Plane & Pilot
Tuesday, July 6, 2010

Is Glass Safer?


Making sense of the NTSB glass-cockpit report


Earlier this year, the NTSB released the findings of a special study that they conducted comparing glass-cockpit aircraft and similar conventional, or “round dial,”-equipped aircraft. The purpose of the analysis was to determine if the “transition to glass-cockpit avionics in light aircraft would improve the safety of their operation.” It’s likely that this study stemmed from the GAMA report that in 2006, 90% of all new piston aircraft were equipped with flat-screen avionics. The investigation also may have been prompted by anecdotal evidence that flying glass is somehow harder or more dangerous than conventional flying—as if Avidyne and Garmin glass displays could be likened to rogue computers such as the HAL 9000 of 2001: Space Odyssey. But after much hoopla, media retort and blog discussions, essentially the NTSB report tells us nothing we haven’t known all along. What’s problematic, though, is that they present their findings in a skewed way leading one to believe, at least in part, the nonsensical notions that glass is more dangerous than round-dial aircraft.

It’s important to look at how the study was conducted to confirm the soundness of the methodology used. We’ve all heard compelling results of studies or polls that later turn out to be proven wrong or erroneous. This often is caused by researchers using poor technique when inquiring into the subject at hand, or worse, making claims that go beyond what can logically be construed from collected data. Furthermore, if a study is comparing apples to oranges, it’s invalid.


A recent NTSB study concluded that glass-cockpit aircraft were no safer than conventional instrument aircraft. Their recommendation for training on specific equipment is necessary to realize the safety potential of glass cockpits.
The NTSB made some stretches that compromise their findings. The premises of the NTSB conclusions rely heavily on data they accumulated through surveys of pilots of both glass and conventional general aviation aircraft in 2006 and 2007. These surveys asked for summaries of how the aircraft were used and the number of hours they were flown. However, only 24.7% of conventional aircraft users and 33% of glass users responded. According to the University of Texas, a minimum of 50% response rate is necessary for sound data when using mail surveys and a minimum of 40% is necessary if the survey is done via e-mail. So probably the most critical component used to calculate the accident rate used in the study, the number of hours flown, was deficient.

Furthermore, the NTSB made no effort to ensure the compatibility of accident circumstances. Simply, they compared apples to oranges. Glass airplanes are generally flown farther than conventional aircraft, deal with different weather conditions, are typically flown by a different pilot cohort, and tend to be used less for instructional flights than round-dial types. Is it fair to compare fatality rates of aircraft flown in IMC versus those flown in VMC, then say, “Aha! Those airplanes that are flown in IMC are more dangerous”? Of course not. Numerous previous studies have shown that flying in IMC is inherently more deadly than in VMC.

The NTSB study looked at 2,848 conventional and 5,516 glass aircraft that were built from 2002 to 2006. There was a wide range of aircraft included in the study, but the big names were Cessna, Cirrus, Diamond and Piper. Then the accident rates of these aircraft from 2002 to 2008 were analyzed. A total of 141 accidents occurred in conventional aircraft and 125 in glass. So which is safer? On the surface, the conclusion is obvious—glass: There were 125 accidents among 5,516 glass airplanes versus 141 accidents among 2,848 conventional aircraft. Unfortunately, it’s not fair simply to just count accidents nor is it fair to calculate an accident-per-airframe ratio. Why? Because if 5,000 of those glass airplanes were sitting in the hangar the whole time, we might change our opinion on their safety. So, accident rates per 100,000 hours are calculated to level the playing field.



1 Comment

Add Comment