Linear and Nonlinear Valence Functions: A Behavioral Decision‐Making Assessment
通过实验对比两种线性和四种非线性效价函数的预测能力，发现近40%的受试者表现出非线性函数，强调在期望理论中正确指定效价函数形式的重要性。
ABSTRACT In Vroom's [45] original formulation of expectancy theory, the relationship between affect and perceived instrumentality was assumed to be linear. Others have suggested that such a relationship may be better modeled by a nonlinear, utility‐type function [30]. The current research contrasts the predictive ability of two linear and four nonlinear functions. Using four levels of McClelland's [26] needs for achievement, affiliation, and power as instrumentalities, 101 subjects provided more than 12,900 decisions on the valences of jobs in a behavioral decision‐making experiment. Nearly 40 percent of the subjects exhibited nonlinear valence functions. The results emphasize the need to specify the appropriate functional form of the valence component to enhance predictive accuracy and to prevent misspecification problems.