Cross-Validation of the Computerized Adaptive Screening Test (CAST)
Abstract
The Computerized Adaptive Screening Test (CAST) was developed to provide an estimate at recruiting stations of prospects' Armed Forces Qualification Test (AFQT) scores. The CAST was designed to replace the paper- and-pencil Enlistment Screening Test (EST). The initial validation study of CAST indicated that CAST predicts AFQT at least as accurately as EST and that it is more efficient to use (Sands & Gade, 1983). This report summarizes the findings from a cross-validation study of CAST that used the following procedure. Prospects' CAST scores were recorded by recruiters in U.S. Army recruiting stations and were subsequently matched by social security number to applicant tapes from Military Entrance Processing Stations to obtain AFQT scores and relevant demographic data. These data were examined using regression, discriminant function, and cross-tabulation analyses to provide recruiters with information on how to interpret prospects' CAST scores. The correlation between CAST scores and AFQT scores was .80 for the cross-validation sample, whereas in the initial validation sample the correlation was .85. These data indicate that CAST scores are very good predictos of AFQT scores.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1984
- Accession Number
- ADA163148
Entities
People
- Paul A. Gade
- Rebecca M. Pliske
- Richard M. Johnson
Organizations
- U.S. Army Research Institute for the Behavioral and Social Sciences