Evidence Based Assessment – Policy, Principle and Practice Perspective

Proceedings of The 5th International Conference on Modern Approaches in Science, Technology & Engineering

Year: 2019

DOI: https://www.doi.org/10.33422/worldcre.2019.08.459

[Fulltex PDF]

Evidence Based Assessment – Policy, Principle and Practice Perspective

Vineet Joshi, Sadhana Parashar and Priyanka Sharma



Application of general principles of drawing conclusion based on evidence forms the foundation of Evidence-based Assessment (EBA). EBA is not a novel method or technique or model rather it is a systematic approach of constructing educational assessment based on evidence-driven reasoning from learning sciences, cognitive sciences, measurement models and other sources of evidence from assessment related disciplines. Educational assessments of all kinds try to draw reason from test takers and draw conclusions about their knowledge and abilities. Rooting these conclusions on evidence and using them to support conclusions is what forms the core of EBA. Over the years, several assessment methods, including latent trait theories like IRT have evolved to improve the validity of conclusions drawn from the test scores, in a more scientific manner. This paper discusses the significance and usage of variety of evidence from various sources, at different stages of educational testing to improve the quality of the test, especially validity of the conclusion drawn on the basis of the test scores about test takers. It is not limited to, but briefly touches the principle and practices of Evidence-Centered Design (ECD), and tries to establish linkages at various points of the testing cycle. It extends to using evidence from measurement models from Classical Test Theory and Item Response Theory. The intent is to produce usable evidence and present evidence in simpler form to users, who may be great subject experts and instruction designers or policy makers but may not be great measurement experts.

Keywords: Testing, Evidence, Psychometrics, ECD, Measurement Models.