Predicting Success in Primary Flying School From the Canadian Automated Pilot Selection System: Derivation and Cross-Validation

Publisher Website
Google Scholar
Abstract
This research derived and cross-validated an equation predicting outcome in Primary Flying School (PFS—the initial stage of Canadian Forces (CF) pilot training—from the Canadian Automated Pilot Selection System (CAPSS), a computer-driven, moving-base flight simulator recording up to 250,000 instrument readings per candidate. For the derivation study, 225 pilot candidates operated CAPSS before entering PFS. A pool of 20,700 summary measures (SMs) from 880 data sets was generated from the raw data. Through several cycles of analyses these measures were reduced to a single prediction equation. The equation was cross-validated on a new sample of 172 pilot candidates. The CAPSS equation significantly predicted PFS outcome for the cross-validation sample, r = .47, p < .00005. Classification analysis indicated that 79.4% of the candidates selected by CAPSS actually passed PFS, compared to 69.8% for the current selection system. Only 13.3% of those who could have passed were rejected, whereas the current system is estimated to reject 57.0% of qualified applicants. It is estimated that using CAPSS instead of the current selection system will double the number of PFS graduates from a group of applicants.

This publication has 2 references indexed in Scilit: