Using Item Response Theory to Evaluate Measurement Precision of Selection Tests at the French Pilot Training

Abstract

In pilot selection settings, decisions are often based on cutoff scores. In item response theory the measurement precision of a test score can be evaluated by its degree of information. We investigated whether the maximum of test information corresponded to the cutoff zone for 10 cognitive ability tests of the current French civil air transport pilot selection (n = 577). An item response theory model was fitted to the data. The 6 adequately fitted tests' test information curves were generally maximal in the cutoff score zone. Several ways to improve the absolute level of precision of tests in pilot selection situations are discussed.