[Rd] (PR#7225) propogation of rounding error for t.tests with unequal sample sizes (PR#7225)

Tucker-Kellogg, Gregory Gregory.Tucker-Kellogg at mpi.com
Mon Sep 13 22:49:16 CEST 2004


Thanks, I'm sure our users will welcome it.  

This is also the fastest bug response I've ever seen on any software,
and much appreciated.

GTK

>-----Original Message-----
>From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk] 
>Sent: Monday, September 13, 2004 4:14 PM
>To: Tucker-Kellogg, Gregory
>Cc: R-bugs at biostat.ku.dk; r-devel at stat.math.ethz.ch
>Subject: RE: [Rd] (PR#7225) propogation of rounding error for 
>t.tests with unequal sample sizes (PR#7225)
>
>
>I am seeing NaNs, which is more sensible.
>
>We will add a test for stderr small compared to the larger group mean.
>
>On Mon, 13 Sep 2004, Tucker-Kellogg, Gregory wrote:
>
>> 
>> My apologies for not including the transcript in the 
>original message:
>> 
>> > xx <- rep(2.10,80)
>> > groups.balanced <- as.factor(c(rep("one",40),rep("two",40)))
>> > groups.unbalanced <- as.factor(c(rep("one",44),rep("two",36)))
>> > t.test(xx ~ groups.balanced)
>> 
>> 	Welch Two Sample t-test
>> 
>> data:  xx by groups.balanced
>> t = 0, df = 78, p-value = 1
>> alternative hypothesis: true difference in means is not equal to 0 
>> 95 percent confidence interval:
>>  -4.004247e-16  4.004247e-16 
>> sample estimates:
>> mean in group one mean in group two 
>>               2.1               2.1 
>> 
>> > t.test(xx ~ groups.unbalanced)
>> 
>> 	Welch Two Sample t-test
>> 
>> data:  xx by groups.unbalanced
>> t = -4.1007, df = 54.298, p-value = 0.0001388
>> alternative hypothesis: true difference in means is not equal to 0 
>> 95 percent confidence interval:
>>  -1.322366e-15 -4.539911e-16 
>> sample estimates:
>> mean in group one mean in group two 
>>               2.1               2.1 
>> 
>> 
>> 
>> >-----Original Message-----
>> >From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
>> >Sent: Monday, September 13, 2004 4:04 PM
>> >To: Tucker-Kellogg, Gregory
>> >Cc: r-devel at stat.math.ethz.ch; R-bugs at biostat.ku.dk
>> >Subject: Re: [Rd] (PR#7225) propogation of rounding error for 
>> >t.tests with unequal sample sizes (PR#7225)
>> >
>> >
>> >What are you seeing here?  I get identical results in the two cases.
>> >
>> >On Mon, 13 Sep 2004 tucker at mpi.com wrote:
>> >
>> >> 
>> >> # users would benefit from a warning about the behavior in the #
>> >> groups.unbalanced case below.  A propogation of difference is the 
>> >> variance # calculation is leading to an apparently significant 
>> >> difference in # means, even though the numeric values are all 
>> >> identical.  Obviously upon # inspection the difference in 
>means is 
>> >> meaningless, but users might be warned # if all of the 
>input data is 
>> >> identical
>> >
>> >That's not so easy to achieve, given rounding error, but I am
>> >seeing 0 variance in both cases.
>> >
>> >> xx <- rep(2.10,80)
>> >> 
>> >> groups.balanced <- as.factor(c(rep("one",40),rep("two",40)))
>> >> groups.unbalanced <- as.factor(c(rep("one",44),rep("two",36)))
>> >> 
>> >> t.test(xx ~ groups.balanced)
>> >> 
>> >> t.test(xx ~ groups.unbalanced)
>> >> 
>> >> 
>> >> 
>> >> <<insert bug report here>>
>> >> 
>> >> 
>> >> 
>> >> --please do not edit the information below--
>> >> 
>> >> Version:
>> >>  platform = sparc-sun-solaris2.8
>> >>  arch = sparc
>> >>  os = solaris2.8
>> >>  system = sparc, solaris2.8
>> >>  status =
>> >>  major = 1
>> >>  minor = 9.0
>> >>  year = 2004
>> >>  month = 04
>> >>  day = 12
>> >>  language = R
>> >> 
>> >> Search Path:
>> >>  .GlobalEnv, package:methods, package:stats, package:graphics,
>> >> package:utils, Autoloads, package:base
>> >> 
>> >> 
>> >> 
>> >> This e-mail, including any attachments, is a confidential business
>> >> communication, and may contain information that is confidential, 
>> >> proprietary and/or privileged.  This e-mail is intended 
>only for the 
>> >> individual(s) to whom it is addressed, and may not be 
>saved, copied, 
>> >> printed, disclosed or used by anyone else.  If you are 
>not the(an) 
>> >> intended recipient, please immediately delete this e-mail 
>from your 
>> >> computer system and notify the sender.  Thank you.
>> >> 
>> >> ______________________________________________
>> >> R-devel at stat.math.ethz.ch mailing list
>> >> https://stat.ethz.ch/mailman/listinfo/r-devel
>> >> 
>> >> 
>> >
>> >-- 
>> >Brian D. Ripley,                  ripley at stats.ox.ac.uk
>> >Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
>> >University of Oxford,             Tel:  +44 1865 272861 (self)
>> >1 South Parks Road,                     +44 1865 272866 (PA)
>> >Oxford OX1 3TG, UK                Fax:  +44 1865 272595
>> >
>> >
>> 
>> 
>> 
>> This e-mail, including any attachments, is a confidential business 
>> communication, and may contain information that is confidential, 
>> proprietary and/or privileged.  This e-mail is intended only for the 
>> individual(s) to whom it is addressed, and may not be saved, copied, 
>> printed, disclosed or used by anyone else.  If you are not the(an) 
>> intended recipient, please immediately delete this e-mail from your 
>> computer system and notify the sender.  Thank you.
>> 
>> ______________________________________________
>> R-devel at stat.math.ethz.ch mailing list 
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>> 
>> 
>
>-- 
>Brian D. Ripley,                  ripley at stats.ox.ac.uk
>Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
>University of Oxford,             Tel:  +44 1865 272861 (self)
>1 South Parks Road,                     +44 1865 272866 (PA)
>Oxford OX1 3TG, UK                Fax:  +44 1865 272595
>
>



This e-mail, including any attachments, is a confidential business communication, and may contain information that is confidential, proprietary and/or privileged.  This e-mail is intended only for the individual(s) to whom it is addressed, and may not be saved, copied, printed, disclosed or used by anyone else.  If you are not the(an) intended recipient, please immediately delete this e-mail from your computer system and notify the sender.  Thank you.



More information about the R-devel mailing list