomial proportions, and this is the
basis upon which confidence intervals are constructed for LRs. Some
software gives rather poor approximations to the interval. In order to
calculate precise confidence intervals for LRs, you might like to try
http://www.statsdirect.com/update.htm and install the software. Please let
the list know about this if you find it helpful.
Kind regards
Iain Buchan
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|