Yes, you catch my thought completely. A solution I used is to enumerate all combination of A, B and C to obtain all possible values of (A-B)/(A-C) and also enumerate all combination of D and E to get all possible D/E, then apply Wilcoxon-Mann-Whitney rank test to compare these two groups of values. By this method, if the lengths of A,B,..,E are 5, then we will have 5x5x5=125 values of (A-B)/(A-C) and 5x5=25 values of D/E. Such enumerated data may inflate the sample size and make the test more significant than the true. I don't know does it reasonable? -Weibo On Tue, Aug 24, 2010 at 4:41 PM, Moshe Olshansky <m_olshansky@yahoo.com>wrote: > Do you mean that A,B,...,E are same length and you want to apply t-test to > compare > (A-B)/(A-C) to D/E? > Technically this is straightforward, but even if A,B,...,E are normally > distributed, this is very much not true for (A-B)/(A-C) and D/E, so it won't > be appropriate to apply t-test (uness A,C and E are constants). > > > --- On *Tue, 24/8/10, Weibo Xie <xwbmail@gmail.com>* wrote: > > > From: Weibo Xie <xwbmail@gmail.com> > Subject: Re: [BioC] Five variables, ABCDE, how to check whether (A-B)/(A-C) > is equal to D/E? > To: "Moshe Olshansky" <m_olshansky@yahoo.com> > Cc: "Bioconductor" <bioconductor@stat.math.ethz.ch> > Received: Tuesday, 24 August, 2010, 6:09 PM > > > > Thanks for your quick reply! I think it's like 'equal means' in Two- Sample > t-Test. > > -Weibo > > On Tue, Aug 24, 2010 at 3:58 PM, Moshe Olshansky <m_olshansky@yahoo.com<http: mc="" compose?to="m_olshansky@yahoo.com"> > > wrote: > > What exactly do you mean by "equal"? > > --- On Tue, 24/8/10, Weibo Xie <xwbmail@gmail.com<http: mc="" compose?to="xwbmail@gmail.com">> > wrote: > > > From: Weibo Xie <xwbmail@gmail.com<http: mc="" compose?to="xwbmail@gmail.com"> > > > > Subject: [BioC] Five variables, ABCDE, how to check whether (A-B)/(A-C) > is equal to D/E? > > To: "Bioconductor" <bioconductor@stat.math.ethz.ch<http: mc="" compo="" se?to="bioconductor@stat.math.ethz.ch"> > > > > Received: Tuesday, 24 August, 2010, 5:45 PM > > Hello everyone, > > > > I have 5 variables, denoted as A to E, each with several > > replicates. Is > > there a reasonable statistical test to check whether > > (A-B)/(A-C) is equal to > > D/E? > > > > Thanks for your help, > > > > -Weibo > > > > [[alternative HTML version deleted]] > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor@stat.math.ethz.ch<http: mc="" compose?to="Bioconductor@s" tat.math.ethz.ch=""> > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > > > > > [[alternative HTML version deleted]]