ABSTRACT
This chapter describes an approach to the design of an authoring system intended to support nonassessment experts in the design of problem-solving assessments. The approach is based on an explicit representation of how the interactions among problem-solving variables modulate task complexity. Performance Assessment Links in Science (PALS) has been one of the few efforts to adopt a systematic vetting procedure to ensure that the assessments housed at the site meet a set of standards related to validity and reliability. More commonly, assessment sites are primarily warehouses for tasks with undefined technical properties. The assessment framework is based on Baker and O'Neil's approach to designing problem-solving assessments using computer technology. This approach first characterizes three types of problem-solving tasks: a task in which an appropriate solution is known in advance, a task in which there is no known solution to the problem, and a task that requires an application of a given tool set to a broadly ranging set of topics.
