An Adaptive Testing Strategy for Achievement Test Batteries.
Abstract
An adaptive testing strategy is described for use with achievement tests which cover multiple content areas. The testing strategy combines adaptive item selection both within and between the subtests in the multiple-subtest battery. A real-data simulation was conducted in order to compare the results from adaptive testing with those from conventional testing, in terms of test information and test length. Data for the simulation consisted of test results for 365 fire-control technicians on a paper-and-pencil administration of a 232-item achievement test which was divided into 12 subtests, each covering a different content area. Correlations between subtest scores from adaptive and conventional testing were .90 or higher for eleven of the twelve content areas. An information analysis showed that for all 12 subtests, the subtest information curves from adaptive testing were essentially identical to the corresponding subtest information curves from conventional testing. On the average, the number of items administered with adaptive testing was half as many as was required with conventional testing; the shortest adaptive test battery used 18% of the total number of items in the conventional test, while the longest used 80%. The adaptive testing strategy, therefore, provided a considerable reduction in test length and virtually no loss in precision of measurement when compared with the conventional administration of the achievement test battery. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1977
- Accession Number
- ADA046062
Entities
People
- David J. Weiss
- Joel M. Brown
Organizations
- University of Minnesota