Published by the MIT News Office at the Massachusetts Institute of Technology, Cambridge, Mass.
Published by the MIT News Office at the Massachusetts Institute of
Technology, Cambridge, Mass.
NEW STUDY Crowded Planes More Likely To Crash, Researchers Say By Elizabeth Thomson News Office Crowded airplanes-those that are at least three-quarters full-are five times more likely to suffer major crashes than less-crowded aircraft, two MIT researchers have found in a study on US domestic jet flights. The work indicates that air travel is safest during off-peak hours when planes are less crowded. "It's already cheaper to fly at off-peak hours, and it's probably also safer," said Arnold Barnett, a professor of operations research and statistics at the Sloan School of Management. The researchers stress that their results do not mean that a high passenger load in itself will cause a plane to crash. It might, however, increase the probability that some other related event-such as a longer takeoff run-will cause a crash. The study, by Professor Barnett and Todd Curtis, who received his SM in management from the Sloan School earlier this year, was published in the October 1991 issue of Flight Safety Digest. The work was funded by the Federal Aviation Administration. Professor Barnett and Mr. Curtis, who is now a systems safety engineer with Boeing, analyzed major domestic jet crashes on established US airlines during the period from 1975-89. A major crash was defined as one that resulted in the deaths of at least 20 percent of the passengers on board. The researchers found that in the 10 major crashes over that period, on average 84.7 percent of the seats on the flights were occupied. In contrast, 59.4 percent of seats were occupied on the average domestic jet flight over the same period, according to US Department of Transportation data. The seat occupancy rate for the crashes was therefore more than 25 percentage points higher than for the average domestic flight. "Even when account is taken of the sampling error in data with only 10 observations, the probability of getting a 25-point excess by chance alone is well below one in 1,000," the researchers wrote in Flight Safety Digest. They subjected the data to two statistical tests, with the same results. The work by Professor Barnett and Mr. Curtis also suggests that wide- body jets are six times more likely to crash than smaller planes. According to the researchers, such jets are used in only seven percent of US domestic jet flights, yet three of the 10 planes that crashed were wide-body jets, for 30 percent of total crashes. "The discrepancy between 30 percent and seven percent implies that wide- body jets suffered roughly six times as many major crashes per million flights as did smaller jets," the authors wrote. Professor Barnett and Mr. Curtis point out that a high passenger load in itself will not cause a plane to crash, but crowding could be associated with greater risk that some other event will cause a crash. For example, they wrote, "fully loaded aircraft go further down the runway before lifting off, climb more slowly and have higher stall speeds than more lightly loaded aircraft. These circumstances reduce the time and flexibility that the pilot . . . has to respond to emergencies at takeoff." In addition, crowded planes are more likely to be the targets of terrorists. Crowded planes are also most common during rush hours and peak seasons, which include their own risks. For example, midair and runway collisions may be higher at rush hour when more planes are operating. Further, "the busy summer vacation season coincides with the period when weather hazards such as thunderstorms are most numerous," Barnett and Curtis wrote. As a result, "disproportionate numbers of air travelers may be at the wrong place at the wrong time." The authors acknowledge that the study is based on a small number of crashes, but used statistical methods that took full account of the small sample sizes. Commenting on the study's finding that planes in major crashes had nearly twice as many passengers aboard as average domestic jets, Professor Barnett said: "The chance that such an excess arose by coincidence is less than that of tossing a coin 12 times in a row and getting heads every time." Barnett and Curtis conclude that a more even distribution of passenger loads across flights could reduce the risk of domestic jet travel. However, they wrote: "One must be realistic. Business travelers will not fly at 2am, and trips home for Christmas will not be replaced by visits at Halloween." Nevertheless, "the possibility that off-peak flights may be safer . . . could increase their attractiveness to some passengers." The authors emphasize that during the period they studied, "established US domestic air carriers amassed the best aggregate safety record in the history of civil aviation." To that point, the study estimates that a US air traveler who took one domestic jet flight each day could on average go 14,000 years before dying in a fatal crash. "The risk is very low, but it might be even one-third lower if passenger loads were more evenly distributed across flights," Professor Barnett said.