34607 - How can I compare means in PROC SURVEYMEANS?

Usage Note 34607: How can I compare means in PROC SURVEYMEANS?

PROC SURVEYMEANS does a univariate analysis, and so it does not provide any tests for comparing means of different variables or domains. You can use PROC SURVEYREG to compare means instead.

Example

The following DATA step creates a data set named SAMPLE, representing a stratified and clustered sample design. Measurements are obtained on three variables, X1, X2 and X3, and there is a domain variable D with two levels.

   data sample;
   w=1;
   seed=273497;
   do str=1 to 2;
     do clus=1 to 3;
       do i=1 to 5;
         d=ceil(2*ranuni(seed));
         X1=rannor(seed);
         X2=rannor(seed);
         X3=rannor(seed);
         output;
       end;
     end;
   end;
   drop seed i;
   run;
   proc print data=sample(obs=10);
   run;

Here is the partial printout of the data set

Obs	w	str	clus	d	X1	X2	X3
1	1	1	1	2	-0.01593	-0.73217	0.57644
2	1	1	1	1	0.01562	1.86961	0.13164
3	1	1	1	1	-0.91482	-0.05238	-1.37444
4	1	1	1	2	-0.28570	-0.26941	1.24729
5	1	1	1	1	-0.81229	0.78182	-0.55051
6	1	1	2	2	0.36365	-0.89002	-1.23039
7	1	1	2	1	0.73405	-0.89658	0.15576
8	1	1	2	1	-0.78285	-0.19153	0.64189
9	1	1	2	1	-0.55411	0.74675	0.01340
10	1	1	2	2	0.15389	-0.69313	0.68290

PROC SURVEYMEANS can compute the overall and domain means and their standard errors as follows.

   proc surveymeans data=sample;
     domain d;
     cluster clus;
     strata str;
     weight w;
     var X1 X2 X3;
   run;

Statistics
Variable	N	Mean	Std Error of Mean	95% CL for Mean
X1	30	-0.234304	0.298386	-1.0627556	0.59414701
X2	30	-0.104348	0.277145	-0.8738267	0.66513116
X3	30	0.157001	0.162645	-0.2945731	0.60857487

Domain Analysis: d
d	Variable	N	Mean	Std Error of Mean	95% CL for Mean
1	X1	14	-0.417031	0.438644	-1.6349016	0.80083945
	X2	14	0.177790	0.330187	-0.7389553	1.09453485
	X3	14	0.056237	0.198463	-0.4947848	0.60725972
2	X1	16	-0.074418	0.293380	-0.8889709	0.74013411
	X2	16	-0.351218	0.225153	-0.9763423	0.27390598
	X3	16	0.245169	0.290883	-0.5624521	1.05278986

In order to use PROC SURVEYREG for variable means comparisons, first reshape the data so that it is arranged for an ANOVA. That is, so that all the measurements are in a single variable and there is a class indicator variable. This DATA step creates the analysis variable MEASUREMENT containing the X1, X2 and X3 values and the class variable GROUP.

   data sample2;
     set sample;
     array num (*) x1 x2 x3;
     length group $ 8;
     do i=1 to dim(num);
       measurement=num[i];
       call vname(num[i], group);
       output;
     end;
     drop x1 x2 x3 i;
   run;
   
   proc print data=sample2(obs=10);
   run;

Here is the partial printout of the reshaped data.

Obs	w	str	clus	d	group	measurement
1	1	1	1	2	X1	-0.01593
2	1	1	1	2	X2	-0.73217
3	1	1	1	2	X3	0.57644
4	1	1	1	1	X1	0.01562
5	1	1	1	1	X2	1.86961
6	1	1	1	1	X3	0.13164
7	1	1	1	1	X1	-0.91482
8	1	1	1	1	X2	-0.05238
9	1	1	1	1	X3	-1.37444
10	1	1	1	2	X1	-0.28570

To set up an equivalent analysis in PROC SURVEYREG, specify the same design and weight statements that you had in PROC SURVEYMEANS. The model response variable is MEASUREMENT, the set of all analysis variable values from the reshaped data, and the independent class variable is GROUP, the ID variable also created in the reshaped data. It is convenient to specify a no-intercept model with the NOINT option, so that the parameter estimates are equal to the analysis variable means. The following CONTRAST statement in PROC SURVEYREG compares the means of X1, X2 and X3.

   proc surveyreg data=sample2;
     cluster clus;
     strata str;
     weight w;
     class group;
     model measurement=group/noint solution vadjust=none;
     contrast 'Ho: X1=X2=X3' group 1 -1 0, 
                             group 1 0 -1;
   run;

Regression Analysis for Dependent Variable measurement

Estimated Regression Coefficients
Parameter	Estimate	Standard Error	t Value	Pr > \|t\|
group X1	-0.2343043	0.29838563	-0.79	0.4762
group X2	-0.1043478	0.27714538	-0.38	0.7257
group X3	0.1570009	0.16264467	0.97	0.3891

Note:

The denominator degrees of freedom for the t tests is 4.

Analysis of Contrasts
Contrast	Num DF	F Value	Pr > F
Ho: X1=X2=X3	2	0.55	0.6144

Note:

The denominator degrees of freedom for the F tests is 4.

Note that the VADJUST= option is specified. VADJUST=DF by default in PROC SURVEYREG, which uses the degrees of freedom adjustment factor (n-1)/(n-p) in the G matrix. This results in different variance estimates in PROC SURVEYREG than those obtained in PROC SURVEYMEANS. By using VADJUST=NONE, the same variance estimates can be obtained in PROC SURVEYREG. The VADJUST= option was added in SAS 9.1. Prior to SAS 9.1, you would have to use the %SMSUB macro to compute contrasts for variables or domain levels.

You can compare the domain means using the original data set. This code compares the mean of X1 between the two domains.

   ods select parameterestimates contrasts; 
   proc surveyreg data=sample;
     cluster clus;
     strata str;
     weight w;
     class d;
     model X1=d/noint solution vadjust=none;
     contrast 'X1: D1 vs D2' d 1 -1;
   run;

Regression Analysis for Dependent Variable X1

Estimated Regression Coefficients
Parameter	Estimate	Standard Error	t Value	Pr > \|t\|
d 1	-0.4170311	0.43864384	-0.95	0.3956
d 2	-0.0744184	0.29337965	-0.25	0.8123

Note:

The denominator degrees of freedom for the t tests is 4.

Analysis of Contrasts
Contrast	Num DF	F Value	Pr > F
X1: D1 vs D2	1	0.59	0.4864

Note:

The denominator degrees of freedom for the F tests is 4.

Additional help on writing contrasts is presented in Examples of Writing CONTRAST and ESTIMATE Statements.

Operating System and Release Information

Product Family	Product	System	SAS Release
Product Family	Product	System	Reported	Fixed*
SAS System	SAS/STAT	64-bit Enabled Solaris
		64-bit Enabled HP-UX
		64-bit Enabled AIX
		Windows Millennium Edition (Me)
		Windows Vista
		Microsoft Windows XP Professional
		Microsoft Windows Server 2003 Standard Edition
		Microsoft Windows Server 2003 Enterprise Edition
		Microsoft Windows Server 2003 Datacenter Edition
		Microsoft Windows NT Workstation
		Microsoft Windows 2000 Professional
		Microsoft Windows 2000 Server
		Microsoft Windows 2000 Datacenter Server
		Microsoft Windows 2000 Advanced Server
		OS/2
		Microsoft Windows 95/98
		Microsoft® Windows® for x64
		Microsoft Windows XP 64-bit Edition
		Microsoft Windows Server 2003 Enterprise 64-bit Edition
		Microsoft Windows Server 2003 Datacenter 64-bit Edition
		Microsoft® Windows® for 64-Bit Itanium-based Systems
		OpenVMS VAX
		z/OS
		ABI+ for Intel Architecture
		AIX
		HP-UX
		HP-UX IPF
		IRIX
		Linux
		Linux for x64
		Linux on Itanium
		OpenVMS Alpha
		OpenVMS on HP Integrity
		Solaris
		Solaris for x64
		Tru64 UNIX

* For software releases that are not yet generally available, the Fixed Release is the software release in which the problem is planned to be fixed.

Type:	Usage Note
Priority:
Topic:	Analytics ==> Analysis of Variance Analytics ==> Regression Analytics ==> Survey Sampling and Analysis SAS Reference ==> Procedures ==> SURVEYMEANS SAS Reference ==> Procedures ==> SURVEYREG

Date Modified:	2010-03-30 14:32:45
Date Created:	2009-01-27 12:30:44

E-mail This Page

Knowledge Base

Usage Note 34607: How can I compare means in PROC SURVEYMEANS?

Example

Operating System and Release Information

E-mail:
Password