The results of standardized tests are misused and often do more harm than good, according to William W. Cooley, professor and researcher at the University of Pittsburgh (Pennsylvania). The results seldom assist anyone in improving instruction. Too many out-of-school variables such as social-economic status influence test scores. Achievement is influenced by family background and what the student is exposed to in class; knowing these variables will tell more than tests can. Students tend to learn what they are exposed to; therefore, tests must measure what is being taught in order for them to be useful. Too often this is not the case. The effectiveness of the teaching or curriculum is questioned when, in fact, what is being taught is not being tested. In evaluating programs, schools seldom use tests that measure what the developers claim they are putting across in their materials. Additionally, the tests do not measure the knowledge that a student already possesses or how much time is spent in learning. Summative evaluation isn't answering the questions it is supposed to. All of this is already apparent to practitioners. Given the present state of evaluation, trained classroom observers are much better equipped to explain what works than any available test. (DC)