A Comparison of Multi-Attribute Generalizability Strategies

Publisher Website
Google Scholar
Abstract
Two generalizability theory strategies are appropriate for analyzing multiple attributes: the first is an integrated design that includes the attributes as a facet; the other involves a strategy of analyzing variance and covariance components on "specific" attributes. Sums of squares, mean squares, and variance components of the integrated design are shown to be simple functions of their specific design counter parts. Of particular interest is a relationship showing that integrated design variance components not involving attributes are identical with the average respective covariance components across attributes. An illustration is provided to demonstrate the relationships, strengths, and weaknesses of each strategy.

This publication has 3 references indexed in Scilit: