[R] anova and post hoc multicomparison tests

Rogers, James A [PGRD Groton] James_A_Rogers at groton.pfizer.com
Fri Sep 24 15:23:12 CEST 2004


Your comments are a compliment to R.

Undoubtedly other software is preferable if you want to do
Student-Newman-Keuls or Fisher's "protected" LSD (ANOVA F-test followed by
unadjusted T-tests). Perhaps the reason is that neither Student-Newman-Keuls
nor Fisher's "protected" LSD is a valid multiple comparison procedure.
Student-Newman-Keuls does not even control the probability of making at
least one false assertion of inequality (which is the almost the minimum one
could ask of a multiple comparison procedure). For details, including
examples of where these methods fail, see:

Hsu, J.C. (1996). Multiple Comparisons: Theory and Methods. Chapman & Hall.

If you want to use R to perform valid multiple comparisons, such as
Dunnett's MCC or Tukey's HSD, see the function TukeyHSD and also the
multcomp package. 

Jim Rogers

> Message: 78
> Date: Fri, 24 Sep 2004 10:54:55 +0200
> From: "BLANCHER Guillaume" <G.BLANCHER at isa-lille.fr>
> Subject: [R] anova and post hoc multicomparison tests
> To: <r-help at stat.math.ethz.ch>
> Message-ID: <F2B8DADA7A384D48925C74D52D68AE702484DC at srv-exch.isa.net>
> Content-Type: text/plain;	charset="us-ascii"
> Hello everyone, 
> Like a lot of people, I have been looking for functions in R doing ANOVA
> (ok) and performing multicomparisons (like Student-Newman-Keuls, etc.).
> As I have been a little bit disappointed, I have bee looking through the
> net for such "open source" softwares. I found one in:
> http://www.statpages.org/miller/openstat/OS4.html
> I have begun to use it, and it seems good and simple to understand (as
> for a non-specialist like me).
> Sorry for R, but I prefer OpenStat4 to R for ANOVAs and post hoc tests.
> Guillaume

LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}

More information about the R-help mailing list