Police dogs as divining rods

(This piece was originally posted at the BrandX blog.)

A local couple has been murdered in their own home. You, a respected officer of the law, are responsible for tracking down their killer. But leads are short and public pressure to solve the case is building by the day. Your career’s on the line here. What do you do?

If you were a cop in 17th century France, the answer would have been: find a guy with a magic stick. Indeed, in 1692 French police enlisted a peasant named Jacques Aymar-Vernay in the search for the perpetrator of a double homicide. And it paid off. Aymar-Vernay, who had gained a national reputation for his claimed ability to use a y-shaped stick, a “divining rod,” to locate sources of water, claimed his divine branch had fingered a 19-year-old man with a hunchback as the killer. The man was tortured to death.

A few years later, not surprisingly, Aymar-Vernay was outed as a fraud.

Today, we laugh at our primitive ancestors and their naive belief that a divining rod -- shown by study after study to be no more reliable than chance -- could actually be used to track down criminals. But the joke’s on us: we’re still using those magic sticks, except now they have four legs and are covered in fur.

In last week’s episode, we highlighted one case out of Virginia where the reliability of a drug-sniffing dog named “Bono” was called into question after it was found that, of the 85 times the dog had signaled there were drugs in a vehicle, drugs were only found 22 times. What we soon figured out, though, is that it wasn’t the fault of poor Bono but his human handlers, just as it wasn’t the stick that was at fault for sending a French hunchback to his death.

As researchers at UC Davis observed, police dogs will often signal that there are drugs in a car not because there are, but because that’s what they think their handlers want. And that leads to a lot of false positives. In Australia, one analysis found drug-sniffing dogs (or rather, their handlers) got it wrong 80 percent of the time. And the Chicago Tribune found that police dogs were wrong in 56 percent of the cases it analyzed -- and in 73 percent of cases involving Hispanic drivers, indicating that the dogs are being used to rationalize racial profiling in the war on drugs.

But that hasn’t stopped the police from relying on dogs for a simple reason: like the divining rod of old, the dogs lend a pseudo-scientific rationale to whatever it is one wants to do, be it finding a suspect to pin a murder upon or allowing one probable cause to search a vehicle. Indeed, in the “Bono” case a judge ruled that it didn’t matter the dog was wrong 74 percent of the time, according to one news account, because of “other factors, including the dog’s training and flawless performance during re-certification sessions” -- and, presumably, because ruling the other way would mean throwing out a whole lot of other cases. And we wouldn’t want something as silly as scientific evidence to get in the way of a conviction, no matter the century.