The number of judgments in a cluster, n(n-1) divided by 2, refers to the maximum number of pairwise comparisons in a cluster. However, Expert Choice can compute priorities from an eigenvector equation when there are missing judgments. At a minimum, there needs to be at least n-1 judgments. In order for this to work, there must also be a spanning set of judgments with a path from any element to any other element in the cluster.
There is a trade-off between time to make judgments and accuracy. If each of the n-1 judgments in a spanning set were perfectly accurate, there would be no need for any more (redundant) judgments. However, this is not the case – particularly when pairwise verbal judgments are made. One method to employ with Expert Choice is to elicit the n-1 judgments on the diagonal of the matrix, representing the elements in a cluster as a minimum spanning set, and perhaps the diagonal above consisting of an additional n-2 judgments.
