Ken McElroy Online

School Testing, the Scientific Way

Ken McElroy
February 26, 2001

Politicians from Democratic California Governor Gray Davis to Republican President George W. Bush have made student testing in public schools a centerpiece of their education plans. President Bush's plan would require testing of children every year in math and reading from the third through the eighth grades, and would tie federal education funds to test results.

Many critics of standardized tests believe that testing comes at the expense of learning, that students are "taught to the test," and they are drilled with test-taking strategies rather than learning core subject matter.

This seems to be a valid concern. It is quite possible that a student could learn enough to score well on a standardized test, but not know anything that is not on the test, but that they nevertheless should know to be well-educated, productive citizens. There was some evidence of this phenomenon in Texas - a Rand Corporation issue paper released in October 2000 found that major improvements by Texas students in tests administered by the state are not reflected in national exams given to the same students. The Bush campaign at the time disputed the paper's conclusions.

These concerns do not argue against testing however. They only argue against certain testing methodologies or implementations. The focus must remain always to educate the students, and any testing is merely one tool to measure student progress. Testing must not be an end in itself.

The first objective is to develop a curriculum, an outline of knowledge that every student is expected to have upon completion of each grade level. Only after we decide what students should know can tests be developed to measure what they do know.

When a new drug is being tested, the scientific method requires that the studies be done "blind." That is, the study subjects don't know who is getting the real drug, and who is getting a placebo. The objective is to measure real world results without introducing any bias due to the testing process itself. Of course, we aren't going to give some kids a "placebo" test, but we can use some aspects of the scientific method in school testing to measure real world knowledge rather than the ability to score well on a particular test.

The students and even their teachers should not know in advance the details of test format or content. All they would have in advance would be standard curriculum outlines detailing what students would be expected to know about various subjects. Neither the teachers in the classroom nor the students should even know very far in advance precisely when the testing would be conducted. Without prior knowledge, teachers would be forced to teach subject matter more broadly, not narrowly tailored to a particular test.

Once you decide that students in the fifth grade need to know long division, for example, you develop a test that accurately and fairly measures ability to do long division, and then administer the test to students. There is no need to tell teachers or students specific test methods or content, only that they will be required to know long division. There wouldn't be any classroom time preparing for the test, teachers would have to teach the subject. What could be simpler?

Of course, there will be some objections. Some may only be concerned with showing improved test scores, even if those scores do not mirror any real increase in student achievement beyond the test itself. This will be especially true if teacher compensation or school funding are tied to test scores. But parents, educators, political leaders, and others who really want to measure what children are learning in school should look to the scientific method of testing. Students have had enough placebos in school already.