Generally speaking, there are two aspects in the judgment of the scale value of attitude. The one is that in which the scale values are judged in terms of agreement with the statements of attitude, and the other is that in which they are judged in terms of opposition to the statements. L. L. Thurstone has pointed out the reliability of the experimental scale value S dependent on agreement with the statements of attitude, but the present writer has been thinking that the attitude measurement in terms of the experimental scale value S' dependent on the opposition to the statements would be none the less effective, and that for that reason it may be of great necessity to find out a new formula which is able to measure the attitude in terms of the continuous variable which has both directions mentioned above. The attitude score L by R. Likert, though without any scientific and statistical foundations, has been relatively valid and reliable scoring on the basis of the attitude continuum with both directions of agreement and opposition. The present writer has experimentally examined the validity and reliability of such attitude indices as S, S', L, S-S' and S/S', and found out that S-S' is the most valid and reliable of them all. However, it must be admitted that S-S' is not free from some defects as its calculation is very complex and the opinions selected in its scale are quite few. In order to correct these defects, the present writer thought it best to apply S, S', and S-S' to the attitude scale measured by Thurstone's method of apparently equal intervals. Based on the experimental results, the present writer has concluded that Sr-S'r is the most valid and effective of, all these subjective rating scale values, Sr, Sir and Sr-S'r, measured by the method of apparently equal intervals. When examined statistically, S-S' has a significant difference. Such significant differences, existing not only in the area of attitude measurement but also in the form of perceptional judgment, necessarily call for further psychological investigations and interpretations.
It was previously found that the swings of the galvanometer in P. G. R. were closely associated with those words that were related to their past experiences on their present problems and conflicts, when the subjects were quietly told a story. Now two experiments have been so designed as to solve the problem of whether the galvanometer swings only in regard to words in a story given or not: first, to find the relation between the impressive words in a story read by the subject and the P. G. R. responses upon those words when they appear in a story told, secondly, to examine the P. G. R. responses upon the same words as given in different contexts. The results of the first experiment indicate that the expected P. G. R. responses were made to the impressive words and also that the responded words were related to the subjects' needs, the present problems or their characters, etc. It was found from the results of the second experiment that the P. G. R. responses were irrelevant to the same words given indifferent contexts and some swings were not associated With the important parts of the stories. The above findings point the way toward the possibility of finding a clue for the personalty diagnosis in the analysis of those words in a story told that are associated with the specific responses of P. G. R. despite of the fact that the P. G. R. responses are not always relevant to the words in a narration.