information about the attainment of individual students, as well as comparative information about how one individual performs relative to others. This information may be used by state- or district-level administrators, teachers, parents, students, potential employers, and the general public. Because largescale assessments are typically given only once a year and involve a time lag between testing and availability of results, the results seldom provide information that can be used to help teachers or students make day-to-day or month-to-month decisions about teaching and learning.
As described in the National Research Council (NRC) report High Stakes (1999a), policy makers see large-scale assessments of student achievement as one of their most powerful levers for influencing what happens in local schools and classrooms. Increasingly, assessments are viewed as a way not only to measure performance, but also to change it, by encouraging teachers and students to modify their practices. Assessment programs are being used to focus public attention on educational concerns; to change curriculum, instruction, and teaching practices; and to motivate educators and students to work harder and achieve at higher levels (Haertel, 1999; Linn, 2000).
A trend that merits particular attention is the growing use of state assessments to make high-stakes decisions about individual students, teachers, and schools. In 1998, 18 states required students to pass an exam before receiving a high school diploma, and 8 of these states also used assessment results to make decisions about student promotion or retention in grade (Council of Chief State School Officers, 1999). When stakes are high, it is particularly important that the inferences drawn from an assessment be valid, reliable, and fair (American Educational Research Association, American Psychological Association, and National Council on Measurement in Education, 1999; NRC, 1999a). Validity refers to the degree to which evidence and theory support the interpretations of assessment scores. Reliability denotes the consistency of an assessment’s results when the assessment procedure is repeated on a population of individuals or groups. And fairness encompasses a broad range of interconnected issues, including absence of bias in the assessment tasks, equitable treatment of all examinees in the assessment process, opportunity to learn the material being assessed, and comparable validity (if test scores underestimate or overestimate the competencies of members of a particular group, the assessment is considered unfair). Moreover, even when these criteria for assessment are met, care must be taken not to extend the results to reach conclusions not supported by the evidence. For example, a teacher whose students have higher test scores is not necessarily better than one whose students have lower scores. The quality of inputs—such as the entry characteristics of students or educational resources available—must also be considered. Too often, high-stakes assessments are used to make decisions that are inappropriate in light of the limitations discussed above.