ABSTRACT
In one way or another, the four chapters on which I comment here are devoted to making tests meaningfully connected with processes of learning and performance of cognitive tasks—more meaningfully, that is, than may have been the case for earlier generations of tests. Actually, making tests meaningfully connected with learning is hardly a new idea. It was present even in some of the earliest modern discussions of standardized achievement tests, as in a text by Monroe, DeVoss, and Kelly (1917). The idea of testing for understanding was central in the work of Ralph Tyler and his associates in developing the highly innovative educational achievement tests used in the so-called Eight-Year Study of the Progressive Education Association (Smith & Tyler, 1942). The theme of making tests to measure understanding rather than merely factual knowledge occurs every so often in the series of handbooks of educational measurement sponsored by the American Council on Education (Lindquist, 1951; Linn, 1989; Thorndike, 1971), but without great emphasis or a singular focus. The problem, it seems, is that it has been very difficult for test constructors—whether they be full-time professionals at the job, or teachers or professors—to move away from easier approaches and to develop evaluation procedures that will assess students' deep comprehension of concepts in mathematics, science, history, literature, or whatever. Developing such procedures requires much effort, time, and creativity—elements that are often in short supply in educational measurement work.
