CSI and cadaver dogs - some facts and statistics.

[pathfinder73]

You seem to have overlooked the bit which confirms that Martin Grime had not at that time acted as an expert witness in the US.

**snip

During a visit to Parker's home back in September 2007 Grime said he and Eddie sniffed around their garage.

"He immediately gave a positive bark response within the garage between a truck parked to the left of the entrance and a boat parked to the right," Grime said.

Grime added Eddie did not seem interested in the vehicles but in a scent that was wafting in the air, based on the way the dog held his nose upward. Grime said Eddie then "hit" on an abandoned house next door. Testimony shows that house was never repaired after a fire gutted the inside and killed a child several years ago.

During lengthy cross-examination Grime said there is no evidence to show Eddie smelled anything incriminating against or linked to Mr. Parker. Like Higgins, Grime said cadaver dogs can only prove useful when there is other evidence that corroborates the dog's "hits."

The FBI has a keen interest in the outcome of this case. If Parker is convicted the case could pave the legal way for future prosecutions where there is no evidence other than dog "hits" in connection with a person accused of murder.

Toward the end of the day Judge Wood learned that while Grime has international acclaim he has never testified as an expert witness in the United States.

http://www.scentevidence.com/2009/07/dog-debate-at-center-of-murder-case.html

Sept 2007
"He immediately gave a positive bark response within the garage between a truck parked to the left of the entrance and a boat parked to the right," Grime said.

Sept 2010
Theresa Parker's family and friends will finally be able to put the Georgia 911 dispatcher to rest after her skeletal remains were found scattered along the Chattanooga River.


He said that, after Theresa Parker's friend reported to police that she was worried about Theresa in March 2007, two Walker County sheriff's deputies found an empty house and looked inside the Parkers' garage when they weren't allowed.

On the left side of the garage, they found Sam's LaFayette Police Department vehicle. On the right side, where Theresa's Toyota 4Runner should have sat, they found nothing. They also found Sam Parker's truck outside the garage, and days later they found the 4Runner back in its place -- though no one ever saw Theresa again.

When Theresa's family reported her missing, members of the Georgia Bureau of Investigation asked Sam Parker where he was the night his wife had last been seen. He told them he had been cruising in his truck.

But investigators knew that wasn't true because the truck had been home when deputies checked on the Parkers. The inconsistency in Sam Parker's story was a key point during a September 2009 trial in which he was found guilty.

You don't fool Eddie. If there's cadaver scent he will detect and alert to it.

[slartibartfast]

I note you have failed to address the mathematical reasoning on the square of probabilities and are merely being abusive.

Very sorry, been having a very enjoyable evening out, will hopefully get time to address your misconceptions tomorrow.

[mercury]

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.

So? They remain as circumstial evidence. There is no such thing as a cadaver dog alerting correctly to remnant cadaver scent from a decomposing body which is probative. Not yet, anyway. Except in police training or scientific studies.

[slartibartfast]

OK let us ignore dogs.

Take a bag of pebbles that has eighty per cent red spheres and twenty per cent blue spheres.
Picker one draws five balls on average they will draw four them in the ratio 4:1 and can predict that the bag contains balls in a ratio of four to one with an uncertainty of p
If picker two repeats the same exercise then the uncertainty would fall to p + x – it is more likely to be right.

Now consider a bag containing balls which has eighty per cent hollow spheres and twenty per cent solid spheres.
Picker one draws five balls on average they will draw them in the ratio of 4:1 and can predict that the bag contains balls in a ratio of four to one with an uncertainty of p
If picker two repeats the same exercise then the uncertainty would fall to p + x – it is more likely to be right.


Now consider a bag which includes spheres each of which is either red or blue and independently each is hollow or solid.
Balls are withdrawn by a neutral party.
Picker one  only sees the balls and determines their ratio of red to blue balls.
Picker two who is blind only weighs the balls and determines whether the balls are solid or hollow.

Now their probability decreases according to the same matrix

   T   T   T   T   F
T   P   P   P   P   N
T   P   P   P   P   N
T   P   P   P   P   N
T   P   P   P   P   N
F   N   N   N   N   N

Proportion of correct calls is 16 out of 25

This doesn't actually mean anything, it's just pseudo claptrap.

[slartibartfast]

Well its copied from a textbook. If you don't recognise a distribution matrix you are no statistician, merely a shady politician using words to avoid sensible argument.

Maybe you forgot to include the explanations that went with it. Good at cut and paste aren't you.

[OxfordBloo]

Maybe you forgot to include the explanations that went with it. Good at cut and paste aren't you.

It is not cut and paste. It is adapted from the text book I used some fifty years ago and is amended to make it easier to understand for those who do not understand statistics. The matrix gives every possible outcome for a test of two variables with a probability of 80%.

I note you do not address the mathematics, but resort to individual attack.

[OxfordBloo]

maybe things have moved on in fifty years and once youre seventy odd your marbles cant be trusted theyre encrusted lol even though you still feel the need to pontificate your time is ovef use whats left of it well mate

Actually the idea fell out of favour for some reason - probably because little in the real world relies on such parallel tests. But it was resurrected some thirty years ago to show how Big Parma was using and misusing statistics to bolster their claims to efficacy. When the parallel trial analysis was applied it showed that many drugs were not as efficacious as had been thought.

[mercury]

We all know big pharma lie for profits and are morally defunct I know I worked for them for 8 years

Dogs and their handlers who are looking for dead people or signs of them are not the same goodnight

[OxfordBloo]

We all know big pharma lie for profits and are morally defunct I know I worked for them for 8 years

Dogs and their handlers who are looking for dead people or signs of them are not the same goodnight

The statistics are not used by dogs or handlers but by people assessing the reliability of dogs.

Statistics of this kind are irrefutable.

[OxfordBloo]

Bumped for slartibartfast to share his knowledge of statistical analysis by addressing the mathematics rather tan trying to shoot the messenger.

[ShiningInLuz]

What?   You missed all the media reporting on where, when and why they were digging up parts of Praia da Luz?    Do you honestly believe they weren't looking for remains?

Just as well they didn't follow the Sun story then.  Four sites, all wrong.  Mechanical diggers? What hilarity.

Mind you, it sold newspapers.

And I'm still with Slartibartfast on probability theory.

Whilst the fact that the dogs were not used in such a manner renders the debate academic.

[OxfordBloo]

Just as well they didn't follow the Sun story then.  Four sites, all wrong.  Mechanical diggers? What hilarity.

Mind you, it sold newspapers.

And I'm still with Slartibartfast on probability theory.

Whilst the fact that the dogs were not used in such a manner renders the debate academic.

If you dispute the mathematics of the sttisticalmdistribution, I am happy to defend it.

Simple denial without proof of error is just personal belief.

[Alice Purjorick]

If you dispute the mathematics of the sttisticalmdistribution, I am happy to defend it.

Simple denial without proof of error is just personal belief.

Would you like to tell us precisely what your mathematics has proved. In simple terms please as I get confused. I once read a whole book on maths absolutely convinced it was The Operator j what dunnit only to find out on the last page it were Curly Dee. So you see my dilemma.

[Mr Gray]

Would you like to tell us precisely what your mathematics has proved. In simple terms please as I get confused. I once read a whole book on maths absolutely convinced it was The Operator j what dunnit only to find out on the last page it were Curly Dee. So you see my dilemma.

probably best just to accept it's all over your head then

[OxfordBloo]

Would you like to tell us precisely what your mathematics has proved. In simple terms please as I get confused. I once read a whole book on maths absolutely convinced it was The Operator j what dunnit only to find out on the last page it were Curly Dee. So you see my dilemma.

If you have two parallel but separate decisions to make, the probabilities of each test need to be multiplied together, reducing the accuracy of the two parallel tests.

If you have a series of tests, the second test improves the accuracy of the two tests.

So, if you have two blood dogs and bother alert in sequence, the probability of them being wrong is reduced from 4 in 5 to 4 in a hundred.

Consider the matrix for this case

Dog one on average is wrong once in five alerts as is dog two. We need to examine all possible permutations of correct and incorrect alerts.

Each dog alerts R R R R W

Showing this as a matrix we have

                      DOG TWO
DOG ONE. 
                 W R R R R
                  W R R R R
                  R W R R R
                  R R W R R
                  R R R W R
                  R R R R W

There is only one place in the matrix where both are wrong so the probability is 96% correct increased from 80%


Now consider two different tests- see the previous descriptions for either Eddie and Keela, or the choice of spheres either red or blue and hollow or solid

The matrix now is



R R R R W
R R R R W
R R R  R W
R R R R W
W W W W W

giving an accuracy of 16 out of 25 or 64%