Determining Mean Predicted Performance for Army Job Families
Abstract
The present study is designed to obtain mean predicted performance (MPPs) for the 9- and 17-job families, using composites based on 7 ASVAB tests, using a triple cross validation design permitting completely unbiased estimates of MPP. While the authors have previously computed MPPs for 9 and 17 family composites, they have not been computed for composites that have had all hierarchical effects removed by a transformation to the Army conventional standard score (ACSS) scale (with its use of equal means and equal standard deviations). The specific research objectives are as follows: 1. To compute regression weights for the 7 ASVAB tests to form assignment composites corresponding to the two alternative second-tier structures (9 or 17 families) and to determine the classification efficiency in terms of MPP that would result from the use of all positive weights and the conversion of the composite scores into the ACSS scale. Weights are corrected first for unreliability of the criterion and, then, for restriction in range effects due to assignment from an Army input population to MOS samples. The weights are applied to test scores of independent samples to obtain back (biased) and cross (unbiased) MPPs. 2. To obtain MPPs for the two sets of job families for the youth population as described in (1) above. This involves a correction due to assignment from the Army input population into Army jobs, and then a separate restriction in range correction due to selection from the youth population into the Army. 3. To compare MPPs for the two sets of job families for the Army Input/youth populations. 4. To evaluate the relative value of the two sets of job families taking into account MPPs and composite validity coefficients, used in establishing cut scores for the ACSS scale.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2003
- Accession Number
- ADA410612
Entities
People
- Cecil Johnson
- Joseph Zeidner
- Susan Weldon
- Yefim Vladimirsky