There is a FDR test by Benjamini and Hochberg (1995), which is less
conservative than Bonferroni. More info in:
Benjamini, Y. and Hochberg, Y. (1995). Controlling the false discovery rate:
A practical and powerful approach to multiple testing. Journal of the Royal
Statistical Society, Series B, 57, 289-300
-----Original Message-----
From: Arnold Levinson [mailto:[email protected]]
Sent: Friday, June 14, 2002 10:48 AM
To: Statalist
Subject: st: Testing proportions
Colleagues,
I have two statistical questions and a Stata question.
1. One by one, I want to segregate each category of a variable and test it
the remaining observations combined --using ethnicity as an example, blacks
against non-blacks, Hispanics against non-Hispanics, etc. The description of
results would be "significantly different (or not) from all others
combined." I think the test should be adjusted for multiple comparisons (do
others agree?), but I don't think Tukey (single pairwise) or Scheffe
(simultaneous pairwise) would be appropriate. Bonferroni arguably might be
applicable but conservative. Has the test I'm contemplating been defined?
2. Now consider that the data are from a complex-sample survey. I'm not
aware of any methods for multiple comparisons that allow cluster/pweight
adjustments (I'm less concerned about strata, they typically increase
efficiency).
3. Now, if there are answers to 1 & 2, is there a way to apply a multiple
comparison adjustment to a survey command? (For that matter, is there a way
to use single post-estimation tests, e.g. -- svylc -- or -- svytest --
following svytab?)
Thanks for any help.
Arnold
Arnold H. Levinson, PhD
Assistant Professor
Department of Preventive Medicine and Biometrics
University of Colorado School of Medicine
303.777.8801
[email protected]
fax 303.239.3394
*
* 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/
