ABSTRACT
There are many ways to address the design of automated scoring systems for complex constructed response tasks. This volume illustrates a variety of such methods, as well as their human scoring alternative, in terms of EvidenceCentered Design (ECD; see Mislevy et al., chap. 2, this volume). Some of these methods are familiar statistical techniques applied in innovative ways, whereas others are relatively new methods to the field of educational measurement. A common theme across all of these chapters is leveraging advances in technology and measurement methods to accomplish assessment goals that were previously impossible. This chapter continues that theme by presenting an application of Bayesian networks (BNs, also called belief networks, Bayesian belief networks, influence diagrams, relevance diagrams, causal probabilistic networks, and Bayes nets) for the evidence accumulation process of automated scoring of a web-based assessment prototype of computer networking ability that uses computerized simulation tasks.
