The results of the study on social factors of 50 Korean youngsters in Japan who have committed offenses are as follows;

1. In most of their families, the value system of Confusianism is dominant. Most youngeters adapting to this system have a strong selfcontrol over committing offenses. On the other hand, in the families in which the transition from the first generation to the second generation is being occurred, youngsters have a rather weakened control.

2. The pressure of discrimination from which they suffer differs in accordance with the condition of their residence and the type of occupation in which their parents engage. In case the first generation people recide in a small concentrated area and live in their own pattern of life, the pressure of discrimination to the second generation becomes maximum. Many of Korean youngsters who are brought up in such an environment are inclined to have a strong hostility against Japanese in general.

3. In spite of the conditions mentioned above, however, the majority is trying to solve the conflict as a marginal man by means of L. Child’s “apathetic” reaction.

The purpose of the present study was to test the validity of the prediction table, which was introduced in “The prediction of Juvenile Delinquency” (1961) by T. Tatezawa.

This prediction table was applied to two groups of delinquent children, ranging in age from 8 to 15, who had been treated in Hokkaido Central Child Guidance Center. One group was composed of 100 boys who had been confined in training school, and the other group was composed of 100 boys who had not fallen again into delinquency for 18 monthes after the first delinquency.

These 200 cases were rated on the seven social factors of the original prediction table, i. e. employment of father, affects of father, supervision by mother, discipline by mother, cohesiveness of family, subject preference, and companions.

The results were as follows. The two score-class table, cut between 349 and 350, as suggested in the Tatezawa report, identified correctly 99 per cent of the training school samples, but also identified incorrectly 82 per cent of the first delinquent samples as the recidivism ones.

The accuracy of prediction (58.5%) counted from our results was extremely low as compared with the original accuracy of prediction (93.4%). Between Tatezawa’s samples and ours, of course, there were some differences of personal and environmental conditions. But, as a result of examining these conditions, we could not help coming to the conclusion that in making the original table for the prediction of recidivism it had not been fit to make use of non-delinquent juveniles as the control group.

Since seven social factors adopted in the Tatezawa prediction table were also significantly different between our groups, we attempted to modify the sub-categories of “employment of father” and the weighted failure scores of all constructed factors. and made a new prediction table of an accuracy of 80.5 per cent.