Generalizability Theory (G-Theory)

Generalizability Theory (G-Theory)

1. Emphasizes the estimation of random effect variance components

   • Requires that investigators define the condition of measurement of interest to them

2. Error sources are twofold

   • One that included rater main effects, and one that excluded such effects

3. The sources of measurement error include cultural content, psychological task set, topastic or guessing error, and so forth

   • A given measure or score is a sample from a hypothetical universe of possible measures= A score is a multi-factorial concept

4. Two stages or phases

   • G-study
     • studying the sources of variance
   • D-study
     • has to do with a decision study

5. G-Theory is the administration of the test under operational (i.e., real test use) conditions as well as determining the magnitude of the sources of variance

   • Includes
     • specification of sources of variance, simultaneous identification of the magnitude of these sources, and implementation of these estimates to test interpretation and use
   • Purpose
     • The systematic identification and the empirical examination of sources of variance simultaneously

6. Three major sources of variance that envelop all other types of variance

   • universe score variance, systematic variance, and random or residual variance

7. A unified concept of validity is composed of six different aspects

   • the content aspect, the substantive aspect, the structural aspect,, the generalizability aspect, the external aspect, and the consequential aspec