On May 13, 2010, at 9:24 PM, Frank E Harrell Jr wrote:
On 05/13/2010 08:16 PM, Shi, Tao wrote:
Hi Prof. Harrell,
Could you please elaborate on why chi-square test is more
appropriate in this case? Thank you very much!
...Tao
Exact tests tend to not be very accurate. Typically their P-values
are too large. See
@Article{cra08how,
author = {Crans, Gerald G. and Shuster, Jonathan J.},
title = {How conservative is {Fisher's} exact test? {A}
quantitative evaluation of the two-sample comparative binomial trial},
journal = Statistics in Medicine,
year = 2008,
volume = 27,
pages = {3598-3611},
annote = {Fisher's exact test; $2\times 2$ contingency table;size
of test; comparative binomial experiment;first paper to truly
quantify the conservativeness of Fisher's test;``the test size of
FET was less than 0.035 for nearly all sample sizes before 50 and
did not approach 0.05 even for sample sizes over
100.'';conservativeness of ``exact'' methods;see \emph{Stat in Med}
\textbf{28}:173-179, 2009 for a criticism which was unanswered}
}
Lest you be concerned that Frank is selectively citing the literature,
here are a few more citations demonstrating problems with "exact
methods", starting with the citation in the prop.test help page:
http://www.stats.org.uk/statistical-inference/Newcombe1998.pdf
And a few others:
http://www.jstor.org/pss/2685469
http://www.ine.pt/revstat/pdf/rs080204.pdf
http://goliath.ecnext.com/coms2/gi_0199-4388181/Evaluation-criteria-for-discrete-confidence.html
http://www.math.ist.utl.pt/~apires/PDFs/AP_COMPSTAT02.pdf
Some R methods:
http://finzi.psych.upenn.edu/R/library/PropCIs/html/add4ci.html
http://finzi.psych.upenn.edu/R/library/pairwiseCI/html/pairwiseCImethodsProp.html
http://finzi.psych.upenn.edu/R/library/Epi/html/ci.pd.html
http://finzi.psych.upenn.edu/R/library/binMto/html/binMtoMethods.html
http://finzi.psych.upenn.edu/R/library/pairwiseCI/html/pairwiseCI.html
----- Original Message ----
From: Frank E Harrell Jr<f.harr...@vanderbilt.edu>
To: r-help@r-project.org
Sent: Thu, May 13, 2010 5:35:07 AM
Subject: Re: [R] read table for Fisher Exact
On 05/12/2010 03:31 PM, visser wrote:
i have 2 groups i want to
compare: group A and group B
each group contains let's say 20
patients
i want to perform a Fisher Exact test on genotype
distribution
so, see if there is a sign diff in genotpe
frequency/distribution (#AA, #AB,
#BB) between group A and B
not
for 1, but for 1000 different genes
my question: how should i
build my table so i can do:
test<-
read.table("table1.txt")
fisher.test(test)
i know a lot
is still missing in the syntax, but i do not know what. any
help would
be soo much appreciated!!
Note that in this case, Fisher's exact test has
a good chance of being
less accurate than an approximate Pearson chi-square
test.
--
Frank E Harrell Jr Professor and Chairman School of Medicine
Department of Biostatistics Vanderbilt
University
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
David Winsemius, MD
West Hartford, CT
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
