Dear list, If I know the standard error for k1 and k2, is there anything I can call in R to calculate the standard error of k1/k2? Thanks. Jun [[alternative HTML version deleted]]
On Mar 2, 2012, at 4:47 PM, Jun Shen wrote:> Dear list, > > If I know the standard error for k1 and k2, is there anything I can > call in > R to calculate the standard error of k1/k2? Thanks.This does not appear to be a well-posed question yet, and it is arguably more a statistics question than a coding question. Perhaps if you posted it at: http://stats.stackexchange.com/ ..... with more detail about the background for this question (especially _how_ you know these bits of information, but also what might be the purpose of this effort) you might get a more complete answer. -- David Winsemius, MD West Hartford, CT
On 12-03-02 4:47 PM, Jun Shen wrote:> Dear list, > > If I know the standard error for k1 and k2, is there anything I can call in > R to calculate the standard error of k1/k2? Thanks.No, because it depends on the joint distribution of k1 and k2. Even if you knew they were independent, that would not be sufficient (though you could use the delta method to get an approximation in that case; look it up). Duncan Murdoch> > Jun > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at 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.
On Mar 2, 2012, at 7:05 PM, Duncan Murdoch wrote:> On 12-03-02 4:47 PM, Jun Shen wrote: >> Dear list, >> >> If I know the standard error for k1 and k2, is there anything I can >> call in >> R to calculate the standard error of k1/k2? Thanks. > > No, because it depends on the joint distribution of k1 and k2. Even > if you knew they were independent, that would not be sufficient > (though you could use the delta method to get an approximation in > that case; look it up).A nice article with useful information on three approaches to this problem appeared in BMC Medical Research Methodoogy: "Methods for confidence interval estimation of a ratio parameter with application to location quotients", by Beyene and Moineddin. http://www.biomedcentral.com/1471-2288/5/32 Both were at Department of Public Health Science, University of Toronto, Toronto, Ontario, Canada, when this appeared in 2005. I thought they might have been neighbors of yours, Duncan, but I looked at a map and see that my understanding of Ontario geography is not particularly accurate.> > Duncan Murdoch > >> >> Jun-- David Winsemius, MD West Hartford, CT
On 03/03/12 13:35, David Winsemius wrote:> > On Mar 2, 2012, at 7:05 PM, Duncan Murdoch wrote: > >> On 12-03-02 4:47 PM, Jun Shen wrote: >>> Dear list, >>> >>> If I know the standard error for k1 and k2, is there anything I can >>> call in >>> R to calculate the standard error of k1/k2? Thanks. >> >> No, because it depends on the joint distribution of k1 and k2. Even >> if you knew they were independent, that would not be sufficient >> (though you could use the delta method to get an approximation in >> that case; look it up). > > A nice article with useful information on three approaches to this > problem appeared in BMC Medical Research Methodoogy: > > "Methods for confidence interval estimation of a ratio parameter with > application to location quotients", by Beyene and Moineddin. > http://www.biomedcentral.com/1471-2288/5/32<SNIP> Chapter 6 of "Beyond Anova" by Rupert G. Miller, Chapman & Hall, 1997, might also be of interest. cheers, Rolf Turner
deltamethod function in package msm may help (but bear in mind the warnings/admonitions/recommendations of other helpers) HTH Rubén -- Rubén H. Roa-Ureta, Ph. D. AZTI Tecnalia, Txatxarramendi Ugartea z/g, Sukarrieta, Bizkaia, SPAIN -----Original Message----- From: r-help-bounces@r-project.org on behalf of Jun Shen Sent: Fri 3/2/2012 10:47 PM To: R-help Subject: [R] Calculation of standard error for a function Dear list, If I know the standard error for k1 and k2, is there anything I can call in R to calculate the standard error of k1/k2? Thanks. Jun [[alternative HTML version deleted]] ______________________________________________ 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. [[alternative HTML version deleted]]