The applicability of generalizability theory to social work research and practice

Social workers have traditionally relied upon classical reliability theory (CT) to evaluate the consistency of the measures that they develop and use. Generalizability theory (GT), which was developed by Cronbach, Gleser, Nanda, and Rajaratnam, has many advantages over CT, however. While CT lumps error into one aggregate, GT allows the researcher to identify and estimate the magnitude of multiple sources of error. Once these sources have been estimated, GT provides a means for measurement designs to be modified to minimize error. This paper outlines he advantages of GT over CT, and emphasizes how GT better allows for the expression of social work values in measurement. The addition of GT to social work curricula is recommended.