[1] http://www.slu.edu/Documents/law/Law%20Journal/Archives/LJ56...
> Dog sniffs are “sui generis,” the Court has held—they’re unique in that they don’t reveal any information about the contents of the object sniffed except the presence of contraband, as to which you have no privacy right. Thus, if the police have a dog ready to sniff your car when they pull you over for a traffic violation, you have no basis for objecting to the sniff. And, of course, if the dog does alert to the car, that is probable cause, so the police can then search the whole car. That’s what the officer wanted to do with Jay-Z, but the K-9 unit wasn’t there when he was pulled over, and was late arriving.
* Actual statistics are rather hard to come by in large part because from a cops standpoint a false positive is not that big a deal.
The officer finished with what they were doing and started to walk away and the person returned. Shortly after the person entered the line of sight of the officer the dog, who was constantly looking at the officer's face, 'triggered'-- wagging its tail and moving excitedly towards the person; whom the officer immediately detained.
When the crazy door to door dog searches case was before the supreme court, I took some time to read a bit about the subject; there is virtually no structural protection or really any review that prevents the dogs from just being a laundering mechanism for their handlers (perhaps unconscious) suggestions. Though they do have true positives to the extent their accuracy is characterized at all, they're not very reliable, and very little is done to analyze or control for false positives (and many officers that handle them seem to believe there are basically no real false positives-- that instead false triggers must be on 'residue'), though there seems to be no scientific basis for that belief. (Though there easily could be; a reasonable practice would be to log and track all triggers by each dog and the results in order to monitor their performance and establish their their level of both type-1 and type-2 errors).
Unfortunately, the courts are continuing to allow wide leeway for something that appears to be almost pseudo-science-- ascribing near magical properties to the dogs-- presumably because we like the result and it sounds reasonable; evidence be damned.
http://en.wikipedia.org/wiki/Clever_Hans#The_Clever_Hans_eff...
"The Clever Hans Effect has also been observed in drug sniffing dogs. A study at University of California Davis revealed that cues can be telegraphed by the handler to the dogs, resulting in false positives."