Generalizability theory was developed as an overarching structure with which to evaluate the dependability of psychological measurements. It is meant as an alternative to the correlational analyses which are typically used in classical test theory. It is also meant to provide a unifying theory for evaluating the psychometric soundness of a test. Both reliability and validity are considered to be part of the same conceptual framework. The value of the generalizability score is the result of manipulating different sources of error. The statistical analysis model used is ANOVA.
In developing a generalizability model for evaluating a clinical assessment instrument, the researcher would first determine the conditions or facets which are deemed to be relevant to that assessment instrument. For example, if the relevant facets are temporal stability and examiner influence, the experimental design would involve obtaining test scores on more than one occasion and...
References and Readings
- Shavelson, R. J., & Webb, N. M. (1991). Generalizability theory: A primer. Newbury Park: SAGE.Google Scholar