Abstract: The aims of this study were to explore if the loadings on the first un rotated factor can be taken as a way to evaluate the task complexity of the balance scale task and to explore the validity of the prior analysis on the balance scale task based on the Relational-Representational Complexity Model. 264 children from grade 4, 5 and 6 were presented with a computerized balance scale task. There were 51 items and six item types, including weight problems, distance problems, weight distance problems, conflict weight problems, conflict distance problems, and conflict balance problems. Results showed that: (1) the loadings of all balance scale items on the first un rotated factor and the difficulty of those items had no significant positive relation. It might be that task complexity could be quantified by the loadings on the first un-rotated factor only when the difficulty of task was moderate. (2) The analysis of task complexity based on the Relational-Representational Complexity Model could predict task difficulty (R²=0 95), which indicated that the model was valid in analyzing the complexity of the balance scale task.

Key words: Relational-Representational Complexity Model, task complexity, task difficulty, factor analysis