Abstract
An algorithm for nonmetric internal unfolding analysis of a preference matrix is presented. It is based on the absolute value principle and intends to achieve a maximal mean rank-correlation between the rows of the data matrix and the corresponding rows of the distance matrix computed from the geometric representation. Using simulated data and a real data example, namely preferences for family compositions, the algorithm is compared with MINIRSA, an algorithm for unfolding based on the transformational principle.