【摘要】 针对已有规范性决策研究与描述性决策研究尚不能对多参考点共存情形属性价值评价予以全面、直接决策工具支持的不足,通过引入竞争对局及其偏好得分概念提出了基于目标导向效用理论和对局效用比较的新规范性价值函数,在此基础上通过结合范围—频率(RF)理论模型,给出了能够兼顾规范性与描述性两个决策框架的规约性目标导向多参考点属性价值模型及评价方法。相对于规范性目标导向期望价值理论而言,新价值函数在全面科学反映形式多样的决策者偏好结构上具有更为宽广的适用性。相对于RF模型等行为决策理论而言,目标导向多参考点属性价值模型及评价方法克服了行为决策理论模型因其描述性框架性质而不能直接支持决策者属性价值评价的不足,能够对多决策参考点共存情形直接指导决策者的属性价值评价。另外,目标导向多参考点属性价值评价方法将决策目标看成是直接影响方案评价与选择的决策变量,因此也具有发展与深化多属性决策理论的科学价值。模拟验证分析与方法应用表明:目标导向多参考点属性价值评价方法不仅具有广泛的应用适用性,而且在价值评价上也展示出了较强的科学合理性。更多还原

【Abstract】 To overcome the shortage of current normative and descriptive studies of decision-making,i.e.,less capability of providing direct comprehensive supports for attribute value evaluation with multiple reference points（MRPs）,two concepts including competitive games（CGs）and CG preference scores（CGPSs）are incorporated with the target-oriented utility/value theory（TOUVT）,by which a new normative attribute value function based on TOUVT and the game-preference comparison,shorted as NAVF-T-GPC,is thus presented.Based on NAVF-T-GPC and the range-frequency（R-F）model,aprescriptive linear programming（PLP）model as well as its corresponding appraisal approach,considered as a merge of normative decision-making with descriptive decision-making,is built for directly aiding decision makers with MRPs to evaluate attribute values.According to the presented approach,when a specific performance（x） lies in(x_r,x_r+1),its attribute value is given by v*（x）=[∑r_s=1Δ*（x_s）p（x_s）]/[∑s=1RΔ*（x_s）p（x_s）],where x_s（xr）denotes the s^th(r^th)reference point or target object in the target attribute interval[x_L,x_H],s=1,…,R,r=1,…,R-1,Δ*（x_s） reflecting the CGPS of x_s is determined by the PLP model,and p（xs）represents the occurrence probability of xs.Note that the PLP model embodies such key academic thoughts that the attribute value of performance x（x_L≤x≤x_H）,behaviorally disclosed by the R-F model,is only a linear approximation to that determined by NAVF-T-GPC.Compared with TOUVT,in which the CG-PSs of different attribute performances are supposed to be equal,NAVF-T-GPC is more suitable for reasonably reflecting various decision-makers’complex preference structure since unequal CGPSs related with MRPs are permitted.Compared with behavioral decision-making theories such as the R-F theory/model,which can not be directly applied to real-world decisions because of its descriptive nature,the presented approach is capable of directly aiding decision makers with MRPs.Besides,the approach is built from a new viewpoint of taking the decision goal as a variable directly affecting alternative evaluation and alternative selection,and thus of importance to develop the multiple attribute decision-making theory.Simulation analysis and application comparison show that the presented approach can be widely applied to real-world decisions and has advantages in reasonably evaluating attribute values over the current subjective method based on equally-spaced value judgments.更多还原