Hello,
I read a book about statistics in psychology. The authors use SPSS. They
talk about post hoc tests after ANOVA finds significant effects:
- Gabriel's procedure (for equal or slightly different sample sizes)
- Hochberg's GT2 (for different sample sizes)
- Games-Howell procedure (for populations with unequal variances)
I could not find them in R. Do they not exist in R or are there any
equivalents?
I know that I can use Tukey HSD if the sample sizes are equal and the
variances are homogeneous.
Thanks!
Claus
On Mon, 28 Nov 2005, Claus Atzenbeck wrote:> I read a book about statistics in psychology. The authors use SPSS. They > talk about post hoc tests after ANOVA finds significant effects: > > - Gabriel's procedure (for equal or slightly different sample sizes) > - Hochberg's GT2 (for different sample sizes) > - Games-Howell procedure (for populations with unequal variances) > > I could not find them in R. Do they not exist in R or are there any > equivalents?I saw a message today by Brian Ripley at <http://www.r-project.org/nocvs/mail/r-help/2002/0565.html> stating that there are only a few multiple comparison tests in R. How do you calculate post hoc multiple comparisons tests with R for normal distributed samples with different variances? Would you claim that I savely can use TukeyHSD even for unequal variances? Claus
On Tue, 29 Nov 2005, Claus Atzenbeck wrote:> How do you calculate post hoc multiple comparisons tests with R for > normal distributed samples with different variances?In order to make it more visible, I have created an overview that shows my decision about what test I use. It is available at <http://cs.aaue.dk/~claus/temp/TestOverview.pdf>. What can I change that I still can use R to perform my test? (At the very bottom of the overview you see note about the missing tests in R.) Thanks for any comment. Claus
What would be nice would be an R routine that automatically implements this flowchart. On 11/30/05, Claus Atzenbeck <claus.atzenbeck at freenet.de> wrote:> On Tue, 29 Nov 2005, Claus Atzenbeck wrote: > > > How do you calculate post hoc multiple comparisons tests with R for > > normal distributed samples with different variances? > > In order to make it more visible, I have created an overview that shows > my decision about what test I use. It is available at > <http://cs.aaue.dk/~claus/temp/TestOverview.pdf>. What can I change that > I still can use R to perform my test? (At the very bottom of the > overview you see note about the missing tests in R.) > > Thanks for any comment. > Claus > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html >
>Gabor Grothendieck <ggrothendieck at gmail.com> writes: > >> What would be nice would be an R routine that automatically >> implements this flowchart. > >I'd recommend learning about p.adjust and the multcomp package >*instead* of following the flowchart.Another approach to correcting for multiple testing is implemented in the pacakge qvalue by Storey (see CRAN and also http://faculty.washington.edu/~jstorey/qvalue/). Storey JD. (2002) A direct approach to false discovery rates. Journal of the Royal Statistical Society, Series B, 64: 479-498 Storey JD. (2003) The positive false discovery rate: A Bayesian interpretation and the q-value. Annals of Statistics, 31: 2013-2035 Storey JD, Taylor JE, and Siegmund D. (2004) Strong control, conservative point estimation, and simultaneous conservative consistency of false discovery rates: A unified approach. Journal of the Royal Statistical Society, Series B, 66: 187-205 (The references can be accesseed from the web-site) HTH's Neil -- "The surest way to make a monkey of a man is to quote him." - Robert Benchley