consortiums

PROJECT 13

SELECTION AND CLASSIFICATION

Classification and Selection for Personnel Applications Using A Data
Envelopment Analysis Approach

Jose H. Dula
School of Business Administration,
University of Mississippi

Donna Retzlaff-Roberts, James Van Scotter
University of Memphis

Proposed Grant $100,890

Abstract:
This project will develop and test a new approach for improving the effectiveness of

personnel selection and classification decisions. There are three data characteristics that cause difficulty for existing classification methods (1) unbalanced group sizes, (2) unequal misclassification costs, and (3) non-normal data. Preliminary research suggests a hybrid method incorporating data envelopment analysis and linear programming discriminant analysis (DA/DEA) is effective in this difficult situation and outperforms other methods. Further research is needed to fully develop and test this promising methodology. Results will suggest ways DA/DEA may alleviate long-standing selection and classification problems.