Generalizability Theory

Learning Objectives

·          Understand when it is appropriate to use Generalizability Theory (section 1)

·          Understand how the variance, using various designs, is partitioned (section 2)

·          Understand how to calculate measurement error using Generalizability Theory (section 3)

·           Be able to calculate “reliability coefficients” using   Generalizability Theory in crossed, nested, and fixed study designs (section 4)

Section 1

Definitions

Generalizability Theory (Shavelson & Webb, 1991)

a statistical theory that describes how multiple sources of error in measurement can be estimated separately in one analysis and allows the investigator to consider numerous applications of an instrument

Facet

a potential source of variation

Object of Measurement   

source of variation arising from systematic differences among ability of subjects in the study

the variation reflects differences in knowledge, skills, etc.

Next