The primary goal of the design of experiments is to determine the maximum amount of information about a system with the minimum amount of effort. A well-designed experiment guides the experimenter in choosing what experiments actually need to be performed. From the resulting measurements, the experimenter can determine the effects on performance of each individual input factor, and the effects of their interactions. The form of the experimental design also allows a quantitative evaluation of the error inherent in the experimental measurements relative to the overall system response.

A key assumption behind the design of experiments is that there is a nonzero cost associated with performing an experiment. This cost includes the time and effort required to gather the necessary data, plus the time and effort needed to analyze these data to draw some appropriate conclusions. Consequently, it is important to minimize the number of experiments that must be performed while maximizing the information obtained.

Good experiment design allows the experimenter to

• isolate the effects of each individual input variable,

• determine the effects due to interactions of the input variables,

• determine the magnitude of the change in the system's output due to the experimental error, and

• obtain the maximum amount of information with the minimum amount of effort by limiting and controlling the number of experiments that must be performed.

Types of experiments

The simplest design for an experiment varies one input (factor) while holding all of the other inputs constant.