Flying tens of tonnes of metal, suitcases and people is no easy feat, especially in a steady and safe manner. But between technological solutions, regulations and safety checks, this feat is in our world an everyday reality, with aircraft ranking as one of the safest means of transport ever. Nevertheless, accidents do happen, and sometimes what makes them more likely are factors that are difficult to explain at first glance. For example, an accident is more likely to happen if the pilot is driving the aircraft, rather than the co-pilot. Tom Becker, of the University of London, and Peter Ayton, of the University of Leeds, recently conducted a study to investigate this precise phenomenon. The results were presented by Ayton at the conference SPUDM 2025 (Subjective probability, utility & decision making) that took place in September at the IMT School.
Differences in roles
Let's take a step back: in most aircraft, there are two people responsible for flying it. They are divided into two types of roles, one hierarchical and one operational. The hierarchical division is between pilot (or pilot-in-command, PIC) and co-pilot (or second-in-command, SIC). The operational division, on the other hand, assigns the roles of pilot flying (PF, the person who actually flies the aircraft) and pilot monitoring (PM, the person who checks the status of the aircraft and must report deviations or anomalies, and intervene if necessary). Both the pilot and the co-pilot can take on the roles of pilot flying and pilot monitoring, and in fact they usually alternate between them when going back and forth along a route.
For decades, studies and analyses have been carried out to understand whether and how the way crew members interact, including their role, influences flight safety and quality. Starting with research conducted by NASA in the 1970s, it is known that some aircraft accidents have been caused or aggravated by human errors related to decision-making processes and the ability of pilots to coordinate. As early as the 1980s, pilots underwent specific training in human error management, called Crew Resource Management (CRM), which was fully integrated into pilot training in the 1990s.
Even with all these precautions, the human factor still remains, and the phenomenon at the heart of the new study. Already another investigation by the US National Transportation Board (NTSB) in 1994 observed that in 30 out of 37 cases of accidents with “human” causes, the pilot was in the role of pilot flying, i.e. he was actually flying the aircraft. Other, equally old studies also point to this fact.
Looking for an explanation in the accident database
But why would an accident be more likely to occur when the driver is driving, rather than the co-pilot? To understand this, Ayton and Becker first wanted to confirm the validity of the phenomenon by analysing a much higher number of accidents than in previous studies. They then tested three hypotheses, three possible explanations for this phenomenon, also to understand how it could possibly be mitigated.
First, the two researchers, using the database provided by the Jet Airliner Crash Data Evaluation Centre (JACDEC), excluded incidents that were not relevant, such as those due to technical malfunctions or those on small aircraft, usually piloted by a single person. From this selection (2,293 events) it was possible to establish the roles of pilot and co-pilot for 841 events, the basis for the rest of the study. The analysis confirmed the initial observation: more than 70 per cent of the accidents took place with the pilot in the pilot flying role, and 76 per cent of the fatalities always occurred under these conditions.
To confirm that the phenomenon is linked to the human factor, the operating conditions of the aircraft were also assessed, i.e. whether the aircraft was technically in normal flight condition, with no emergencies on board, or whether there were technical problems or emergencies such as smoke or flames. Another factor taken into account was preventability, i.e. the realistic possibility for pilots to prevent the accident by simply behaving differently. These two factors were assessed for the various events, and independently confirmed by a panel of experts.
It turned out that 72 per cent of the analysed events occurred during a normal operating condition, and that even almost 88 per cent of the events would have been preventable by pilots. Looking only at fatalities, as many as 96.3 per cent occurred in preventable accidents.
Do hierarchical relationships, experience, cognitive effort count?
Having confirmed with a larger number of studies that the assignment of roles between pilot and co-pilot influences the probability of an aircraft accident, the question remains as to why. Three possible factors that could explain this phenomenon have been highlighted from previous studies. The first, the one most frequently cited in other similar studies, concerns the hierarchical difference between pilot and co-pilot: given this asymmetry, a co-pilot in the pilot monitoring role (i.e. one who is supervising but not actually flying) might refrain from expressing observations and concerns. Another factor might be the difference in experience: the co-pilot, usually less experienced than the pilot, might be a less capable and aware pilot monitor. Ultimately, the difference in cognitive load could also play a role: if the pilot, in addition to being in a hierarchically superior role, has the actual piloting role, the total load of these two roles could reduce his situational awareness and impact his decision-making capabilities.
To test these hypotheses, it is necessary to have measurable variables on which to test them. For the hierarchical difference, Hofstede's “Power Distance Index (PDI)” was considered, which measures, in people with less power within organisations and institutions, the acceptance of power and inequality. This is a somewhat cultural value, which varies from country to country: in the UK and Germany, for example, it is 35, while in Austria it is 11. In practice, this value indicates that, for example, Austria is culturally characterised by a population that is less distant from power, and somewhat less ”subservient“ and complacent to inequality than the first two. Considering this index, if the hierarchical difference were relevant to explain these accidents, pairs of pilots and co-pilots from different countries should show different results in terms of accidents.
Regarding the difference in experience, one can look at the number of hours flown by the pilot and co-pilot, both in general and for the specific type of aircraft involved in the accident.
Finally, the cognitive impact can be studied through the age of the people involved, since several studies show an age-related decline in performance. The phenomenon should therefore be more pronounced for older people, and age should be a predictor of both flight test performance and accident frequency.
Of the 841 initial events, for only 144 it was possible to collect all the information necessary to test the various hypotheses, to see which variables have predictive power with respect to the possibility of an aircraft accident.
A paradigm shift for security
From the analysis, the hypothesis linked to the hierarchical difference, at least as measured by the PDI value, is disproved, just as the difference in experience, measured in flight hours, and the pilot's age do not appear to be predictors of an accident. Instead, the age of the co-pilot seems to be relevant for younger co-pilots, although this is not something that depends on their experience, which, as seen above, does not seem to come into play.
The final interpretation therefore suggests that there is an effect due to the hierarchical difference, but that this is not so much related to an actual difference in experience, but to a typical age difference between pilot and co-pilot, which makes the monitoring role more difficult for the co-pilot.
How to solve, or at least mitigate, this problem? The two authors of the study suggest that the solution is to ensure that the person in the pilot role is never also in the pilot flying role, i.e. actually driving. One way to achieve this, while maintaining the necessary alternation in driving on the various routes, is to abolish the co-pilot rank and have two pilots with the same rank in the cockpit, who can alternate roles, so that the one acting as pilot has the pilot monitoring role, with the other acting as pilot flying. This is a non-trivial paradigm shift, considering that this configuration of roles has been in place in aviation for decades, but one that, according to the data, would be crucial in avoiding accidents and saving lives.
Jasmine Natalini
