The difference in sampling frames of RDD telephone survey after redefinition was verified. Previously, sampling frames of the RDD telephone survey were largely separated between the working exchange and list-assisted designs. However, the telephone number system has changed in recent years. This research, reflecting those changes, redefined the working exchange design in order to divide the working exchange frames into two categories: one including telephone numbers belonging to carriers other than NTT and the other including telephone numbers belonging to only NTT. The verification of the difference in the sampling frames confirmed the following two findings: first, as for the list-assisted design, comparing with the NTT-working exchange, its coverage was low at most by approximately 3-4%, while efficiency in physical inspection was promoted. Second, the samples collected in the NTT-working exchange frame were different only in their attribution and reply results. This verification proved the efficiency of the list-assisted design; however, it was also discovered that a population exists that could be collected only by NTT.
Since the telephone (RDD) poll of the day is carried out in the call center, which is highly managed, the risk of operator's error generation is low. However, this report suggests the risk of operator's error generation due to intervene of operator's judgment. It is the case that the questionnaire is not carefully examined. Now therefore, the question with an unclear meaning with a long sentence and/or an open-answer question must be avoided when there is not enough time for developing a thorough questionnaire.
In Internet surveys, the responses for rating scale questions can be influenced by appearance of the rating scale on a computer screen. To examine the effects of the way of displaying response items on a computer screen of the internet survey, the author has conducted an experiment using 16 kinds of questionnaires. Each of the questionnaires includes same questions but the choice items of 7-point rating scales were displayed in different manners: vertical or horizontal layout, ascending or descending order, numbered or worded labels, bipolar or unipolar scales. By using the result of the surveys, he mainly compares top-3 box ratings. In the comparisons, he finds some order effects and other effects: higher top-3 box ratings by displaying the rating scales in ascending order than that of descending order and higher top ratings by displaying numbered labels with bipolar scales than that of worded labels. However, he doesn't find any significant difference from whether to display the scales vertically or horizontally.
The article by The Examination Committee of Criteria for ‘Obesity Disease’ in Japan and Japan Society for the Study of Obesity (2002) established cut-off points of waist circumference at 85 cm for males and 90 cm for females as criteria of obesity disease. Their article has also become a basis for criteria of metabolic syndrome in Japan; however, their article has various problems on statistical aspects. How the criteria on waist circumference should vary was investigated through reexamination based on descriptions in their article. First, their article obtained criteria on visceral fat area (VFA) from data in which males and females were pooled, while our reexamination suggested that we should use separate criteria on VFA between males and females. Second, their article inappropriately used regression lines to estimate waist circumference corresponding to VFA cut-off points. Our reexamination with errors-in-variables models suggested alternative cut-off points of waist circumference at 87 cm for males and 85 cm for females. Our simulation confirmed that the criteria by their article might lead to inappropriate diagnosis which is strict for males and easy for females.
This study investigated the visibility of character colors in connection with background colors. Each subject observed the randomly assigned text whose background and character colors were chosen from the eight basic colors, i.e., black, blue, cyan, green, magenta, red, white, and yellow, on a plasma display, and rated how easy it was to see it on a five-point rating scale. A maximum likelihood asymmetric multidimensional scaling was applied to the data, and the four-dimensional Okada and Imaizumi's (1987) model was found to be the optimal model. As a result of the weighted orthogonal Procrustes rotation, the symmetric part of the optimal model was congruent with the Euclidean distance in the dimensionally shrunk or expanded YUV color space with the additional dimension on which red, magenta, green, and cyan were opposite to the others. As regards the skew-symmetric part, magenta and blue were shown to match the character colors better than the background colors.
The main objective of this paper is to investigate the structure of survey response data on nuclear power generation. Previous studies have shown about the relationships between the responses at the individual level and at the aggregate level in some panel surveys. Some studies have reported about the stability of marginal distribution in spite of response fluctuations at the individual level. I focus on the stability of marginal distribution of response data in a panel survey where an accident of nuclear plant happened between the first wave and the second wave. Among others, I analyzed the stability of marginal response by classifying the changes of individual responses into two factors: individual fluctuation and substantial changes. I confirmed the stability of marginal response distributions of simple tabulations as well as the response structure represented in Hayashi's Quantification Method III, in spite of individual response fluctuations.
Okumura (2007) proposed a simulation-based method of sample size determination for hierarchical linear models using posterior predictive distribution. In this article, Okumura's method is applied to a social psychological research which focuses on the difference of self-evaluation score of mathematics achievement in TIMSS 2003 between Japanese and American 8th grade students. The result says that we can keep high statistical power and can estimate the difference in high precision using a sample of about 240 students, and the size of the difference is considered to be the one that cannot be ignored from the psychometrical viewpoint.