Originally Posted by Traveler
Your posts are always worth reading!
humor is the gadfly on the corpse of tragedy
Furthering the hijacking.... Just making sure.... you do realize that there are NO grizzlys in Colorado, right? Furthermore, I've been hiking and backpacking in CO for 40 years now, and have seen 5 bears, all timid. Hiking 5 months on the Appalachian trail, I saw 11 bears. Just sayin... bears are not any more of a problem in CO than on eastern trails, probably way less of a problem.
BTW, I have respect, but no real fear of black bears, but when it comes to Grizzlies.... No thanks! Na-ga-da.
Actually I didn't know that. I just assumed that since you have mountain lions and at least one rattle snake every 50 feet, then surely there are grizzlies waiting to kill me, too.
I've been to CO twice -- once to help daughter and SIL move, once to dog sit while they were on vacation. Going again this Thanksgiving, though.
I've hiked Amicaloa to Daleville, saw two bears -- one ran when it saw me, the other looked at me for a looooong time and kinda ambled off the trail in a way that left me wondering where in the brush and rodo-hell it was ahead.
Well reasoned.
That said, the murder rate in my state is less than 2 per 100,000. And the murder rate for people killed by a stranger around here is much lower than that.
But the good news is that the people in Kansas City MO can rest easy in the knowledge that their community is as safe as the AT!
Here's where people differ in their thinking: I am never scared, but I am almost always able to protect myself. I don't give a hoot about statistics, because trouble comes in stores, on roads, parking lots, trails, parks, your own driveway, your house, your workplace, your church, etc. If you happen to be one of the people who are violently accosted, believe me, it won't matter to you where you are or what you are doing, or the likelihood of trouble. Relative probability according to statistics won't matter at all to you at that point. Police can arrive after an incident and clean up and maybe collect evidence. Not much help. A while ago, a high-ranking official in my state responded to an inquiry saying that Virginia State Parks are safe. Safe is relative to how you are impacted. No place is safe unless you are OK with being somewhat safe - relative to somewhere else.
I've learned....
That a smile is an inexpensive way to improve your looks.
Well said! I share your views,am always as prepared as I can be depending on where I am.No,I would not expect to necessarily win against an aggressor like the murder that occurred this summer on the trail but one thing is for sure,I would at least be in a position to resist if a speedy retreat was not possible.
Neat. I used to tutor advanced statistics.
I'm glad someone else has the duty to try to explain where some of the calculations have gone wrong in this thread.
This year there was a guy who got at least to Shaws with a gladius.
Though to be fair, he slack packed almost the entire trail, so his gladius wasn't actually with him while he was hiking.
As I think of this more, most people murdered in the US are murdered by people they know. Perhaps estranged lovers, a drug deal gone bad, gang turf wars, etc. Since my ordinary life in civilization carries practically no such danger, I consider myself to be statistically much safer than average. I'd guess that if you're reading this right now you also are a person living a safer than average life (let's face it, there probably aren't many street gang members reading this).
But on the AT it seems that almost all known murders are by strangers which means that all of us kind of share an equal risk with the exception than men and women likely do not have equal risk. Although the risk on the AT may not be substantially worse than the average in the US, most of us, when hiking go from below average risk to average risk, which means our risk is greater on the trail than in every day life.
Nevertheless, I hope to see some of you on the trail next spring on my next hike.
Would love to hear your more correct analysis.
(Yours is very well reasoned and articulated, FlyPaper.
Here is a larger question.
Would one or more of the 7 thru hikers who was murdered on the AT have acted any differently after meeting their soon-to-be killer if they had known the AT murder rate was actually 50% higher than Chicago’s (using your 31/100k number) — rather than the 1 in 6 million figure the ATC likes to trot out after each tragedy?
It should be hard to fathom that 7 Thru hikers were killed on the AT. Not rationalized as a product of large numbers.
For every 3000 thru hikers that joyously celebrates on top of Katahdin, one didn’t because he/she was murdered many miles into their thru hike.
Sorry folks, those attempts at any sort of statistical analysis are all way off, not even close to the right order of magnitude. No, I will not attempt to present any of my own, not worth doing and it would fall on deaf ears anyway. Any analysis has to use the metric "hiker miles" or even more accurately "hiker hours". A quick glance sees none of this; If I missed someone doing this, I apologize, this thread blossomed in silliness quickly, and I have no desire to sift through to find the meaningful posts.
Carry anything you want, big knives, guns, machetes, who cares (RYOH R=ruin)
I stand by my analysis.
Also, while I didn't mention "hiker hours", I did mention "person-years". For my purposes, "person" and "hiker" would be interchangeable. And if you need help converting hours to years, I can send you a link.
Consider the following math "word problem":
There have been 20,479 completed thru-hikes. If 23% of thru-hike attempts have been successful, how many thru-hike attempts have their been?
4eyebuzzard worked this problem and came up with 90,000 as an estimate. One can squabble about the 23%, but 4eyebuzzard's methodology for estimating thru-hike attempts was 100% SPOT ON. I have two high school students in my house and both of them can solve this problem. In fact, I think they both could have solved it while in 5th grade. But the response from the other math expert on this thread was the following:
Why would you limit this to successful thru hikers only? You're comparing apples to oranges.
I took this further and scaled the number based on "person-years" and used that to compare to publish statistics from cities around the US. I stand by my methodology. My apologies for not converting "person-years" to "hiker-hours" for you.
If you have something useful to add, I'd be interested I'd be interested in hearing it. But I don't think the air of condescension is warranted.
Personally, if you do produce an analysis that finds an answer off by more than an order of magnitude, I'd expect to be able to shoot holes in it quite easily.
OK! So, I'll push at your analysis a bit... lets work together here and try to come up with a reasonable number, sorry if I was blunt, I wasn't picking on you specifically. And please if I stop, how about you stop with the insults (your "word problem")?
First of all, what IS your number? I would assume the "number" we're looking for here is the probability of being murdered on the AT for a thru hike, right? Or maybe I'm missing some point on what you specifically are saying.... This thread is about danger on the AT, if you're talking about something else, then I'm NOT talking about your analysis.
Next, what have Thru-hikers got to do with it (maybe we agree here?)? The AT is dominated by hikers, not thru hikers. Please look up the total hiker-hours on the trail over the last 45 years before you start play with stats. If you want to use "person years", that's fine (yes hiker=person), but how would you convert hiker hours to hiker years? Using the factor 365.25*24 would be way off, because the vast majority of hikers on the trail are day hikers, spending what, 6-8 hours? Combine those with thru hikers (and section hikers) who spend, say an average of 18-20 hours a day on the trail (remember town stops!), the average might be 7-9 hours, so I would use 365.25*8 or so for a conversion from hiker hours to hiker years. How about we just say "hiker days"?
Edit: Aha, NOW I see your detailed analysis down in post 67, I didn't see it earlier, I only saw your more recent post. I'll take a closer look later today, gotta do something more useful first!
Sorry for the insults.
The goal is to compare the danger of life on the trail by converting to a form used for typical US cities (murders per 100,000 per year). I think we're on the same page that we don't count total successful thru-hikers nor total thru-hike attempts, but we find hiker time on the trail. As far as time in the city, I would count that, but I could respect a choice to not count that too. Personally, if a thru-hiker got killed while doing laundry in Damascus, the news would be reported as "thru-hiker killed while thru-hiking". And I think all agree we must count thru-hike attempts, not thru-hike completions because no one murdered while thru-hiking completes that hike and we have no way of knowing whether they would have been among those who complete the hike or not.
For simplicity, I've narrowed it down to just thru-hikers (only counting thru-hiker time on trail and thru-hiker murders). There have been section hikers murdered too, but we don't count their time on the trail nor any homicides of non-thru hikers. The number given of 7 thru-hikers killed while thru-hiking is something I accept, but if someone wants to challenge that number with more careful analysis, I'm open to input. Clearly it is not 1 and it is not 50. 7 seems about right.
One can separately analyze danger to section hikers and day hikers and aggregate the numbers at the end, but estimates are harder to attain for time spent on the trail. Perhaps adding section hikers and day hikers would change the numbers significantly up or down, but I think it is still interesting to to calculate just for thru-hikers. Personally, I would guess that section hikers share the same risk profile, but probably not day hikers. As a section hiker, I consider myself to be at similar risk to that of a thru-hiker.
You ask what is my number. "I would assume the "number" we're looking for here is the probability of being murdered on the AT for a thru hike, right?"
Primarily my number is "homicides per 100,000", just like we find for Kansas City and Baltimore so we can make an apples-to-apples comparison, which must take into account total time on the trail being much less than the average time a city dweller spends in their city. This could be converted to probability of being murdered on an AT thru-hike by scaling the total time spent. You'd probably want to assume a completed thru-hike when getting that number because obviously the odds of being murdered on a complete thru-hike is higher than the odds of being murdered if you quit at Blood Mtn.
To estimate the total danger on the trail, one could do these steps.
1. Estimate the danger that thru-hikers experience.
2. Estimate the danger that section-hikers experience.
3. Estimate the danger that day-hikers experience.
4. Aggregate them with appropriate scaling based on proportional population-time to get a "homicides per 100,000 for person year".
One could also approach it this way...
1. Estimate average time on the trail for all hikers.
2. Find total homicides for all hikers.
3. Divide and scale accordingly to get a final "homicides per 100,000 for person year".
I think the second approach is what you're leaning toward and that both of these approaches are an attempt to estimate the same final value.
I have been focused on the first approach, and within the first approach, only step #1. While step #1 does not give the answer to how dangerous is the trail, it is arguably something more relevant and interesting to any potential thru-hiker. As a section hiker, I'd just guess my risk profile is similar to that of a thru-hiker. A day-hiker is probably a lot different, although I wouldn't know whether it would be higher or lower.
Not going to quote your post FP, too much screen real estate... I understand where you're coming from, but I see a couple of serious flaws in yours and others' arguments, at least from my perspective.
First, in terms of "danger for a hiker per hour (day, month, year) on the trail" I see no actual evidence that section/thru hikers are in any more danger. Am I missing something? Is there evidence that, say, someone who camps along the trail vs. just hikes for a day is in any more danger? I suspect there is some correlation, but is there evidence to this? When lacking evidence, but suspecting something is correlated, I'd start with a factor of 2, just to start, meaning you're twice as likely to die hiking and camping vs. just hiking. Very flaky, but really, we're just looking for an order of magnitude here.
One serious flaw is the statistical significance of a very small number of murders (7 in 45 years, right?) vs. the thousands in a big city per year. This one little fact is probably a deal killer in terms of any meaningful results, but heck, we'll ignore it for now (not being sarcastic).
Also be very careful about your thru-hiker count. Lots of us, myself included, don't register either starting or finishing my long hikes. I probably should, it would help trail organizations know better about trqaffic, but I just get lazy and don't. This probably only makes a 30-40% difference; I quote that number after discussing this with the ATC some years ago, they estimate about 2/3rd register. This difference is minor w.r.t. order of magnitude estimates.
What I do not see is any way to compare safety on the trail to safety in the big, relatively dangerous cities like Baltimore or Chicago. None of the arguments that I read below make sense to me in trying to form this comparison. I'll think more about it.
I believe the only reasonable way to estimate the danger to any given hiker in terms of his chance of getting murdered for any given length of time on the trail can only be done using a Monte Carlo analysis. I'm probably copping out saying this, but I'm biased because that's what I did for roughly half of my 30 year engineering career in the rocket biz. Basically when it's too hard to figure out a given statistic, just make some rough assumptions, throw them in a pot, make a couple billion random draws and collect the results. It's really easy to do, in fact, next bad weather day, maybe I'll try just that.
Back to work....
I agree that there is no evidence that section/thru/day hikers have a different risk profile. I think I posted my most recent post while you were writing your post. I would guess that spending the night in camp is the more risky time and this seems to align with the reported crime data also. Section hikers and thru-hikers share this risk, but not day hikers. But, for my analysis I've focused only on thru-hikers, the rest being side conversations that are interesting. One could convince me that the trail is safer than I've estimated because day-hiking is so common and so much safer than overnight hiking. But I'm still more interested in the danger to thru-hikers and section-hikers.
Thru-hiker count, I agree is suspect, but probably very useful for rough estimates as you say. Also, I have thrown out the estimate on my own that the average thru-hiker spends 3 months on the trail. That's a rough number and probably overestimates a bit. With numbers that rough, it's hard to squabble about other rough estimates.
And the weakest is probably the very small number of 7 murders. You take what you get. There's nothing better to use. We know we are getting at best make a rough estimate. And I jumped in this thread in the first place to counter the claims that the stats prove how safe the trail is compared to the city. The stats, to the extent they are able to feed estimates point the other way. At least that is my contention.
As far as Monte Carlo, I don't see how that will help, but perhaps I'm missing something. If you simulate a billion thru-hikes assuming that 7 out of 90,000 get killed, you'll end up getting the ratio of 7 out of 90,000. From where I see it, the initial assumption will determine the outcome.
Sorry, I just don't get it. I just don't understand why you separate the small number of thru hikers from the huge population of hikers, it seems to be this totally skews any possible results, hence why I made my initial brash and rude remarks.
How I see a sequence of arguments is this:
Take the last 45 years.
Estimate the total number of hiker-hours on the AT. Just for grins, pick a wild guess for now, lets say 45*5,000,000*8 (45 years, 5 million average hikers on the AT a year, hiking an average of 8 hours each). We can refine this later if necessary. This would be 1.8 billion hiker-hours.
Take the number of murders, apparently 7.
So the rate of murder is 7 murders per 1.8 billion hiker hours including everyone, day hiker, section, thru, whatever.
Estimate the time on the trail for a successful thru hiker. Let's say 150 days at 20 hours a day (allowing for town stops one out of every 6 days, and we'll ignore his/her danger of getting to/from town). Call it 3000 hiker hours, meaning a successful thru hiker is on the trail roughly 3000 hours.
So the chance of a thru hiker getting murdered on the AT is 7 murders * 3000 hiker hours / 1,800,000,000 hiker hours (in the last 45 years) = .000016 murders per successful thru hiker, or call it 1 in 60,000 chance.
If one wants to claim that a person spending a night during his/her hiker-hours is in more danger, that's fine, I doubt if it's a factor of 2, but call it that, so call it 1 in 30,000 chance of getting murdered on a thru hike.
I really am curious though, what makes any of us think that being murdered is more likely if you are camping? Being dark and scary isn't enough of an argument. Does anyone know, of the 7 murders, how many of these murders occurred at night with a camping victim? (tent, shelter, whatever). We all know the most recent one was indeed a camper, in his tent, poor soul.
That's my rough number. 1 in 60,000 (or 30000 if we think camping makes it twice as likely, which is a stretch). It seems pretty different from anything I see below, like 7 out of 90000 (or one in 12000), but I do admit it is in the same order of magnitude.
I'm gone for the next 5-6 hours, can't comment anymore until then.
(side note: the previously mentioned Monte Carlo analysis would take into account where along the trail hikers were, his/her age, gender, that sort of thing, it would't be too hard to construct a series of a billion random hikers, nuff said, I just cop out and do this sort of thing with engineering problems, easier than thinking too much)
I'm retired, no one's paying me to spend the time to gather the data properly and to look for patterns. Like you said, in another post. You're doing word problems, at an 8th grade level maybe (probably even lower, the kids are learning a lot more math at earlier grades these days.)
Your person hours idea isn't a terrible way of looking at it. But, you have to start at the beginning and decide if the initial question even makes sense. No one starts a hike with 100% certainty that they'll be a successful thru-hiker. There are questions on what makes a specific hiker more, or less vulnerable, there are questions on the motive of the murderers, there need to be control groups, the sample size of murders is tiny!, and that creates it's own problems with statistical significance.
My initial point is correct however, as there are already quite a few people tossing out variants of the 7 out of 100,000 number as of it's gospel. It isn't.
How to correct this number and this conclusion. I can't even explain everything wrong with it without hours of effort. At a minimum you'd have to take a solid college statistics course or two to even understand what's involved. Here's a link to just a basic structure of how to conduct a study.
http://gchang.people.ysu.edu/class/pstudy.htm
If I were world emperor, every High School would require a statistics course, a formal logic course and a debating course.
Edit: My posts sometimes come across as know it all, so I'll just state that there's a hell of a lot I don't have a clue about. Bad statistics just happens to be a peeve of mine.
I'm not a statistician, though I do have a degree in Math Education (high school algebra and geometry). I haven't tried to understand all the points made above. I just wanted to point out what I think might be a hole in the considerations:
Where are the criminals? Given that nearly all hikers I've met (day, section, or thru) appear to be decent law-abiding people, there would appear to be very few individuals on the trail who have murderous inclinations. Criminals tend to stay in cities. Yes, I know small towns have criminals also. If the people who wish to kill aren't on the trail [very much], what difference does it make how many miles, or hiker-hours, or person-years, or whether I sleep at home or on the trail? If the bad people aren't there, I'm safe from bad people.
So shouldn't the calculations take into account the distribution of bad people among the population?
Posting Permissions
