Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Nick Cox <n.j.cox@durham.ac.uk> |
To | "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu> |
Subject | st: RE: mean comparison |
Date | Wed, 17 Nov 2010 13:07:56 +0000 |
What are your grounds for saying that one variable is normally distributed and the other not? An optimistic take on this question is that the t-test is not as sensitive to non-normality as many appear to believe. If you run a t-test _and_ e.g. Mann-Whitney you may find similar P-values. But the latter is not a test of whether means are different. I don't know what non-parametric tests for comparing means you have in mind. A pessimistic take is that if the distributions are very different, reducing comparison to a test on means may be an over-simplification, or even quite the wrong question. Getting very different P-values could be consistent with that. Plotting a graph is usually called for too. Nick n.j.cox@durham.ac.uk dk I have data for two variables for one sample, want to compare means of them, i can use t test or non-parametric test, but the problem is that my one variable is normally distributed and other is not, hence which test I should use, * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/