13.2 Bayesian Network

-According to the National Research Council report knowing what students know (2001), assessment is an evidentiary system comprises of Cognition , Observation, and Interpretation. -Proficiency levels of examinees may not be measured directly. -The target of assessment is a particular information about an examinee’s knowledge, skills, and abilities, formulated with a certain purpose in mind. -Observations can be made based on examinees’ performance on tests with specific measurement objectives. -Examinees’ proficiency levels are cognitive and reside within the examinee’s mind; they are not directly observable. -Based on the performance on a test (evidence/observations), inference can be drawn on the examinee’s proficiency levels.

13.2.1 Evidence Centered Design

The ECD was developed to systemized the cognitive diagnosis as an “effective and valid evidentiary systems”.

ECD system is dependent on few issues i.e., we need to develop clear understanding of the purpose of the assessment, use of assessment results, and about the constructs of the assessment.

In general, ECD system requires the following;

-Identification of the constructs to be measured, and their expected associations. -Identify observable behavior which may represent constructs and generate items or activities to identify the constructs. -Design the evidence and inference engine to infer about students’ abilities. -The ECD components mostly represent how students develop knowledge and competence within a domain. -Furthermore, a high-quality assessment structure requires coordinated combinations of the ECD components.

Therefore, conditional dependencies among the ECD components are important, and hence, Bayesian network modeling comes in handy.