What is randomization?

Randomization is a technique that is used to balance the effect of extraneous or uncontrollable conditions that can impact the results of an experiment. For example, ambient temperature, humidity, raw materials, or operators can change during an experiment and inadvertently affect test results. By randomizing the order in which experimental runs are done, you reduce the chance that differences in experimental materials or conditions strongly bias results. Randomization also lets you estimate the inherent variation in materials and conditions so that you can make valid statistical inferences based on the data from your experiment.

Suppose you work for an offset printing company and want to maximize the effectiveness of the bookbinding technique. You can control factors such as glue temperature, paper type, and cooling time. However, you cannot control humidity, which can affect how quickly the glue sets. Or, perhaps there are other "unknowns" that cannot be easily controlled or measured. For example, the bookbinding machine might not apply consistent pressure.

When you create a designed experiment, Minitab automatically randomizes the run order, or ordered sequence of the factor combinations, of the design. For example, a 2-level full factorial design based on the bookbinding example yields the following results, (which will vary because of randomization):

If you do not randomize, the run order and the standard order are the same. There may be situations when randomization leads to an undesirable run order. For instance, in industrial applications, it may be difficult or expensive to change factor levels. Or, after factor levels are changed, it may take a long time for the system to return to a steady state. Under these conditions, you may not want to randomize. Alternatively, you may want to randomize with a split-plot design in order to minimize the level changes.

If you want to re-create a randomized design with the same run order, you can choose a base for the random number generator. Then, when you want to re-create the design, you use the same base.