BOX 2–1 Example of a Cognitive Model of Learning for Assessing Children’s Problem-Solving Rules
Siegler (1976) examined how people develop an understanding of the components underlying the principle of torque in balance-scale problems. He presented children of different ages with the type of balance scale shown below, which includes a fulcrum and an arm that can rotate around it. The arm can tip left or right or remain level, depending on how weights (metal disks with holes in them) are arranged on the pegs on each side of the fulcrum. However, a lever (not shown in the figure) is typically set to hold the arm motionless. The child’s task is to predict which (if either) side would go down if the lever were released.
Two variables influence the outcome: (1) the amount of weight on each side of the fulcrum and (2) the distance of the weight from the fulcrum. Thus the keys to solving such problems are to attend to both of the relevant dimensions and to combine them appropriately by using the multiplicative relationship of weight times distance. On the basis of his research, together with the known tendency of young
need not take into account every subtlety and complexity about learning in a domain that has been uncovered by cognitive research. Instead, what is being proposed in this report is that assessment design be based on a representation or approximation of cognition that is both consistent with a richer psychological perspective and at a level of detail sufficient to accomplish the job of assessment. Any model of learning underlying an assessment will be a simplification of what is going on in the mind of the examinee and in the social situation within which the assessment takes place. As described and illustrated more fully in Chapter 5, the point of basing assessment on a cognitive model is to focus the assessment on those competencies that are most important to measure in light of the desired inferences about student learning.
Finally, if the goal of basing assessment on an appropriate model of learning is to be realized, cognitive models will need to be developed for a broader range of the curriculum. Currently, cognition and learning are con-