Abstract
The purpose of this study was to test the hypothesis that high (low) self-esteem children modeled after a lower (higher) SR criterion model in a multiple condition. Based on their self-esteem rating scores 48 third-graders were selected and divided randomly into two modeling conditions: a HL condition in which subjects viewed a video showing both an H model (high SR criterion) and an L model (low SR criterion), and an M condition that showed children only an M model (middle SR criterion). Both groups contained high and low self-esteem children. The number of tokens taken as self-reward in a task which was the same as that shown on the video was analysed. The following were the main results. The high self-esteem group in the HL condition took more tokens for their high score than the high one in the M condition. But the low self-esteem group in the HL condition tended to take fewer tokens for their high score than the low one in the M condition. In the HL condition, the low self-esteem group took fewer tokens than the high one. These results supported the hypothesis.
