Classifying Decision Strategies in Multi-Attribute Decision-Making: A Multi-Dimensional Scaling and Hierarchical Cluster Analysis of Simulation Data

Previous studies on decision strategies in multi-attribute decision-making (MADM) have primarily relied on computational simulations to assess strategy performance under varying conditions, with particular emphasis on comparisons to the weighted additive rule (WAD) and on evaluations of the cognitive effort required. In contrast, considerably less attention has been devoted to examining the consistency of decision outcomes across different strategies or to developing a systematic classification of strategies based on outcome similarity. To address this gap, the present study investigates the characteristics of decision strategies by analyzing the concordance rates of choices made under identical conditions, along with measures of decision accuracy and information-processing effort. We conducted a hierarchical cluster analysis and applied multi-dimensional scaling (MDS) to a choice concordance matrix derived from simulations using the Mersenne Twister method. In addition, linear multiple regression analyses were performed using the MDS coordinates as predictors of both decision accuracy and cognitive effort. The cluster analysis revealed a primary bifurcation between two major groups: one centered around the Disjunctive (DIS) rule, and another encompassing compensatory strategies such as WAD. Notably, although the Lexicographic (LEX) rule is traditionally considered non-compensatory, it exhibited high similarity in choice patterns to compensatory strategies when assessed via concordance rates. In contrast, DIS-based strategies produced markedly distinct choice patterns.