Accession Number : ADA323122

Title :   Selection and Prediction Using Latent Variable Models,

Corporate Author : CALIFORNIA UNIV LOS ANGELES

Personal Author(s) : Huang, Li-Chiao ; Muthen, Bengt

PDF Url : ADA323122

Report Date : DEC 1996

Pagination or Media Count : 149

Abstract : The purpose of this study is to investigate the use of factor scores for improving predictive validity in personnel selection. Recent studies have shown that general ability is a good predictor of future performance. However, specific factors may be important indicators for personnel selection as well. In the latent variable context, the hierarchical model which considers general and specific factors simultaneously offers possibilities for solving prediction problems efficiently and parsimoniously. This type of modeling is useful in determining the relative importance of general and specific factors as predictors of the criterion variable, and for improving person-job match. Two approaches are used in the current investigation: an artificial data is used to study the prediction of future performance, and a real data application using the Army Project A and the Marine Corps JPM enlistment and performance data is used to study the practical implications of the use of specific factors. The results of this study show that using specific factors in addition to the general factor as predictors provides better selection decisions. The illustration using the real data analyses suggests that including specific factors in predicting hands-on performance for most of the jobs under consideration creates gains in terms of sensitivity, specificity, and proportion of correct decisions.

Descriptors :   *PERSONNEL SELECTION, *RANK ORDER STATISTICS, MATHEMATICAL MODELS, ARMY PERSONNEL, PERFORMANCE(HUMAN), COGNITION, STATISTICAL DATA, ENLISTED PERSONNEL, MONTE CARLO METHOD, STATISTICAL SAMPLES, REGRESSION ANALYSIS, MATHEMATICAL PREDICTION, MARINE CORPS PERSONNEL, JOB ANALYSIS, APTITUDE TESTS, SCORING, CROSS CORRELATION.

Subject Categories : Personnel Management and Labor Relations
      Statistics and Probability
      Psychology

Distribution Statement : APPROVED FOR PUBLIC RELEASE