Poll

Peer Reviewed Research suggests that Scent Dogs of all types have a maximunm combined accuracy of about 90%

I Understand and Accept this
18 (34.6%)
I believe Scent Dogs are more accurate than this
12 (23.1%)
I am not sure
4 (7.7%)
I don't believe Scent Dogs generally are that accurate
18 (34.6%)

Total Members Voted: 38

Voting closed: February 25, 2014, 04:23:56 AM

Author Topic: Poll - Scent Dogs Accuracy  (Read 66861 times)

0 Members and 1 Guest are viewing this topic.

debunker

  • Guest
Poll - Scent Dogs Accuracy
« on: April 28, 2013, 12:52:28 PM »
Additionally for the true possible marking of Cadaver scent, as this involves two reactions rather than one, the accuracy is 9/10 squared or about 8 out of ten- one suggestion in five will be an error.
« Last Edit: October 27, 2013, 11:21:04 AM by John »

Offline Centaur

Re: Poll - Scent Dogs Accuracy
« Reply #1 on: April 28, 2013, 12:58:32 PM »
I don't believe you can put a figure on the accuracy as there are to many external factors which have to be taken into account.  If you gave every dog the same test then compared the results then I would accept the findings but sporadic results combined are worthless.

debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #2 on: April 28, 2013, 01:02:21 PM »
I don't believe you can put a figure on the accuracy as there are to many external factors which have to be taken into account.  If you gave every dog the same test then compared the results then I would accept the findings but sporadic results combined are worthless.

If every peer reviewed test of dogs showed a maximunm accuracy of about 90% then it is quite valid to assume that no dog is an outlier with accuracy over a long p[eriod any better than the average unless the are specifically tetsed. That is how science works.

There is no evidence that Eddie is better than any other dog- in fact there is no reliable unbiased evidence at all available about Eddie's accuracy

Offline Centaur

Re: Poll - Scent Dogs Accuracy
« Reply #3 on: April 28, 2013, 01:18:50 PM »
Thats what I mean it is all based on the results provided by people who have their own agenda to succeed and make lots of cash.

debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #4 on: April 28, 2013, 02:37:44 PM »
Thats what I mean it is all based on the results provided by people who have their own agenda to succeed and make lots of cash.

You misunderstand debate and the scientific method.

No-one hags so far produced a peer reviewed experiment where scent dogs have a greater than 90% accuracy in real life situations. It seems to be the best available estimate- some being as low as 35% and a few as high as 90%.

My views are immaterial- anyone else is welcome to provide a reference for any published experiment with better proof of accuracy.

That is how science and forensics works.

Redblossom

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #5 on: April 28, 2013, 02:55:15 PM »
This any use? See table, third row of results, PPV, much higher than 90%

http://ksgarvin.files.wordpress.com/2008/04/garvin.ppt

Link to full paper

http://www.pawsoflife.org/Library/HRD/Oesterhelweg%201998.pdf

Not sure if that is what you are looking for, but interesting all the same

PS The poll question cant be that relevant in this case as we are not interested in scent dogs of all types just the one type
« Last Edit: April 28, 2013, 03:46:54 PM by Redblossom »

debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #6 on: April 28, 2013, 03:47:32 PM »
This any use? See table, third row of results, PPV, much higher than 90%

http://ksgarvin.files.wordpress.com/2008/04/garvin.ppt

Link to full paper

http://www.pawsoflife.org/Library/HRD/Oesterhelweg%201998.pdf

Not sure if that is what you are looking for, but interesting all the same

There is no statistical analysis giving any form of confidence intervals. This is probably because the small scale of the experiment would have produce confidence intervals which were indicative of the experiment not being well designed enough to draw any scientific conclusion.

That said, even in this non in vivo experiment, the accuracies are usually lower than 100%.

THis piece of research needs to be interpreted against contrary evidence from well designed experiments with full statistical analysis.

WE have here an poorly designed and badly interpreted experiment which produces results outside the envelope for well desig need and interpreted experiments.

Redblossom

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #7 on: April 28, 2013, 03:51:25 PM »
Well, in that case, I give over to the scientifically knowledgable here who know what you are talking about, and can respond,  I don't as it's over my head








debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #8 on: April 28, 2013, 04:05:57 PM »
Well, in that case, I give over to the scientifically knowledgable here who know what you are talking about, and can respond,  I don't as it's over my head

Properly designed and well analysed experiments have what are call confidence intervals associated with them. These give a rating for how much an experimental results are a better prediction than mere chance. The more individual tests there are, the higher the confidence interval will be.
 
When no confidence interval is stated, it is a sign that there is likely to be a problem with that issue- usually caused by too few individual tests.

For instance if we do a coin toss experiment (which should be 50/50 with a fair coin) initial results with less than a hundred tests are likely to deviate from this 50/50 result, but experiments with hundreds of tests will tend to return to the expected mean.

Low test number experiments do not allow valid conclusions.

Redblossom

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #9 on: April 28, 2013, 04:08:55 PM »
OK thanks, I understood that time. So time to test more cadaver dogs then
 8((()*/



debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #10 on: April 28, 2013, 04:11:06 PM »
OK thanks, I understood that time. So time to test more cadaver dogs then
 8((()*/

Or look at the other better designed experiments.

Redblossom

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #11 on: April 28, 2013, 04:12:33 PM »
OK thanks, I understood that time. So time to test more cadaver dogs then
 8((()*/

Or look at the other better designed experiments.
I will if you link me to one on cadaver dogs


debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #12 on: April 28, 2013, 04:23:49 PM »
OK thanks, I understood that time. So time to test more cadaver dogs then
 8((()*/

Or look at the other better designed experiments.
I will if you link me to one on cadaver dogs

Most research has been done on a variety of scent dogs because cadaver dogs are a more recent development, and because there is no reason to assume any difference between different target sniffer dogs.

Any research on cadaver dogs tends to be small scale, badly designed and often done outside peer review and by people with a particular expectation of results.

The paper you quote is probably the best (of a bad job lot) on cadaver odor detection.

Well designed and interpreted experiments on scent dogs vary from 0% (Where no target scent was used, but handlers "knew" which objects were "contaminated", through 35% for in vivo drug detection dogs, to a maximum of 90%.

I use tbe90% figure to be as fair as possible to dog worshippers. My best guess is that dogs are rarely better than 80% accurate.

Offline Angelo222

Re: Poll - Scent Dogs Accuracy
« Reply #13 on: April 28, 2013, 04:44:25 PM »
What about all the times the dogs failed to detect anything when there was in fact something to find?  Are these episodes classified as fools errands or are they conveniently overlooked?   Let's face it a spaniel's hooter and what it can achieve is not an accurate science and more akin to a circus act.

Those who promote this ability or whatever it is called do so in order to embelish their own credibility and line their pockets.  How many of these dogs are winging their way to Bangladesh to do their bit?
De troothe has the annoying habit of coming to the surface just when you least expect it!!

Je ne regrette rien!!

debunker

  • Guest
Re: Poll - Scent Dogs Accuracy
« Reply #14 on: April 28, 2013, 04:48:58 PM »
What about all the times the dogs failed to detect anything when there was in fact something to find?  Are these episodes classified as fools errands or are they conveniently overlooked?   Let's face it a spaniel's hooter and what it can achieve is not an accurate science and more akin to a circus act.

Those who promote this ability or whatever it is called do so in order to embelish their own credibility and line their pockets.  How many of these dogs are winging their way to Bangladesh to do their bit?

Totally agree but when arguing with fanatical dog supporters it is necessary to proceed one step at a time.