Safety in Numbers
The proactive use of data will propel the aviation industry towards its target of zero accidents
The industry’s 2009 accident rate was the second-lowest ever. But that figure is merely a stepping stone for greater improvement.
The great strides already taken—including the IATA Operational Safety Audit (IOSA), the Safety Audit for Ground Operations (ISAGO), Safety Management Systems and the Runway Excursion Toolkit—are being supported by new initiatives.
In late March, IATA, together with the International Civil Aviation Organization (ICAO), the US Federal Aviation Administration (FAA), and the Commission of the European Union (EU), signed a Declaration of Intent to exchange safety data. The move will make a safe industry even safer and reduce the number of duplicated audits of state safety systems and airlines. Initial sharing of audit information will begin in early 2011.
“The FAA knows that the use of data to identify safety trends is the key to preventing future accidents,” says an FAA spokesperson. “In the United States, the Commercial Aviation Safety Team (CAST) has already reduced fatal commercial accidents by 83% (1997-2008). That group has now moved away from a forensic approach—looking at the major causes of past accidents—to a proactive approach using inflight and other data to predict trends. Sharing data globally will help governments and industry reduce accidents while keeping pace with a changing industry.”
IATA will share aggregated IOSA information, allowing it to be cross-referenced with audits from the other parties to enhance overall safety efforts. “IATA has a lot of data from nearly 900 IOSA audits,” says Guenther Matschnigg, Senior Vice President, IATA Safety, Operations and Infrastructure. “However, we don’t do physical ramp inspections. Both the EU and, to a lesser extent, the FAA, do include actual spot-checks. What they report will complement IOSA findings and provide a complete picture, allowing for effective corrective actions.”
The concept of data collection and dissemination does not stop there. IATA has also launched its Global Safety Information Center (GSIC). This combines multiple safety databases such as IOSA, operational safety reports in the Safety Trending Evaluation Analysis Data Exchange System (STEADES), ISAGO, and the Flight Data Analysis (FDA) program.
The amount of data collated in the GSIC is extraordinary. STEADES alone has more than 110 airline members. It contains over 550,000 reports, and these are increasing at a rate of more than 80,000 per year.
All IATA member airlines can now access the GSIC’s full set of information at gsic.iata.org. “At some point, we will extend access to non-member airlines and other aviation stakeholders,” says Matschnigg. “It will allow airlines to benchmark themselves against world best practice and rectify any shortcomings.”
Bernard L’Esperance, Head of Quality, Safety and Occupational Health, Air Seychelles, agrees that it is “a considered and interwoven approach to improve safety. ICAO, national authorities and airlines cannot introduce a measure to improve safety without first justifying it and conducting a risk-management exercise,” he says. “This is where data collection and availability play a key role in improving safety while also ensuring the buy-in of the industry.”
The right questions
As L’Esperance implies, having information on tap is only part of the equation. Using it intelligently is the key to success. There is some truth in the old adage that there are lies, damn lies and statistics. Given the volume of data in aviation, it could probably be used to arrive at any pre-conceived conclusion.
In fact, pre-conceived ideas are one of the many challenges that arise when addressing data analysis. It is normally referred to as confirmation bias. This means there is a tendency to look for supporting evidence for a theory. Contamination, whereby proximate but irrelevant information is allowed to influence analysis, is another psychology term. And overconfidence in calibration means the best case scenario inexplicably becomes the most probable explanation.
Overcoming these challenges requires the right questions to be asked. “Before any trending or data analysis is started, it is important to determine your main areas of risk, and what occurrences are of concern,” says L’Esperance. “Once these have been established, the analysis can start to determine factors at play in these areas. This analysis would be further used to determine the trend. By structuring the conduct of these trends, it follows that the correct information is gleaned without biasing the data.”
Drilling down is key. Initially, runway excursions are established as a significant cause of incidents and accidents. Then the causes of the excursions are determined. The final result is a prioritized action list.
Even without intelligent analysis, there are positives in amassing data. The availability bias is overcome—no longer will decisions be made on information at hand rather than information required. Induction, where general rules are postulated from insufficient data, is similarly belittled.
Two points warrant further investigation. Another psychology term—bystander apathy—looks at situations where individual responsibility is abdicated in a crowd. Could so much industry data overwhelm the safety decisions individual airlines must make on a daily basis?
Matschnigg stresses the opposite is true. “Data sharing helps decision making,” he insists. “There is no point in collecting data for its own sake. That’s why GSIC will be made as freely available as possible. Through skilful analysis, through sharing, data can be turned into action. Data doesn’t make us safer. But understanding the data and acting where it tells you to, does.”
The matter of whether anecdotal evidence and small but significant facts might get lost in such a wealth of data also needs addressing. L’Esperance is convinced that data collection only reaffirms all valid sources of information.
For example, when a pilot knows an aircraft doesn’t feel right, he “will start to analyze this feeling and his actions, the position of switches and so on,” L’Esperance says. “He will also attempt to recall all the past occurrences, discussions with other pilots and industry information to determine possible scenarios.”
Garuda Indonesia President and CEO Emirsyah Satar agrees. “Information from highly skilled, experienced personnel is always worth considering,” he says. “But that information can be further explored, and the accuracy verified, by cross-checking against industry facts.” Data backs up intuition. It turns a hunch into a hypothesis.
Risk or uncertainty?
One other crucial point is addressed by the data sharing concept: the difference between risk and uncertainty. Insurers have long-winded explanations but essentially risk is something you can measure. The uncertain and the unknown are exactly that—immeasurable.
A rare event is impossible to tabulate against probability. Holding information in silos only increases the chances of atypical events. Combining data moves the industry further into the measurable arena. It turns the unknown or exceptional into calculable risk. And that can be brought under control.
“There is no competition when it comes to safety,” says Giovanni Bisignani, IATA Director General and CEO. “Cooperation is the way forward. We have a common goal of zero accidents and zero fatalities. The safety data from audits and oversight programs contains important parts of a whole picture. Agreeing to put this data together is a big step forward.”
More on safety at www.iata.org/safety
Lessons in data management
The catastrophic failure of Long-Term Capital Management (LTCM) in the late 1990s highlights the need to understand data properly.
Headed by Nobel Prize winning economists, it seemed the hedge fund could do no wrong. Based on a complex mathematical formula, the fund delivered returns of 40%-plus in its first two years. But its financial models were based on only five years’ data. It didn’t include the stock market crash of 1987, for example. The events of 1998, which culminated in the Russian default, exposed the deficiency. LTCM went bust in a spectacular manner. Its calculations said such a sharp market plunge shouldn’t have happened in the entire history of the universe. Nor had it updated its model to include the globalization trend. To its surprise, markets move in sync.
Aviation is a global industry and it understands that one incident affects all stakeholders. Integrating all available information will ensure that air travel becomes safer