[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: kwallis2 Output organization

From   "Nick Cox" <>
To   <>
Subject   RE: st: kwallis2 Output organization
Date   Sun, 15 Mar 2009 16:39:38 -0000

-kwallis2- is a user-written program by Herve Caci on SSC dating from

(The Statalist FAQ carries a permanent request that you make clear where
user-written stuff you refer to comes from.) 

As Roy Wada points out, quite what is wanted here is indicated only

But David and Martin have correctly identified the main issue:
-kwallis2- does not leave enough stuff in memory to make this kind of
thing possible without programming. If this kind of program were being
written today by an experienced Stata programmer, it would made be
r-class and more results would be left in memory. 

David is not quite right: -kwallis2- defines three globals via 

mac def S_1 `KW'
mac def S_2 `maxGrp'-1
mac def S_3 `KW'/`gTied'

but this presumably is not going to be enough to do what is wanted here.
-mac def- is an old way of saying -global-, in this case.) 

Martin makes a slip here: when he says compiler, he means interpreter.
Only Mata has a compiler. But, as pointed out often and often, "setting"
-version- does not itself rule out using later stuff. 


Martin Weiss

Seems you are out of luck on this one. As David says, it does not leave
behind its results, so no postestimation tool can pick up any of them. I
remember Nick saying that -ereturn post- was not around in version 5 -
which is the version to which kwallis2 sets the compiler. 
Your only option seems to be to -log- the output and copy from the log
file. Or - if you are so adventurous - to redo kwallis2 to post its

David Airey 

For that to be possible without additional programming, kwallis2 would
have to leave behind stuff in return list, a matrix, etc. I don't think
it does this.

type -viewsource kwallis2.ado- to see...

Narasimhan Sowmyanarayanan

> I am using the kwallis2 procedure to do Multiple comparisons between
> groups. I am comparing some 27 groups. I was wondering if there was a
> way to get the output of this procedure in a tabular format and if
> someone knew a trick for that ? I wanted to learn if there was a
> smarter way to do it in STATA than do manual work. I guess I was
> looking at some way analogous to getting table outputs for regression
> ?

*   For searches and help try:

© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index