Dirty Old Drug-sniffin’ Dog
Dogs are very good at some things. They can pull sleds over snow and herd sheep. They can chase escaped convicts and retrieve downed ducks. Dogs can lead the sight-impaired through city streets and make great pets.
What they cannot do well is sniff out illegal drugs. Defense attorneys, it’s time to Daubert the dog.
Scientific evidence, from fingerprints to blood-typing to DNA identification, has been used in judicial proceedings for over a hundred years. But scientific evidence must be based on real science (well, Duh!)
For scientific evidence to be admissible, three elements must be present: the underlying basic science must be generally accepted, the process or procedure used must accurately demonstrate the ultimate evidentiary issue, and the proper procedures must have been followed in the instant case.
The news media –especially the tabloids – have been full of psychics called in to help solve crimes; but the evidence of these so-called psychics never appears in court. Science has shown conclusively that their mystical powers do not exist. On the other hand, when DNA testing was introduced in 1984, it was firmly grounded in both genetic theory and laboratory practice. It quickly became the gold standard of individual identification.
Is dog sniffing more like psychics or DNA identification? Actually, it is probably more like lie detection.
Polygraphs (lie detectors) have been around since the 1920s. They are widely used in employment situations and non-judicial law enforcement, but in spite of this wide-spread use, polygraph test results are still not admissible in court. The underlying science is sound. Pulse and respiration rates and galvanic skin response (sweating) can be accurately measured and they can be correlated to the testee’s emotional reaction to a question asked. However, those reactions cannot reliably be correlated to that testee’s truthfulness in response to those questions. If asked “did you kill your mother?” his reactions will be the same whether he did in fact kill her or is simply reacting emotionally to his mother’s death.
Like the polygraph, the dog’s nose is based on solid science. The neuroanatomy and physiology of the amazing canine olfactory system is thoroughly studied and known. What is still unestablished is any correlation between that science and the ability to reliably detect illegal drugs.
Even the police seem unsure of their claims. One police dog handler is quoted as saying that dogs detect illegal drugs, not any particular drug. That statement is like saying that humans can smell food, but cannot tell gingerbread from onions. Even the human nose, poor odor detector that it is, can distinguish the dank of marijuana from the acerb bite of amphetamine.
Although a few studies of dogs’ drug-detecting abilities have been conducted, only a few of those have been done by independent researchers not conducted to police agencies. Most of those studies have not been properly blinded (keeping the dog handler ignorant from the existence and placement of any test objects) or double blinded (in which the score recorder is also kept ignorant until after the test is completed. Double blind testing is the scientific norm for these kinds of experiments. Even this procedural laxity has not produced dogs scoring significantly better than chance.
Dog searches, as currently conducted, cannot even be evaluated retrospectively. They are unlike fingerprints, which have been compared in millions of cases in over a hundred years without a single case of identical prints appearing. The results of a dog search are not a simple yes or no. The result set has four spaces: true positives (the dog alerts and drugs are found), false positives (dog alerts, no drugs found), true negatives (dog does not alert and no drugs are present), and false negatives (dog does not alert, but drugs are present). As searches are currently conducted, records are kept only for true positive results: the drugs are confiscated and the possessor is arrested. If the result is a false positive, a search takes place, but when no drugs are found no further action is taken. When the result is negative (true or false), no arrest is made and, in the case of false negative searches, the drugs remain undetected and in circulation. No measure of the accuracy of a dog’s work may be made unless all four outcomes are accounted for. In the case of an unsymmetrical distribution of the objects, as when only a small percentage of the population has drugs, the number of false positives and false negatives can easily exceed the number of true positives. [For a more thorough explanation of this issue look at my earlier post, “Mammograms, Drug Tests, and Bayes”.]
A study central to a case now under appellate review in Florida presents another serious question: Are drug-sniffing dogs just a modern replay of Hans the Counting Horse? Hans the Horse was an entertainment sensation of the nineteenth century that was revealed as a long-running fraud. His owner toured him around Europe demonstrating the horse’s ability to count and to do simple addition by tapping with a front hoof. However, someone finally thought to look at the trainer instead of the horse. The trainer was (probably unconsciously) counting along with the horse and the horse was watching the trainer. When the trainer stopped counting, so did the horse.
The Florida case involves an experimental test of drug dogs. Samples of illegal drugs were concealed in a test locale and the dogs were used to detect them. As usual in these tests, the dogs did slightly better than chance, but when the data were examined, an interesting test artifact came to light. The hidden drug samples were marked with red cards so they could be retrieved when the test was completed. The red cards would not affect the dogs, which are color-blind. But the analysis revealed that all of the drug samples discovered by the dogs were marked by cards visible to the dogs’ handlers. Was this trial simply a sophisticated example of Hans the Counting Horse?
The role of the dog handlers in that study suggests the importance of administration and oversight in the application of science to law enforcement. Recent scandals have rocked the Harris County Probation Department in Houston and the Massachusetts State Crime Laboratory in Boston. In both of those cities laxness, carelessness, and possibly outright fraud have forced the prosecutors and courts to reject the results of tens of thousands of urine drug tests and at least hundreds, the actual number still being unknown, of convictions are now under review. Many, if not most of these will have to be set aside. Compared to dog searches, urine tests are very routine and controlled. They take place in supervised workspaces under the direct supervision of supervising officials. Dog searches occur in public spaces, not in police stations or laboratories. They are conducted by individual police officers without direct on-site supervision. If tightly managed routine urine tests can be as corrupted as the Houston and Boston stories reveal, then dog handlers acting without supervision or even more open to failed procedures.
Defense lawyers, line up your experts. Voters, hold your prosecutors and judges to account. It’s time to end the Havoc and releash the Dogs of Drug War.
Dirty Old Drug-Sniffin’ Dog
(Apologies to Johnny Cash)
Well he's not very friendly to look at
Oh he's shaggy and he roots like a hog
And he's always bustin' my stash house
That dirty old drug-sniffin' dog
I'm gonna stomp your head in the ground
If you don't stay out of my green house
You dirty old drug-sniffinin' hound
Now if he don't stop tearin' my plants up
Though I'm not a real bad guy
I'm gonna get my rifle and send him
To that great grow house in the sky
You're always hangin' around
But you'd better stay out of my stash house
You dirty old drug-sniffin' hound
[If anyone has a guitar and wants to unleash their inner Johnny Cash, I would love to have a recording of this little ditty.]