Whenever I read about these tests, I am reminded of an
old joke: "Standards are great. There are so many
to choose from."
The effort that goes into choosing and implementing
these tests would often be better directed at producing
a informative graphical display. I do understand that
some people are limited by bosses and/or reviewers who
insist on them and/or a P-value culture.
Given the maze of ideas and procedures, and the
uncomfortable feeling that in practice people often
choose a test on rather unsatisfactory grounds,
any golden thread to get you out of the maze is
welcome. One thread more attractive than most has
already been discussed in the Stata Journal:
SJ-3-2 st0035 . . . . . . . . . . . Multiple-test procedures and smile plots
(help smileplot if installed) . . . R. Newson & The ALSPAC Study Team
Q2/03 SJ 3(2):109--132
implements multiple test procedures to control either
the family-wise error rate or the false discovery rate;
graphical tool for summarizing a set of multiple analyses
Nick
[email protected]
David Airey
> In general, with ANOVA post-hoc testing, Stata does not provide
> "canned" solutions, other than the few available in "oneway". Enough
> information is left behind by "anova" for the statistician to
> program
> these tests, however. If generalized to more complex models than
> "oneway", it might make a good command and Stata Journal article!
>
> See online help and manuals for "test", "lincom", and "anova
> postestimation" if you have Stata 9. Also, see "ereturn list", after
> an anova.
>
> Point your statistician to "http://www.stata.com/support/faqs/stat/
> test1.html" for a head start!
>
> Anyone have the reference to the "Waller-Duncan" test? Is this the
> multiple range test?
>
> Duncan D B. Multiple range and multiple F tests. Biometrics 11:1-42,
> 1955.
>
> http://en.wikipedia.org/wiki/Duncan's_new_multiple_range_test
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
