Up in the air on solid ground : examining the validity of tests, procedures and assumptions in commercial pilot selection

Study I evaluated the latent structure of the selection battery and supported a two- factor model in which written tests primarily reflected a general cognitive factor, while operational tasks contributed a distinguishable second factor. A refined confirmatory model (CFI = . 86, RMSEA = . 029) reduced the interfactor correlation to r = . 44, with differentiation driven by task demands (specifically, the degree to which subtasks emphasized speeded perceptual-motor execution versus complexity).

The CFI of .86 falls below the conventional .90 threshold; however, in large samples, CFI is sensitive to trivial degrees of misspecification that carry no practical consequence, and the RMSEA of .029 indicates close fit by any conventional standard.</p><p dir="ltr">Study II examined sex differences in test profiles and intercorrelations and found not only mean differences across several abilities but also stronger associations between manual spatial ability and multitasking and perceptual speed among women, indicating a more integrated ability profile for female candidates.

Study III tested whether high-school grades in Swedish, English, and Mathematics predicted admission and training outcomes and found no meaningful predictive utility; null results were unaffected by range restriction. Study IV reported a decade-long validation of the full selection system at LUSA (N = 171 admitted candidates; core analytic sample N = 144). A global suitability rating assigned by the aviation psychologist at the close of the individual test stage of the process predicted training performance and improved cross-validated explained variance (between the selection variables and pilot training outcomes) from R2 = .03 to .15; the corrected association strengthened from r = . 39 to r = . 51.

The rating was essentially uncorrelated with written test scores, suggesting that it captured information not reflected in those measures.</p><p dir="ltr"><b>Conclusions:</b> Validity in commercial pilot selection is best understood as a property of an integrated, construct-driven system rather than of individual predictors. The studies demonstrate a distinguishable operational ability factor alongside general cognitive ability; structural differences in ability profiles between male and female candidates; negligible predictive utility for academic grades in this operational context; and meaningful incremental validity for structured expert observation in a cohort where test score variance was compressed by multi-stage pre-screening.

Together, the findings support a view of pilot selection in which cognitive tests, operational assessments, and expert judgment contribute complementary information at successive stages; with construct validity as the prerequisite for defensible predictive inference for each selection component. These findings carry implications and highlight methodological considerations for how selection systems should be designed, evaluated, and adapted in the ever-evolving environment of the cockpit.</p><h3 dir="ltr">List of scientific papers</h3><p dir="ltr">I. <b>Lager, E.</b>, Sorjonen, K., & Melin, M. (2026).

General and Operational Abilities in Pilot Selection: Validation of a Two-Factor Model. [Submitted]</p><p dir="ltr">II. <b>Lager, E.</b>, Sorjonen, K., & Melin, M. (2024). Gender differences in operational and cognitive abilities.

Frontiers in Psychology, 15, 1402645. <a href="doi.org/10.3389/fpsyg.2024.1402645" target="_blank" rel="noreferrer">doi.org/10.3389/fpsyg.2024.1402645</a></p><p dir="ltr">III. <b>Lager, E.</b>, Sorjonen, K., & Melin, M. (2025). Selection in the Operational Domain Requires More Than Grades: High School Marks do not Identify High-Flyers. International Journal of Selection and Assessment, 33(4), e70026. <a href="doi.org/10.1111/ijsa.70026" target="_blank" rel="noreferrer">doi.org/10.1111/ijsa.70026</a></p><p dir="ltr">IV. <b>Lager, E.</b>, Sorjonen, K., & Melin, M. (2026).

From Aptitude to Altitude: A Decade-Long Validation of Multi-Stage Selection Predictors for Commercial Pilot Training. [Submitted]</p>