Poll - Scent Dogs Accuracy

[Aiofe]

youre entitled to your opinion but it didnt answer my question

All tests have a reliability and an unreliability, even dogs. One test for cadaver dogs showed that there was a thirty percent false positive if handlers 'knew' which targets were 'contaminated'.

Good enough to me to show quite a high unreliability.

[Redblossom]

All tests have a reliability and an unreliability, even dogs. One test for cadaver dogs showed that there was a thirty percent false positive if handlers 'knew' which targets were 'contaminated'.

Good enough to me to show quite a high unreliability.

wheres the link for this test and 30 per cent  out of 100 is not quite high by any standard

To have any argument whatsoever you will need to know the test results  of EDDIE in this case.....you dont have them, therefore no argument.....

You are also libelling Grime here but thankfully its water off a ducks back......as it has no basis in any fact

[Aiofe]

wheres the link for this test and 30 per cent  out of 100 is not quite high by any standard

To have any argument whatsoever you will need to know the test results  of EDDIE in this case.....you dont have them, therefore no argument.....

You are also libelling Grime here but thankfully its water off a ducks back......as it has no basis in any fact

There are no tests on Eddie or Keela done by an independent authority; so we must assume that the best they can have been is similar to other scent dogs.

I am not libelling Grime in any way. Where have I libelled him?

[Aiofe]

This article is not peer reviwed but is interesting.

Two cadaver dogs were tested. One for only three tests, one for 14. 3 tests is far to few to draw any firm conclusions, and 14 is far to small to draw any rigorous analysis of accuracy.

But note that the "14 test" dog had one false positive and one false negative- an error rate of about 14%- accuracy of about 86%. One test in the fourteen was a failure to ID scent that was there. One test in the 14 the scent was "identified" when it was not there.

There note:

"With an overall successful positive diagnostic ratio of id of blood scent of 17.5 for Frankie
and a positive diagnostic ratio of 7.6 for Buster and with negative ratios of 38.5 and 16.4 respectively,
it is clear that certain dogs have the ability to detect human blood scent in principal in both trial and
scenario training. Any conclusions must be drawn carefully as by the nature of this project there were
many limitations mainly due to lack of number of dogs to compare and the limitations of number of
trials that could be done at any one time and within the overall time frame of this project"

http://www.staffs.ac.uk/assets/Simon%20Newbery_tcm44-19866.pdf

[LagosBen]

I haven't voted because anything dogs alert to has to be backed up with evidence.

[Aiofe]

Here is the really frightening result, and it is peer reviewed and statistically valid with a complete analysis!

They set out to test the 'Clever Hans' effect. Clever Hans was a horse that could apparently do advanced arithmetic. Asked questions it would hoof the ground the correct number of times. It was shown beyond doubt that all the horse was doing was reacting to the audience and handler's body language and stopped hoofing when it saw the expectation- if it was asked a question out of site of the handler and audience that knew the answer, it could not do arithmetic.

This current examination was most unfair, but very indicative. No scents were used at all! But the handlers not only knew that the experiment was to test for the scent, but also believed that certain identifiable items were imbued with the scent. Their dogs magically identified the scent reliably- but only reliably with the handler's supposed 'knowledge'. The handlers were unconsciously cuing the dogs.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3078300/

[Aiofe]

An analysis of the US approach to Cadaver Dogs evidence- concluding that a reaction is not stand alone evidence, but requires further support:

http://www.policek9.com/Fleck/Cadaver%20Dogs.pdf

"The question that remains is the cadaver or human remains dog. Is this dog a
human scent dog or a contraband substance dog?
There is no Federal or State case law, that I’m aware of, that addresses an alert
from a cadaver or human remains dog as probable cause to obtain a search
warrant, search or arrest.
Therefore, if the cadaver or human remains dog is placed in the contraband
substance dog group, that alert would stand alone. If the cadaver or human
remains dog is placed in the human scent dog group, that alert would have to be
corroborated by other evidence.
Based upon review of the case law below, an alert from a cadaver dog is only
reasonable suspicion. The dog alert must be corroborated by other evidence. "

[Aiofe]

The Komar tests:

http://www.ncbi.nlm.nih.gov/pubmed/10097372

"A 1998 study by Debra Komar at the University of Alberta, Edmonton, analyzed and interpreted the effectiveness of eight dog and handler teams at locating simulated animal scattered or scavenged human and animal remains among snow and leaf cover. Tests consisted of "blind searches" or trials in which handlers did not how many items to search for or where they were hidden. The items included dry human and animal bone, and gauze and small articles of clothing soaked in human decomposition fluids and then dried. The study revealed considerable variation in the success rates of the dog-handler teams. The individual dog-handler teams had success rates ranging from 55 to 95% over the trials, and the overall recovery rate for the trials was 81%."

Note range of 55-95 with median at 81%- two in ten reactions were errors on average. Note- peer reviwed and statistically valid.

http://www.researchgate.net/publication/13194566_The_use_of_cadaver_dogs_in_locating_scattered_scavenged_human_remains_preliminary_field_test_results

[Aiofe]

The outlying study by Oesterhelweg et al.

This is often quoted and is the most supportive of all published papers. It gives accuracies of about 90% over a small number of tests that are not statistically valid- no confidence limits are given.

Even on this basis, one in ten reactions is false.


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

[AnneGuedes]

Here is the really frightening result, and it is peer reviewed and statistically valid with a complete analysis!

They set out to test the 'Clever Hans' effect.

What has this to do with Prof Harrison MBE who supervised the canine searches in PDL ?

[AnneGuedes]

There are no tests on Eddie or Keela done by an independent authority; so we must assume that the best they can have been is similar to other scent dogs.

I am not libelling Grime in any way. Where have I libelled him?

This is not true. Those dogs were certified on a regular basis. Remember they were called by Prof Harrison MBE, the UK specialist in missing persons.

[AnneGuedes]

The Komar tests:

http://www.ncbi.nlm.nih.gov/pubmed/10097372

"A 1998 study by Debra Komar at the University of Alberta, Edmonton, analyzed and interpreted the effectiveness of eight dog and handler teams at locating simulated animal scattered or scavenged human and animal remains among snow and leaf cover. Tests consisted of "blind searches" or trials in which handlers did not how many items to search for or where they were hidden. The items included dry human and animal bone, and gauze and small articles of clothing soaked in human decomposition fluids and then dried. The study revealed considerable variation in the success rates of the dog-handler teams. The individual dog-handler teams had success rates ranging from 55 to 95% over the trials, and the overall recovery rate for the trials was 81%."

Note range of 55-95 with median at 81%- two in ten reactions were errors on average. Note- peer reviwed and statistically valid.

http://www.researchgate.net/publication/13194566_The_use_of_cadaver_dogs_in_locating_scattered_scavenged_human_remains_preliminary_field_test_results

This study isn't much relevant for the McCann case, since the dogs didn't alert outside.
About longevity of human primary odours in a controlled environment
http://newyorksearchandrescue.org/download/Scents%20and%20Sense-Ability%20K9%20article.pdf

[Aiofe]

What has this to do with Prof Harrison MBE who supervised the canine searches in PDL ?

Nothing

[Aiofe]

This is not true. Those dogs were certified on a regular basis. Remember they were called by Prof Harrison MBE, the UK specialist in missing persons.

Certified does not include a valid test regime. It is an assessment of training and competence, not of accuracy. No such tests for Eddie or Keela have been published.

[Aiofe]

This study isn't much relevant for the McCann case, since the dogs didn't alert outside.
About longevity of human primary odours in a controlled environment
http://newyorksearchandrescue.org/download/Scents%20and%20Sense-Ability%20K9%20article.pdf

Inside or out does not matter. What does is the low success rate.

The highest claimed rate of one false result in ten; I suspect it is much lower.

The Clever Hans experiment is the most worrying indicating considerable handler cuing.