Abram G. Bayroff
Published: 1967
Total Pages: 28
Get eBook
In the conduct of continuing research to improve screening and induction techniques, the INPUT QUALITY Task has explored the potential contribution of programmed testing particularly with respect to use of the branching technique which tests the examinee with items appropriate to his own ability level. The present publication reports on the trial administration of two branching tests by means of a computerized system with teletypewriter input/output developed by the National Bureau of Standards. A specific objective was to determine the comparability of computerized branching tests and conventional paper-and-pencil tests with reference to reliability, test score information, and test construction rationale. The two specially constructed 8-9-item branching tests (verbal and arithmetic reasoning) and corresponding conventional 40- and 50-item tests of the Army Classification Battery (ACB) were administered, in counter-balanced order, to a sample of 102 enlisted men. Statistical analyses of score results and observations of response patterns are presented. Findings indicated substantial correlation of the short branching tests with their longer conventional counterpart tests (r = .83 and .79, greater than would be expected with equally short conventional tests). Findings also reinforced the research promise of branching tests and indicated the need for reexamination of classical test theory. (Author).
