[R] Posthoc tests for ANOVA

[Frank E Harrell Jr]

To add to the strong points made by Brian Ripley, there are many choices of multiple comparison adjustment methods, and the choice is pretty much arbitrary.  
Also see 

@ARTICLE{coo96mul,
  author = {Cook, Richard J. and Farewell, Vern T.},
  year = 1996,
  title = {Multiplicity considerations in the design and analysis of clinical
          trials},
  journal = J Royal Statist Society Series A,
  volume = 159,
  pages = {93-110},
  annote = {multiplicity; multiple endpoints; multiple treatments; p-value
           adjustment; type I error; argues that if results are intended to be
           interpreted marginally, there may be no need for controlling
           experimentwise error rate}
}

Frank Harrell


On Wed, 23 Jan 2002 09:27:11 +0000 (GMT)
Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:

> On Wed, 23 Jan 2002 Jan_Svatos at eurotel.cz wrote:
> 
> > Dear List,
> >
> > are there post-hoc tests like Scheffe, LSD, etc. available after ANOVA test
> > is performed with significant F-statistic?
> > I have tried
> > help.search("Scheffe"),
> > but "No documentation found" (and I have most of packages installed).
> > Probably there are such tests in R, and I am just searching badly...
> 
> This is usually known as `multiple comparisons', and there is not
> much in R.  There are
> 
> 1) Function Tukey in package Devore5
> 2) pairwise.t.test and friends in package ctest, including p.adjust.
> 
> but as S-PLUS users keep pointing out, multicomp there has much more.
> 
> > My second question is: Which test/method I should use for ANOVA-like test
> > on 0-1 variables
> > (leading to multiple binomial variables instead of Normal)?
> 
> Not sure I understand.  Anova is about a single response.  Perhaps you
> mean logistic regression and the analysis of deviance?
> 
> Many of us would feel that testing is not important here: what is
> important is to estimate sizes of effects and perhaps to select predictive
> models.  Classical multiple comparisons deals with only a small part of
> the selection effects, in particular ignores model choice.
> 
> -- 
> Brian D. Ripley,                  ripley at stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,             Tel:  +44 1865 272861 (self)
> 1 South Parks Road,                     +44 1865 272860 (secr)
> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
> 
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
> _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._


-- 
Frank E Harrell Jr              Prof. of Biostatistics & Statistics
Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences
U. Virginia School of Medicine  http://hesweb1.med.virginia.edu/biostat
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._