USE OF ARMY SCHOOL SAMPLES IN ESTIMATING ACB TEST VALIDITY
Technical research note
ARMY BEHAVOIR AND SYSTEMS RESEARCH LAB ARLINGTON VA
Pagination or Media Count:
Results are reported of the reexamination of the statistical method currently employed in estimating the predictive effectiveness of the tests of the Army Classification Battery ACB when validity data are restricted to a segment of the entering population with which the tests are used. Difference ratios were computed for partial correlation coefficients and for regression weights. Computations were based on data obtained on 2480 enlisted men at the beginning of their first enlistment. For analysis, the total sample was subdivided into 12 MOS-identified subgroups. It is anticipated that a projected replication based on more recent data will yield results for clear-cut decision on appropriateness of subgroup results on noncognitive scales for estimating validity by methods commonly used with linearly related cognitive tests.