Abstract
On investigating the questionnaire-type diaguosis of personality, we find a methodological problem. It is that responses to the test items of the questionnaire-type instrument are not mutually independent, but a correlative reflection, as a whole, of the subjects' inner state.
For the function of the each item to subjects is determined by the subjects' inner state, it does not always agree with the test-maker's expectations If this is proved by the further investigation, this methodology of diagnosis of personality will lead to a serious problem.
In this paper, we dealt with one of the most standardized questionnaires YATABE-GUILFORD PERSONALITY INVENTORY, all the items of 12 scales of which the inventory is made were correlated with each other item.
Results of our investigation follow.
1) In the correlation coefficients among all items, the number of comparatively high correlations in the different scales were greater than the correlations in the same one.
2) As a result of the centroid factor analysis of correlation matrix on all the items, we interpreted 75.2% with three factors which were identified. And these factor loadings did not correspond with the 12 scales of Y-G personality inventory.
3) According to the result of factor analysis by principal axis method of each scale, we explained two scales above 70% by the first factor, but one of them has a high rate of contribution in second factor.
4) As the result of re-test of the internal consistency of all the scales, we find no scale of which all the items correlate above 0.5 with scale score.
These results clearly suggested that the methodological problems are important. Nevertheless we will continue with further investigation toward the determination of measurable scales on making clear the nature of the problem.
