Using near misses to improve risk management decisions

Abstract

In the process safety literature it is often claimed that the analysis of near‐miss incidents can improve process safety performance. However, empirical research has demonstrated that near miss events are interpreted as marginally successful, leading to riskier behavior due to lower perceived risk analyses. In other words, experiencing a near‐miss incident can result in the false conclusion that the original risk was over‐estimated. To be effective, the analysis of near‐misses must be grounded in an objective evaluation of the event, the outcome, and the severity of the alternative outcome's consequences. In this article, several examples are presented where a near‐miss incident preceded a serious incident with significant human injury or property damage. If the near‐miss had been investigated, properly evaluated, and appropriate corrective actions been implemented, the more severe incident would not have occurred. A recurring theme in these examples is the underestimation of consequence severity, that is, the misinterpretation of the near‐miss as a high probability, low‐consequence severity scenario rather than a narrowly averted low‐probability, high‐consequence severity scenario. A deliberate consideration of the lower probability, more severe consequence scenario would have facilitated a more thoughtful evaluation of corrective actions. A methodology based on game theory is presented for evaluating the impact of near‐misses on risk perception. A game‐theoretical model is used to demonstrate the need for shifting payoffs away from lagging process safety indicators and toward leading indicators. An equally important outcome of this analysis is the effective communication of the risk at operational and managerial levels of the organization. © 2013 American Institute of Chemical Engineers Process Saf Prog 32: 322–327, 2013

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.