* Example for paired sample testing problem;
options pagesize=53 linesize=76 pageno=1;
data keybrd;
 input time1 time2;
 diff=time1-time2;
 label time1="Completion Time with Smaller Keyboard"
       time2="Completion Time with Larger Keyboard"
       diff="Difference in Completion Times"
       ;
 datalines;
   59.48  85.73
   53.11  41.25
  111.22  45.81
   71.13  52.23
   52.95  46.63
  101.66 102.05
   51.02  35.36
   73.38  52.83
   53.56  50.96
   80.57  58.28
   65.31  59.27
   79.20  73.27
   80.48  61.63
   57.13  44.87
   73.82  58.99
   86.35  43.06
   54.04  53.78
   57.89  45.53
   67.73  45.75
   57.02  54.31
   38.06  46.32
   60.81  69.80
   82.26  52.50
   43.83  43.49
   32.83  32.90
   35.15  32.35
   39.26  33.48
   32.57  30.33
 run;
title "Testing for Equal Population Mean Completion Times";
proc print label;
 var diff time1 time2;
proc univariate normal;
var diff;
histogram/normal;
probplot / normal(mu=est sigma=est w=1 l=1);
run;
* You can use the procedure TTEST instead of proc univariate;
* With TTEST, there is no need to produce variable "diff";
proc ttest;
paired time1*time2;
run;