Objective. To test whether a data-driven, in-depth statistical assessment of official accident reports could identify ways to improve general aviation safety beyond the individual safety recommendations each report contains.

Method. Mixed methods. 983 publicly available field-investigation reports (UK n=493, SA n=490) were coded for pilot age, licence type, total and recent flight time, aircraft type, phase of flight, CICTT occurrence category and injury severity; hypotheses were tested statistically. A pilot survey administered via social media tested pilots' perceptions of the findings, and structured interviews with pilots, instructors, engineers and student pilots explored areas of interest in depth.

Findings. Correlations in accident causation, training accidents and phase of flight between the two countries; significantly more powerplant-failure occurrences in South Africa; adverse trends between injury severity and pilot age, total experience and recent experience, with low-recency and lower-total-time pilots suffering more severe injuries; and a statistically higher likelihood of fatal injury for pilots aged 51 and older, despite identical licence revalidation requirements for all ages. Loss of control in flight was the dominant fatal occurrence category.

Application. These findings inform Andre's aviation risk assessment models, the recency-weighted approach to pilot experience in underwriting, and Andre's writing on loss of control and the "impossible turn".