This is a very good thread, giving us some good insights into the professionalism and decision making processes of police and the courts. It's one of the big bennies of these boards.

Iain and Thomas, I hope you will not abbreviate your posts.

Thomas,
What was your source on the above post?
I am always interested in all sides. It helps me prepare for the next time I end up sitting in the witness box at a supression hearing either testifying about my dog or someone elses.

I don't know where Mr. Bird got the dogs he used to conduct his research, I know that if I had a dog that was only 20% proficient, he would be someones pet. Understandably, the ACLU would would have the same results. More than one aspect of the dog has to be viewed. Secondly, while I'm very familiar with the opinions by the USC, and the 9th Circus (intentional spelling)they all deal with issues of the dog's response being probable cause. So now we are defining probable cause. The salient point, sniffing by a trained dog is not a search. It's the response of the dog that can cause a search. Using Mr. Birds figures, the dog had 10,000 opportunities to be correct. Again, using his figures the dog was incorrect 199 times. On my abbacus that is 1.9%. How's that go, figures lie, liars figure. I guess it's how you want to read the numbers.

DFrost

Kevin, here is a link to the source of the ACLU letter ACLU-NC Letter In Opposition to Use of Dogs in Drug Searches on School Campuses

Thank you.
Musashi and Sun Tzu both said study the way of your enemy (paraphrased). Here the way is deception.

Originally posted by Thomas Mincher:
Kevin, here is a link to the source of the ACLU letter ACLU-NC Letter In Opposition to Use of Dogs in Drug Searches on School Campuses

In regards to "figures lie, liars figure", in this case they are not lying with the figures, the math is accurate and actually does a good job of indicating the problem with large scale random searches(as I alluded to in my earlier posts). Of course the question is how does it play on the jury and how does it make you look on the witness stand. In that light let me explain the math.

The dog in the example is correct 98% of the time. But when you put the percentages in to real numbers with real humans, what you find is that out of 10,000 people the dog indicated that 248 of them had drugs but in reality only 49 of those 248 people did have drugs. So in this case, the dog is only correct about 20% of the time. So what’s the problem, how can the dog be both 20% correct and 98% correct? The true answer is relatively simple, and it goes to what I was trying to point out earlier. The problem is not the dog ( as a tool the dog is right 98% of the time), the problem is that the tool (the dog) is being misused and because the dog is being misused the gross number of failures are larger than they should be. Of course the question is, how is the dog being misused? That answer is pretty simple to, the person using the dog targeted the wrong population. In the population the dog was sniffing, only 1 out of every 200 people was carrying or using drugs. IF you as a dog handler are targeting that population not only are you wasting your department’s resources but you are really p'o almost 200 honest innocent civilians who probably now see you and your dog in a whole different light and that is not good for you, your department, or the dog/breed. Worse than that, because you have poorly used a good tool, you have also thrown into doubt how good that tool really is. The fact is most people don't understand that math, but where the rubber hits the road you stopped/searched almost 250 people and only snagged 50; that is what the jury is going to hear.

So lets look how a good debater would handle this in court. --

"officer, your dog is certified by the state to be correct 98% of the time, yet when YOU are handling the dog he was only right 20% of the time. Now I don't think the dog is capable of lying, so obviously you must be lying when you say your dog alerted to drugs on all these people. Good people of the jury, it is obvious that officer so and so is targeting these group of people because he does not like their "type". We need to send a strong message to rouge officers like this and their departments. Yes, they found drugs in my defendant’s car, but the officers handling of the dog already proves he will fake evidence, how do we know that he did not plant the drugs in the car. My client is not guilty, he never had the drugs, all of this is a result of a corrupt officer and a corrupt department that besmirch my client and 250 other innocent people. Lets send a strong message by finding my client innocent, and by placing a large cash judgment against the department and the police officer personally."


NOW THAT would be lying with statistics, in reality the only thing the math indicates is that the improper tool (the dog) was used in an improper environment (a population with low drug abuse). But don’t expect the jury to understand that.

In any case, it all goes back to my original point, if your department is planning on doing dog sniffing check points you run the real risk of making lots of innocent people very angry with you, your dog and your department. Not only that but you may actually place cases against real bad guys in jeopardy. If you are smart you will use your dog only when there is a reasonable suspicion of wrong doing, no matter what the courts have decided is “legal” In short, if its something George Washington would horse whip you for doing to the constitution, then don’t do it, no matter how “legal” or “needed” it may currently be.

I thought I would add the following in hopes of clearing that math aspect up a bit more.

Lets get down to what the important statistic really is, namely how many people are you stopping/searching because of the dog and what percentage of those stops and searches were truly justified by the presence of drugs. Constitutional issues aside, if you are hassling more innocent people with intrusive searches than you are catching bad guys, then you are fighting a losing public relations battle; you may be hurting your chances in front of a jury; AND you are not making effective use of the dog.

In that light, let’s call that statistic the Effective Search Ratio (ESR). When you do the math there are only two variables that make up the ERS (effective search ratio). The first variable, and the one here most people are familiar with, is how often the dog is wrong or right. In the ACLU example the dog was assumed to right 98% of the time. The other variable in calculating the ERS is the percentage of the “sniffed” population that is actually innocent. In the ACLU example they found that approximately 99.5% of the time people were not carrying drugs. The one thing that should really catch your eye here is that the dog's accuracy rate is LOWER than the population's innocence rate. Any time the dog is less accurate than the population is innocent, you will be snagging more innocent people then you will be snagging criminals. It is the functional equivalent of tying to measure inches with a yardstick, in other words its not the right tool for the job.

In the ACLU example, with the dog 98% accurate, and the population at 99.5% innocent, the effective search ratio ERS (the number times contraband was found divided by the number of searches made) was 20%. When the population’s rate of innocence matches the dog’s rate of accuracy (98% in this case) then the ERS would be 50% (meaning 1 out of every 2 searches would find contraband). Again holding the dog at 98% accurate, if the population is 90% innocent then the ERS jumps to 84% (meaning 84% of the time you will find contraband). And finally, when only 50% of the population is innocent, the ERS jumps to 98%.

So that means if your dogs nose is accurate 98% of the time and you want your searches to find contraband 98% of the time, then you had better be targeting a group of people in which half of them are carrying drugs.

The math of the situation is inescapable, it really shows why it is important for the officer to weed out the innocents by using his own good street smarts, it also shows how bad you will look in court if you try mass searches on mostly innocent people. The other very interesting thing about all of this is that, the officer had better keep very accurate records about the dog’s search history, because proving that numbers were “fudged” would be a very easy thing to do.

In any case, I hope I cleared things up a bit.

Thomas

well here I ago again, but I just wanted to add the following thought:

Again, lets pick a dog that is right 98% of the time, now lets have that dog sniffs a group of people where 92% of the people are truely innocent (and by truely innocent I am talking about people who don't do drugs, never had done drugs and truely have no drug residue anywhere on them or near them). In that case the Effective Search Ratio would only be 80%. That means 80% of the time that your dog indicated on this population you would find drugs, the other 20%
of the people you searched are COMPLETELY INNCOCENT. I think it underscores the need not to jump to assumptions about people just because the dog indicated on them but nothing was found.

Actually, the math is correct, only in your interpretation. Understandably, it's the point you are trying to make. Be that as it may. I don't deal with assumptions, mythical numbers or a general consensus of some statistician. When in court, I deal with actual numbers, facts and real live bad guys. I'll answer the defense attorney as I have many times, truthfully. I'll present the records of my dogs, testify to thier completeness and let the chips fall where they may. With that, viva la Supreme Court, and good day sir.

DFrost

not quite, math is math. But I will give it one more shot.

Say that there were no drug users, the population is 100% innocent (and there is no contamination etc), examine that population with a dog who is 98% accurate. Well obviously the dog is never going to have a chance to be right and have every opportunity to be wrong. That is what happening with the numbers, the more innocent the population is, the greater the opportunity the dog has to be wrong and the less opportunity it has to be right. There is nothing mystical or magical about that, it’s just a fact, and that is exactly what can make a good dog's numbers look so bad.

So what are the risks,
1. The dog's actual “on the street” hit to miss ratio ends up being bad enough to draw into question "probable cause" and the evidence gets thrown out.

2. The evidence remains but the poor track record sways the jury. (as I pointed out in a previous post)

3. Innocent citizen backlash.

Now one could chose to live with that, and just wait till SCOTUS completely eliminates the concerns of risks 1 & 2 above. As far as risk 3 is concerned, the way children are getting used to searches, frisking and metal detectors in school, that will soon no longer be an issue either.

But as for TODAY, if you want to improve your dog's actual street track record, the absolutely easiest thing you can do is to search only when you have you have some level of suspicion. If you are searching a group of people where only 1 in 100 is carrying drugs, even a very good dog will have a horrible street record. If you use your street smarts to weed out the innocents so that bad guys number 8 out of 100, the dog’s street record will go from 30% to 80%. On the other hand, If you want to ruin your dog’s credibility, set up a check point in a low drug use area.

On an interesting side note, the other thing the data tells us is that a dog that can make a 99% accuracy rate should be worth a LOT more than a dog that can only do 98%. A 99% accurate dog should have an Effective Street Record (ERS) that is twice as good as a 98% dog under the same conditions. In short, a 99% dog is twice as good as a 98% dog, and probably should be priced accordingly.