Direction of impact of factors not known

We often know how factors influence the target criterion. It is obvious what values of a factor promote or inhibit the success of the variants. Comparing different states of factors, we can say

"The more the better" or

"The less the better".

We use assumptions on how each factor affects the quality of a variant and on how factor variations act on the qualitiy of a variant. Building on those assumptions, it is possible to compare hypothetical alternatives and even to put unobserved alternatives into a ranking. When there is such detailled knowledge on the object of analysis, the FACTORFINDER-Software is of little further use.

Despite existing assumptions on how each factor acts there may be the need to apply the FACTORFINDER-Procedure: It reduces the information on the importance of each factor to just one value and thereby increases the readability of the results.

In an early stage of analysis, it is unknown what factor promotes or inhibits the success of an alternative. In this case, using the FACTORFINDER-Software makes sense as the software identifies factor-value-pairs that lead to acceptable alternatives.

Let us look at a single variant of acceptable success, a combination of certain input factors: We do not know, what values of a factor promote success and what values inhibit success. The only fact we know is, that all goal-adverse values are not important enough to let that alternative fail. If we observe different values of a factor in different acceptable variants, there are two possibilities left: Either this factor is critically important and it does not tolerate adverse values. In this case, each value observed must positively act on the target criterion. The other possibility is this factor is less than critically important. That grants some tolerance to goal-adverse values.

The FACTORFINDER-Procedure does not require any assumptions on the direction of action of factor variations. As well, there are no assumptions on the semantic distance between different values necessary. There is no metric of congruence required. In fact, comparison only states difference or identity.