Example 7.5 Randomized Complete Block Design

Note: See A Randomized Complete Block Design in the SAS/QC Sample Library.

In a randomized complete block design (RCBD), each level of a treatment appears once in each block, and each block contains all the treatments. The order of treatments is randomized separately for each block. You can create RCBDs with the FACTEX procedure.

Suppose you want to construct an RCBD with six treatments in four blocks. To test each treatment once in each block, you need 24 experimental units. The following statements construct the randomized complete block design shown in Output 7.5.1:

proc factex;
   factors Block / nlev=4;
   output out=Blocks Block nvals=(1 2 3 4) randomize(12345);
run;
   factors Treatment / nlev=6;
   output out=RCBD
          designrep=Blocks
          randomize(54321)
          Treatment cvals=('A' 'B' 'C' 'D' 'E' 'F');
run;
quit;
proc print data=RCBD;
run;

Note that the order of the runs within each block is randomized and that the blocks (1, 2, 3, and 4) are in a random order.

Output 7.5.1: A Randomized Complete Block Design

Obs Block Treatment
1 3 F
2 3 D
3 3 C
4 3 A
5 3 B
6 3 E
7 2 C
8 2 D
9 2 F
10 2 B
11 2 E
12 2 A
13 1 C
14 1 F
15 1 B
16 1 E
17 1 A
18 1 D
19 4 A
20 4 D
21 4 C
22 4 F
23 4 E
24 4 B