A Man's Best Friend
The state highway patrolman spots the car with Michigan plates eastbound on the Ohio Turnpike. He sees the car change lanes without putting on its turn signal, so that's enough for a stop. Talking to the driver and checking his license takes a little while, and the cop draws it out so that the other officer, the one with Rover, the drug-sniffing dog, can get there. Rover takes a stroll around the car, then agitatedly begins pawing at the trunk. The cops open the trunk, and find nothing.
Two days later, the exact same scenario unfolds, and this time the cops find five sealed bags containing seven kilos of Columbia's finest non-coffee export. The defense files a motion to suppress, but the law is that the dog sniff isn't a search, and the dog's alert provides probable cause to conduct a search of the trunk. End of story.
Of course, the judge never knows about what happened two days earlier.
This problem is endemic in 4th Amendment cases. It's what called sample selection bias: people draw conclusions from a non-representative sample of data. We talk about the "high societal costs" of the exclusionary rule because that's all we see. If the police stop a car for no reason, find drugs, and the drugs are thrown out, we know about that because that case makes it to court. If the police don't stop ten cars for no reason because they know anything they find will be thrown out, that certainly has value in a society which prizes liberty. But we never hear about those ten cars, because those cases don't make it to court. Similarly, it's much easier to conclude that drug-sniffing dogs are reliable when the only cases judges hear are ones where the dog's alert led to drugs being found.
The Supreme Court dealt with dog alerts back in 2005 in Illinois v. Caballes. Caballes was stopped for driving six miles over the speed limit, and, when he refused to consent to a search of his vehicle, a drug dog was brought to the scene. The dog alerted, marijuana was found in the vehicle, and the Court upheld the search. The only issue really addressed by the Court was whether the dog sniff constituted a search, but there's language in the opinion noting the reliability of dogs in detecting narcotics. Justice Souter dissented, arguing that "the infallible dog is a creature of legal fiction."
After reading Radley Balko's recent article in Reason.com, "The Mind of a Police Dog," Souter seems to have the better argument. Balko's basic point is that perhaps the dog's strongest trait is an ability to read the cues of its master; as Balko explains:
For dogs descended from lines bred for protection and companionship, this talent makes sense. A dog adept at distinguishing friend from foe was likely to be kept around and bred, and one very good way to tell friend from foe is to read your master's body language.
How this translates to drug-sniffing is disclosed in a study which involved 18 dogs and their handlers: each had to conduct eight searches of rooms of a church, for drugs or explosives. The handlers were told that each search area would contain up to three target scents, and that in two cases those scents would be marked by pieces of red paper. They weren't told that two of the targets contained decoy scents -- unwrapped hidden sausages -- to encourage false alerts by the dog. The 144 walkthroughs produced 225 total alerts.
All of them were false; none of the search areas contained scents of either drugs or explosives. More interestingly, false alerts were made twice as often at target sites designed to fool the handlers (the red paper) than to fool the dog (the sausages). In short, the dog wasn't detecting drugs or explosives, it was detecting whether its handler believed there were drugs or explosives. In fact, although not of course mentioned in the study, the best indication that dog alerts are affected by the conscious or unconscious clues of their handlers is a recent Pentagon study which disclosed that explosive-detecting dogs are better than any technology the Defense Department has come up with to find bombs in Iraq and Afghanistan. Why? Because their handlers have no idea where those bombs are, so the dogs can concentrate on using only their olfactory senses.
A lot of this isn't really new. In his dissent, Souter relied extensively on various studies which showed that dogs gave false positives anywhere from 12 to 60% of the time, and that 80% of US bank notes were contaminated with sufficient traces of cocaine to cause a trained dog to alert.
There's a counter-argument to this, though. Keep in mind we're talking about probable cause. What's the quantum of proof to establish that? It certainly isn't 100%, or even 51%; that's the quantum for preponderance of the evidence, and we know that probable cause is below that. So even if a dog gives false positives 60% of the time, isn't that sufficient to cross the bar of probable cause?
That's what the Ohio courts have held. Typical is the 9th District's decision three years ago in State v. Barbee, which relied on an earlier case to hold that an alert was sufficient to establish probable cause because it provided a "fair probability" that drugs would be found. (In that earlier case, the evidence had included stats that drug dogs gave positive alerts "only" 61% of the time. Bad argument.) In fact, a number of courts have followed the 6th District's holding in State v. Nguyen that "real world" evidence about drug dogs' reliability isn't even relevant; all the state need do is prove the dog is properly trained and certified, which may be done through "documentary proof." In other words, if the prosecutor plops down the certification form, it's the end of discussion.
Whether Balko's article and the church study change any of that is the question. There's a big difference between a 39% false positive result (as in the earlier 9th District case) and a 100% false positive result, as in the study. The possibility of "handler bias" adds another level to the argument, and a number of courts don't restrict evidence on reliability to the extent that the 6th District does. The much more widespread use of police cruiser cameras offers the possibility of detecting cues that the handler gave the dog, as described in this article.
One thing's for sure, though. It might be a good idea for the courts to take another look at the whole issue. And for defense attorneys to do that, too, and not just assume that because Rover said there were drugs, that's enough to justify a search.