David R. Henderson  

Numeracy and Risk in Air Travel: A Personal Account

Choice Architecture in the Caf... The Roots of Signaling Denial...

Months ago, my wife and I made plans to fly from San Francisco to Honolulu on Sunday, July 7. Of course, we didn't know that the day before there would be a big airline crash in San Francisco. Should that one incident have changed our thinking about flying? Of course not. Air travel is incredibly safe and one sample point doesn't change that fact.

Here's what I found encouraging, though. Not only did my wife not waver in her decision to fly out of San Francisco, but also, the only discussion we had about the Asiana Airlines, other than the amazingly small loss of life, was about whether it would cause our flight to leave late. She understood that the incident had no connection to our risk. Moreover, just observing people in the international terminal from which we flew, I didn't notice anyone who seemed unduly afraid. Of course, there's selection bias: those who were unduly afraid may not have shown up. Still, what was refreshing also was the talk generally in the media about how safe airline travel is. There seemed to be real numeracy, at least on this issue.

Even when we took off and looked down at the Asiana wreckage beside one of the runways, I didn't hear people gasp. To the extent people commented, it seemed to be human curiosity at work.

I'm very grateful that there do not appear to be any videos of the plane crashing. Had there been such videos, many people's fear would probably have been higher. Sometimes, apparent numeracy is only skin-deep.

Comments and Sharing

COMMENTS (10 to date)
JohnC writes:

I recall reading an article re transportation safety, and I believe the week after a major incident was the "safest" time to travel travel (by plane, and also train, bus, or boat), possibly because everyone was extra careful.

I dimly recall that "safest" referred to both major incidences (crashes) and also minor "safety events" (e.g, near-misses).The latter are rare, but they do happen from time to time (even though the passengers, and often the pilots, are oblivious).

Eelco Hoogendoorn writes:

There is at least one such video, that I have seen. Made from too far away to see any people tumbling over the runway, luckily, but it very well captures what is going on with the plane itself. Quite impressive to see such a big object make such a thud without completely splintering to pieces right away.

Indeed, flying and the 777 are very safe modes of transportation regardless. As often, it appears to be a human error at play here. Id be interested in seeing how the fatalities break down between pilot and mechanical failure.

Damien writes:
She understood that the incident had no connection to our risk.

100% true if we're dealing with objective risk. Flying is just as safe as before, at least for people flying out of SFO (flying in might be a different story if the PAPIs are still out).

But it'd also be rational for people to now revise their beliefs about how likely a crash is. If, for instance, people learned through the news coverage that adherence to best CRM practices varies a lot between airlines, and that many pilots don't get much hand-flying experience, it'd be rational to now consider that a crash is slightly more likely. Not because it's objectively more risky to fly but because they now have a more accurate perception of the risks.

It'd still be ridiculous to be afraid of flying, though.

Bostonian writes:

Holman Jenkins has a column in today's WSJ http://online.wsj.com/article/SB10001424127887323823004578595323218709446.html

"Bring in the Robots: Why accidents are rarer and humans more often at fault"

which concludes as follows:

"Regulators and lawmakers had better get cracking on drafting rules to permit robo-controlled transportation. That includes the long-delayed opening up of the airspace to unmanned vehicles (which this column first wrote about in 2005). Meanwhile, we should expect those accidents that do occur increasingly to arise from the awkward transition in which "deskilled" operators are still in the loop and computers haven't completely taken over yet."

Jack PQ writes:

I second Damien: although flying is very safe, I think it is normal to do Bayesian updating, and the Asiana Air flight is new information. However, it is not obvious the Bayesian updating would mean we conclude flying is riskier than we thought. On the contrary, seeing that the crash landing led to only minimal loss of life, thank goodness, we might conclude that flying is even safer than we thought.

Damien writes:

Jack PQ: Thanks, you're right! Silly that I didn't think about it since, in practice, that might even be more likely. I suspect many people do have the idea that whenever a plane crashes, you're basically done for. Spot on!

Thomas Strenge writes:

David, your class on numeracy was one of the most profound educational experiences of my life. I still quote regularly the statistical comparison of likelihood of death in car crashes (as backed by NTSB) versus terrorism (i.e. how many fully laden passenger jets would terrorists have to blow up to equal the risk of dying in a US highway car crash). The comparison always amazes and takes the wind out of most TSA supporters' (few as they are) sails. All the best!

Eric writes:

@Jack PQ:

Oddly, that's exactly the Bayesian update I did. I decided plane flight was (very) slightly safer than I had believed a few days ago.

David R. Henderson writes:

@Thomas Strenge,
Thanks. That's nice to hear.

jb writes:

actually there is a video of the crash, though from a considerable distance:


Very thought-provoking commentary too!

Comments for this entry have been closed
Return to top