Improvements in Personnel Selection with Neural Networks: A Pilot Study in the Field of Aviation Psychology

This article discusses problems of data combination in personnel selection. In a study on selecting candidates for pilot training, 82 participants were tested. The predictor variables were attention, vigilance, reactivity under stress, and spatial and numerical ability. The evaluation of the applicants after training was used as the criterion. To test the predictive validity of the test battery neural networks, linear discriminant analysis and logistic regression analysis were used. The results indicate that neural networks are applicable even when the discriminant analysis produces biased results due to violations of its assumptions, and also outperform the logistic regression analysis in terms of separability of correct and incorrect classifications based on the height of the classification probability.