REGRX: A Computerized Stepwise Regression Algorithm with Residual Analysis.
Abstract
Regression analysis programs are commonly used to develop prediction systems for Air Force personnel research, e.g., a system for accurately predicting future job performance of enlisted individuals based on information obtained during their Air Force careers. This information might include such predictor variables as aptitude or ability test scores, biographical data, and physical attributes. Requirements for the development of such prediction systems within the technical programs of the Air Force Human Resources Laboratory (AFHRL) are numerous. REGRX, a stepwise regression algorithm with residual analysis, functions as a subroutine in the computer-based TRICOR utility correlation and regression system. The comprehensive statistical/mathematical methodology comprising REGRX has served as a powerful tool in the development of Air Force prediction systems. This paper contains the technical details that are necessary for the user to take complete advantage of the analytical capabilities of REGRX. This information includes an in-depth discussion of the REGRX mathematical/statistical methodology with associated computational formulas and printed output samples. Therefore, this document aids researchers in executing and interpreting the results of the REGRX subroutine. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1983
- Accession Number
- ADA123987
Entities
People
- Janos B. Koplyay
- Larry K. Whitehead
- Walter G. Albert
Organizations
- Air Force Research Laboratory