Fixing what’s wrong with testing in K-12 education

Read, watch or listen to the news and you’ll frequently hear concerns expressed about educational testing, particularly that there is too much of it in our schools. Students, parents and teachers don’t see the value and, indeed, often believe there’s harm caused by the types of standardized assessments typically used in schools today. The harm perceived by parents is significant enough that some have chosen to have their children opt out of taking state tests.

Those assessments are perceived to take too much time from instruction and, perhaps much worse, cause students and teachers to focus their attention on only a narrow portion of the intended curriculum. This curricular narrowing has been said to result in an overemphasis on facts and procedures to the exclusion of how, when, and why to use those facts and procedures in problem solving. Such assessments are used primarily to provide accountability information that is useful to local, state, and federal policy makers, but offer very little value for classroom purposes.

If thoughtfully designed, however, assessment can be a positive force for teaching and learning, and ETS’ CBAL research initiative is based on that idea. Our goal is to illustrate new approaches to technology-based testing in the hope that some of those approaches will be picked up by state testing programs and the companies that help to implement those programs.

CBAL stands for “Cognitively Based Assessment of, for, and as Learning.” This name highlights three key goals which we believe any modern assessment system must strive to achieve. The first goal is to document what students know and can do. Such documentation is essential if local, state, and national policy makers are to understand the condition of education and act to improve it.

The second goal is to help teachers plan instruction and students regulate their learning. Different types of tests and assessment processes are best suited to these two goals; standardized tests are more attuned to the former goal and informal classroom processes to the latter one. One innovation of CBAL is designing these two very different approaches to assessment so that they work together as a system. They reinforce one another by being drawn from the same curriculum standards and by providing complementary information, more finely detailed for the classroom teacher and more broadly described for the policy maker.

What about assessment “as” learning? CBAL’s third goal is to provide a worthwhile educational experience for students and teachers as part of assessment, something that standardized tests have never been designed to achieve, but that we believe can now be accomplished by such assessments. Assessments should be designed to model good instructional practice for the teacher and good learning practice for the student.

Successful learners have internal criteria against which they evaluate their own work as well as the work of others. CBAL’s experimental English language arts assessments include such criteria. For example, we provide students with a variety of source materials on a topic—which may include text, video, and audio. We then give them criteria for what constitutes a good argumentative essay, present them with an example of a faulty one, and ask them to identify how that essay violates the criteria and where those violations occur. We then ask the students to construct their own essays from the provided sources. These same criteria appear repeatedly in CBAL standardized assessments, as well as in CBAL classroom assessment and teacher professional development materials. We present the criteria repeatedly so that students will internalize them by making them a “habit of mind.” We hope teachers will internalize them too, incorporating them into their routine instructional practice.

A final innovation that CBAL has pursued is to measure curriculum standards both broadly—which current tests do—and deeply, something state assessments have been widely criticized for not doing. One way that CBAL tries to achieve this goal is by including questions that are very much like the ones teachers prefer for instruction. CBAL English language arts and math assessments use a “mini-project” structure that not only targets facts and procedures, but also the conceptual understanding needed to use those facts and procedures in real-world contexts. These mini-projects may call upon students to use computer simulations, integrate information from multiple sources, and critically evaluate websites—activities they routinely do as part of classroom learning and that many professionals do in the workplace.

An important lesson from seven years of CBAL research is that we can’t measure student learning both broadly and deeply with a short test. Creating assessments that measure subject matter areas well and that exemplify good practice takes more student time, not less. But if assessment can be made to be an educational experience for students and teachers, that additional time could provide very significant learning returns.

Randy Bennett, Educational Testing Service

Randy Bennett is the Norman O. Frederiksen Chair in Assessment Innovation at ETS.