 Poster: A snowHead
|
|
Adrian, must buy an Iphone.... NOW!!! LOL!!!
|
|
|
|
|
|
Obviously A snowHead isn't a real person
Obviously A snowHead isn't a real person
|
|
|
|
Well, the person's real but it's just a made up name, see?
Well, the person's real but it's just a made up name, see?
|
|
|
|
You need to Login to know who's really who.
You need to Login to know who's really who.
|
|
Thanks colinmcc but I knew the date on that page when I first read it last spring plus I thought its contents made its nature obvious. Also I thought that RPF asked a tounge-in-cheek question.
|
|
|
|
|
|
Anyway, snowHeads is much more fun if you do.
Anyway, snowHeads is much more fun if you do.
|
| Quote: |
Also I thought that RPF asked a tounge-in-cheek question.
|
You thought correct 
|
|
|
|
|
|
You'll need to Register first of course.
You'll need to Register first of course.
|
That app is completely useless without Facebook/twitter status updates!
Tux
|
|
|
|
|
|
|
|
| ed123 wrote: |
Some more thoughts about the 'study'.
No details on how far away from the start of the search the test beacon was.
Simulated avalanche 30m x 50m. ie small and just about within range of most digital beacons.
If the subjects started at the edge then primary search- for a signal would be fast with all beacons.
If started further away -- ie 50-70m let alone 100-200m then some of the analogue but not digital beacons might be in range and reduce time of primary search.. The paper does not mention this aspect at all- which is a serious omission.
Needs that detail- and detail for example of time for primary, secondary and fine search etc.
I wouldn't base anything on the paper as presented in the thread. |
Bottom line is that a relative beginner with an S1 will hone in with much more speed and reliability than with the basic analogue devices. This was just an interesting bit of testing not science.
The problem with the S1 is that it was released too early and has picked up a bad reputation. Now we have people chipping in with out-of-date information. Also, resorting to accusation of fanboyism is unhelpful.
|
|
|
|
|
|
|
|
|
PJSki, I think that the way the data was gathered and presented gave the appearance of a scientific study with meaningful results. But either only half of it has been presented (which would be a worry) or it was done very badly.
|
|
|
|
|
|
You'll get to see more forums and be part of the best ski club on the net.
You'll get to see more forums and be part of the best ski club on the net.
|
ed123, the appearance of a scientific study? Don't be ridiculous. 
It is what is is and is plain for all to see.
|
|
|
|
|
|
|
|
PJSki, humm
let's see; Title, Authors with affiliations, Abstract, Methods and Materials, Conclusions, Discussion,References- data
looks like a duck, walks like a duck sounds like it
just not a very good one.
|
|
|
|
|
|
snowHeads are a friendly bunch.
snowHeads are a friendly bunch.
|
|
also the authors themselves describe it as a 'study'.
|
|
|
|
|
|
And love to help out and answer questions and of course, read each other's snow reports.
And love to help out and answer questions and of course, read each other's snow reports.
|
ed123,
| Quote: |
2009 Transceiver Test
Assessing the new Three Antenna Transceivers |
Note the word TEST. Note the absence of the word SCIENTIFIC. And the fact they used logical methodology doesn't mean they were passing this off as scientific research. So I'm afraid you're just spinning - probably because you don't like the results.
But as you seem to care so much about it, I guess you must be right. 
|
|
|
|
|
|
|
|
|
PJSki, Recon you must have one then
|
|
|
|
|
|
You know it makes sense.
|
|
|
|
Otherwise you'll just go on seeing the one name:
Otherwise you'll just go on seeing the one name:
|
PJSki, you are getting more and more confused.
First it is not a scientific study, now it is a test (although the authors say it is a study) , then you say there are results- but there aren't any! That is the whole point- or lack of.
|
|
|
|
|
|
Poster: A snowHead
|
PJSki, however- it's Christmas ffs!
I think we are at cross purposes.
I come from a medical / scientific background and appraise medical journal articles as part of my job- and examine others in the ability to do so. So I can critique a study like this.
But I'm only a keen amateur skier at best and whilst I own a beacon or two I've never used one in anger nor ventured that far from the security of a resort without a guide.
I think you are coming from an opposite position.
My point is that the paper is a poor one and doesn't actually say anything (and anyway a difference of 30s or so is not that important compared to extraction).
I think your point is that having used lots of beacons the S1 is best as far as you are concerned. I couldn't say- I've never used one- but this study/test/paper says nothing about it.
Here in th eUK it is Boxing day- a day traditionally preserved or eating too much cold turkey and rows with relatives. Happy Christmas!
|
|
|
|
|
|
Obviously A snowHead isn't a real person
Obviously A snowHead isn't a real person
|
| ed123 wrote: |
PJSki, you are getting more and more confused.
First it is not a scientific study, now it is a test (although the authors say it is a study) , then you say there are results- but there aren't any! That is the whole point- or lack of. |
| Quote: |
Results
In the hands of a knowledgeable user there was only a small difference in search times
between the two-antenna and the newer three-antenna transceivers. The fastest was the
Ortovox d3 (avg 1 minute 13 sec), then the Tracker (1minute 35 sec), the Pieps (1 minute
36 sec), the Ortovox S1 (1 minute 38 sec) and the Mammut (1 minute 47 sec). The
slowest was the Ortovox F1 (at a still respectable 1 minute 56 sec). The standard
deviation (STDEV) shows how much variation there was in the times and a low number
suggests a more consistent ease of use, ranging from the Ortovox S1 at :23, Ortovox d3 at
:31, Tracker at :32, Pieps at :36, Mammut at :47, and the Ortovox F1 at 1:07. In the hands of an
expert user there appears to be little difference in the results using the multi-antenna transceivers
and all are faster than the F1 which is nearing two decades of use.
The results from the naive users are very interesting! Six out ten testers failed to find the
two targets using the Ortovox F1 in the allotted time. All the naive users did much better
using the two and three-antenna transceivers. The fastest was the Ortovox S1 (2 minutes
59sec), the Ortovox d3 (3 minutes 21 sec), Mammut (3 minutes 25 sec), the Tracker (3
minutes 26 sec,) and the Pieps (3 minutes 34 sec). |
I think you'll find the method they employed yielded the above results, plus the tables. They used a device for measuring the passing of time in increments we on Earth call minutes and seconds.
Anyway, call it what you like, it makes not pretense at being scientific.
|
|
|
|
|
|
Well, the person's real but it's just a made up name, see?
Well, the person's real but it's just a made up name, see?
|
In the Yuletide sprit of confrontation my point is that there are no statistically significant results or P values.
They make plenty of pretence of being scientifc.
|
|
|
|
|
|
You need to Login to know who's really who.
You need to Login to know who's really who.
|
| ed123 wrote: |
In the Yuletide sprit of confrontation my point is that there are no statistically significant results or P values.
They make plenty of pretence of being scientifc. |
Yes there are. I would have expected to see little differences in the pros' times across all models (which is what we see) but a marked difference with the novice users. The fact that many of the novices were unable to locate within the time with the older models but succeeded with the newer ones is significant. The fact that the S1 had the best average time less significant, but I would have put money on that result.
The S1 coming top should be a significant result for all those who have been going around saying it's rubbish. But sadly, and most unscientifically, these people seem to be going into denial in the face of the test results.
|
|
|
|
|
|
Anyway, snowHeads is much more fun if you do.
Anyway, snowHeads is much more fun if you do.
|
|
PJSki, no there aren't any stats on significance given- I know this.
|
|
|
|
|
|
You'll need to Register first of course.
You'll need to Register first of course.
|
PJSki, sorry - I am with ed123 on the quality of the study. It does follow a classical scientific experimental reporting pattern, and does attempt to look like a scientific paper.
Whilst it is at least better than most of the stuff we see posted here, there is nothing really about what the statistical power of the study is, etc etc in the paper. There is an attempt to have two different groups and a degree of "crossover" in the design, and SDs are given, but no more than that. Usually for most normally distributed variables we would be looking for perhaps 2SD +/- ranges before a significant p value had been reached that the results were not by chance. I personally do not believe that there were enough participants to allow a statistically significant difference to be shown - and for us to be confident that the results were not superior for one beacon by chance. (eg tossing 5 heads in a row shows a pattern - but doesn't mean the coin is fixed...)
It is certainly a step in the right direction, but at best the paper is a decent observational paper which contains some interesting findings and poses further questions. It certainly isn't the last word, but it is probably better than anything I have seen posted about beacons before.
However I have just bought a new S1... 
|
|
|
|
|
|
|
|
| ed123 wrote: |
| PJSki, no there aren't any stats on significance given- I know this. |
So there is nothing, according to you, significant about novices running out of time with the older units but succeeding with the news ones?
|
|
|
|
|
|
|
|
stoatsbrother,
| Quote: |
| However I have just bought a new S1 |
Why the S1?
|
|
|
|
|
|
You'll get to see more forums and be part of the best ski club on the net.
You'll get to see more forums and be part of the best ski club on the net.
|
|
|
|
|
|
PJSki, because a number of people I have skied with who use it like it. Because I think it does handle multiples well. Because I suspect that the study linked to here would have been significantly positive if it had been powered properly to reach an answer.
Playing around with it and a barryvox I am not finding it hugely faster and need to practice the pinpointing more.
However I suspect that for single burials what we really need to be practicing to save time is search teamwork and digging.
|
|
|
|
|
|
snowHeads are a friendly bunch.
snowHeads are a friendly bunch.
|
|
ed123, yep, you two mean different things.
|
|
|
|
|
|
And love to help out and answer questions and of course, read each other's snow reports.
And love to help out and answer questions and of course, read each other's snow reports.
|
OK, but the question remains:
So there is nothing, according to you, significant about novices running out of time with the older units but succeeding with the news ones?
|
|
|
|
|
|
|
|
PJSki, it is not necessarily statistically significant. If there were 100 participants it might be easier to know that this wasn't a chance happening. In drug trials companies will often be quite selective about the outcomes they publish in small sample studies.
The study does not prove that the differences do not exist, but nor does it prove that they do not genuinely exist at a level beyond chance.
|
|
|
|
|
|
You know it makes sense.
|
I am very sad.
I have just done the stats on the first 2 columns of data presented- but made a mistake th efirst time and am repeating them
Last edited by You know it makes sense. on Mon 28-12-09 23:13; edited 1 time in total
|
|
|
|
|
|
Otherwise you'll just go on seeing the one name:
Otherwise you'll just go on seeing the one name:
|
| stoatsbrother wrote: |
PJSki, it is not necessarily statistically significant. If there were 100 participants it might be easier to know that this wasn't a chance happening. In drug trials companies will often be quite selective about the outcomes they publish in small sample studies.
The study does not prove that the differences do not exist, but nor does it prove that they do not genuinely exist at a level beyond chance. |
Disagree. 10 is enough of a sample when comparing old and news transceivers in the hands of novices. It's also what I've found in the 10 or so tests that I've done. We aren't tossing a coin here, so there's no need for a large sample base. The differences in these devices are known and understood already, so it's only a matter of measuring the difference. I don't need to test the sprinting ability of 100 80 something men to know that they are a lot slower at sprinting than 100 20 something men. The question isn't are these devices easier to use for novices but rather how much easier are they.
|
|
|
|
|
|
Poster: A snowHead
|
I am very sad.
I have just done the stats on the first 2 columns of data presented
The P value for the difference between the average time faster D3 and average time slower S1 in the hands of professionals, using a 2 tailed paired students T test was greater than o.05. Meaning that it was not satistically significant.
In the case of the fastest (the S1) and slowest digital beacons (DSP) when used by novices, using unpaired 2 tailed T tests between the P value was 0.22 and not at all significant.
|
|
|
|
|
|
Obviously A snowHead isn't a real person
Obviously A snowHead isn't a real person
|
PJSki, I personally agree with your view that the S1 is better for novices and have put my money where my mouth is. But the study linked to here does not prove it. Statistical and scientific methodology does not always look obvious and what looks obvious isn't always true.
If you can generate sufficient p values from the data I'd love to see them. I think the data supports but does not prove the conclusions drawn.
[edit] I see ed has run the numbers. If you doubled the participants and added in other trials may be you might get down to so value of below 0.05. It is a bit surprising that they reported SDs but not p values.
|
|
|
|
|
|
Well, the person's real but it's just a made up name, see?
Well, the person's real but it's just a made up name, see?
|
| ed123 wrote: |
| I am very sad. |
Yes I'd agree with that. Whilst you maybe correct in 'rubbishing' the stats, you are also insinuating that the overall accuracy is in question with respect to the S1.
Buy / borrow / whatever, an S1, take yourself with a largish group to the avy park and then draw your own conclusions. If you want to make things difficult you can use the S1 in analogue mode which makes it no different from the F1 etc.
The outcome of the 'study' is exactly in line with my experience.
I think this thread has become confusing for people looking to buy a transceiver, in my experience, the S1 is much easier to use in digital mode than the 'rest' and exactly the same as others in analogue mode. The problem is that it is nearly £400, twice the price of some of the competition, hard for some to justify. Like jackets, people say TKMax is best until they buy a decent one.
|
|
|
|
|
|
You need to Login to know who's really who.
You need to Login to know who's really who.
|
| ed123 wrote: |
I am very sad.
I have just done the stats on the first 2 columns of data presented
The P value for the difference between the average time faster D3 and average time slower S1 in the hands of professionals, using a 2 tailed paired students T test was greater than o.05. Meaning that it was not satistically [sic] significant.
In the case of the fastest (the S1) and slowest digital beacons (DSP) when used by novices, using unpaired 2 tailed T tests between the P value was 0.22 and not at all significant. |
Are you doing all the results or just the bits that support your argument?
|
|
|
|
|
|
Anyway, snowHeads is much more fun if you do.
Anyway, snowHeads is much more fun if you do.
|
allanm, you can get it for less than that. I did. But I agree that trying one yourself is the correct thing to do.
Pjski, actually what he has done is test two hypotheses without prejudging the answer then reported the results. The study did not present its primary endpoint and then picked the outcome which looked most significant without performing the necessary stats.
Again, I think the S1 is better and would love to see a bigger more powerful study.
|
|
|
|
|
|
You'll need to Register first of course.
You'll need to Register first of course.
|
|
stoatsbrother, this isn't a drugs test. Anyway, the difference in times for novices using new devices over old ones is significant. People don't need a peer reviewed scientific paper on which to make the purchase decision. After all, you didn't.
|
|
|
|
|
|
|
|
PJSki, Depends on what you mean by "significant".
Does it look a significant difference on first glance? Yes. Is it as impressive when you look deeper if ed is right? No.
Having said which - the data where the users did not find both targets are more difficult to quantify statistically. If the hypothesis was "are you less likely as a novice searcher to find both targets in a given time with an F1 rather than an S1" I suspect the numbers already given may already give proper proof.
What the p values ed123 calculated suggest is that for the tests for novices there is roughly a 20% chance that search time results differences could just be coincidence.
If someone repeated the study and got a similar result you could probably reduce that to the below 5% chance which is usually taken in most scientific fields as to suggest that the result is likely to be genuine. The trouble is the authors neither clearly state nor prove a hypothesis. They make some useful observations which do not reach the level of proof.
Another example. I believe that helmets do actually reduce skier deaths, but the low baseline death rate per skier per year and the difficulties of eliminating confounding variables (do more cautious skiers choose helmets? Do they ski in more dangerous places etc etc...) mean that I don't think any workable study which can prove this will ever be doable.
So yes - I make my choices based on more than studies (and this one does not say there is no difference - it just doesn't prove there is). But when someone publishes a study which does appear to be trying to be scientific - it does need to be read critically to see if it really does what it says on the tin.
|
|
|
|
|
|
|
|
| PJSki wrote: |
Are you doing all the results or just the bits that support your argument? |
No - I did what I could.
First I did the first 2 columns of the first table- as a start.
Then I did the S1 against the slowest digital transciever- no difference.
the authors of the study should have done this- and more and then included it. THis i swhat would be done in any scientific paper- which this has all the appearance of being. THere are several papers about transciever use.
| stoatsbrother wrote: |
| It is a bit surprising that they reported SDs but not p values. |
exactly- this is where I am coming from- which is I think from a totally different position from most posters here- I am commenting about the quality of the paper and the very poor and in my view misleading stats. I am not commenting at all about the subjective experience of using any of these devices- only about what can and cannot be said about them from paper in the OP and no more.
|
|
|
|
|
|
You'll get to see more forums and be part of the best ski club on the net.
You'll get to see more forums and be part of the best ski club on the net.
|
Whilst playing with the numbers is an interesting academic exercise, I think it may be looking for false significance. The study says that they had a simulated avalanche 30 by 50 metres and buried the transmitters in the run out area. There is no info on how big the run out area is, or even whether it overlaps the 30 by 50 area. For each search a random pair (ie two) of the five transmitters was turned on. So there is an unknown distance to travel, whilst searching, from the start to the first detected transmitter, plus an unknown distance from there to the second transmitter. There is no info on how close they had to be to a transmitter to have "found" it. (All the practise session I have had involved actually touching the transmitter, by hand or by probe.)
Each transceiver was tested by 15 guides and 10 students. There are 10 different pairs of the five transmitters. Given random transmitter choice it would be easy, with these numbers, for the transmitter locations used to be skewed in favour of some transceivers and against others.
The sentence "Two different test transmitters were switched on for every test" is used for both test groups. It could be correct for the students tests, but I do not see how for the guides tests. Perhaps this sentence should be interpreted as no transmitter was used for two consecutive tests.
Could a statistician answer this question for me? There are 10 possible exercises. Each of 10 (or 15) people perform an exercise chosen randomly from the set and their time (ie duration) is recorded. How many significant digits can I infer from the results?
Sorry, I am not a statistician so I am not sure how best to phrase the above question. But I see three digits of precision being stated in the recorded times and in the derived values. I think that there is insufficient source data to believe that the derived values are accurate to three digits.
|
|
|
|
|
|
|
|
|
Adrian, nor was there any mention of the fact that the first candidates would have made tracks!
|
|
|
|
|
|
|
|
FAQ
