Consensus among experts in the technological forecasting by the Delphi method is here shown to be interpretable as a leading indicator or rather an antecedent factor of a national consensus and therefore as an intersubjective probability of choosing a particular policy among prospective policies when experts are properly selected. The condition under which this interpretation is valid is shown to hold in several Delphi surveys. Hence a logical foundation is obtained to fill two gaps: one gap between subjective opinions and an objective event of technological breakthroughs, and the other gap between the distributed opinions and a single event. Thus the statistical analysis of consensus is now given a meaning and the statistical methods are discussed to measure the consensus among experts in terms of, e.g., distribution matchings of their opinions. Experiences in technological and social forecastings and assessments are reported.
The use of penalty marks in competitive examinations is discussed, using two new indices called gap coefficient and inversion rate, which are evaluated under certain assumptions on the distribution of the true scores. Although some assumptions are rather arbitrary, it is shown at least that in some cases penalty marks can reduce the possibility for inferior examinees to get higher scores than superior ones.
This paper develops Bayesian Statistical Procedures for problems in the “Errors-in-the -Variables” Model. First, we derived the exact marginal posterior distribution for the regression coefficients in the model, and suggested an approximation to the exact pdf. Secondly, we derived the optimal prediction equation in the sense that it minimizes the expected loss for prediction.
A method of sensitivity analysis is proposed to detect the influential observations in Hayashi's third method of quantification (Hayashi, 1956). It evaluates the changes of the eigenvalues and the scores assigned to the categories and/or individuals due to a small change of the weights for a single or multiple individuals (or categories) by using the perturbation theory of eigenvalue problems.
This paper shows how one can measure the overall efficiency of organizational decision -making to yield maximum organizational outputs under many constraints of economic, political and institutional variables. The case study taken in this paper includes Japan's three public corporations, namely, the Japan Tobacco & Salt Public Corporation (JTS), the Nippon Telegraph & Telephone Public Corporation (NTT) and the Japan National Railways (JNR). In contrast to most economists' approaches to the study of public corporations in terms of the concept of “market failure, ” this study introduces a new concept of organizational behavioral, “inertia”, on economic decision-making, and next attempts to measure an organizational efficiency in a similar sense of Harvey Leibenstein's “X-efficiency.” To formulate such a behavioral inertia, this study applies the state-space modeling, a common modeling technique in the modern control theory.