Simplified DINDSCAL

INDSCAL (INDividual Differences SCALing) model is commonly used to study the individual differences in three-way data based on doubly centered set of matrices of squared dissimilarity measures between a range of stimuli. An alternative approach, called DINDSCAL (Direct INDSCAL) directly analyzes the set of squared dissimilarity matrices. The advantage of this approach is that the fitting of the zero diagonal entries of the dissimilarity matrices is avoided. Both, INDSCAL and DINDSCAL, require solution of difficult optimization/estimation problems and rely on iterative algorithms. Here we consider some alternative models to INDSCAL and DINDSCAL, which adopt certain simplifying constraints. As a result, the produced solutions are approximate, but the algorithms are not iterative and are very fast. Thus, they may be advantageous for analyzing large data with or without constraints.