Generalizability theory
Keyword(s):
This chapter is a detailed review of generalizability theory (G theory), an extension of classical reliability theory originally devised by Cronbach. The basic concept is that any measurement contains multiple sources of error, and through the use of G theory these various sources can be calculated in a single study. This permits the researcher to examine the relative magnitude of different sources of error and the relation among them. The power of the method rests in its ability to use these estimates to devise optimal strategies for distributing observations. That is, G theory can be used to determine how a fixed number of observations should be distributed across raters, occasions, or cases to optimize reliability.