Effect of Reducing the Number of Tests in the Armed Services Vocational Aptitude Battery (ASVAB)

摘要

The major goal of this research is to determine the effect on ASVAB classification efficiency of dropping the Numerical Operations (NO) and Coding Speed (CS) tests from the battery. If it could be shown that removing either or both of these tests would not significantly reduce the mean predicted performance (MPP) of ASVAB classification or would not affect gender and racial fairness of the battery, then removing these tests would become defensible. Two data sets were used. The first set consisted of 150 job families containing 260, 000 first-tour enlistees ,and the second set (a subset of the first) consisted of 66 job families containing 83,000 observations. The criterion data were Skill Qualification Test (SQT) data from FY1987-89. Findings indicate that a significant loss in MPP would be incurred by reducing the 9-test battery to a 7- test battery by removing NO and CS. The loss is 6.2 percent for the 150 job family data set and 8.1 percent for the 66 job family data set. The loss incurred from dropping NO is much less than from dropping CS, but the combined loss of dropping both tests is significantly greater than dropping either test alone. Dropping NO and CS would result in a greater loss of MPP for females than for the total sample. Dropping the two tests would not only increase gender unfairness of the battery, but would significantly reduce the accuracy of performance prediction for females. The pattern of loss for blacks is not as clear.