There have been few studies which examine the relationship between family images based on the adjective rating scale and maladjusted behaviors in adolescence. The biggest one in several reasons is that the SD method standard doesn't seem appropriate to assess a family image clinically, at least as far as the factors extracted in several precedent studies are concerned. Therefore, the purpose of this study is following: The first (study 1) is to compose a bipolar adjective rating scale to assess a family image. The second (study 2) is to examine the relationship between family functioning image and tendency to maladjustment in adolescence by the scale established in study 1.

　　Study 1: In this study, in terms of mental function, suppose family function is defined as two dimensions, “affection” and “socialization”, the former is regarded as “emotional bond among family members (Kizuna-kan)” and the latter is “flexible discipline for children (Jyunan-sei)”. The scale consists of 16 pairs of adjectives corresponding to the two contents. The results of factor analysis, G-P analysis, and test-retest method targeted for 468 high school students (1st and 3rd grade; 179 boys and 289 girls) show sufficient internal validity of the scale.

　　Study 2: According to the family image score, the same subject has been divided into four groups. Significant difference can be seen at p＜.01 level statistically as a result of one-way analysis of variance where each score of “tendency of PST-Ⅲ concerning asocial/antisocial behaviors” is a dependent variable, and groups are factors. As a result of multiple comparisons by the LSD method, there is a parabolic relationship between groups in two dimensions. That is, it is confirmed that subjects in the middle image level have the least tendency toward maladjustment, just like Olson's hypothesis of Circumplex Model (1979).