Human factors analysis of postaccident data: Applying theoretical taxonomies of human error

Abstract

Human error is involved in nearly all aviation accidents, yet most accident reporting systems are not currently designed around any theoretical human-error framework. As a result, subsequent postaccident databases generally are not conducive to traditional human factors analysis, making the identification of interventions extremely difficult. To address this issue, this study utilized 3 conceptual models of information processing and human error to reorganize the human factors database associated with U.S. Navy and Marine Corps aviation accidents between 1977-1992. All 3 taxonomies were able to accommodate well over three quarters of the pilot-causal factors contained in the database. Examinations of the recoded data revealed that procedural and response-execution errors were most common, followed by errors in judgment. However, judgment errors were more frequently associated with major than with minor accidents. Minor accidents, on the other hand, were associated more with procedural errors than were major accidents. This investigation demonstrates that existing postaccident databases can be reorganized using conceptual human-error frameworks, which may allow for previously unforeseen trends to be identified.