Résumé. Les méthodes d'inférence de probabilités subjectives que sont l'amplitude directe et la bissection ont été comparées par l'auteur en ce qui a trait à l'exactitude, telle que déterminée par l'environnement (fréquence relative), des probabilités résultant de l'application de ces méthodes et à l'acceptabilité des méthodes d'inférence. Chaque participant à l'expérience a utilisé les deux méthodes et attribué à chacune d'elles une cote d'acceptabilité. L'auteur a mis à la disposition des sujets un logiciel d'inférence interactive, comprenant l'affichage graphique sous forme d'histogramme des probabilités selon les deux méthodes d'inférence, afin de les aider et d'assurer une comparaison plus raffinée des méthodes. Des étudiants de deuxième cycle diplômés en gestion, bien renseignés tant sur le contexte de la tâche que sur les probabilités subjectives, se sont prêtés à l'expérience. L'inférence directe a donné des probabilités subjectives beaucoup plus exactes que la bissection et s'est révélée beaucoup plus acceptable aux yeux des participants.

The direct magnitude and bisection subjective probability elicitation methods are compared based on the environmental (relative frequency) accuracy of the resulting probabilities and the acceptability of the elicitation methods. Each subject utilized, and provided acceptability indications for, both methods. Interactive elicitation software, including histogram‐like graphical displays of the probabilities for both elicitation methods, were provided to aid the subjects and provide for a more refined comparison of the methods. Second‐year graduate business students, knowledgeable about both the task context and subjective probabilities, completed the experiment. Direct elicitation of probabilities provided significantly more accurate subjective probabilities than did the bisection method and the former method also was significantly more acceptable to the subjects.

Given the objective of describing quantitatively how people use financial data to generate financial judgments, second-year graduate business school students were provided with data for each of 60 companies in an experimental setting, and an estimate of the change in security price over a one-year period was obtained from each subject for each security. Properties of the cognitive models people used to generate the price changes were inferred within the context of the Brunswik Lens Model. Topics considered included judgmental accuracy and consistency, types of models used including consideration of linear vs. nonlinear information processing and interjudge agreement. Reported results include empirical evidence on subjective security price change distributions, an encouraging degree of accuracy, a tendency to overestimate the prices of "low-priced" securities, evidence consistent with linear processing of financial data, depiction of distinct clusters of cognitive models and limited interjudge agreement. Suggestions for future research also are indicated.

Robert E. Hamilton, William F. Wright, Internal Control Judgments and Effects of Experience: Replications and Extensions, Journal of Accounting Research, Vol. 20, No. 2, Part II (Autumn, 1982), pp. 756-765

Robert K. Eskew, William F. Wright, An Empirical Analysis of Differential Capital Market Reactions to Extraordinary Accounting Items, The Journal of Finance, Vol. 31, No. 2, Papers and Proceedings of the Thirty-Fourth Annual Meeting of the American Finance Association Dallas, Texas December 28-30, 1975 (May, 1976), pp. 651-674