tag:blogger.com,1999:blog-6710487119650146215.post1933794817062516625..comments2017-03-27T01:51:05.632-07:00Comments on R Tutorial Series: R Tutorial Series: One-Way ANOVA with Pairwise ComparisonsJohnhttp://www.blogger.com/profile/05331039307550313006noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-6710487119650146215.post-12533729718194234692011-02-07T20:55:58.303-07:002011-02-07T20:55:58.303-07:00Hi Dave,
The tests can yield different results, b...Hi Dave,<br /><br />The tests can yield different results, because they use different statistical techniques to calculate significance. I am not qualified to make recommendations about the techniques or explain them from a statistical standpoint (only how to execute them in R and generally apply them in certain contexts). However, I will have a post coming up in a few weeks that demonstrates several methods for conducting pairwise comparisons in R, including HSD.<br /><br />I recommend talking to a professional statistician about the merits and circumstances surrounding the different tests. Also, consider that different fields have different standards and that practical application rarely meets the ideals and rigors that statisticians would like to see. Therefore, you should also consult a senior member of your field and the published research to get an idea of what others are doing.<br /><br />JohnJohn M. Quickhttp://www.blogger.com/profile/05331039307550313006noreply@blogger.comtag:blogger.com,1999:blog-6710487119650146215.post-387552227146319362011-02-07T17:17:03.159-07:002011-02-07T17:17:03.159-07:00what is the advantage of using pairwise.t.test as ...what is the advantage of using pairwise.t.test as opposed to a tukey honest significant difference test here. This came up <a href="http://stats.stackexchange.com/questions/6912/how-could-a-tukey-hsd-test-be-more-signif-then-the-uncorrected-p-value-of-t-test" rel="nofollow">here</a><br /><br />But I am unclear as to when it is appropriate to use a post-hoc integration of the interacting variables to do the t-testdavebridgeshttp://davebridges.github.com/noreply@blogger.com