The observation corner of the assessment triangle represents a description or set of specifications for assessment tasks that will elicit illuminating responses from students. In a tutoring session, for example, the observation framework describes what the learner says and does, does not say and do, or says or does with specific kinds of support or scaffolding. In a formal assessment, the observation model describes examinee products, such as written or oral responses or the choice of a distractor for multiple choice items. In assessment, one has the opportunity to structure some small corner of the world to make observations. The assessment designer can use this capability to maximize the value of the data collected, as seen through the lens of the underlying beliefs about how students learn in the domain.
For example, on the basis of the cognitive model presented in Box 2–1, Siegler (1976) designed situations to observe which rules, if any, describe how a child is solving balance-scale problems. Asking children how they solved the problems might appear to be the simplest strategy, but Siegler believed that answers to such questions could either overestimate or underestimate children’s knowledge. The answers would give a misleadingly positive impression if children simply repeated information they had heard at home or in school, whereas the answers would give a misleadingly negative impression if children were too inarticulate to communicate knowledge they in fact possessed. In light of these considerations, Siegler formulated an observation method that he called the rule assessment method to determine which rule a given child is using (see Box 2–2).
The tasks selected for observation should be developed with the purpose of the assessment in mind. The same rich and demanding performance task that provides invaluable information to a teacher about his tenth-grade class—because he knows they have been studying transmission genetics for the past 6 weeks—could prove impenetrable and worthless for assessing the knowledge of the vast majority of students across the nation. Large-scale assessments generally collect the same kind of evidence for all examinees; thus observations cannot be closely tied to the specific instruction a given student has recently experienced.
Every assessment is based on certain assumptions and models for interpreting the evidence collected from observations. The interpretation corner of the triangle encompasses all the methods and tools used to reason from fallible observations. It expresses how the observations derived from a set of assessment tasks constitute evidence about the knowledge and skills being assessed. In the context of large-scale assessment, the interpretation method is usually a statistical model, which is a characterization or summarization of patterns one would expect to see in the data given varying levels of student