School Testing

By John H Barton

This week students across the State are in the throes of the annual testing ritual. Each student in grades 2-11 will be tested on basic subjects of reading/writing and math and a few others based on grade level.

These exams, made up of norm-referenced and criterion-referenced components, will take 2 hours per day over about 7 days and form the STAR test. On testing days students will be encouraged to eat a good breakfast and bring a favorite toy or a special pencil to school.

Districts will reschedule the lawn mowers and quiet construction. But the value of the tests is suspect. They have limited use in the classroom, are statistically untenable, and are coalesced into an almost meaningless API score. Most importantly they have become the focus of teaching rather than a tool for improvement. And in a year in when schools are being asked to slash their budgets these tests will cost the State over $120 million. From a district, school and classroom perspective these tests have limited value as they are only marginally aligned to standards and do not test art, drama, music, or citizenship. In short they do not accurately reflect what we do and what we teach - a central tenet of a valid assessment tool.

Further a single test given at a single time representing an entire year's work is an unstable tool. Scores will vary as a result of illness, challenges at home, random noise and the like. Finally the data arrives at most Districts in late summer making diagnostic efforts difficult. Much of the test, the SAT 9, is scored on a norm-referenced basis. Thus it is designed to create a bell shaped distribution with 50% of students scoring above and below the 50th percentile - with most centered around the 50th percentile. Thus when the Governor announces each year that 51% of students have scored above the 50th percentile and that is not good enough he speaks of nothing except his ignorance of the test. The test was designed to create this result; it is statistically impossible, across a large enough sample, for more than 50% to score above the 50th percentile. In short the state's goal for student improvement + all above the 50th percentile is patently impossible with the SAT 9. Fortunately the test is slowly changing to criterion referenced components but we will lose much of the longitudinal data in the change.

The centerpiece of these tests is the Academic Performance Index or API. This score is the result of a weighting and averaging exercise of all scores across a school to create a school wide score. This statistical high wire act creates a score that is easy to repeat but has no value to educators. After all students learn - not schools. It does, however, mean a lot to the State.

Op/Ed on Testing * John Barton for Assembly

The API is so important that many schools are discouraged from teaching art and music or even social studies and science as those subjects take time away from test preparation. In many Districts the entire curriculum has been realigned to prepare students to perform well on the tests rather then to perform well in life - turning education and assessment on their heads. Valid assessments help determine whether the curriculum is effective and, in turn, suggest alternate courses; they ought not be the end in and of themselves.

To face down a record shortfall the Governor proposes to cut back on teacher training, textbook purchases, and other programs that support teaching and learning. But not one penny, except for performance awards, has been cut from testing. And the costs will grow as the testing is just beginning. Starting with the class of 2004 all high school students must pass the California High School Exit Exam to graduate. They must also pass individual subject area exams. All of this will be on top of locally given assessments, and the PSAT, SAT and Advanced Placement tests. By some estimates student in high school will have 80 hours of tests each year.

As a school board member I know that assessment is crucial to guiding the learning of students. Valid assessments are highly informative and crucial to continued improvement. But to be informative they must reflect actual learning, have something approaching immediacy and be focused enough to suggest areas for improvement. The tests being imposed upon my children and their classmates fail these criteria and need re-examination. Palo Alto Unified, which routinely scores well, has the leadership ability and perhaps the responsibility to begin resisting this testing regimen and encourage a more responsible one. Changes toward criterion referenced tests are good but we must reduce the overall testing load, receive results more qui