[R] bug in the ctest package: binom.test

[Kurt Hornik]

>>>>> Peter Dalgaard BSA writes:

> Mai Zhou <mai at ms.uky.edu> writes:
>> R 0630 for windows
>> 
>> > library(ctest)
>> > binom.test(7,10,p=0.3, alternative="two.sided")
>> 
>> returns a p-value of =< 2.2e-016 and a warning
>> 
>> In Splus 3.4 
>> 
>> > binom.test(7,10,p=0.3, alternative="two.sided")
>> 
>> returns a p-value of 0.0106
>> 
>> 
>> I think it is the 
>> 
>> max(v[v<=(1+eps)*PVAL])       causing the problem...
>> 
>> max() of an empty vector.......
>> 

> Exactly. And conspiring with the lack of a sanity check in format.pval
> causing it to print -Inf as < 2.2e-016...

> The fix is as simple as max(0,v[v<=(1+eps)*PVAL])
>                             ^^

I noticed this problem last week.  However, my fix is different, because
binom.test() now computes p-values using the ``sum of probs of all y not
more likely (``more extreme'') than the observed x'' principle (which is
also what fisher.test() now does consistently).

An updated version of ctest will shortly be uploaded to CRAN.

-k
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._