Especially in the first stages of research, it is not always clear which of many possible factors are relevant. Inclusion of all factors results in enormous factorial designs, which can cause practical problems because of the large number of subjects involved. Fractional design offers a solution to this problem by reducing the number of factor combinations of a design and, therefore, treatments (or runs) and cases. There are two situations in which the use of fractional design is recommendable (Landsheer & van den Wittenboer, 2000). The first is the elimination of possibly irrelevant factors from a large number of possible candidates.The second is the estimation of main factors and sometimes two-way interactions when higher order interactions can be considered negligible in designs with a large number of factors.The reduction of a full design is accomplished by confounding effects. Effects that are confoundedare noted as an equivalence,for instance A 5 B * C * E 5 B * D * F 5 A * C * D * E * F, which means that main effect A is confounded with the two three-way effects BCE and BDF and with the five-way effect ACDEF. For the totality of these confounded effects, there is only one estimate of variance. When this estimate is not significant, the conclusion that all confounded effects are negligible is warranted. When this estimate is significant, it can be considered to be an estimate of factor A, assuming that the higher order effects BCE, BDF, and ACDEF are negligible.Higher order effects can never be estimated when using a fractional design.Most fractional designs allow all main factors to be tested, although in some cases, the two-way interaction effects can be estimated next to the main effects. The price to be paid is the impossibility of estimating higher order interaction effects. If these higher order interactions are assumed to be zero, they can be eliminatedfrom the confounded effects, allowing for a valid estimation of the magnitude of main effects and possibly two-way interactions. Main effects are estimable when they are confounded only with higher order interactions, but not with each other. Twoway interactions can be estimated when they are confounded neither with each other nor with the main effects, but with only higher order interactions. The main effects and two-way interactions can be estimated, if the higher order interactions are known. In general, when the higher order effects are known in advance, they can be subtracted from the total of estimated variance. However, in most cases, the higher order effects are assumed to be zero.Fractional design has evolved in the field of quality control. When a new product is made, there are many unknown possible shortcomings in the production process. Mesenbrink and Lu (1994), who studied the quality of a wavesoldering process, gave an example of such a study. In the The computer program Fractional Design Wizard creates fractional factorial designs that are costeffective and especially useful for discarding irrelevant factors from a large number of possi...