Fractional factorial design

In the statistical design of experiments is meant by a partial factor plan (English: fractional factorial design) a test plan that by playing only a certain fraction of the factor combinations contained in a complete test plan.

A complete experimental design for six factors at two levels, for example, would require 26 = 64 test runs. For economic reasons, it is in such a case is often resorted to some kind of factor map of the 26-3 or 27-4, etc.. The spelling 26-3 indicates here that, of the total of 26 possible test runs only a fraction, namely 1/8 ( 2-3), is performed ( 64: 8 = 8). To select the test runs in the Sub- factor plan eight called an orthogonal array is often used.

The disadvantage of sub- factor plans is the thereby occurring blending of factor effects and interactions. This becomes all the more noticeable the smaller the selected fraction of the complete test plan. To describe the type and amount of confusions occur every part factor plan is a resolution class (English: resolution number) assigned.