TRANSPOSE Procedure

Example 6: Transposing Data for Statistical Analysis

Features:

COPY statement

VAR statement

This example arranges data to make it suitable for either a multivariate or a univariate repeated-measures analysis.

The data is from Chapter 8, “Repeated-Measures Analysis of Variance,” in SAS System for Linear Models, Third Edition.

Program 1

options nodate pageno=1 linesize=80 pagesize=40;

data weights;
   input Program $ s1-s7;
   datalines;
CONT  85 85 86 85 87 86 87
CONT  80 79 79 78 78 79 78
CONT  78 77 77 77 76 76 77
CONT  84 84 85 84 83 84 85
CONT  80 81 80 80 79 79 80
RI    79 79 79 80 80 78 80
RI    83 83 85 85 86 87 87
RI    81 83 82 82 83 83 82
RI    81 81 81 82 82 83 81
RI    80 81 82 82 82 84 86
WI    84 85 84 83 83 83 84
WI    74 75 75 76 75 76 76
WI    83 84 82 81 83 83 82
WI    86 87 87 87 87 87 86
WI    82 83 84 85 84 85 86
;

data split;
   set weights;
   array s{7} s1-s7;
   Subject + 1;
   do Time=1 to 7;
      Strength=s{time};
      output;
   end;
   drop s1-s7;
run;

proc print data=split(obs=15) noobs;
   title 'SPLIT Data Set';
   title2 'First 15 Observations Only';
run;

Program Description

Set the SAS system options. The NODATE option suppresses the display of the date and time in the output. PAGENO= specifies the starting page number. LINESIZE= specifies the output line length, and PAGESIZE= specifies the number of lines on an output page.

options nodate pageno=1 linesize=80 pagesize=40;

Create the WEIGHTS data set. The data in WEIGHTS represents the results of an exercise therapy study of three weight-lifting programs: CONT is a control group, RI is a program in which the number of repetitions is increased, and WI is a program in which the weight is increased.

data weights;
   input Program $ s1-s7;
   datalines;
CONT  85 85 86 85 87 86 87
CONT  80 79 79 78 78 79 78
CONT  78 77 77 77 76 76 77
CONT  84 84 85 84 83 84 85
CONT  80 81 80 80 79 79 80
RI    79 79 79 80 80 78 80
RI    83 83 85 85 86 87 87
RI    81 83 82 82 83 83 82
RI    81 81 81 82 82 83 81
RI    80 81 82 82 82 84 86
WI    84 85 84 83 83 83 84
WI    74 75 75 76 75 76 76
WI    83 84 82 81 83 83 82
WI    86 87 87 87 87 87 86
WI    82 83 84 85 84 85 86
;

Create the SPLIT data set. This DATA step rearranges WEIGHTS to create the data set SPLIT. The DATA step transposes the strength values and creates two new variables: Time and Subject. SPLIT contains one observation for each repeated measure. SPLIT can be used in a PROC GLM step for a univariate repeated-measures analysis.

data split;
   set weights;
   array s{7} s1-s7;
   Subject + 1;
   do Time=1 to 7;
      Strength=s{time};
      output;
   end;
   drop s1-s7;
run;

Print the SPLIT data set. The NOOBS options suppresses the printing of observation numbers. The OBS= data set option limits the printing to the first 15 observations. SPLIT has 105 observations.

proc print data=split(obs=15) noobs;
   title 'SPLIT Data Set';
   title2 'First 15 Observations Only';
run;

Output 1

Split Data Set

                 
               SPLIT Data Set                                1
                           First 15 Observations Only

                     Program    Subject    Time    Strength

                      CONT         1         1        85
                      CONT         1         2        85
                      CONT         1         3        86
                      CONT         1         4        85
                      CONT         1         5        87
                      CONT         1         6        86
                      CONT         1         7        87
                      CONT         2         1        80
                      CONT         2         2        79
                      CONT         2         3        79
                      CONT         2         4        78
                      CONT         2         5        78
                      CONT         2         6        79
                      CONT         2         7        78
                      CONT         3         1        78

Program 2

options nodate pageno=1 linesize=80 pagesize=40;

proc transpose data=split out=totsplit prefix=Str;

   by program subject;
   copy time strength;

   var strength;
run;

proc print data=totsplit(obs=15) noobs;
   title  'TOTSPLIT Data Set';
   title2 'First 15 Observations Only';
run;

Program Description

Set the SAS system options.

options nodate pageno=1 linesize=80 pagesize=40;

Transpose the SPLIT data set. PROC TRANSPOSE transposes SPLIT to create TOTSPLIT. The TOTSPLIT data set contains the same variables as SPLIT and a variable for each strength measurement (Str1-Str7). TOTSPLIT can be used for either a multivariate repeated-measures analysis or a univariate repeated-measures analysis.

proc transpose data=split out=totsplit prefix=Str;

Organize the output data set into BY groups, and populate each BY group with untransposed values.The variables in the BY and COPY statements are not transposed. TOTSPLIT contains the variables Program, Subject, Time, and Strength with the same values that are in SPLIT. The BY statement creates the first observation in each BY group, which contains the transposed values of Strength. The COPY statement creates the other observations in each BY group by copying the values of Time and Strength without transposing them.

   by program subject;
   copy time strength;

Specify the variable to transpose. The VAR statement specifies the Strength variable as the only variable to be transposed.

   var strength;
run;

Print the TOTSPLIT data set. The NOOBS options suppresses the printing of observation numbers. The OBS= data set option limits the printing to the first 15 observations. SPLIT has 105 observations.

proc print data=totsplit(obs=15) noobs;
   title  'TOTSPLIT Data Set';
   title2 'First 15 Observations Only';
run;

Output 2

In the following output, the variables in TOTSPLIT with missing values are used only in a multivariate repeated-measures analysis. The missing values do not preclude this data set from being used in a repeated-measures analysis because the MODEL statement in PROC GLM ignores observations with missing values.

TOTSPLIT Data Set

                              TOTSPLIT Data Set                               1
                           First 15 Observations Only

   Program Subject Time Strength  _NAME_  Str1 Str2 Str3 Str4 Str5 Str6 Str7

    CONT      1      1     85    Strength  85   85   86   85   87   86   87
    CONT      1      2     85               .    .    .    .    .    .    .
    CONT      1      3     86               .    .    .    .    .    .    .
    CONT      1      4     85               .    .    .    .    .    .    .
    CONT      1      5     87               .    .    .    .    .    .    .
    CONT      1      6     86               .    .    .    .    .    .    .
    CONT      1      7     87               .    .    .    .    .    .    .
    CONT      2      1     80    Strength  80   79   79   78   78   79   78
    CONT      2      2     79               .    .    .    .    .    .    .
    CONT      2      3     79               .    .    .    .    .    .    .
    CONT      2      4     78               .    .    .    .    .    .    .
    CONT      2      5     78               .    .    .    .    .    .    .
    CONT      2      6     79               .    .    .    .    .    .    .
    CONT      2      7     78               .    .    .    .    .    .    .
    CONT      3      1     78    Strength  78   77   77   77   76   76   77