2021-08-04T00:59:46Zhttps://nagoya.repo.nii.ac.jp/oaioai:nagoya.repo.nii.ac.jp:000020942021-03-01T21:27:37Z<原著>3 相因子分析の適用上の諸問題SOME PROBLEMS IN THE APPLICATION OF THREE-MODE FACTOR ANALYSIS村上, 隆MURAKAMI, Takashi後藤, 宗理GOTO, Motomichi辻本, 英夫TSUJIMOTO, HideoThe main purpose of this study is to investigate the effectiveness of three-mode factor analysis in practical use. Some illustrative examples of the application of this method to real date are presented and some problems in the application processes are considered. In this study, three-mode factor analysis is interpreted as a special case of ordinary two-mode factor analysis, that is, as a factoring method with additional constraints on either the factor loadings or the factor scores. Concretely, Tucker's model is modified slightly in order to interpret the results of three-mode factor analysis, and an indexρ is introduced for estimating the efficiency of the three-mode method compared with the two-mode method. The data used as examples are classified into two types; the time series type and semantic differential (SD) type. The former set comes from the longitudinal study of social attitudes of secondary school students, and latter sets consist of the ratings of Chopin's music and Rorschach cards on 20 scales respectively. The main results are as follows : (1) In the analysis of time series data, three-mode factor analysis can reveal the factor structures consistent in time while the ordinary method can not find out these structures. Furthermore, three-mode factor analysis can provide the information about the changes of factor structures as a function of time. (2) Three-mode factor analysis is a useful method to treat the SD type data. This method considers the individual differences and it can distinguish between the mean profiles and the individual differrences. (3) Under the appropriate conditions, the efficiency of this method reaches 60% or more. Core matrix shows a meaningful pattern in any case. In general, three-mode factor analysis is considered to be useful as an exploratory method to understand the global structures of the data and to find the interesting hypotheses.国立情報学研究所で電子化したコンテンツを使用している。名古屋大学教育学部1978-12-28jpndepartmental bulletin paperhttps://doi.org/10.18999/bulfep.25.19http://hdl.handle.net/2237/3517https://nagoya.repo.nii.ac.jp/records/209410.18999/bulfep.25.1903874796名古屋大學教育學部紀要. 教育心理学科251939https://nagoya.repo.nii.ac.jp/record/2094/files/KJ00000137111.pdfapplication/pdf1.8 MB2018-02-16