ABSTRACT
As discussed previously, three basic tools in the experimental design process are replication, randomization, and blocking. We will begin using them in this chapter and continue using them throughout the book. Recall that replication refers to repeating experimental runs in order to collect multiple measurements of the experimental phenomenon of interest. The replication process enables one to estimate the true mean value of a population with both greater precision and greater accuracy. It also allows estimation of the magnitude of experimental error. Randomization of experimental units and/or treatments is a precaution against bias. Each experimental unit, then, is equally likely to be assigned any one of several treatments. For example, virtually all statistical methods require that observations (or experimental errors) be independently distributed random variables, and randomization makes this assumption valid. Finally, blocking is a technique used to increase the precision of an experiment by controlling for an extraneous variable within a population sample. Blocks consist of sample items grouped to be more homogeneous with respect to that variable. For example, a simple block may combine microbial counts from the left and right sides of a test subject used in a preoperative skin preparation evaluation. This is because one would expect the population counts to be more similar in numbers on the left and right sides of an individual than counts among different individuals. The samples within a block undergo the experimental treatments with randomization occurring only within that block, and posttreatment results are compared with each other only within that block. In our 82simple example using two products, each would be assigned randomly to either the left or right side of the subject. This process reduces experimental error, making the statistical test more powerful—that is, more likely to reject a null hypothesis when it is false.
