Multifacet Universes of Generalization and D Study Designs

Abstract

In this chapter it is assumed that all facets in the universe of admissible observations can be considered infinite in size and, consequently, the G study variance components have been estimated for a random model. Given these assumptions, this chapter provides procedures, equations, and rules for estimating universe score variance, error variances, generalizability coefficients, and other quantities for random and mixed model D studies with any number of facets. The random model equations and rules are quite general. The mixed model procedures are somewhat simplified; more general procedures are treated in Chapter 5.