Search

In recent years a substantial body of research findings (based almost solely on responses to the Dilemmas-of-Choice instrument as the measure of risk), commonly referred to as the "risky shift" phenomenon, has suggested that groups take greater risks than individuals. This interpretation, if generalizable. would replace the commonly held belief that groups make decisions that are more conservative than. or the average of, individual decisions. This paper examines, empirically and analytically, some major limitations of the generalizability of the risky shift phenomenon. The major concerns of this paper are analysis of (1) the conventional method of statistically analyzing the data generated in risky shift studies, and (2) the measure of risk almost universally used by investigators of the risky shift phenomenon. In order to accomplish this task several experiments were designed. The conventional statistic used in the risky shift literature is analyzed and inferences appropriate to it are explored. The conventional test supports the risky shift findings. A revised measurement scale and alternative statistical testing procedures are discussed. The revised scale permits a continuum of response rather than the discrete choices permitted with the Dilemmas-of-Choice instrument. Results using the alternative methods of statistical analysis and the revised scale are presented. These results do not support the previous findings that groups accept a level of risk significantly greater than that acceptable to the average group member deciding alone. These findings strongly suggest that generalization of the risky shift phenomenon may be limited to individuals and groups using the Dilemmas-of-Choice instrument as the measure of risk. [ABSTRACT FROM AUTHOR]

An inductive inference problem will often be structured so that the target variable (hypotheses) are logically distant from the observable events (data). In this situation it may be difficult or impossible to assess the probabilistic connection between them, but it may be possible to decompose the problem through the use of intermediate or explanatory variables. That is, it will often be possible to assess the likelihood of the observed data given some intermediate variable, and the likelihood of that intermediate variable given another, and so on, until the hypotheses of interest are reached. Inferences which incorporate one or more intermediate variables are called hierarchical, cascaded, or multistage inferences. The present paper presents a normative model for the solution of the general hierarchical inference problem. The formulation begins with a formal description of the hierarchical inference tree, including a discussion of various simplifying conditional independence assumptions. The solution is first derived for three special case models of differing structure, and then the algorithm for the general solution is given for two cases: one in which the conditional independence assumptions have been made and one in which they have not. [ABSTRACT FROM AUTHOR]

This paper discusses the application of decomposition rules to Bayesian man-machine diagnostic inference systems. An experiment is reported which investigates the use of such decomposition rules in dealing with conditionally dependent data. The experiment, unlike its predecessors, employs an intuitive inference task where the veridical odds can be closely estimated. One major conclusion of this research is that logically equivalent decomposition rules are not necessarily of equal difficulty. Consequently, recommendations as to the superority of a particular decomposition rule should not be made; the systems designer's choice of a particular decision rule should be based on the types of data and conditional dependencies encountered in his particular situation. Such considerations may lead to the adoption of hybrid decomposition rules of a type not previously used. [ABSTRACT FROM AUTHOR]

This paper presents the results of an experiment designed in investigate the relationship between intuitive Bayesian point est inial and theoretically optimal Bayesian point estimates for various probability distributions (normal distributions) and loss functions (linear and quadratic loss functions). The results indicate that this relationship depends somewhat on the parameters of the distribution and loss function. Overall. however, the subjects performed the task of intuitive estimation and prevention very well. On the average, the intuitive estimates were quite close to the optimal estimates despite the complexity of the decision-making problem. [ABSTRACT FROM AUTHOR]