Hello, I am analysing several samples whose sizes are from 9 to 110. I would like to test their distribution with R, whether they are normal or not. I wonder which test for normality from R should I use . Thank you for help. Samuel Samuel BERTRAND Doctorant Laboratoire de Biomecanique LBM - ENSAM - CNRS UMR 8005 151, bd de l'Hopital 75013 PARIS Tel. +33 (0) 1 44 24 64 53 Fax +33 (0) 1 44 24 63 66 [[alternative HTML version deleted]]
A qqplot is a good raw test to look quickly the normality of a distribution. best A.S. ---------------------------- Alessandro Semeria Models and Simulations Laboratory Montecatini Environmental Research Center (Edison Group), Via Ciro Menotti 48, 48023 Marina di Ravenna (RA), Italy Tel. +39 544 536811 Fax. +39 544 538663 E-mail: alessandro.semeria at cramont.it
shapiro.test is also relevant. -- Jonathan Baron, Professor of Psychology, University of Pennsylvania Home page: http://www.sas.upenn.edu/~baron R page: http://finzi.psych.upenn.edu/
Hello Samuel, Regardless of some more fundamental problems (see below), a test to "prove" normality based on a sample of 9? - Fugetaboutit. Knut At 10:20 2004-02-06 +0100, I wrote:>... > >It may be tempting to interpret a non-significant result of a statistical >test as to verify the hypothesis, e.g., as to verify that the distribution >of the data is Gaussian. Unfortunately, a non-significant test is merely >non-conclusive (Popper KR, Wien: 1937), so one would have to test for >equivalence, e.g., as TOST (two one-sided tests). To do this with the a >test for normality (Shapiro, Lillifors, ...), however, it may be difficult >to come up with a justification for an equivalence limit. > >...At 09:32 2004-02-17 +0100, you wrote:>Hello, > >I am analysing several samples whose sizes are from 9 to 110. >I would like to test their distribution with R, >whether they are normal or not. >I wonder which test for normality from R should I use . > >Thank you for help. > >Samuel BERTRAND >Doctorant >Laboratoire de Biomecanique >LBM - ENSAM - CNRS UMR 8005 >151, bd de l'Hopital >75013 PARIS >Tel. +33 (0) 1 44 24 64 53 >Fax. +33 (0) 1 44 24 63 66Knut M. Wittkowski, PhD,DSc ------------------------------------------ The Rockefeller University, GCRC Experimental Design and Biostatistics 1230 York Ave #121B, Box 322, NY,NY 10021 +1(212)327-7175, +1(212)327-8450 (Fax) kmw at rockefeller.edu http://www.rucares.org/clinicalresearch/dept/biometry/
Thanks to everyone for the responses! Samuel> SUMMARY :MY QUESTION :>>Hello, >> >>I am analysing several samples whose sizes are from 9 to 110. >>I would like to test their distribution with R, >>whether they are normal or not. >>I wonder which test for normality from R should I use . >> >>Thank you for help. >> >>Samuel BERTRAND >>Doctorant >>Laboratoire de Biomecanique >>LBM - ENSAM - CNRS UMR 8005 >>151, bd de l'Hopital >>75013 PARIS >>Tel. +33 (0) 1 44 24 64 53 >>Fax. +33 (0) 1 44 24 63 66THE RESPONSES : A qqplot is a good raw test to look quickly the normality of a distribution. best A.S. ---------------------------- Alessandro Semeria Models and Simulations Laboratory Montecatini Environmental Research Center (Edison Group), Via Ciro Menotti 48, 48023 Marina di Ravenna (RA), Italy Tel. +39 544 536811 Fax. +39 544 538663 E-mail: alessandro.semeria@cramont.it ==================================================================== shapiro.test is also relevant. -- Jonathan Baron, Professor of Psychology, University of Pennsylvania Home page: http://www.sas.upenn.edu/~baron R page: http://finzi.psych.upenn.edu/ ===================================================================== You can test normality with the Jarque-Bera test. You can get this test in the package tseries, of Adrian Trapletti. The package is in CRAN, and you can install it directly from within R. Hope it will help you. Agustin "RCU" <aalonso@rcumariacristina.com> ====================================================================== Hello Samuel, Regardless of some more fundamental problems (see below), a test to "prove" normality based on a sample of 9? - Fugetaboutit. Knut At 10:20 2004-02-06 +0100, I wrote: ... It may be tempting to interpret a non-significant result of a statistical test as to verify the hypothesis, e.g., as to verify that the distribution of the data is Gaussian. Unfortunately, a non-significant test is merely non-conclusive (Popper KR, Wien: 1937), so one would have to test for equivalence, e.g., as TOST (two one-sided tests). To do this with the a test for normality (Shapiro, Lillifors, ...), however, it may be difficult to come up with a justification for an equivalence limit. ... "Knut M. Wittkowski" <kmw@mail.rockefeller.edu> ====================================================================== [[alternative HTML version deleted]]