Modelling decision making under risk and uncertainty by novel utility measures

Abstract

In this paper, we will address the classical decision theories (the expected value, EV, and expected utility theories, EUTs) along with their violations, such as the common consequence, common ratio effect, violation of betweenness and the fourfold risk pattern. In particular, a numerical method is proposed to determine the utility function of an individual or group of individuals. The approach depends on the individual's evaluation of the certainty equivalent (CE) of a decision problem under uncertainty. Later, we propose an optimisation model to predict human preference between pairs of reward scenarios in which uncertainty is involved. The optimisation model implements binary logistic regression (BLR). Both SPSS and Excel Solver were used in the optimisation and parameter fitting. The presented model is verified via collected survey and literature studies. It is found that the model is able to explain the violations and serve as a new replica to predict human preferences.