The use of multiple-choice questions (MCQ) on many kinds of qualifying examinations such as the Japanese National Board Examination for Medical Practice has become very common. It is well known that examinees who know little or nothing about the subject material can still score points by random guessing on MCQ tests. The aim of this paper is to make clear the characteristics of MCQ scores from the incomplete knowledge that an examinee has. Using the stochastic model and taking incomplete information into account, we derived a score-distribution based on the knowledge levels of examinees taking an MCQ test, and showed the positive and negative effects of the incomplete knowledge of these examinees.
This paper shows the utility of a psychometric model called the generalized high threshold (GHT) model to cross-national comparative studies of national characters based on public opinion data. This model provides a sort of scalar measure of conformity of each nation to a hypothetical culture called the superculture. The superculture is a “standard” against which those nations are compared for the elucidation of their mutual relationships. Furthermore, a new representation method called the vectorial angular (VA) representation is developed for a graphical representation of the GHT model. In this representation all nations are represented as vectors, and the measures of conformity and the matching rates of responses of cross-national public opinion research are represented by the angles between those vectors. The GHT model and the VA representation are applied to two cases: the comparative study of national characters (France, FRG, Japan, the UK and the USA); the comparative study of opinions on the issues of EC unification (12 European countries). These applications show the utility of the model and representation for both domain-independent and domain-specific comparative studies of national characters at the exploratory phase of research.
A scaling method for qualitative multiple responses or categories corresponding to doses of stimuli is provided. This method is applied to some sets of experimental data, comparing the scores determined by this method and the regression curves estimated between the scores and the doses with those determined by the ridit method and by quantification method III. Further, using certain simulated data, the reproducibility of the original regression curves between the variables generating the simulated data by the curves estimated between the scores and the doses is investigated. These results are also compared with those reproduced by the ridit method and by quantification method III.
A single profile of a solid object contains much information about the shape of the object. Viewing the changing profiles of a moving object provides even greater information about the shape of the object. Few computational models of this process have been applied to the human ability to recover the shape and motion of solid objects from their changing profiles. We propose a theory that relates measurable quantities of changing two-dimensional (2-D) profiles to structural properties of three-dimensional (3-D) surfaces in motion. The relevance of this theory to human perception is shown by relating theoretical predictions to existing psychophysical results as well as additional demonstrations of human recovery of shape from profiles.