Abstract: Both the multiple factor decision-making method based on variable weights (MFDMM-VW) and the Choquet integral model (CIM) for multiple attribute decision-making (MADM) cannot reflect reasonably the decision-maker's point-dependence preference in MADM. To overcome the mentioned shortcoming, a new MADM approach to evaluation and decision making, i.e., the MADM model with relative variable weights (MADMM-RVW), and its corresponding method are presented based on the swing-weighting, the analytic network process, and the relative evaluation thought embodied in data envelopment analysis. Compared with the MFDMM-VW, the MADMM-RVW is able to reflect directly the decision-maker's point-dependence preference structure, and thus overcome the arbitrariness of the MFDMM-VW in reflecting the decision-maker's point-dependence preference, resulted from decision analysts when the MFDMM-VW is applied. Compared with the CIM, the MADMM-RVW can not only reflect the decision-maker's point-dependence preference but also avoid more efficiently the CIM's large-scale estimation problem of decision parameters in many MADM cases. Applied in a case study, the MADMM-RVW is showed to give decision conclusions well consistent with objective existence and the decision-maker's qualitative opinions on preference dependence, and thus be capable of better reflecting the decision-maker's specific preference behaviors.

Key words: MADM, decision making with variable weights, preference dependence, swing weighting, analytic network process, data envelopment analysis