COMPARISON OF COMPUTER-SIMULATED CONVENTIONAL AND BRANCHING TESTS
Abstract
In the usual testing situation, each examinee takes all the items, and item sequence is the same for each examinee. It is possible, however, to have sequential or branching tests in which all examinees do not take te same items and the sequence of item presentation for an individual is some function of his performance on previous items; that is, an item answered correctly is followed by a more difficult item, an item answer incorrectly, by a less difficult item. The rationale for the latter procedure is that presentation of items based on an examinee's past performance allows each individual to take items that are progressively more appropriate to his own level of ability. It is conceivable that such a procedure would reduce testing time, and for a given amount of time would permit more accurate measurement of an individual's ability, principally by reducing opportunities for chance success by low ability examinees' attempting items too difficult for them.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1970
- Accession Number
- AD0707375
Entities
People
- Carrie W. Waters
Organizations
- United States Army Research Laboratory