Date:         Fri, 7 Nov 2008 10:06:25 -0500
Reply-To:     joey m <sasuser.joey.m@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         joey m <sasuser.joey.m@GMAIL.COM>
Subject:      how-to-do question
Content-Type: text/plain; charset=ISO-8859-1

Hi:

I would like to make a Shapiro-Wilk test and then, if significant, make a
transfrmation on my data. But, if Shapiro is not significant I would like to
make a homogeneity of variance test.

I can do that easyly in an interactive mode, but how can I do it in a batch
mode?

I would like to figure out a way SAS can take the probability of Shapiro-W
(out of the proc univariate output) and decide whether to go to my code to
make the transformation or go to my code for homogeneity of var.

Thanks a lot.

PS. I added the code I am playing with.

*

data* raw;

input Group A B Y @@;

datalines;

1 1 1 86 1 1 2 41

1 2 1 50 1 2 2 56

1 3 1 39 1 3 2 35

2 1 1 30 2 1 2 25

2 2 1 36 2 2 2 28

2 3 1 33 2 3 2 30

3 1 1 32 3 1 2 24

3 2 1 31 3 2 2 27

3 3 1 15 3 3 2 19

4 1 1 30 4 1 2 45

4 2 1 35 4 2 2 30

4 3 1 17 4 3 2 18

;

/* RUN ANOVA ON RAW DATA & SAVE RESIDUALS */
*

proc* *mixed* data=raw;

class group;

model Y = group / outpred=residuals;
*

run*;
*

proc* *univariate* data=residuals normal plot;

var RESID;

output out=normal normal=SW probn=probSW;
*

run*;
*

*

/* IF SHAPIRO IS SIGNIFICANT (i.e. le 0.001, NOT NORMAL),

MAKE SQ-ROOT TRANSFORMATION */
*

data* transf;

set raw;

newY=sqrt(abs(y));
*

run*;

/* ELSE

-IF SHAPIRO IS NOT SIGNIFICANT (i.e. gt 0.001, NORMAL),

TEST HOMOGENEITY ON RAW DATA */
*

proc* *glm* data=raw;

class GROUP;

model Y=GROUP;

means GROUP /hovtest;
*

run*;