Using Test Scores for Decision Making

Tests can provide information that facilitates and supports decision making, particularly in the domain of education. However, each type of decision, each type of inference from test data, and each model of decision making carries with it certain constraints and risks. In this chapter I will outline the principles of interaction between these aspects of test-supported decisions, detailing the risks, benefits, and limitations of each. In the first section I present a brief discussion of the various types of decisions in which test data can profitably be brought to bear. These include instructional decisions, guidance decisions, classification/placement decisions, educational selection decisions, licensure and employment decisions, and program evaluation decisions. The second section discusses the various types of validity evidence: content, criterion-related, construct, and instructional/curricular validity. The third section covers the relationship between the type of decision being made and the type of validity evidence required. For example, different evidence is required for guidance decisions than for program evaluation decisions; different evidence is required for selection decisions than for licensure decisions. The fourth section discusses the various models for combining data from different sources to make a single decision. The conjunctive, disjunctive, and compensatory models are described briefly and examples given where one model is more likely to function better than the other models.