It is a propitious time for a move toward coherent assessment systems, Herman observed. The Race to the Top funding, the opportunity for states to sign on to the common core learning standards, and converging confidence in the potential of new kinds of assessments—particularly formative assessments—combine to produce an important window of opportunity. Fortunately, Herman said, there is a strong body of research on which to base new approaches.1
Herman delineated key elements of a coherent assessment system: that it is a system of assessments, not a single assessment; that it is coherent with specified learning goals; and that its components collectively support multiple uses in a valid manner.
On the value of a system, as opposed to a single assessment, Herman noted that most tests used for accountability purposes today target only a limited subset of the learning goals that school systems set for their students. “If we want to know whether kids can write,” Herman observed, “we need something more than an editing test with multiple-choice questions. If we want to know whether kids can innovate, engage in inquiry, or collaborate with others, again, multiple-choice or short-answer tests are not giving the depth of information that we really need.”
This is critically important not only because tests communicate what it is important for students to learn, as was emphasized throughout both workshops, but also because their results will only support sound decision making if they provide a rich picture of what students know and are able to do. Thus, by moving from an exclusive reliance on multiple-choice and short-answer items to systems that also include performance and other kinds of assessments, states can better serve accountability purposes: they will be able to answer questions about important capacities not well addressed by current tests, such as depth of thinking and reasoning, the ability to apply knowledge and solve problems, the ability to communicate and collaborate, and the ability to master new technology. At the same time, the kinds of measures that teachers need on a daily basis in their classrooms are quite different from the annual or throughcourse kinds of measures that policy makers use to monitor progress on a more macro level.
Assessment systems are able to provide that rich picture if they are coherent with established goals for learning. This is not a controversial idea, Herman observed, but she distinguished four types of coherence. The most fundamental type of coherence is that among models of how students learn, the design of assessments, and the interpretation of their results, as illustrated in Figure 5-1. In Herman’s view, the fundamental coherence most states have now is orga-