@article{oai:nagoya.repo.nii.ac.jp:00001531,
 author = {村上, 隆 and Murakami, Takashi and 野上, 康子 and Nogami, Yasuko},
 journal = {名古屋大學教育學部紀要. 心理学},
 month = {Dec},
 note = {We proposed a factor analytical method consisting of a series of procedures as a technique for constructing multiple scales defined as the simple sums of responses to a set of questions, and verified the appropriateness of the procedure at least partially by applying it to a few sets of real life data. Our procedure includes the following steps : 1) Find the matrix of (unrotated) principal component loadings by the eigendecomposition of the covariance matrix between variables instead of the correlation matrix. 2) Determine the number of components based on the method proposed by Velicer (1976) as well as inspecting the scree plot simultaneously. 3) Rotate the loading matrix by the use of the independent cluster rotation given by Harris and Kaiser (1964). We found that the Velicer's method of the determination of the number of components and Harris-Kaiser oblique rotation worked well in general, especially, from the viewpoint of the congruence of the correlations of principal components to those of the constructed scales. However, we did not find any clear evidences that the factoring of the covariance matrix is preferable to that of the correlation matrix. The results suggest that more comprehensive studies may be necessary to demonstrate the prevalence of our method., 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。},
 pages = {99--109},
 title = {心理学的個人差測定尺度構成のための主成分分析の使用について},
 volume = {46},
 year = {1999}
}