Shavelson, R. J., & Webb, N. M. (1991).Generalizability Theory: A Primer. SAGE Publications, Incorporated.http://ebookcentral.proquest.com/lib/harvard-ebooks/detail.action?docID=6636862

Generalizability (G) theory is a statistical framework used to evaluate the dependability of behavioural measurements by examining multiple sources of error simultaneously. Unlike classical test theory, G theory allows for the separate estimation of various error sources, such as differences across test items or testing occasions. This framework utilizes variance components to quantify the magnitude of each error source, aiding decision-makers in determining the necessary conditions for dependable scores. G theory distinguishes between generalizability (G) studies, which identify potential error sources, and decision (D) studies, which use G study information to optimize measurement designs for specific applications. A key aspect is differentiating between relative decisions, focused on ranking individuals, and absolute decisions, focused on an individual's level of performance. G theory ultimately provides a generalizability coefficient, akin to a reliability coefficient, indicating the accuracy of generalizing from observed scores to a person's average score across a defined universe of admissible observations.

keepSave to notecopy_alldocsAdd noteaudio_magic_eraserAudio OverviewschoolBriefing doc