From: Leitch, Matthew C. Sent: Monday, January 24, 2011 6:53 PM To: 'info at network-theory.co.uk' Subject: question about the pt() calculation Hello Thank you for your time. I am a graduate student at the University of Texas Medical Branch, and I was wondering if you could help me with a R program I am writing. I have some data that is stored a file that has 1733 rows and 4 columns. Each row is independent, so I have a loop system so I don't have to manually perform 1733 t-tests. However I got most of it to work, thanks to the "An Introduction to R" book and a helpful colleague. But I am having some difficulty with getting my t-test scores converted to p-values. The pt() test won't take the t object, but I am unsure how to make it create a p-value for each row. If you have any suggestions they would be greatly appreciated. Best Wishes Matt -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: t-practice-script.txt URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110125/a4d34f3f/attachment.txt>
From: Leitch, Matthew C. Sent: Monday, January 24, 2011 6:53 PM To: 'info at network-theory.co.uk' Subject: question about the pt() calculation Hello Thank you for your time. I am a graduate student at the University of Texas Medical Branch, and I was wondering if you could help me with a R program I am writing. I have some data that is stored a file that has 1733 rows and 4 columns. Each row is independent, so I have a loop system so I don't have to manually perform 1733 t-tests. However I got most of it to work, thanks to the "An Introduction to R" book and a helpful colleague. But I am having some difficulty with getting my t-test scores converted to p-values. The pt() test won't take the t object, but I am unsure how to make it create a p-value for each row. If you have any suggestions they would be greatly appreciated. Best Wishes Matt -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: t-practice-script.txt URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110125/73762947/attachment.txt>
Matthew - Others will probably tell you about the folly of performing 1733 t-tests on groups with 4 observations each, but an alternative to your approach would be to use R to solve your problem. (I'm using var.equal=TRUE because that's what you're calculating, but you might consider using the default behaviour assuming unequal variances.) onerow = function(i){ thetest = t.test(noncancer[i,],cancer[i,],var.equal=TRUE) c(thetest$statistic,thetest$p.value,as.numeric(thetest$p.value < 0.05)) } answer = t(sapply(1:nrow(cancer),onerow)) Then answer should have the information that you want. - Phil Spector Statistical Computing Facility Department of Statistics UC Berkeley spector at stat.berkeley.edu On Tue, 25 Jan 2011, Leitch, Matthew C. wrote:> > > From: Leitch, Matthew C. > Sent: Monday, January 24, 2011 6:53 PM > To: 'info at network-theory.co.uk' > Subject: question about the pt() calculation > > Hello > > Thank you for your time. I am a graduate student at the University of Texas Medical Branch, and I was wondering if you could help me with a R program I am writing. I have some data that is stored a file that has 1733 rows and 4 columns. Each row is independent, so I have a loop system so I don't have to manually perform 1733 t-tests. However I got most of it to work, thanks to the "An Introduction to R" book and a helpful colleague. But I am having some difficulty with getting my t-test scores converted to p-values. The pt() test won't take the t object, but I am unsure how to make it create a p-value for each row. If you have any suggestions they would be greatly appreciated. > > Best Wishes > Matt >