data scores;
input Gender $ Score @@;
datalines;
f 75 f 76 f 80 f 77 f 80 f 77 f 73
m 82 m 80 m 85 m 85 m 78 m 87 m 82
;Output Data from SAS Procedures
Many SAS procedures make important statistics available as output data sets, either with an OUTPUT statement, or with an option on the PROC statement. With ODS commands (Output Delivery System), every table produced by every procedure can be returned as a data set.
(This same group of ODS commands also makes it very simple to save your output in many document formats and also to create many statistical graphs with little effort.)
Creating Data Sets with ODS
ODS divides procedure output into components called output objects. Look at the following default output from the TTEST procedure:
proc ttest data=scores;
class Gender;
var Score;
run; The TTEST Procedure
Variable: Score
Gender Method N Mean Std Dev Std Err
f 7 76.8571 2.5448 0.9619
m 7 82.7143 3.1472 1.1895
Diff (1-2) Pooled -5.8571 2.8619 1.5298
Diff (1-2) Satterthwaite -5.8571 1.5298
Gender Method Minimum Maximum
f 73.0000 80.0000
m 78.0000 87.0000
Diff (1-2) Pooled
Diff (1-2) Satterthwaite
Gender Method Mean 95% CL Mean Std Dev
f 76.8571 74.5036 79.2107 2.5448
m 82.7143 79.8036 85.6249 3.1472
Diff (1-2) Pooled -5.8571 -9.1902 -2.5241 2.8619
Diff (1-2) Satterthwaite -5.8571 -9.2064 -2.5078
Gender Method 95% CL Std Dev
f 1.6399 5.6039
m 2.0280 6.9303
Diff (1-2) Pooled 2.0522 4.7242
Diff (1-2) Satterthwaite
Method Variances DF t Value Pr > |t|
Pooled Equal 12 -3.83 0.0024
Satterthwaite Unequal 11.496 -3.83 0.0026
Equality of Variances
Method Num DF Den DF F Value Pr > F
Folded F 6 6 1.53 0.6189SAS produces four output tables:
- “Statistics” (wrapped as multiple lines)
- “Confidence Limits” (also wrapped)
- “T-Tests”
- and “Equality of Variances”
Any table can be saved as a SAS data set once you know the ODS name for the table. But as you can see in the preceding output, these are not always obvious.
Determining the Names of Output Tables
Use the ODS TRACE ON and ODS TRACE OFF statements to determine the names of procedure tables (you can also find them in the SAS Help documentation for each procedure, under “Details”). These give you the table names in the Log:
ods trace on;
proc ttest data=scores;
class Gender;
var Score;
run;
ods trace off;1 The SAS System
Output Added:
-------------
Name: Statistics
Label: Statistics
Template: Stat.TTest.Statistics
Path: Ttest.Score.Statistics
-------------
Output Added:
-------------
Name: ConfLimits
Label: Confidence Limits
Template: Stat.TTest.ConfLimits
Path: Ttest.Score.ConfLimits
-------------
Output Added:
-------------
Name: TTests
Label: T-Tests
Template: Stat.TTest.TTests
Path: Ttest.Score.TTests
-------------
Output Added:
-------------
Name: Equality
Label: Equality of Variances
Template: Stat.TTest.Equality
Path: Ttest.Score.Equality
-------------
NOTE: The PROCEDURE TTEST printed page 1.
What we need are the “Names.”
Redirecting the Output
Once you know the name of the table you want to make a SAS data set from, use an ODS OUTPUT statement to save the results.
The basic specification is
ODS OUTPUT ods-table = dataset;For example, to save the “T-tests” table as a data set:
ods output ttests=tstats;
proc ttest data=scores;
class Gender;
var Score;
run;Below is a listing of the six variables and two observations that got written to the TSTATS data set:
proc print data=tstats;
run; Obs Variable Method Variances tValue DF Probt
1 Score Pooled Equal -3.83 12 0.0024
2 Score Satterthwaite Unequal -3.83 11.496 0.0026Creating Data Sets with PROC Options
Many SAS statistical PROCs also provide options to save values calculated by the procedure into data sets. This is typically done with an option on the main PROC statement, or by including an OUTPUT statement as part of the PROC.
These are not options with PROC TTEST, but they are both options with PROC GLM.
Additional Stats with a PROC Option
If we fit our t-test as a general linear model, we can save the fit statistics by adding the OUTSTAT=dataset option.
This saves the F-test statistics. (Notice that F is the square of the t value above, and the the probability is the same as the classic “pooled” results.)
proc glm data=scores noprint outstat=glmfit;
class Gender;
model score = Gender / ss3;
run;
proc print data=glmfit;
run; Obs _NAME_ _SOURCE_ _TYPE_ DF SS F PROB
1 Score ERROR ERROR 12 98.286 . .
2 Score Gender SS3 1 120.071 14.6599 .002400695OUTPUT Statement
Adding an OUTPUT statement to PROC GLM provides a method of saving fitted values and diagnostic measures.
proc glm data=scores noprint;
class Gender;
model score = Gender / ss3;
output out=fitted predicted=scorehat residual=residual;
run;
proc print data=fitted;
run; Obs Gender Score scorehat residual
1 f 75 76.8571 -1.85714
2 f 76 76.8571 -0.85714
3 f 80 76.8571 3.14286
4 f 77 76.8571 0.14286
5 f 80 76.8571 3.14286
6 f 77 76.8571 0.14286
7 f 73 76.8571 -3.85714
8 m 82 82.7143 -0.71429
9 m 80 82.7143 -2.71429
10 m 85 82.7143 2.28571
11 m 85 82.7143 2.28571
12 m 78 82.7143 -4.71429
13 m 87 82.7143 4.28571
14 m 82 82.7143 -0.71429Last Revised: 7/11/2024