ACCIDENTS IN NORTH AMERICAN MOUNTAINEERING

Hi,
I'm looking for some kind of analysis of climbing accidents over the last twenty years. Has anyone done this using ACCIDENTS IN NORTH AMERICAN MOUNTAINEERING or another. maybe European, resource? Sorry if I'm just being useless at using google. Thanks in advance.

Steph Abegg has done a nice job compiling some stats from ANAM.

stephabegg.com/home/project…

stephabegg.com/home/project…

Thanks

One thing that jumped out at me from these pie charts is the percentage of accidents related to rappelling. When you read the internet forums, it seems to be taken as gospel that more people are injured or killed in rappelling incidents than any other single factor; but this linked analysis shows that rappelling is a primary cause in only about 3% of incidents.

JL

saxfiend wrote:When you read the internet forums, it seems to be taken as gospel that more people are injured or killed in rappelling incidents than any other single factor; but this linked analysis shows that rappelling is a primary cause in only about 3% of incidents.

Good point.
Another category that came out smaller than I might have guessed:
"Equipment Failure"

saxfiend wrote:One thing that jumped out at me from these pie charts is the percentage of accidents related to rappelling. When you read the internet forums, it seems to be taken as gospel that more people are injured or killed in rappelling incidents than any other single factor; but this linked analysis shows that rappelling is a primary cause in only about 3% of incidents. JL

I always thought that most DEATHS were caused by rappelling accidents. Also that most rappelling accidents result in death, and not injury.

If you look at the first two graphs, you can see how few deaths there are per reported accident. Those pie charts include ALL reported accidents, not just deaths. I'd bet that if you drew up new pie charts with only the accidents that resulted in death, you'd see the percentage caused by rappelling go much higher.

What immediately stands out to me is that the amount of deaths has remained fairly constant over the last 30-40 years, and that the amount of reported accidents has decreased substantially since the 80's.

I'd like to know what the spikes in 1976 are from, as well as '80 and '86. Without explanation I'd say that the data is a bit suspect.

Well my paid-work career has been much involved with many kinds of statistical data (including accidents + deaths + natural disasters). My assessment is that there are deep (and mostly unresolvable) problems with making statistical inferences from those AAC mountaineering accident reports.

Most of the key problems are identified by the compiler/interpreter Steph Abegg on this page

One problem mentioned is "missing data" (accidents which are not reported). This is deeply problematic because of what experienced statisticians often call "selection bias". The compiler claims that this problem usually has little effect on percentages or relative comparisons.

But rather there are several important data sets and situations (with practical consequences) where subtle tricky "selection bias" with the missing data has led to completely erroneous inferences (accompanied by fancy formulas and statistical terminology).

Therefore the claim that "selection bias" is not a significant problem for mountaineering accident reports would need to be very carefully argued.

My guess would be that once you include non-death accidents, it's difficult to argue how selection bias would not be a significant problem for making many kinds of practical inferences from the AAC reports.

Ken

edit: name of statistical compiler corrected -- Thanks to kind suggestion by J.Colin.Olson

Ryan Williams wrote:I always thought that most DEATHS were caused by rappelling accidents. Also that most rappelling accidents result in death, and not injury.

Well the second claim has a strong "whiff" of selection bias -- since deaths are more likely to be more widely reported and remembered than non-death rappelling accidents. And accidents that result in only minor injuries might not be reported at all. Some climbers might be so embarrassed about their reputation that they might tell other climbers that some rappeling injury was from a driving accident.

The most recent rappel accident (going off the end of the ropes) at one of my local crags did not result in death. I recall two cases in recent years of very experienced prominent climbers taking falls getting lowered off the end of the rope without death.

Ken

Another deep problem is guessing what percentage of climbing accidents are suicides (or failed suicide attempts).

Or "russian roulette", where someone chose to attempt a climb under conditions which were obviously (and unnecessarily) dangerous -- or "unnecessarily dangerous" in a way not obvious to people other than the climber(s) themselves.

Or deliberately taking a significant risk which was completely unnecessary for completing a climb.

This is not just about climbing -- there are several other activities in life where suicide attempts and "russian roulette" games by depressed people are believed to be included in the death and accident statistics (e.g. non-intoxicated people alone in their car driving at high speed into a concrete bridge abutment).

Climbing can provide familiarity with environments and risky moves which provide temporarily depressed people with (additional) opportunities to inflict death (or substantial probability of death) on themselves.

Some of those kinds of deaths (or serious injuries) surely are present in the AAC accident reports.

Ken

Unfortunately there just aren't any good sources of climbing accident statistics out there right now, which I discovered when I needed some for my thesis research a few years ago. ANAM is a good collection of anecdotes and is useful as far as that goes, but I doubt any accurate conclusions can be drawn from the data as a whole.

Also be aware that there may be inaccuracies in that info - I can only speak from the one case I know about. I was in a climbing accident a few years ago that made it into ANAM. About the only thing they got right was my name and age, as the report was obviously drawn directly from the Daily Camera story which reported my accident. We all know how accurate those stories are likely to be in terms of climbing accidents, so why ANAM used it and represented it as truth is baffling to me. But neither ANAM nor the Daily Camera got any input from me on the story, and neither was interested in my offer to correct the record after they were each published.

That's fine, I'm not really worried about the false report out there - but it did make me suddenly question how accurate everything else in ANAM might be. I wonder what percentage of their accident reports are taken directly from local newspapers who know nothing about climbing and neither newspaper reporter nor ANAM make any attempt to contact the climbers involved to confirm the info.

Note to self: Accidents in North American Mountaineering is nothing more than entertainment.

"Most of the key problems are identified by the compiler/interpreter Stephen Abegg himself on this page"

I'm not involved in this conversation, and honestly gender stuff bores me, but Steph Abegg is a woman not a man. And I guess that next to her climbing resume that makes me.... a little boy.

Em Cos wrote:Unfortunately there just aren't any good sources of climbing accident statistics out there right now

I doubt there ever was, or ever will be. It's really a difficult and costly sort of thing to gather, and the payback for most of it is not simple or obvious.

Still the (flawed) collection of the reports is useful to me. Like Saxfiend's point about Rappelling as a cause of accidents. That struck me because I reasoned that rappelling is one of the more obviously identifiable "causes" of an accident, less open to interpretation or ambiguity than many others, very likely to be included in any public report, and pretty likely to be selected as the "main" cause by the statistical compiler if present.

. (Similar arguments could be made about the "equipment failure" cause which I earlier mentioned -- something which much engages MountainProject forum discussion when it happens) .

Suppose the percentage of accidents with Rappeling as main cause were 60% (instead of 3%). That likely would have made me re-think my selection of climbing routes -- to avoid routes with rappel descents. Of course I (and many people) would then be very interested to know what sort of rappels seemed to be associated with the accidents, or what procedures -- and that would enable me and others to refine our selection of what routes (or procedures) to avoid.

But since the statistics compiled by Steph Abegg from ANAC came out as only 3%, I do not feel any need for that sort of avoidance or investigation. So that's an "actionable" inference from ANAC reports.

On the other hand, it does not make me feel OK with relaxing any vigilance or discipline with procedures in rappeling (since I assume such vigilance is what enables the percentage to be so much less than 60%)

Ken

J.Colin.Olson wrote:"Most of the key problems are identified by the compiler/interpreter Stephen Abegg himself on this page" I'm not involved in this conversation, and honestly gender stuff bores me, but Steph Abegg is a woman not a man. And I guess that next to her climbing resume that makes me.... a little boy.

Thanks for the correction.
I will now fix that in my earlier post.

It strikes me that another actionable finding from the statistical compilation of ANAS reports is the percentage of accidents that took place in the Descent phase of the climbing.

Ryan Williams wrote: One thing that jumped out at me from these pie charts is the percentage of accidents related to rappelling. When you read the internet forums, it seems to be taken as gospel that more people are injured or killed in rappelling incidents than any other single factor; but this linked analysis shows that rappelling is a primary cause in only about 3% of incidents. JL I always thought that most DEATHS were caused by rappelling accidents. Also that most rappelling accidents result in death, and not injury. If you look at the first two graphs, you can see how few deaths there are per reported accident. Those pie charts include ALL reported accidents, not just deaths. I'd bet that if you drew up new pie charts with only the accidents that resulted in death, you'd see the percentage caused by rappelling go much higher. What immediately stands out to me is that the amount of deaths has remained fairly constant over the last 30-40 years, and that the amount of reported accidents has decreased substantially since the 80's. I'd like to know what the spikes in 1976 are from, as well as '80 and '86. Without explanation I'd say that the data is a bit suspect.

In my unscientific sample as a climber since 1974, there have been a lot of waves of popularity for climbing over that period.  One year lots of new climbers which you can see to some extent in summits for places like Devil's Tower.  
Also hard to know percentage of under-reporting which may very. 70s saw some very sketchy ideas that if it wasn't risky it wasn't real.
I tend to think the under-reports aren't as much of an explanation as variation in the numbers of climbers. Washington state alpine accounts for a large number of fatalities and weather plays a big role there.

Welcome to 2017