Understanding Replication and Pseudo-replication

The layout of an experiment is not understood unless the manner of treatment replication is understood. Recognizing when treatments are not actually replicated (pseudo-replication) requires determining the actual experimental units for each treatment. This determination is often more difficult than one might expect. A treatment is replicated when it is applied to two or more experimental units (e.u.'s) in one of two main ways: 1) in a completely randomized design treatments are randomly assigned to e.u.'s. 2) in a randomized block design treatments are randomly assigned to one e.u. per block. Since blocks are often called "reps " the term replication is restricted by some to mean blocks. Unfortunately, this can cause confusion so that a REP term is occasionally included in the ANOVA table of a completely randomized design, when it should be indicated by the error term (e.u.'s nested within treatments). When thinking about the replication of an experiment it is important to recognize what the e.u.'s are (it is possible for different treatments in an experiment to have different e.u.'s, as occurs with the split-plot). An experimental unit is a basic unit of experimental material to which on