Improving the selection of pilot air force candidates using latent trajectories: an application of latent growth mixture modeling

Abstract

Latent growth mixture modeling is a statistical approach that models longitudinal data, grouping individuals who share similar longitudinal data patterns into latent classes. We evaluated the application of this method in a sample of ab initio pilot applicants (N = 297), using longitudinal data collected from a military flight-screening program (where the applicants flew seven required flights), resulting in a final pass–fail outcome. Results showed the existence of a two-class solution (Cluster 1 presented an initially higher performance and contained 75% of the Pass candidates) and the psychomotor coordination and general adaptability showed a significant effect.