Business Rules

Users of this database are urged to exercise considerable care and caution when interpreting the meaning of the data. A number of definitional and statistical problems associated with aviation accident data can result in misleading and even erroneous conclusions when this data is used as a measure of safety performance. These problems include, but are not limited to the following:

Broad Definition: The member states of the International Civil Aviation Organization (ICAO) have agreed upon a global standard definition of an aviation accident. This definition is very broad, however, and would include both a flight attendant receiving a broken arm as the result of an aircraft encounter with turbulence at altitude as well as the runway collision of two wide-body aircraft that resulted in the catastrophic loss of hundreds of lives and one or more aircraft. The breath of this definition raises at least two difficult issues related to the use of accident data as indicator of safety performance.

First, the definition is not a very useful indicator of the severity of an accident, which means that simple counts of accident could be misleading. Historically, the most commonly used method to differentiate accident severity has been to classify an accident based on whether or not it involved one or more fatalities. Recognizing that even this differentiation still results in very broad categories of accidents, a number of other approaches to refine severity further have been attempted. Recently, for example, the United States’ National Transportation Safety Board introduced a new categorization system that divides accidents into either "Major, Serious, Injury, or Damage".

Second, the definition is sufficiently broad that it raises issues about what is aviation or non-aviation activity. These issues follow from the fact that aviation safety engineers and analysts tend to focus on events that involve the design, manufacture and operation of aircraft – domains where they can establish /modify safety standards as required. If an accident event falls outside these domains, even if it clearly was related to some form of aviation activity, some aviation safety analysts and aviation authorities would argue that it should not be counted as an aviation accident. For example, the proponents of this position might argue events involving the injury/death of a ramp agent that steps into a turning propeller or is run over by a tug pushing an aircraft back from an airport gate should not be included in aviation accident counts. The is no consensus among aviation authorities as to how these events (as well as suicides, turbulence events, etc.) should be classified and, therefore, care should be exercised when comparing data from more than one source of accident data.

Accident Rates: Accident counts, by themselves, cannot be used to compare the safety performance of either airlines/operators, aircraft types or segments of the air transportation industry. Airlines that have operated the greatest numbers of flights and flight hours could be expected to have suffered the greatest number of fatal-to-passenger accidents (assuming that such accidents are random events, and not the result of some systematic deficiency). Similarly, the most used aircraft types would tend to be involved in such accidents more than lesser used types. Moreover, the air transportation industry is dynamic and the 100% growth in airline activity over the past fifteen years can result in historical comparisons that are misleading. The method most commonly used to address these issues is to calculate accident rates in terms of accident counts and some measure of aviation activity; i.e., accident counts divided by flight hours, departures, passengers boarded, etc. Unfortunately, such rates can not be calculated for many airlines because reliable activity data is not always available.

Use of Accident Data to Evaluate Airline Safety Performance: Airline accidents are very rare events and the risk of death or serious injury for air travelers is exceedingly small. Prof. Arnold Barnett of the Massachusetts Institute of Technology, using data for the decade 1987-96, has calculated that a passenger faces a death risk of one in eight million. "Stated somewhat differently, if a passenger facing a death risk of one in eight million to chose one flight at random each day, that passenger would, on average, go for 21,000 years before perishing in a fatal crash." His and related research also has found that while accident data may suggest there may be apparent differences in carrier safety records at any particular time, due largely to the infrequent but catastrophic nature of an aviation accident, there is no evidence that such distinctions persist nor that they are predictive of future safety performance.

In summary, considerable caution should be exercised in using the accident data contained in this database to evaluate individual airline safety performance. Definitional problems limit the utility of this data and statistical research indicates that past accident records provide no meaningful information to consumers seeking to make safety-enhancing comparisons for current or future travel choices.