Nobo Thru-Hike Start Date Graph - and a plea

[Seatbelt]

The graph even looks like a "bubble".

[RED-DOG]

Theirs absolutely no WAY to know exactly how many people hike the AT, theirs to many road crossing that a lot of people starts at and at least half the people don't even apply for the 2000 miler status so everything on this thread is just SPECULATION not TRUTH.

Theirs a lot more people that start In Feb and finishes than what's on Laurie P table thingy.

[MrGonzalo]

Fantastic information...
Does anybody know whether there is any statistical information regarding shelter occupancy during the season? While the kick-off information is good, the idea of lessening the impact and avoiding shelter crowding relies on the assumption that all hikers are traveling at a constant and uniform pace.

For example, a 5-monther that decided to wait a week to avoid the March 1st crowd would undoubtedly be dealing with the same crowd just days later. In this situation, would she (and the trail) be better served by starting with the crowd and quickly separating from it? Food for thought.


Sent from my iPhone using Tapatalk

[full conditions]

Theirs absolutely no WAY to know exactly how many people hike the AT, theirs to many road crossing that a lot of people starts at and at least half the people don't even apply for the 2000 miler status so everything on this thread is just SPECULATION not TRUTH.

Theirs a lot more people that start In Feb and finishes than what's on Laurie P table thingy.

At least half the people don't eve n apply for 2000 miler status? Any actual data support this claim or is this just SPECULATION not TRUTH?

[Slo-go'en]

At least half the people don't eve n apply for 2000 miler status? Any actual data support this claim or is this just SPECULATION not TRUTH?

I like to make up statics too, but that's true speculation. I would venture to guess a few more start in Feb and finish then the data Laurie presented shows, but not a lot more.

If an electronic counter were to be set up on the trial near Springer, it would give a reasonably accurate number of NOBO hikers during the prime time of March and April as it would be safe to assume that the majority of those passing by are wannabe thru's

[full conditions]

Laurip has the compiled average of five years worth of data. Unquestionably some folks fell through the cracks along the way but that does not in any way substantially lessen the quality of her data - the graphs speak volumes about when peak travel times are for aspiring thrus on Springer Mountain and, by extension, for most of the AT in Georgia.

[mountain squid]

If an electronic counter were to be set up on the trial near Springer, it would give a reasonably accurate number of NOBO hikers during the prime time of March and April as it would be safe to assume that the majority of those passing by are wannabe thru's

No, I don't think that it would. In addition to the groups and other sectioners that I mentioned in my post, there are also numerous dayhikers. How could an electronic counter determine who was who?

See you on the trail,
mt squid

[Alligator]

The graph says average 2010-2014. LaurieP I think based on your explanation you are saying you have totaled the number of starts by start day based on response from completions and that the total number of completions is similar to 2014. So the number on the graph, say for 4/1, represents the total number of people reporting that start day in your completion records and is based roughly on the total from 2014? That number will be higher or lower depending on the number of starts. (Might be better to use frequencies).

The assumption being made though is that completions are not dependent on start date. Start dates are being extrapolated out to the non-completion group. This is not necessarily supported. The dates for the completions could be in the sweet spot for timing a hike. There are likely fewer completions percentage-wise for earlier months of the year and late starts due to cold and arriving late to BSP. This is not necessarily the correct shape for the graph. I'm not saying it is incorrect but it might be flatter.

Also mentioned is the day of the week. While first of the month and 3/15 may be significant specifically, it would be interesting to see the data evaluated simply as day of the week regardless of month and then by month, what week, what day. For instance starts on the first Sat in March. Anecdotal evidence suggests that weekend days are prime days to start. However, in the data you present, each day of the year can at most represent just one Saturday and Sunday out of five possible days of the week (even with the Leap Year). There might be spikes of starts on weekends that rival the first of the month or 3/15.

[Alligator]

Fantastic information...
Does anybody know whether there is any statistical information regarding shelter occupancy during the season? While the kick-off information is good, the idea of lessening the impact and avoiding shelter crowding relies on the assumption that all hikers are traveling at a constant and uniform pace.

For example, a 5-monther that decided to wait a week to avoid the March 1st crowd would undoubtedly be dealing with the same crowd just days later. In this situation, would she (and the trail) be better served by starting with the crowd and quickly separating from it? Food for thought.


Sent from my iPhone using Tapatalk

Are you addressing the idea of getting hikers to choose less used dates? I think so.

A good many people don't really know how fast they can hike when they get their trail legs. Each start date would have a mix of people of different speeds leaving that day. Faster hikers will catch slower hikers from an earlier date. Slower hikers will fall behind and be caught be faster hikers. However, for any two given start days, the hikers from each day are separated by a lag of X days. If it is a random separation of hikers by start date, the graphs of the completion times would most likely be bell-shaped for each of the two groups but separated between the two means by the lag time. (Roughly) Using your 5-monther example the 5 monthers from Date 2 would be about 7 days behind the 5-monthers that started on Date 1 if Date 2 is 7 days after Date 1. Spreading start dates out can reduce peak days. It would also increase minimum number of hiker days at a location, not so much an issue necessarily.

Another way to think of it is thinking waves. Say you have 100 hikers to start. You could send out three waves of hikers-25 on 3/1, 50 on 4/1, and 25 on 5/1. If the wave represents the hikers, the 3/1 starts will spread out so many days apart, the 4/1 so many, etc. Always a big wave for the middle group. Spread it out over 8 weeks and the waves will be relatively uniform. Spreading out the starts evenly reduces the peak heights of the waves, but it does increase troughs. Then attrition dampens the wave the further out it goes.

[RED-DOG]

Why is it only march and April for a more ACCURATE count it should be JAN/FEB through MAY.

[Alligator]

Why is it only march and April for a more ACCURATE count it should be JAN/FEB through MAY.

Are you responding to my post or to another?

[mountain squid]

Why is it only march and April for a more ACCURATE count it should be JAN/FEB through MAY.

?!?!?

I believe the point of the thread is to encourage hikers to start on dates that are not consistently busy, thereby reducing congestion and impact at shelters and campsites in GA. There are not as many hikers that begin in Jan/Feb. There might be more hikers that start on 1 Apr than all of Jan/Feb . . . .

So please reconsider if you were thinking of starting March 1, March 17, April 1. Also, if you can start mid-week instead of the weekend, that will help even out the flow of hikers too.

See you on the trail,
mt squid

[Speakeasy TN]

This year will be my second attempt. I started on 3/17 in '11 because it was "lucky" and mid-week. This year it falls on Tuesday. I'd bet that the Saturday before or after would be busier. I'd love to think that a mid-week start would flatten the spike. That graph shifted to day of week spikes would be fascinating. I'd bet anything that the "key date" starts are MUCH less of an issue when they fall mid-week.

[chris.ault.37]

We're starting April 14 (Tuesday) in Atkins, Va. Figured we'd really break from the crowd!

[Lauriep]

OK, finally! The January through June chart.

AT Nobo thru-hiker Start Date graph 2010-2014 jan-june.jpg


This is the data drawn from 2,000-milers, who represent about quarter of the people who started. I'm sure it does not mirror the actual boots on the ground precisely, but I think it's the best or perhaps most useful and accessible data available. A significant part of the reason we have our 2,000-miler recognition program is so that we can collect data for this just this kind of use.

As a couple of you have said, the number of starters in January and February are no doubt higher than are reflected in this chart. And the drop-out rate in those winter months is higher. And yes, the whole issue of days of the week adds another layer of complexity and nuance this graph doesn't address. Someone smarter and more patient than me would have to calculate that.

As imperfect as these graphs are at representing the absolute true picture, I think they convey something of the large number of hikers starting at Springer in a fairly compressed time frame, with sharp peaks on some days that causes crowding that is detrimental to campsites and to the type of experience the A.T. is intended to provide.

These graphs are meant to be a wake-up call about the situation now, and to help all of us in the A.T. community think about what things will be like if A Walk in the Woods movie causes the number of thru-hiker starters to climb like it did after the book came out almost 20 years ago.

ATC has been working on a voluntary registration system for thru-hikers these last few months and hopes to have a test version available soon. The registration will be free.

We're taking many of the ideas we heard at our discussion at The ALDHA Gathering in Williamstown, MA and have been posted on various threads at WhiteBlaze. Some we didn't have the time, expertise, or funds to implement this year. We have a group comprised of ATC staff and volunteers, trail clubs, agency partners, service providers and hikers working on this, as well as a representatives from the PCTA and Leave No Trace, Inc.

This year's version is essentially a trial run. We'll be welcoming feedback (constructive is always the most helpful, for us anyway). Some small or easy tweaks we might be able to implement this year, but other things will probably we'll have to till next year to implement.

The more hikers who participate this year, the better. We hope all those folks who have signed up here, on Appalachian Trials, various Facebook groups and elsewhere will be willing to take a few moments to register with ATC.

Then next year we'll have graphs that reflect a higher percentage of the actual thru-hike starters.

The voluntary registration is just one part of what we're doing to get ready for the expected increases:

• hiring/training more ridgerunners and caretakers
• promoting alternative thru-hikes (flip-flops)
• Expanding Leave No Trace education
• Working with guidebook editors and app developers to remove listings of unsustainable non-designated hiker-created sites that were heavily impacted

One app developer is hoping to be able to create a feature that will enable hikers to report how many hikers are at a campsite each night.

[atraildreamer]

At Laurie's request, I converted the chart to pdf format if you want to download it and print it out.

[Kerosene]

I "stuck" this thread in the Class of 2015 forum.

[JeremyB]

This is the data drawn from 2,000-milers, who represent about quarter of the people who started. I'm sure it does not mirror the actual boots on the ground precisely, but I think it's the best or perhaps most useful and accessible data available. A significant part of the reason we have our 2,000-miler recognition program is so that we can collect data for this just this kind of use.

As a couple of you have said, the number of starters in January and February are no doubt higher than are reflected in this chart. And the drop-out rate in those winter months is higher. And yes, the whole issue of days of the week adds another layer of complexity and nuance this graph doesn't address. Someone smarter and more patient than me would have to calculate that.

As imperfect as these graphs are at representing the absolute true picture, I think they convey something of the large number of hikers starting at Springer in a fairly compressed time frame, with sharp peaks on some days that causes crowding that is detrimental to campsites and to the type of experience the A.T. is intended to provide.

I spent entirely too much time going over every trailjournals.com journal for the 2014 AT class. 288 data points. This dataset is also biased, as I think people who are journaling on TJ.com are better prepared than the general thru-hiker population. That being said, the average and median start dates for thru-hikers and people who didn't complete is very similar.

Thru hiker start date
Average – March 17
Median – March 19

Noncompletion hikes
Average – March 20
Median – March 22

Only 1 person started in January and 3 in the first two weeks of February, so I think your data is reasonably accurate even though only thru-hikers are the sample in your data.

Here’s the same chart for 1 year for all TJ.com entries. My data is sloppy because I didn’t differentiate between starting at Springer and Amicalola, so there is some fuzziness there.
2014_AT_Start_Histogram_Daily.jpg
2014_AT_Start_Histogram_Weekly.jpg

[map man]

JeremyB, thank you for compiling the data, crunching the numbers, and creating the illustrations. The comparison of completing thru-hikers and non-completing thru-wannabes was particularly instructive, as a lot of people over the years have mentioned that would be nice to know, but this is the first time I can remember seeing a quantitative attempt to figure it out. You mention concern about trailjournals users being more savvy than the average hiker, and that could be right but I wouldn't think there would be a dramatic difference. I also wonder about the difficulty in figuring out who quit the trail and who kept hiking but just quit journaling, when you were separating the completers from the non-completers.

I also think grouping the starting times by week in a couple of the histograms, instead of by day, really helped filter out some of the noise you get from the daily counts and is really useful.

I know first-hand how much work this project must have been and I for one am truly grateful that you did it.

Edit: I have heard people speculate over the years that the early starters cut down on their chances of completing, but this one year TJ sample doesn't seem to support that -- completers start three days earlier, on average, than non-completers. You wouldn't care to add more years to the database would you?

[Alligator]

I spent entirely too much time going over every trailjournals.com journal for the 2014 AT class. 288 data points. This dataset is also biased, as I think people who are journaling on TJ.com are better prepared than the general thru-hiker population. That being said, the average and median start dates for thru-hikers and people who didn't complete is very similar.

Thru hiker start date
Average – March 17
Median – March 19

Noncompletion hikes
Average – March 20
Median – March 22

Only 1 person started in January and 3 in the first two weeks of February, so I think your data is reasonably accurate even though only thru-hikers are the sample in your data.

Here’s the same chart for 1 year for all TJ.com entries. My data is sloppy because I didn’t differentiate between starting at Springer and Amicalola, so there is some fuzziness there.
2014_AT_Start_Histogram_Daily.jpg
2014_AT_Start_Histogram_Weekly.jpg

That the means (and the medians) are close does not mean that the thruhiker and non-completion distributions are the same. One could be flatter than the other or possibly skewed. Smoothing by week was helpful though. Plotting the two groups side by side in the same chart would potentially illustrate any differences if they exist.