STATEMENT OF ARNOLD BARNETT I am a Professor of Operations Research/Statistics at the Sloan School of Management at the Massachusetts Institute of Technology. My field of specialization is Applied Probability and Statistics. I have compiled and studied aviation accident and safety statistics for approximately 20 years and have published articles on this subject. My curriculum vitae is attached. I was contacted by The New York Times regarding issues that had been raised regarding airline safety following the ValuJet crash. Certain of my observations were set forth in an article that appeared in The New York Times on June 9, 1996. Subsequently, I was asked by ValuJet to provide my views on what proper use may be made of available aircraft safety data and that of ValuJet in particular. Summary A summary of the conclusions that I have reached is as follows: Jet airline travel in the United States -- including that on ValuJet -- has been To date, there has been no information regarding the crash of ValuJet's Flight 592 Should human error by some party be assigned in the future, it must be compared As a starting point, any comparison of current accident/incident statistics involving airlines. Current statistics regarding "accidents" and "incidents," including those set forth in the Federal Aviation Administration ("FAA") May 2, 1996 report, regarding ValuJet and other airlines do not indicate that a passenger was less "safe" flying ValuJet than any other U.S. jet carrier. Because of the safety of ValuJet relative to other forms of low-cost travel, a permanent shut-down of ValuJet would increase the travel safety risks borne by millions of ordinary Americans that have benefited from ValuJet's low fares. While there has been a temporary suspension of ValuJet's operations, a resumption of flights would signify that existing concerns have been addressed. Many ordinary Americans Given what has been written and broadcast over the past six weeks since the ValuJet crash, many Americans have been led to believe that the crash of Flight 592 was a predictable result, a culmination of precursor events and problems involving the airline. Often mentioned have been reports that, before the crash, ValuJet had a higher rate of as those terms are defined by the National Transportation and Safety Board ("NTSB") – than that of other airlines. A review of salient facts and statistics, however, makes clear that many key arguments that have been made in this regard are illogical and unfair. Images from the Everglades crash site may have obscured a fundamental point: jet travel in the United States -- including that on ValuJet - has been extremely safe. During the period in 1996 preceding the crash, U.S. commercial airliners had transported hundreds of millions of passengers without a single fatality. ValuJet, before the crash, had carried 9.7 million passengers aboard its flights (i.e., passenger segments) without their incurring a single fatality or serious injury. Any discussion of the particular statistics involving ValuJet should start by acknowledging that they lie in a range that must be characterized as extremely safe. Based on industry-wide statistics, an American who took a domestic jet flight every day would, on average, have to go 21,000 years before succumbing to a fatal crash. Even the safest automobile drivers are well advised to use commercial jets rather than cars. For each hour of trip time they save by flying, their life expectancy increases 67 seconds, because jets comprise a far safer means of travel. (Barnett, 1991) In the aftermath of the crash, it has been pointed out that ValuJet, between the start of its operations in 1993 and the end of April 1996, has had more "accidents" and "incidents" per 100,000 flights than did U.S. commercial jet carriers as a whole. Mary Schiavo, Inspector General of the Department of Transportation, wrote in Newsweek that she was unwilling to fly ValuJet because of its "many mishaps." The key axiom in Ms. Schiavo's reasoning is that greater involvement in accidents implies a greater risk of what passengers really fear - i.c., involvement in a fatal airline crash. While such a premise is superficially plausible, it is conceptually flawed and devoid of empirical foundation. The same can be said of an FAA Report entitled "Low-Cost Air Carrier Safety Record," dated May 2, 1996, which relies entirely upon comparing the accident rates and runway incursion rates of various airlines. While the report concedes that "any findings about the relative safety of low-cost carriers must be made in very guarded terms due to their limited cumulative exposure," the more important point is that the data used in the FAA report do not support any reliable comparison of "relative safety." One fundamental problem is that the NTSB definition of "accident" encompasses vastly different events. The definition does not confine itself to events that, by their nature, might suggest a greater probability of a future airline crash. For example, the term "accident" includes the situation where a flight attendant standing in a galley during unexpected turbulence breaks an ankle. (This is an actual description of one of the "accidents" with which ValuJet was charged.) On the other hand, the term also includes the situation where an airliner attempts a takeoff without the pilot having deployed its flaps and 1/ As a separate observation, my review of the FAA report indicates certain inconsistencies and ambiguities that remain to be resolved. then crashes, killing all passengers. Each of these two events is counted equally as a single accident. Further, a substantial number of "accidents" include difficulties that the airline could not have foreseen (e.g., clear air turbulence) and events that pose or portend little threat of a major crash. Some entries include injuries to an aircraft or ground workers while the plane is near the loading ramp or in tow. These matters are not unimportant, but they do not clearly relate to the key issue that concerns airline passengers - i.e., the risk of serious death or injury during the regular course of the operations that make an airline distinct -- takeoff, flight and landing. Equally important is that the accident statistics do not reflect highly relevant information about the event. An example illustrates the situation. In 1988 a commercial airliner suffered an accident resulting from a structural failure that practically destroyed the upper half of its fuselage. At the instant of the accident, a flight attendant was blown out of the crippled plane and regrettably was killed. The pilot, however, managed to land the plane without a single passenger death. In assessing the safety of the airline, it is extremely important that a skilled cockpit crew saved the passengers in a situation that could easily have killed them all. Yet, if one relies upon simple "accident" tallies, that important fact is made irrelevant. The airline traveler has several legitimate types of questions about the airline: (1) What events are defined as accidents? -- and particularly are they the kinds of events that suggest some defect in crew performance, aircraft performance, airline maintenance, or another similar factor? (2) Once "accident" is redefined to respond to these concerns, at |