Purpose of Generalizability Analysis

Return

• In a performance test, a subject has to perform a number of tasks.
• The performance is rated against established standards.
• Generalizability Analysis provides answers to three questions:

1. Would the ranking of test subjects hold, if a different, but comparable set of tasks were used on the same set of subjects? (G-Study)
2. Could the task set be used to rank a different, but comparable set of subjects? (G_Study)
3. How to optimize cost effectiveness of a study by changing its design and facet of generalization sample sizes to achieve a given level of generalizability? (D-Study). D-Studies are usually based on pilot data.

• the term 'comparable' implies that the samples are drawn randomly from respective universes of admissible candidates and tasks.