Serve Learning First, Accountability Second
With House passage of the Elementary and Secondary Education Act reauthorization, Congress is well on its way to implementing the Clinton Administration's school-reform plans. As passed by the House, the legislation contains one measure deeply desired by all advocates of assessment reform: elimination of the requirement to administer a norm-referenced test annually to all children in Chapter 1 compensatory education. Because norm-referenced tests have narrowed classroom practice in ways that hinder student learning, this action removes a major obstacle to improving assessment, curriculum, and instruction.
The House bill does require states to evaluate how well schools and districts are educating children served by the new Title 1. As in current law, student assessment is mandated in order to hold systems accountable for student performance and provide grounds for intervention where progress is inadequate.
Specifically, the House bill mandates annual student assessments at three grade levels that will "provide individual scores,'' "be comprised of multiple, up-to-date measures,'' and meet "relevant ... professional and technical standards.'' They are to be based on the content standards, and there is an implication that they will be more performance-based. As the Senate takes up the legislation, it should carefully study the likely consequences, positive and negative, of these assessment requirements.
Inherent in the House legislation is a centralized command and accountability structure through content and performance standards and assessments. But is this accountability model adequate to what should be the essential task, making schools places that help all students do well in a rich and empowering curriculum while participating in a supportive learning community? If the assessments mandated by the House perpetuate essential problems from the old norm-referenced tests, then many school practices and structures that impede learning are apt to remain, and the hoped-for improvements will not occur. Unfortunately, this is an all-too-likely result.
Problems with norm-referenced tests reside in the nature of the instruments and in their impact on curriculum and instruction. First, the tests are referenced to "norms,'' not standards. They are structured for the purpose of sorting students on a curve, not ascertaining whether students have learned what society has deemed important. Second, the actualtests are low-level, measuring discrete bits of information or "basic'' skills through multiple-choice questions. Many fundamental aspects of learning are not measured at all. Third, once the tests are made important through their use for accountability, they cause a narrowing of curriculum and instruction or at least inhibit improvement, both because of the low levels that are measured and because the tests offer a terrible model of instruction and learning--passive ingestion of information to be regurgitated on demand.
The House bill offers only a partial solution to these problems. The theory behind the legislation is that since what is tested is what is taught, new tests based on standards should produce significant changes in the classroom. If the standards include higher-order capabilities, then the assessments should measure them, and teachers should teach them. There are, however, major problems with this approach as developed in the legislation.
By requiring individual scores for accountability purposes and by defining new assessments merely as "up to date,'' the legislation allows states to get by with marginal changes in current tests. This requirement also creates financial and technical problems that could limit use of performance assessments.
Obtaining reliable individual scores through on-demand performance exams requires administering many tasks, which then requires large sums of money to pay for scoring them. While there can be substantial benefits in using such assessments, such as modeling good tasks for instruction, measuring higher-order capabilities, and being a vehicle for professional development, the costs are likely to inhibit their use.
The most likely result will be multiple-choice tests with a few "on demand'' performance tasks. Though some essential capabilities, such as the ability to create knowledge or use it in novel circumstances, will be inadequately measured, test-makers and policymakers are likely to claim that these tests are close enough, and given the costs, good enough.
Because assessments are always selections from the whole, teaching to the test--which is effectively encouraged by the legislation--runs the risk of substituting the measured part for the whole. Limited measurement will perpetuate the tests' harmful effects on curriculum and instruction and continue to inhibit fundamental change. "Enhanced'' multiple-choice items and short-answer questions also will continue to present a model of curriculum and learning that is at odds with in-depth understanding, critical inquiry, and how students actually learn. The negative side of "what is tested is what is taught'' will continue.
The assessment purpose, program evaluation, logically does not require producing individual scores. Sampling provides sufficient data, allows more extensive use of performance tasks that can be useful instructional models, enables broader measurement and less narrowing of the curriculum, and could be based on portfolios. Yet the Administration and a number of states fear that without individual scores students won't try hard, which will result in misleadingly low scores on which schools or districts will be held accountable. This reasoning ignores the fact that currently many students don't care about individual scores unless the stakes are immediate or enormous, neither of which pertains to Title 1 accountability assessments.
Thus, there is no compelling rationale for individual scores on accountability tests. More important, the requirement for individual scores is likely to induce most states to choose mostly-multiple-choice tests as their "up to date'' accountability assessments. In turn, the tests will undermine the attainment of the very standards the assessments are supposed to measure, and with that the basic purpose of the legislation.
A fundamental solution to the problems caused by testing would involve placing assessment in the service of assisting instruction and learning ahead of assessment for accountability. In this approach, the central question would be how to make all schools communities of active learners. The assessment questions would be first, how can assessment help this process; and second, how can we be sure these communities of learners are progressing and that the students are developing a core of important knowledge and capabilities?
An approach based on portfolios, projects, and exhibitions would better enable teachers to help students learn, students to assess themselves, and teachers to improve their practice, as well as provide a rich information base for accountability purposes. The state and district would need to sample from portfolios to validate teacher judgments and provide a check on biased or inadequate teachers or poorly functioning schools. The state could supplement this approach with an on-demand sampling assessment.
The House legislation does not prevent states from providing individual scores via portfolios, but since it offers no real encouragement for using assessments to enhance the learning process, it provides no impetus either. It also is not clear whether using local portfolios for individual scores and state sampling exams for school accountability would be permissible.
A portfolio-based system would require massive professional development and probably could not succeed without simultaneously changing curriculum, instruction, and the structure of the school. To some extent these are the goals of the Elementary and Secondary Education Act, but the connection of teacher improvement to assessment in the legislation is tenuous, at best.
The Senate is unlikely to focus on building communities of learners rather than concentrating on standards and accountability assessments. However, it could make several changes that would enable states and districts to go beyond the conceptual limitations of the House and Administration legislation while still providing accountability:
- Include a better definition of the assessments than "up to date.'' Assessments should comprehensively assess student learning in light of standards and model high-quality curriculum and instruction.
- Drop the requirement for individual scores from accountability assessments while requiring individual-progress data for each student. This could come from local assessments. Thus, states could develop systems in which individual student scores are provided by local portfolios with school and district data provided by state sampling from the portfolios or exams.
- Require districts to explain in the strategic plans called for by the legislation how they would insure implementation of classroom-based performance assessments to help instruction and learning. No specific assessment practices should be mandated.
- Establish interconnections between professional development and assessment. Practices such as group scoring of assessments by teachers and collegial discussion of assessments provide excellent opportunities to improve teaching.
- Require that assessments be fair and unbiased. The House does not require this, falsely assuming that adherence to professional standards of validity adequately addresses the problem. The Senate should also require that learning to assess in an unbiased manner be part of teacher professional development.
As passed by the House, the E.S.E.A. reauthorization provides important gains and opens new opportunities. It is not yet the legislation students need and deserve, but it moves in the right direction. A few key changes by the Senate would make assessment more useful in improving education and eliminate foreseeable harmful consequences.
Monty Neill is the associate director of the National Center for
Fair & Open Testing (FairTest) in Cambridge, Mass., and the
co-author of Fallout From the Testing Explosion.