ON A MODEL OF HUMAN INFORMATION PURCHASING IN OPTIONAL STOPPING AND SEQUENTIAL DECISION TASK

Abstract

A descriptive model of human information purchasing behavior in optional stopping and sequential decision task has been presented by Edwards. From the results of post erior probability estimation experiments, he assumed that the subject replaces Eθ(Z)with Eθ(Z′)=ωEθ(Z)in the optimal conditions of maximizing the expected value, where Eθ(Z)is the expected value of log-likelihood ratio for an observation data and ω is an inefficiency parameter. The value of ω is assumed to lie between0and1.However the model does not satisfy the condition(E(n/d))in the data generating process. Then a revised model is presented in this paper. As a result of examining the fitness of the data from the experiments reported by Wallsten and by others to the revised model, the author found and assumed that the condition of ω used in the model is greater than 1 to obtain better fit the experimental results, while Edward spostulated that ω<1. Finally the relevance of the model is discussed.