CSI and cadaver dogs - some facts and statistics.

[OxfordBloo]

You asserted that only a couple (I presume two)  of the cadaver dog alerts could be atributed to cadaver odour....the rest to blood.

I said thats wrong
There were seven WITHOUT a blood dog cooroboratioon and thats not forgetting a blood dog corroboratuin does not not automatically mean  NOT a cadaver scent in the same place

We know the cadaver dog alerted in several  (not a couple) places either where the blood dog did not alert or where the files do not tell us if it did or not
What we know from the files: ie where both dogs documented as utlised
cadaver dog alert with blood dog alert = 2 places, parents' car  and living room flat 5a
cadaver dog  alert with no blood dog alert = clothes x 3, oarebts' bedroom in fkat 5a  = 4

What we know from the files of only the cadaver dog utlised
Cadaver dog alerted in the verandah of flat 5a
Cadaver dog alerted in the garden of flat  5a
Cadaver dog alerted to a soft toy

On balance the cadaver dog alerted qute a few times, and as I posted yesterday or so not once anywhere else

So,  there you go amigo. I have to run to work now.

Grime states that although Keela alerts very specifically to a small area where blood is found, Eddie was trained to alert to the general odour of death in a wider area. This is normal for rescue dogs as they are trained to follow an odour gradient.

The clothes were bundled together and we know that odour transfers from item to item in close proximity.

The 'couple' of alerts are to 5a and to clothes. The clothes had also been in 5a so a single source could have existed.

It is worth remembering that the scientifically determined best estimate for scent dogs is of the order of 80%. Statistically this error rate is increased for tests that require two alerts, the uncertainty increases by the multiple of each individual test, leading to a joint Eddie/Keela alert is 80% of 80%, or 64%. So each of those alerts has a two out of three possibility of being true or more importantly a one on three chance of being wrong.

So we have moved from a contention that there were multiple alerts indicating cadaver odour to the possibility that it was actually a single source detected with an chance of one in three being totally incorrect.

That is the truth value of the dogs.

[John]

The last time SY were on site in Praia da Luz they were looking for her remains.  I don't call that much of a success.

[OxfordBloo]

Actually no Brietta.  The last time SY were on site in Praia da Luz they were looking for her remains.  I don't call that much of a success.

I must have missed that announcement by Grange. Could you provide a cite for it.

Of course they may not have been searching for Madeleine, but seeking to exclude possibilities and thus increase the likelihood that she was not dead and buried in PdL.

[pathfinder73]

Grime states that although Keela alerts very specifically to a small area where blood is found, Eddie was trained to alert to the general odour of death in a wider area. This is normal for rescue dogs as they are trained to follow an odour gradient.

The clothes were bundled together and we know that odour transfers from item to item in close proximity.

The 'couple' of alerts are to 5a and to clothes. The clothes had also been in 5a so a single source could have existed.

It is worth remembering that the scientifically determined best estimate for scent dogs is of the order of 80%. Statistically this error rate is increased for tests that require two alerts, the uncertainty increases by the multiple of each individual test, leading to a joint Eddie/Keela alert is 80% of 80%, or 64%. So each of those alerts has a two out of three possibility of being true or more importantly a one on three chance of being wrong.

So we have moved from a contention that there were multiple alerts indicating cadaver odour to the possibility that it was actually a single source detected with an chance of one in three being totally incorrect.

That is the truth value of the dogs.

If you're going to show statistics use Eddie's not dogs in general. Eddie was EVRD. He was specially trained with new scientific techniques. Eddie alerted to a body that was in a room for only 1 hour before disposal in the Harron case (forensics found nowt!). He found evidence in the burned out car that forensics missed in the same case. Eddie has proved he finds evidence where forensics don't!

As a lawyer once said to me, apropos another matter, ‘One coincidence, two coincidences – maybe they’re still coincidences. Any more than that and it stops being coincidence.’ (Madeleine)

The specialist training techniques - which are highly confidential - were developed by Eddie's handler Martin Grime, along with the UK's National Policing Improvement Agency (NPIA) and America's Federal Bureau of Investigation (FBI).
They are scientifically based and rely on how dogs smell and the chemicals involved. "An enhanced dog goes through much more training and is a lot more discriminating about smells, basically its nose is super sensitive. Other dogs have to do other police duties but mine work full-time in this area, making them very sharp and highly skilled."

http://news.bbc.co.uk/1/hi/magazine/7263355.stm

Pseudo scent is an artificially chemically produced product that its
manufacturers claim to resemble 'dead body scent'. Although some cadaver
dog trainers have had limited success with its use in training, when tested on
my dogs they showed no interest and it is not used as a training aid for them.

[OxfordBloo]

If you're going to show statistics use Eddie's not dogs in general. Eddie was EVRD. He was specially trained with new scientific techniques. Eddie alerted to a body that was in a room for only 1 hour before disposal in the Harron case (forensics found nowt!). He found evidence in the burned out car that forensics missed in the same case. Eddie has proved he finds evidence where forensics don't!

As a lawyer once said to me, apropos another matter, ‘One coincidence, two coincidences – maybe they’re still coincidences. Any more than that and it stops being coincidence.’ (Madeleine)

The specialist training techniques - which are highly confidential - were developed by Eddie's handler Martin Grime, along with the UK's National Policing Improvement Agency (NPIA) and America's Federal Bureau of Investigation (FBI).
They are scientifically based and rely on how dogs smell and the chemicals involved. "An enhanced dog goes through much more training and is a lot more discriminating about smells, basically its nose is super sensitive. Other dogs have to do other police duties but mine work full-time in this area, making them very sharp and highly skilled."

http://news.bbc.co.uk/1/hi/magazine/7263355.stm

Pseudo scent is an artificially chemically produced product that its
manufacturers claim to resemble 'dead body scent'. Although some cadaver
dog trainers have had limited success with its use in training, when tested on
my dogs they showed no interest and it is not used as a training aid for them.

There are no statistics for Eddie.

All scent dogs are prone to various errors.

The rest of your post does not speak to the issue.

[pathfinder73]

There are no statistics for Eddie.

All scent dogs are prone to various errors.

The rest of your post does not speak to the issue.

It proves that Eddie has solved cases on his own where forensics miss evidence. They confessed to confirm Eddie's alerts. Eddie was EVRD. If death has occurred in a missing person's case Eddie will find that odour. That's his job! He goes in first before Keela to find that odour and then bark alerts. Then it's Keela's turn to find any traces of blood in that alerted area.

[OxfordBloo]

It proves that Eddie has solved cases on his own where forensics miss evidence. They confessed to confirm Eddie's alerts. Eddie was EVRD. If death has occurred in a missing person's case Eddie will find that odour. That's his job! He goes in first before Keela to find that odour and then bark alerts. Then it's Keela's turn to find any traces of blood in that alerted area.

I do not deny that Eddie has been successful.

What is unknown is how accurate he and Keela were in this case. Their serial accuracy is about 64%

You are slightly wrong in your description of the deployment of the dogs.

If both alert it says nothing more about the possibility of cadaver odour being present than was known originally.

If Eddie alerts and Keela doesn't, there is about a two thirds likelihood of this being suggestive of cadaver odour being present and a one in three possibility of it being a false positive.

[Brietta]

Actually no Brietta.  The last time SY were on site in Praia da Luz they were looking for her remains.  I don't call that much of a success.

None of us know what information the investigations have uncovered or what evidence the digs may have provided; I think SY were covering all the bases in their much publicised searches in the environs of PDL; at least if Madeleine McCann's remains had been found her family would have had some closure; in the absence of that discovery she is entitled to be presumed alive and worthy of the continuation of being looked for.

[mercury]

Grime states that although Keela alerts very specifically to a small area where blood is found, Eddie was trained to alert to the general odour of death in a wider area. This is normal for rescue dogs as they are trained to follow an odour gradient.

The clothes were bundled together and we know that odour transfers from item to item in close proximity.

The 'couple' of alerts are to 5a and to clothes. The clothes had also been in 5a so a single source could have existed.

It is worth remembering that the scientifically determined best estimate for scent dogs is of the order of 80%. Statistically this error rate is increased for tests that require two alerts, the uncertainty increases by the multiple of each individual test, leading to a joint Eddie/Keela alert is 80% of 80%, or 64%. So each of those alerts has a two out of three possibility of being true or more importantly a one on three chance of being wrong.

So we have moved from a contention that there were multiple alerts indicating cadaver odour to the possibility that it was actually a single source detected with an chance of one in three being totally incorrect.

That is the truth value of the dogs.

Sorry, but you did not address the facts and frankly not sure what your post was meant to mean

[OxfordBloo]

Sorry, but you did not address the facts and frankly not sure what your post was meant to mean

It means that scientifically and forensically, the dog alerts are not and can never be probative.

There is a massive failure to understand scientific and forensic method here.

[slartibartfast]

It means that scientifically and forensically, the dog alerts are not and can never be probative.

There is a massive failure to understand scientific and forensic method here.

Actually if you look at it the other way,  in 2 separate alerts the probability of an incorrect alert (according to you) is 20% on each therefore the combined probability of an incorrect alert is 4%.

[OxfordBloo]

Actually if you look at it the other way,  in 2 separate alerts the probability of an incorrect alert (according to you) is 20% on each therefore the combined probability of an incorrect alert is 4%.

Your maths is awry.

If you have two serial alerts on which the result depends on both, then an 80% error in either case results in a probability of 64% of error in the serial test.

Simple statistical theory.

[slartibartfast]

Your maths is awry.

If you have two serial alerts on which the result depends on both, then an 80% error in either case results in a probability of 64% of error in the serial test.

Simple statistical theory.

It depends on whether you are looking at the combined probability of the alerts being correct or incorrect?

[OxfordBloo]

It depends on whether you are looking at the combined probability of the alerts being correct or incorrect?

No. Consider a fair coin. One toss has the possibility of fifty per cent for heads. Two serial tosses have the probability of 25% for two heads.

Consider a pair of dice. On the first roll the chances of throwing a six is about 16%. Two rolls of the dice, the chances are 0.025.

Probabilities are transitive.

So for a serial test involving 80% probability of correctness each, the probability is 8/10 times 8/10 or 64%.

[pathfinder73]

No. Consider a fair coin. One toss has the possibility of fifty per cent for heads. Two serial tosses have the probability of 25% for two heads.

Consider a pair of dice. On the first roll the chances of throwing a six is about 16%. Two rolls of the dice, the chances are 0.025.

Probabilities are transitive.

So for a serial test involving 80% probability of correctness each, the probability is 8/10 times 8/10 or 64%.

Irrelevant. They sent their best dogs. Why are they the best? I suggest you research the actual dogs involved in this case.