HKIER

Abstract

This paper reviews three distinct research programs in the field of performance-based assessment. The common goal of these programs represented in this analysis was to understand the technical qualities of performance-based assessment, and to develop design strategies towards such a technology. Score variability was shown to be the major problem for the new form of assessment. A cluster of design strategies were developed in these research efforts to address this problem in particular domains: the use of discourse-specific assignments and task-specific scoring in writing assessment; the use of explanation tasks with the same format and criterion-performance-based scoring criteria in history assessment; and the use of hands-on tasks with the same structure and scoring systems in association with types of tasks in science assessment. Common threads linking these programs suggest a set of design strategies for future efforts to develop quality performance-based assessments. These include: employing task specifications to improve representativeness and comparability of performance tasks and generalizability of the resulting test scores; reframing the objects of measurement to gain fine-grained understanding of performance variation in relation to task demands; and using the theories about content domains being measured to guide decisions with approaches to the design of tasks and evaluative criteria.

摘要

N.A.