Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: rmANOVA and posthoc

From   Joseph Coveney <[email protected]>
To   Statalist <[email protected]>
Subject   Re: st: rmANOVA and posthoc
Date   Fri, 27 May 2005 11:18:08 +0900

Joan Camprodon wrote:

Basic question...

I want to do an analysis using one way repeated measures ANOVA and the 
following posthoc analysis.
As my short STATA experience lets me know, I must use the command anova 
if I want to do rmANOVA. But I see no option to give me a posthoc 
Is there one? Or should I use different commands in a do file?

If the latter, should I use the "oneway var1 var2, bonferroni" or do I 
need to set a series of paired t-tests?


Yes; there is an option in Stata.  You can use the -mtest()- option of 
-test- after -anova- to get adjustments for multiple testing 
(-help anova_postestimation-).  Its use is illustrated in the do-file below.

So, you do not need to set a series of paired Student's t-tests; however, I 
recall reading on SPSS's website a statement by one of their statisticians that 
many experts recommend multiple paired Student's t-tests if you have doubts 
about the sphericity assumption.  

If you're so concerned, then as an alternative you can use -manova- and do 
multiple testing after that.  It, too, accommodates multiplicity in its 
-mtest()- option (-help manova postestimation-).  

Also, take a look at -smileplot-, a set of commands (-findit smileplot-) 
written by Roger Newson and downloadable from SSC.  It addresses multiple 
testing more comprehensively.

Joseph Coveney

drawnorm response1 response2 response3, means(0 0.25 0.5) ///
  corr(1 0.7 0.7 \ 0.7 1 0.7 \ 0.7 0.7 1) ///
  seed(`=date("2005-05-27", "ymd")') n(7) clear
generate byte subject = _n
reshape long response, i(subject) j(time)
anova response time subject, repeated(time)
test, showorder
matrix T = (0,  1, -1,  0, 0, 0, 0, 0, 0, 0, 0 \ ///
            0, -1,  0,  1, 0, 0, 0, 0, 0, 0, 0 \ ///
            0,  0,  1, -1, 0, 0, 0, 0, 0, 0, 0)
test, test(T) mtest(bonferroni)

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index