Analysts never know with certainty the ideal values of the factors. Only the alternatives that are realized are known and whether they reach the target criterion acceptably. These information are the basis for comparisons of the variants. From those comparisons of observed alternatives inferences on the ideal alternatives are drawn.
Those successful variants that make all non-ideal values look tolerable come closest to the concept of ideal variants. Those alternatives which consist only of ideal values are most tolerant to unexpected deviations from ideal. Robust alternatives differ from many examples in unimportant factors. Consequently, the values of the intermediate result vector are low. This renders these alternatives prominent and identifies such an example. By that, the FACTORFINDER-Procedure identifies robust alternatives.
The robust variants have manyfold relations to other acceptable variants by subcritical variance to those examples. They are not neccessarily the examples that perfom best (if there is a more differentiated target function behind the threshold value) and not neccessarily the combination of values that is found most often. Robust alternatives are characterized by tolerating best deteriorations by non-ideal values.