Can Tests Lead the Way To Excellence?
The following is the first in a series of essays this spring that will serve as a prelude to Education Week's first Commentary special. "Toward a National Framework: Goals, Standards, and Tests'' will examine issues raised by a proposed national system of assessments. The special section has been made possible by a grant from the John D. and Catherine T. McArthur Foundation and will appear at the end of the publishing year.
By Gregory R. Anrig
We may be repeating a perennial mistake in education. Namely, that testing is the sure-fire way to produce high-quality instruction and learning in our schools.
If having a lot of tests in our schools was the answer, then we wouldn't have the problem of mediocre achievement that faces American schools today. By any measure, American students are not learning enough for the world that awaits them. The public recognized this 20 years ago. States responded to the public's calls for "back to basics'' in the early 1970's by introducing competency testing. By the end of the 1970's, 38 states had mandated competency testing. By the end of the 1980's, 47 states mandated such testing. The results? Except for some improvement in basic skills at lower levels of performance, especially by minority students, the best that can be said is that students in 1990 had generally returned to average achievement levels of their counterparts in 1970. The trouble with that, of course, is achievement levels in 1970 were nothing to brag about.
Despite this track record, we are again grasping at tests as the certified answer to educational excellence. On top of this, advocates expect the "new assessments'' to be used for promotion, graduation, employment, college admission, public accountability, school choice, financial incentives and penalties for teachers and schools--and for instructional improvement! As with competency testing in the 1970's and 1980's, we again expect one kind of measurement to serve all purposes, including decisionmaking that has high stakes for children and their teachers. Once again, teachers and students will find themselves caught in the middle, especially because so little attention in the education-reform movement is being given to assuring these teachers and students the opportunity to be prepared for what will be expected of them.
As taxpayers, we pay dearly for public schools and have every right to expect both accountability and improvement. However, to expect one test or test system, no matter how carefully designed, to serve both purposes simultaneously may result in neither purpose being well served.
First, as we decide what we want public accountability measurements to do, we should draw upon all appropriate techniques, depending on what we are trying to assess and the most effective methods for doing so. Unfortunately, some take the position that all that exists in testing today is bad, all that is proposed is good; all multiple-choice testing is bad, all performance-based testing is good. I don't see these as verities. Instead, I predict we will end up with a variety of methods, which will grow as we learn more in the years ahead. I recommend the following to policymakers today:
- Test for accountability purposes only at key transition points, such as grades 4, 8, and 12.
- Where numbers of students are sufficient, test on a sampling basis to reduce chances of misusing test scores to pigeonhole students into dead-end groups or tracks.
- Report test results to trace progress, rather than simply using them to rank students, schools, and districts.
- Balance assessment methods to efficiently gather important data at reasonable cost in terms of tax dollars and instructional time.
- Design state accountability assessments to be as complementary as possible to assessments used locally by schools for instructional purposes.
While it is fashionable today to criticize state testing programs, many states developed their tests conscientiously, with broad involvement of teachers, parents, employers, and others. If only to protect themselves legally, they took great care to be sure the tests measured knowledge and skills deemed appropriate for a grade level and consistent with the curriculum taught to students. Certainly mistakes were made, especially when teachers were inadequately involved or where test results were misinterpreted and misused. Nevertheless, a lot was learned about what tests can and cannot do.
Some states--California, Connecticut, Kentucky, and Vermont among them--are leading the way in developing new approaches to accountability testing. The National Assessment of Educational Progress has also become a promising proving ground for new performance assessments feasible for large-scale measurement of achievement trends among states, nationwide, and even internationally. (Náåð has been administered by the Educational Testing Service since 1983 for the U.S. Education Department's National Center for Education Statistics.) These initiatives are important for helping policymakers broaden their concepts of accountability testing.
But what kind of assessment will really help improve learning? If learning is to improve by the year 2000 (or any other deadline), we desperately need reforms that support teachers and students. The prime locus for improved learning must be the classroom and the prime resource for this change must be the individual teacher. To date, this has not been the case. Teachers generally feel bypassed by the education-reform movement and believe that reform is something done to them and their students rather than for them.
If we start with the classroom as the center of concern, I believe we can create effective "instructional assessments'' that are very different from those designed for public accountability purposes. These learning assessments would:
- Be performance-based--projects, portfolios, experiments, and other student-constructed responses--and, increasingly, involve the use of computers.
- Be fully integrated into the classroom curriculum, so that assessment and instruction are no longer seen as distinct functions.
- Provide continuous performance feedback that the student and teacher can use daily to plan next learning steps.
- Actively involve learners in tasks (often working with other students) that have meaning to them and will help develop in-depth understanding of what is being learned.
- Involve the teacher as a coach or mentor, rather than a dispenser of knowledge.
Several experts and measurement centers across the country have been working to develop instructional assessments. E.T.S. has developed prototypes in mathematics, science, writing, art, music, problem solving, computer skills, and critical thinking. Much has been learned from these efforts, and initial reactions of students and teachers have been enthusiastic. For example, a program in the Pittsburgh public school system, called ARTS PROPEL, combines assessment and instruction for 6th- through 12th-graders in music, art, and creative writing. Through development (and assessment) of individual work "portfolios,'' ARTS PROPEL encourages students to express themselves artistically and also to solve problems and develop their work from initial draft to finished product--all key elements of the "critical thinking'' process.
Initially developed by the Harvard University psychologist Howard Gardner, ARTS PROPEL does not dictate a certain curriculum to teachers; rather, it gives them tools for teaching and evaluating, and solid principles that can be broadened to apply to other subject areas. Many teachers say they have learned as much as their students through this process.
A critical aspect of instructional assessment is that it builds on past learning and leads toward future learning in the classroom. It is not a single activity, or test, conducted in isolation for external reporting.
Clearly, we must begin to understand that instructional assessment revolves around an entirely different set of issues and questions than traditional accountability measurement. Generally, instructional assessments ask these basic but decisive questions: What tasks will promote the desired learning? How are the standards for these activities shared with and understood by students so that self-evaluation and improvement can occur? How do smaller learnings fit into a framework connecting to larger learning goals? In what form can this information be organized for ready use by teachers to improve daily instruction?
To date, these questions have been missing in much of the assessment-reform movement. This will continue unless we separate our near obsession with accountability testing from our general desire for improved learning. As a national consensus builds regarding standards for each academic field, it will increasingly become possible to design new kinds of assessments that will help teachers teach better and students learn better. With that kind of resource at hand, teachers can move from the sidelines to the playing field of reform in making the new educational goals and standards a reality in the nation's classrooms.
Gregory R. Anrig is president of the Educational Testing Service in