A SUGGESTED USE OF SEQUENTIAL ANALYSIS IN PERFORMANCE ACCEPTANCE TESTING
Abstract
The advantages, limitations, and means of using sequential analysis in performance acceptance testing are discussed. Sequential sampling appears to be useful in testing whenever (1) testing time per test item is high in relation to the time required to score each test item, (2) the test is primarily designed to determine whether a person passes or fails, (3) there is a need for testing more than about 100 persons on the same test, either in 1 group or in a number of groups, and (4) there is negligible correlation between items when criterion scores are held constant. Sequential sampling takes item difficulty and item discrimination into account when discrimination score values (norms) are established. Sequential sampling can readily be adapted to changing standards of accepting people, with no revision of the norms previously set up. Ordinarily, sequential sampling will give about the same accuracy as a fixed-length test, with about half of the testing time and about half of the testing cost.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1953
- Accession Number
- AD0019058
Entities
People
- Rupert N. Evans
Organizations
- University of Illinois Urbana–Champaign