A type of experimental design in which each block contains one representative of each treatment, the treatments being allocated to random positions in the block. Each block is thus a complete replication but allocation of treatments to columns is random. The columns are therefore not complete replications since they may contain two or more representatives of one treatment and none of another. Randomized blocks may be used, for example, when dealing with identifiable inherent variation. Compare Latin square. |
