Validation of Dichotic Listening and Psychomotor Task Performance as Predictors of Primary Flight Training Criteria: Highlighting Relevant Statistical Issues.

摘要

A statistical evaluation of the automated dichotic listening (DLT) and psychomotor tasks (PMT) indicated that both contributed to the prediction of primary flight training criteria. Prior to the main analyses, the extreme skewness-squadron differences in flight grades were removed by tranformations based on z-scores. Primary flight grades were highly correlated with the psychomotor scores (r's between -.26 and -.41) and moderately related with the dichotic listening scores (r's between -.22 and -.28). These r's were significant at an experiment wise alpha of .05. Multiple regression analysis indicated an even stronger validity coefficient when a combination of the performance measures was used (R = .442). Furthermore, the 19.5% of flight grade variance accounted for by the performance based tests was largely independent of the 16.6% variance accounted for by a combination of current selection tests and demographic variables. For the pass/fail criterion, a statistically optimal combination of DLT/PMT variables, selection tests scores, and demographic variables was specified that could be used to identify individuals who are relatively more likely to attrite. Classification matrices illustrate how such predictions could reduce attrition.