Kodo Keiryogaku (The Japanese Journal of Behaviormetrics) Vol. 40(2013) No. 2

The reliability of the reference value of a clinical test is lowered if the sample size of the study data was not sufficient. The purpose of this study was to indicate the reliability of the reference value by displaying not only the point estimate, but also the interval estimate. The rest electrocardiograms of 1276 cases of healthy Japanese men were used for the present study. To set the upper limit of the QT reference value, we used the bootstrap method. As we supposed beforehand,the range of confidence interval widened in both extremes of RR where the sufficient number of study cases could not be collected. As we showed in this study, the upper limit of reference value (the parameter in the upper tail of distribution) could be comparatively easily estimated with the confidence interval by using the bootstrap method.

This paper presents some notes on question items and psychological scales used in social surveys. We review mainly the longitudinal and cross-national comparative survey data collected by the Institute of Statistical Mathematics over half a century, focusing several items concerning subjective social class, sense of satisfaction, and sense of trust as practical examples. In particular, we note on meaningfulness, appropriateness, and stability of psychological scaling on question items as well as quality of statistical representative sampling, as the necessary conditions of scientific research. These notes may suggest how to improve the studies on psychological scaling of social surveys which contradictory results are frequently reported due to misunderstandings on sampling and scaling.

Four research methods (free association, sentence completion tests, cartoon completion tests, respondent association network) were compared in order to clarify the characteristics of methods to collect brand association. The result of an experimental study indicates the characteristics of each method. In this study past studies relating to collecting brand association were reviewed and an experimental study was conducted. The result of the experimental study indicates the relationship between the purpose and method of collect brand association and touches on practical use in business. At the end of the study, future challenges and the pros and cons of each of the methods covered are discussed.

After importance of primary education was internationally agreed at the World Educational Conference in 1990, primary school enrolment was drastically increased in many developing nations. However, the children called “last 10%,” who are left behind after 90% school enrolment was achieved, are assumed to live particularly in rural area. Thus, policy analysis today is expected to assess their specific needs, and in order to analyze the effects from different life environments surrounding each individual, the methodology for hierarchical analysis is needed. However, such methodology is not yet generalized in the field of education policy analysis in developing countries. Therefore, it is urgently necessary to accumulate the case studies of hierarchical models. The purpose of this paper is to introduce the new methodology of hierarchical analysis into this field of study by proposing a needs assessment model for Tanzanian pupil's literacy development, with a special attention to hierarchical environmental factors. This study hypothesizes that each individual child lives under two different hierarchical environments, “educational environment at individual level” and “public life environment at district level,” and analyzes how these hierarchical environmental factors affect on Tanzanian children's literacy development, by using Hierarchical Generalized Linear Modeling.