Date: Mon, 21 Dec 1998 17:30:29 -0000
Reply-To: David Thompson <dtas@netcomuk.co.uk>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: David Thompson <dtas@NETCOMUK.CO.UK>
Subject: Re: P-values from Proc GLM
Content-Type: text/plain; charset="us-ascii"
You can get p-values from PROC GLM in version 6 using PROC GLM OUTSTAT=dset;
David Thompson
Consultant Statistician
David Thompson Applied Statistics Tel +44 (0) 1703 620808
105 Leigh Road Fax +44 (0) 1703 620807
Eastleigh e-mail dtas@netcomuk.co.uk
Hampshire
SO50 9DR http://www.netcomuk.co.uk/~dtas/index.htm
United Kingdom
-----Original Message-----
From: John Iwaniszek <jiwanisz@QCHN.QUINTILES.COM>
Newsgroups: bit.listserv.sas-l
To: SAS-L@VM.MARIST.EDU <SAS-L@VM.MARIST.EDU>
Date: 21 December 1998 12:56
Subject: Re: P-values from Proc GLM
>The best way (before version 7) to get p-values, etc. is from proc mixed
>using the make statement.
>
>john
>
>
>
>
>
>
>Jules Bosch <jxb@BELLATLANTIC.NET> on 12/19/98 08:52:44 AM
>
>Please respond to Jules Bosch <jxb@BELLATLANTIC.NET>
>
>To: SAS-L@VTVM1
>cc: (bcc: John Iwaniszek/QCHN/Quintiles)
>Subject: Re: P-values from Proc GLM
>
>
>
>
>Boyd Newlin wrote:
>>
>> Hello,
>>
>> It is it possible to output the p-values from proc GLM into a dataset?
>> If so what is the code?
>>
>
>=======================
>
>I believe PROC GLM will produce a SAS data set but it may not
>contain the p-values. But, Output from the procedure does
>contain the P-values. One approach to getting the P-values is
>the following which I developed a few years back, with PROC GLM
>help from a statistician. It was developed in the UNIX
>environment using SAS 6.11, I think.
>
>Simply stated, PROC GLM output is sent to an external file
>(LEVENE) and then read into a SAS data set using the INFILE
>statement in the data step following the PROC GLM.
>
> /* THIS OPTIONS STATEMENT IS IMPORTANT */
> options nocenter;
>
> /* REDIRECT PROC GLM OUTPUT TO AN EXTERNAL FILE */
> filename levene '$HOME/levene';
> proc printto print=levene new;
> run;
>
> /* RUN ANOVA USING LEVENE'S TEST FOR HOMOGENEITY OF VARIANCE */
> title1 'ANOVA Running Levene''s Test';
> proc glm data=work.levene;
> by phaseday;
> class groupid;
> model rbwt=groupid /ss3;
> run;
>
> /* REDIRECT OUTPUT BACK TO DEFAULT */
> proc printto;
> run;
>
> /* READ EXTERNAL GLM OUTPUT FILE "LEVENE" TO GATHER KEY INFO */
> data work.levene2
> (keep=phaseday df1 f_value p_value df2 mark animalno
> source groupid);
> infile levene missover pad;
> length animalno $20 source $8;
> animalno='ttt';
> source='levene2';
> groupid=999;
> input word1 $12. @;
> retain phaseday df1 f_value p_value mark;
> if word1=:'PHASEDAY=' then
> phaseday=input(scan(word1,2,'='),3.0);
> else if (word1='Model') then
> do;
> input df1 ssq msq f_value p_value; /* <--- P-values */
> if (p_value>0.01) then
> mark='NS';
> else
> mark=' S';
> end;
> else if (word1='Error') then
> do;
> input df2;
> output;
> end;
> run;
>
>The above "data work.levene2" step reads the output lines from
>the PROC GLM run.
>
>One of the more important aspects of all this is which option
>(CENTER/NOCENTER) is in effect. The above was written with the
>NOCENTER option in effect for SAS Output. It was part of an AF
>application. After extensive testing and some time in production
>the code began to fail (and so did my breathing) when some users
>had the CENTER option in effect as they opened the app. The
>problem was resolved, of course, with the above "options nocenter"
>statement.
>
>HTH.
>
>Jules Bosch
>
| 
