The tired crew that beat the fresh one

In the mid-1980s, with airline crews flying longer duty cycles, NASA's Ames Research Center set out to measure what fatigue did to flying. Clayton Foushee and colleagues recruited twenty volunteer twin-jet crews and put each through a full-mission simulation of a short-haul flight, realistic enough to include routine operations and an unexpected mechanical abnormality that pushed the crew's workload up. Ten crews flew the simulation in the pre-duty condition, after at least two days off and, as the airline's scheduling worked, mostly meeting each other for the first time. Ten flew it in the post-duty condition, within two to three hours of finishing a three-day cycle of short-haul flying together, with eight to ten hour days.

The post-duty crews were, as expected, significantly more fatigued. The surprise was in the performance measures. Rated by an expert observer on dimensions relevant to flight safety, the tired crews performed significantly better than the rested ones. The individual pilots in them made more errors, as fatigued people do, but the crews caught those errors and corrected them before they mattered. Analysis of the cockpit recordings showed the post-duty crews communicated significantly more, which earlier research had already found to be a good predictor of crew performance. Three days together had turned them into a crew. The rested pilots, each competent, were still a set of individuals.

The finding fits what accident investigators had seen. A National Transportation Safety Board review of major airline accidents between 1978 and 1990 found that most involved crews on their first day flying together, and many on their first flight. Hackman, who studied more than three hundred crews across ten airlines, found that airlines nonetheless scheduled pilots as interchangeable units, with crews re-formed for almost every flight, and that the military organisations which kept crews together were treated as an unaffordable ideal. The study's reach extends past aviation: Amy Edmondson has described it as the finding that set her on the research that became psychological safety, when she went looking for whether the same held in hospitals.

Every reorganisation is costed by headcount, and headcount is the one thing that doesn't change when you move people between boxes. What changes is the thing NASA measured: the error-catching that a team builds by working together, which exists between people rather than inside them and disappears the moment they're separated. The tired crews had it after three days. The reshuffle removes it in an afternoon, and the cost arrives later, as mistakes that a familiar team would have caught, and never gets attributed to the reorganisation that caused them.

Two practical rules follow. When you're drawing the chart, list the pairs and teams the change dissolves, next to the headcount, and treat each as a cost to be justified. And when a team has to be new, because of a hire or a project, treat its first weeks the way the accident data treats a crew's first day: as the period of highest risk, with more checking, not less, until the team has caught a few of its own errors.

Source: Foushee, Lauber, Baetge and Acomb, Crew Factors in Flight Operations, Part 3: The Operational Significance of Exposure to Short-Haul Air Transport Operations, NASA Technical Memorandum 88322, 1986. The accident review: National Transportation Safety Board, A Review of Flightcrew-Involved Major Accidents of U.S. Air Carriers, 1978 Through 1990, Safety Study NTSB/SS-94/01, 1994. Hackman's crew research: J. Richard Hackman, Leading Teams, 2002. Edmondson's account: Amy Edmondson, The Fearless Organization, 2018.

Leading Teams

J. Richard Hackman, 2002

On the conditions that make a team work, from the researcher who spent a career finding that the design of the team matters more than the talent in it. The cockpit research runs through it.