Suppose there exist several variants and we believe in one valid relevance hypothesis for each dimension to be analysed. The value that stands for the relevance of each dimension enables us to compare alternatives. We can judge the plausibility of their success.
When we compare two variants, these variants differ in one or more dimensions. The sum of the relevance values of all dimensions that differ shall be called variance. The variance is either subcritical or it surpasses the threshold value of 1.
The relevance hypothesis looks plausible, if we can find to each acceptable example in the base at least one other successful variant, which differs only in factors where the respective relevance values add up to a sum of less than 1. That means those variants differ in factors of an importance that adds up to a less than critical value.
On the other hand, the relevance hypothesis looks implausible, if an successful alternative exists, where no other successful variant with a less than critical variance can be found. That means, either the relevance hypothesis is wrong, or there are variants missing in the example base, or both. This discovery should lead to further investigation in the object to be analysed.