In Study 1, 23 pupils nominated by their teachers as seriously aggressive were assigned to criterion group, and 567 non-aggressive to contrast group. They were among 1701 third to sixth grade pupils at eight elementary schools with 59 teachers. A problem-aggression scale was constructed, with 13 items that best discriminated the two groups. In Study 2, teachers rated aggressiveness of their pupils with a 5-point scale. Participants were 224 third to sixth grade pupils and eight teachers. Reliability of the teacher ratings was .93. The problem-aggression scale had a correlation of .46 with the ratings, and its test-retest reliability coefficient was .85. Measurement accuracy was high on the higher end of the scale when data of 814 pupils were analyzed with IRT. In Study 3, intervention with assertiveness training was experimentally examined. Thirtyeight and 35 pupils in the third grade were assigned to experimental and control groups, respectively. Assertiveness training had a weak therapeutic effect on the low aggressive pupils when a 3-way ANOVA was performed on the pre- and post-intervention scores.

Key words: aggressiveness, criterion-based scale, assertiveness training, elementary school children