Performance standards clearly differentiate among levels. Performance descriptors should be easy to apply to collections of student work. When they apply the descriptors for the performance levels, teachers, parents, and students should clearly see why certain sets of student exemplars or student profiles are assigned to one performance level and not to another.
Performance standards are grounded in student work but not tied to the status quo. The system should reflect the major concepts and accomplishments that are essential for describing each level of performance. Student work that reflects the diverse ways in which various students demonstrate their achievement should be used to inform the descriptions during various stages of development, illustrating where students should be as a result of the educational process rather than where they are now.
Performance standards are built by consensus. The system of standards must be arrived at by the constituency who will use them. It must be built around agreed-upon statements of a range of achievement with regard to student performances. Not only should teachers and students understand the standards, but the “end users,” such as colleges and universities, technical schools, and employers, should also understand what performance standards mean for them.
Performance standards are focused on learning. Performance descriptors should provide a clear sense of increased knowledge and sophistication of skills. Descriptors that simply specify “more advanced” at each successive level are not particularly useful. The “more” should be clearly described or defined to show progression of learning. Cut scores on assessments must be based on this learning, and exemplars of student work should illustrate learning at each level.
SOURCE: Hansche (1998). Reprinted by permission of the Council of Chief State School Officers.
that systems of achievement standards should have, as conceived by Hansche, are shown in Box 4-5.
Three of the four components of the assessment system—labels, descriptors, and exemplars—should be created before assessments are developed. Bond (2000) argues that test development can be improved if test developers are given copies of the achievement standard descriptors and exemplars of what satisfactory performance looks like before test development begins. Test developers can use the content specifications, the assessment framework, and the performance descriptors to create items that assess all levels of performance.
The cut score—the numerical cutoff marking the divide between levels of performance deemed acceptable for particular purposes—is defined in the context of a particular instrument and is usually developed after the assessment is