An extra comment is that by suppressing the data -- ms(none)- -- you are
losing much of the potential value of the plot. Regression people who
leave the data off a scatter plot showing a regression line are usually
regarded as missing the plot (so to speak). Why are ANOVA people allowed
to get away with it?
Nick
[email protected]
-----Original Message-----
From: Nick Cox
Sent: 01 October 2009 18:31
To: '[email protected]'
Subject: RE: Foreach Loop with anovaplot
-anovaplot- is a user-written command from SJ 4-4. Please remember to
say, as requested in the FAQ, whenever you are using a user-written
command and where it comes from.
The error message indicates what the help for -anovaplot- explains:
"With the second syntax [that used here: NJC], a varlist is supplied,
which must be a permutation of the predictors previously used."
So, the short answer is: Make sure your call to -anovaplot- is legal.
I don't know, however, why you are using this second syntax at all. As
your loop is over
anova <something> exam_nb clinic
a call to -anovaplot- without specifying any variables at all would seem
to be what you want, i.e. -anovaplot-'s first syntax.
Nick
[email protected]
Diego Bellavia
I am trying to build a simple script to loop into
some ANOVA for repeated measures tests and
graph them with anovaplot.
The script is the following:
#delimit ;
foreach X of varlist age blood_p cholesterol...
{;
anova `X' exam_nb clinic, repeated(exam_nb);
anovaplot exam_nb, scatter(msym(none));
};
#delimit cr
Where of course clinic is the ID of the subject, and exam_nb the ID of
the
single observation.
When I run it as a command everything works fine but as soon as I put
this
in a .do file, I see the following message:
"exam_nb not a permutation of exam_nb clinic"
that makes sense to me but when I try to add clinic in the anovaplot
command I get some weird graph I cannot understand...
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
