| A retrospective case-control survey discovered candidate risk factors for pilots having aviation accidents, incidents and/or violations during 1987-1988. Risk factors included personal disposition, flight background, and demographic variables. A randomly selected sample from Federal Aviation Administration data bases of 1,049 civil aviation U.S. male pilots received anonymous questionnaires. Response rate was 46.3%. A check of threats to validity found negligible bias arising from non-response, exclusion of respondents, missing data, and power requirements. The final sample size was 394. Variables were rarely normally distributed, but transformation usually improved prediction. Analyses were conducted by univariate inspection of variables, comparison of cases and controls using bivariate parametric and non-parametric tests on original and transformed data, and multivariate analyses of case-control status using selected variables. Logistic and multiple regressions tested predictors in tiered models: (i) personal disposition variables, (ii) personal dispositions and demographics, and (iii) personal dispositions, demographics, and flight backgrounds. Additional logistic regression tests were conducted within strata of three flight background variables: Experience, Exposure, and Skill. Within-case bivariate and some multiple regression tests compared more severe with less severe cases. Severity of case status outcomes included: (i) type of occurrence, (ii) Pilot-in-Command or not, (iii) severity of accident/incident, (iv) pilot- vs. other-caused occurrence, (v) number of occurrences ever had, and (vi) rates of occurrences had, using experience and years flown as denominators. Environmental context predictors appeared in separate multiple regression models using occurrence rates outcomes. Certain variables consistently predicted case status offering a Risk Profile: high Thrill and Adventure Seeking, many ailments, not married, many years flown, many recent hours flown, and possibly, Eastern FAA Region. Stratified and within-cases analyses suggested several predictors of case status including low Superego, few safety behaviors, Private certificate, Personal flying, less education, and fewer pilot friends. The Five Hazardous Thoughts measure had little validity and requires refinement. Future research on theoretical and preventive issues should focus on risk-taking, conscientiousness, judgement, and learning. Methodological recommendations include using measures with improved psychometric properties and validity. |
