Tr. Carretta et H. James, US AIR-FORCE PILOT SELECTION TESTS - WHAT IS MEASURED AND WHAT IS PREDICTIVE, Aviation, space, and environmental medicine, 67(3), 1996, pp. 279-283
Selection of military pilots has long included the use of multiple apt
itude test batteries. Although the names and appearances of the tests
used in pilot selection vary, most are to a large extent measures of g
eneral cognitive ability, or g. This is consistent with the central ro
le played by cognitive ability measures in the prediction of numerous
job training (r = 0.43) and performance (r = 0.34) criteria. Measures
of specific cognitive abilities (e.g., verbal, quantitative, spatial,
perceptual speed) have shown little incremental validity beyond g (inc
rease in correlation of about 0.02). The incremental validity, beyond
g, of measures of pilot job knowledge (e.g., aviation concepts, instru
ments, principles and terms; increase in correlation beyond g of about
0.08), psychomotor abilities, and personality scores (increase in cor
relation beyond g between 0.02 and 0.04) also has been small, but sign
ificant. The unavoidable requirement to reason in responding to test m
aterial causes gto be measured. In broad-ability-range samples; the po
sitive correlations of the measures demonstrate that general cognitive
ability is always present as a higher-order factor. Future measures o
f pilot aptitude may include tests based on cognitive components, chro
nometric methods, neural conductive velocity, or Ether methods. These
measures, despite their appearance, have been shown to mostly measure
g. Subsequently, we expect that future U.S. Air Force pilot selection
tests will mostly be measures of g and will, therefore, continue to be
predictive of performance.