10/25/08

Columbus Dispatch

Thomas M. Stephens

School accountability, as practiced in Ohio, is a major barrier to improving public education, because it’s based upon a limited understanding of how youngsters learn and because its costs are excessive in both time and money: The state’s testing costs are $87.8 million a year, paid to out-of-state vendors. Gov. Ted Strickland held forums around the state all summer in an effort to link better schools with Ohio’s economic growth — a laudable objective but fraught with many barriers. The heavy reliance on high-stakes testing gives a distorted view of the quality of our schools, wastes valuable instructional time and resources and discourages teachers from personalizing their instruction.

Strickland has concerns about testing, including the Ohio Graduation Test. Students must pass all five sections of the test to receive a diploma. They get seven chances to pass, beginning in 10th grade. But the test’s predictive validity is unknown; policymakers can’t determine how well the results correlate with students’ future success, academic or otherwise.

The so-called value-added feature of the testing purports to control for students’ differences that are acquired outside the schools’ purview. But teachers instruct students in groups, and trying to control for differences statistically is questionable when instruction is time- and test-bound.

High-stakes testing forces a narrow approach to teaching and learning. Instead of improving instruction, the negative and unfair consequences of testing demean teachers, who feel they have to force-feed students to help them pass tests. Just try teaching a core subject to a class in which students range widely in their rates and levels of learning. Then try preparing them for tests whose results might smear you and your school’s reputation. If many students fail, you’re labeled a bad teacher.

The value-added feature draws upon statistical theory and uses equations to allow for students’ individual differences so as to estimate each student’s present and future performances. Prediction formulas try to account for missing data of those students who move. Among low-income families, mobility is high, adding to the complications. About 65 percent accuracy is considered good for long-term predictions, leaving 35 percent unaccounted for. Should critical decisions be made with this margin of error?

Here’s one alternative to high- stakes testing that assures all test results are instructionally useful to teachers and also inform students and parents:

• Students develop portfolios of their work, starting in the early years. These collections can include results of their creative accomplishments and the objective results of their achievements.
• Randomly selected students and their teachers are observed in the classrooms. Their written work is also analyzed. Performances are evaluated with immediate results available to them. This approach, called applied behavior analysis, is data-based and has wide application in schools. Trained observers are available from educational service centers and universities throughout Ohio. Random selection of students, over time, can provide an accurate estimate of effectiveness of schools and teachers.
• Core subjects use end-of-course tests created from the instructional content. Students who don’t demonstrate mastery can continue studying the material until they reach a satisfactory level. Time and teaching vary with students’ needs. Unlike the state-mandated tests, these will provide valuable results for re-teaching and learning and will be done in shorter time spans.

All of these methods and materials can be prepared at a reasonable cost in Ohio, and expertise is available within the higher-education system for this purpose.

Students’ and schools’ performances can be elevated without hammering them and their teachers.

Thomas M. Stephens is professor emeritus, College of Education and Human Ecology, Ohio State University, and executive director emeritus, School Study Council of Ohio.