Aviation, Safety and Numbers: When Numbers Replace Purpose
By Florin Necula, PhD | Accountable Manager & Licensed Aircraft Engineer
Sep 13th, 2026
We trust numbers in aviation and usually for good reason. We need accurate calculations to operate aircraft and reliable data to understand recurring problems. But I wonder whether that trust sometimes extends too far when we assess people, organisations and operational safety.
Numbers feel reassuring because they turn a complicated situation into something that appears clear and comparable. A rate, percentage or green indicator is easier to present and defend than a conversation about uncertainty, trade-offs and local conditions. This may explain why organisations continue to add measures: numbers promise control, even when they describe only part of the work. The danger begins when the clarity of the number is mistaken for the reality behind it.
A department achieves its targets and the dashboard turns green. Before calling that a success, we should be able to explain what changed in the operation. Otherwise, we risk allowing the target to replace the purpose of the work.
Lessons from education and healthcare
Reading The Tyranny of Metrics, two examples stayed with me. Muller (2018) describes how judging schools by examination results can push teachers to concentrate on what will be tested. Results may improve while less attention goes to developing curiosity and independent thinking. The school becomes better at producing the measured result, but the quality of education is a broader question.
In healthcare, Muller discusses surgical performance reporting. Publishing success rates can encourage doctors to avoid patients whose conditions make a poor outcome more likely. The figures may improve because the difficult cases have been excluded. That leaves a troubling question about what happens to the patients who most need help (Muller, 2018).
These examples show how measures can influence professional decisions. The consequences depend partly on what happens when someone misses the target. A figure used to explore a problem can acquire a different meaning when it determines rewards or exposes people to criticism.
Li and Evans (2022) examined this through a review of 41 papers and analysis of interviews involving 147 participants in healthcare performance management. They identified unintended consequences including fixation on measures, gaming and administrative burden.
Metrics also consume resources. Data must be collected, checked, entered, reported and explained, and this process can grow into a job of its own. Li and Evans (2022) identify administrative burden as an unintended consequence of performance incentives. The practical test in aviation is whether the information changes a safety decision; if it does not, the time and attention spent producing it may be diverted from the operational objective.
The same questions in aviation
Consider a maintenance organisation responding to a finding about poor shift handovers. A procedure is revised and staff receive a briefing. The action is recorded as complete within the deadline. But are the next shift’s engineers now receiving the information they need? We have to look at the handovers themselves to find out.
Now imagine an airline that treats fewer go-arounds as evidence of better performance. The figure falls and the target is achieved. Perhaps approach preparation has improved, or operating conditions were more favourable. But could crews also have become more reluctant to go around because they know how the decision will appear in the figures?
A go-around may be the soundest decision a crew makes that day. Any assessment needs to examine the circumstances and the quality of that decision. These are illustrative scenarios, but they expose a practical weakness: an improved result can have several explanations, with very different implications for safety.
Occurrence reporting presents a similar difficulty. Fewer reports may reflect improved conditions. They may also reflect disappointment with previous responses or reluctance to speak up. If reporting falls after a change in leadership, that deserves a conversation with staff before it earns a favourable assessment.
Aviation researchers have raised related concerns. Kaspers et al. (2019) identified limitations in reliance on adverse-event measures and insufficient validation of many safety process indicators. Their review questioned how confidently those measures could be linked to safety outcomes. Rare accidents also make it difficult to establish the effectiveness of an individual organisation’s arrangements from its accident record alone.
What the research supports
More recent work suggests that confidence should remain proportionate to the evidence. Ferrante di Ruffano et al. (2024) reviewed 48 occupational safety studies for Lloyd’s Register Foundation. Although positive findings were common, they judged the evidence for causal effects of leading indicators on safety outcomes to be weak. More than 80% of studies had moderate or high risk of bias. This leaves uncertainty about effectiveness; it does not establish that leading indicators are useless.
The practical implication is that calling something a leading indicator does not establish what it predicts. Counting safety briefings may tell us how often they took place. To judge their value, we also need to understand what staff learned and whether that knowledge helped them deal with the conditions they encountered.
Comparisons need similar care. Jegan et al. (2025), reviewing ethical considerations around healthcare incentives, identified tensions involving fairness and professional values that depended on institutional circumstances and measurement design. This raises a useful question for aviation: how much of the difference between organisations reflects their performance and how much reflects the work they undertake?
Two airports handling similar traffic volumes may face different weather, infrastructure and traffic complexity. Converting event counts into rates accounts for some differences in exposure. It still leaves work to do before the figures support a fair comparison. Context determines what we can reasonably conclude.
Keeping the purpose in view
People do not need to falsify records for targets to distort priorities. They notice what management rewards and what they are repeatedly asked to explain. If discussions centre on overdue actions, obtaining closure evidence can become more urgent than checking whether the action worked. Management helps create the behaviour it later evaluates.
Each Safety Performance Indicator should therefore be connected to a clear operational concern. Managers should understand what the measure captures, what it leaves out and what decision it supports. Meeting the target should prompt an explanation of the operational effect, supported by evidence from the workplace.
For handovers, that means checking whether important information reaches the incoming shift. For go-arounds, it means understanding the circumstances and supporting an appropriate decision. Where staff accounts disagree with the dashboard, the discrepancy deserves investigation. Professional judgement can also be mistaken, so both sources of information should remain open to challenge.
There are aspects of work that remain difficult to express adequately in a score. A colleague who notices hesitation and takes time to help may prevent a misunderstanding that never becomes an occurrence report. Surveys and observations can reveal some of this contribution, but a numerical summary will always leave questions about the situation behind it.
We should expect managers to know the figures and understand the work. When a target has been met, the discussion should reach the people whose activity it represents. Ask them what changed and whether it helped. Their answers may tell us how much confidence the green dashboard deserves.
References
Ferrante di Ruffano, L., Watkins, D., Miller, P., & Bishop, E. (2024). Scoping review of the effectiveness of leading indicators for improving safety outcomes in occupational safety and health settings: What is the state of the art? Report for Lloyd’s Register Foundation.
Jegan, R., Rose, A., & Dierickx, K. (2025). Institutional financial incentives in healthcare: A review of normative considerations. BMC Medical Ethics, 26, Article 92.
Kaspers, S., Karanikas, N., Roelen, A., Piric, S., & de Boer, R. J. (2019). How does aviation industry measure safety performance? Current practice and limitations. International Journal of Aviation Management, 4(3), 224–245.
Li, X., & Evans, J. M. (2022). Incentivizing performance in health care: A rapid review, typology and qualitative study of unintended consequences. BMC Health Services Research, 22, Article 690.
Muller, J. Z. (2018). The tyranny of metrics. Princeton University Press.
