Hello, If you determine the means of x treatments and see that one is larger than the others, can you use a sample normal to determine how statistically significant the difference is? or would contrasts by a better tool? How would one go about to do this in R? Thanks for any help (since R is new to me), Igor [[alternative HTML version deleted]]
This is a multiple comparisons problem, so no. You need to use TukeyHSD or the multcomp package. You can test contrasts via t-tests only if you select they before looking at the data. Time to consult a good book on the subject? On Wed, 5 Nov 2003, Igor Roytberg wrote:> If you determine the means of x treatments and see that one is larger > than the others, can you use a sample normal to determine how > statistically significant the difference is? or would contrasts by a > better tool? How would one go about to do this in R? > > Thanks for any help (since R is new to me),-- 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
As Prof. Ripley said, the test on contrast is valid only if you selected the contrast before seeing the data. Using the same data for both hypothesis generation and hypothesis confirmation is extremely hazardous! Andy> From: Igor Roytberg [mailto:ir1982 at u.washington.edu] > > Hello, > > If you determine the means of x treatments and see that one > is larger than the others, can you use a sample normal to > determine how statistically significant the difference is? or > would contrasts by a better tool? How would one go about to > do this in R? > > Thanks for any help (since R is new to me), > > Igor