Alternative Approaches to High-Stakes Testing

Abstract

UNDER No Child Left Behind (NCLB), the federal government requires state governors, superintendents, and school principals, through a regime of annual testing, to demonstrate to the taxpaying public that education dollars are being used effectively to improve student achievement. Developing, administering, and scoring the required assessments call for highly specialized skills and experience that states often lack, leading them to hire outside testing companies and consultants. Student performance on these tests--given at the end of the year to all students in grades 3 through 8--determines rewards and sanctions for schools, teachers, and students. But these standardized tests were not designed for accountability purposes, and experience in isolating and measuring the effects that schools have on student learning is rare. This test-based accountability system is being controlled by people who may know how to develop standardized achievement tests but know very little about the institutional realities of accountability--and even less about how to improve instruction in schools. For the most part, the assessments currently in use are not capable of accounting for the many non school factors that influence test scores, including student background, home environment, poverty level, and English-language proficiency. Such standardized tests are not always reliable measures of what is learned in the classroom, according to assessment expert James Popham. (1) And now NCLB has established these single, end-of-year tests as the dominant measures of school success or failure. HIGH-STAKES TESTING Tests used to make high-stakes decisions, especially in light of the fact that they are given only once or twice a year, necessarily have to meet a set of strict requirements. High-stakes tests must be sensitive--that is, they must be capable of determining changes in achievement related to instructional improvements. The validity of an accountability system depends on designing the right tests. These tests should meet the following criteria: (2) 1. Assessments used to measure student mastery of specific content must include clear descriptions--brief, jargon-free, and teacher-friendly--of what is going to be assessed. Classroom teachers need these descriptions to understand in detail what is expected of their students. 2. Effective assessments focus on a modest number of significant curricular aims, drawn from content standards. The selected content standards clearly must be of major importance. 3. An instructionally sensitive test to be used for accountability purposes must report student performance in a way that enables teachers to know what aspects of their instruction need to be improved and what aspects are working well. The tests currently being used to satisfy NCLB have been judged by the majority of teachers to be instructionally insensitive--incapable of measuring the effects of instruction on student performance. This can lead to tragic results because high-stakes tests can distort instruction and may encourage teachers to teach to the Teaching to a bad test and spending months on drill and skill may boost scores but surely ends up turning off students. The NCLB mandates for AYP (adequate yearly progress) and public reporting of results put enormous pressure on students, teachers, principals, and superintendents to raise test scores. This pressure can lead, in extreme cases, to cheating. The more general response is for teachers to practice on past (and even future) versions of the tests and to restrict instruction to just those subjects that will be tested; this is known as item teaching. These outcomes of high-stakes testing distort the traditional ideal of the teacher as one who makes every effort to achieve the goals of the curricula without regard to any particular test. The real blame for inappropriate forms of teaching in response to testing lies not with teachers but with state and national policy makers who create accountability systems centered on ever-higher test scores (AYP) with little regard for how these scores relate to better learning. …