Accession Number:

ADA280023

Title:

Comparison of a Back Propagation Artificial Neural Network Model with a Linear Regression Model for Personnel Selection

Descriptive Note:

Interim rept. 1 Oct 1990-30 Sep 1991

Corporate Author:

NAVY PERSONNEL RESEARCH AND DEVELOPMENTCENTER SAN DIEGO CA

Personal Author(s):

• Wilkins, C. A.
• Sands, W. A.

Report Date:

1994-05-01

Pagination or Media Count:

16.0

Abstract:

This exploratory research examined the usefulness of an Artificial Neural Network-Back Propagation ANN-BP model predicting dichotomous criteria in personnel selection. The results are quite encouraging. Under a wide variety of circumstances, the ANN-BP model was superior to the Ordinary Least-Squares- Linear Regression OLS-LR model in predicting curvilinear relationships. My once was OLS-LR superior. When the underlying relationship was linear, there were no significant differences. This is somewhat remarkable, given that the OLS-LR model was designed to perform optimally in the linear case. Artificial neural networks, Back propagation, Personnel selection.

Descriptors:

• *NEURAL NETS
• *MATHEMATICAL PREDICTION
• *LINEAR REGRESSION ANALYSIS
• *PERSONNEL SELECTION
• MATHEMATICAL MODELS
• MODELS
• PROPAGATION

Subject Categories:

• Personnel Management and Labor Relations
• Numerical Mathematics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE