[18790] in s-news-athena
[S] Function for Exact Statistics
daemon@ATHENA.MIT.EDU (James Janisse)
Thu Jul 29 17:08:28 1999
Message-Id: <37A0BEE7.CDE3D0EB@med.wayne.edu>
Date: Thu, 29 Jul 1999 16:51:51 -0400
From: James Janisse <jjanis@med.wayne.edu>
Mime-Version: 1.0
To: S-news S-PLUS <s-news@wubios.wustl.edu>,
Daniel Barth-Jones <barthjon@umich.edu>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
I need to obtain exact confidence intervals on the ratio of two binomial
proportions (aka - the relative risk). The methods need to be exact for
two reasons: 1) I have a considerable number of occurrences of zero
cells which cause division by zero problems, when this occurs my point
estimates are infinite, but it would still be possible to obtain a lower
confidence interval. 2) The sample sizes are small enough to bring the
asymptotic results into question. The package StatXact is able to
perform these calculations but I have over 10000 pairs of proportions to
work on. I would prefer to write a program in S-Plus that would
calculate the confidence intervals for me instead of running StatXact
over 10000 times.
Has anyone written an S-Plus function to obtain exact confidence
intervals. I could easily modify the code to run the function on my
data set. I am using S-Plus 2000 and Windows 95.
Two relevant references for these procedures are:
1) Santer & Snell 1980 JASA 75:386-394
2) Berger & Boos 1994 JASA 89:1012-1016
I will be on vacation next week and will summarize all responses when I
return. Thanks again for you help with this.
James
------------------------------------------------------------------------
James J. Janisse
Department of Psychology
Center for Healthcare Effectiveness Research
Wayne State University, Detroit MI USA
------------------------------------------------------------------------
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
