Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: RE: Stata has disappointing support of contrasts/multiple comparisons in mixed ANOVA


From   "Feiveson, Alan H. (JSC-SK311)" <alan.h.feiveson@nasa.gov>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: Stata has disappointing support of contrasts/multiple comparisons in mixed ANOVA
Date   Tue, 11 Mar 2008 11:11:17 -0500

For starters, -anova- should allow a simple command option that would
automatically

(1)evaluate each contibuting basic contrast for at least main effects
and two-way interactions (simple differences for main effects, all
combinations of 2nd differences for two-way interactions) and give
standard errors using an appropriate variance estimate obtained from
fitting the model, and

(2)perform an adjustment for multiple comparisons with a choice of
techniques

As an approximation for more general contrast testing, the "mtest"
option after test should at least have the option for the user to
specify an error term in the model to use. 

Al Feiveson


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: Tuesday, March 11, 2008 9:52 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: Stata has disappointing support of contrasts/multiple
comparisons in mixed ANOVA

Disappointment and embarrassment being noted, what would be your top
candidates for inclusion? 
(And will others then agree, showing a consensus among people who use
this kind of thing on what is justifiable and useful?)  

StataCorp has a certain historical timidity about implementing stuff
when the practitioners of a field seemingly can not even decide on
procedures and every expert contributes a different approach. Rightly or
wrongly, multiple comparisons has that reputation. 

Nick
n.j.cox@durham.ac.uk 

Feiveson, Alan H. (JSC-SK311)

I have been repeatedly asked by people who I convinced to buy Stata how
to do multiple comparisons of interaction effects under more
"complicated" anova models such as those with at least one factor random
and possibly with repeated measures, etc. Apparently Stata does not have
a built-in command/option for multiple comparisons for anything but a
one-way model and does not even support approximate methods for such
comparisons other than using xtmixed and creating all the contrasts and
tests "by hand". Even then, further processing has to be undertaken to
account for multiple testing.

Embarassingly enough, I have had to refer people to SAS to do these
types of comparisons without a lot of customized programing. Am I
missing something or is Stata really woefully deficient in this area?


*
*   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/

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index