Stabilization and Task Definition in a Performance Test Battery
Abstract
Performance testing under unusual environmental circumstances almost always involves repeated-measure designs. Most tasks, however, show practice effects with repeated administrations, effects that may appear in the group mean, the variance among subjects, or the correlations over subjects among trials or repeated testings. There comes a point in many tasks after which practice no longer produces changes in performance; as we will put it, the task stabilizes. Our criteria for stabilization are: the group mean no longer increases, or increases at a slow and regular rate; the variance among subjects no longer changes; and the correlation with earlier trials remains the same from one stabilized trial to the next; finally, the correlation among stabilized trials is constant. Stabilization in this sense is virtually essential to a performance test battery. When it is absent, practice and environmental effects are confounded; interpretation becomes very difficult; and problems of design are greatly complicated, in some cases impossibly so. It is desirable, therefore, to determine in advance whether or not a task stabilizes with practice and, if so, how long it takes. It is additionally desirable that a task be well defined, that is, that it stabilize at a high level; preferably the average correlation among stabilized trials should be greater than .90. The present report concerns ten tasks each of which was practiced for 15 days by either 18 or 19 subjects.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1980
- Accession Number
- ADA099987
Entities
People
- Marshall B. Jones
Organizations
- Penn State Milton S. Hershey Medical Center