Predicting Aircraft Pilot- Training Success: A Meta-Analysis of Published Research

Abstract

Results are given from a meta-analysis of validities for aircraft pilot-selection measures. Sixty-eight published studies were identified for the 1940-to-1990 period, from which 468 correlations were extracted for a cumulated sample of 437,258 cases. The method proposed by Hunter and Schmidt (1990b) was applied to produce a bare-bones analysis. Mean sample-weighted correlations, estimates of true variance, and confidence intervals were computed. Several classes of predictors were found to have confidence intervals that did not include zero, indicating possible generalizability of validities. For the most part, however, the variance accounted for by sampling error alone was small. The effects of moderator variables (including nationality, service, decade of publication, and aircraft type) were evaluated. Of these, decade of publication was most consistently correlated with obtained validities and was associated with a decline in average validities over the five decades of studies examined. Limitations on interpretation of the results and problems associated with the analysis and interpretation of data from the published reports are discussed, and the range of correlations that might be expected from a composite of the groups of predictors that were examined is reported.